three foraracprotein arerequiredfor ofaracin … and mncl2 to a final concentration of 20 ,m],...

5
Proc. Nati. Acad. Sci. USA Vol. 85, pp. 1749-1753, March 1988 Biochemistry Three binding sites for AraC protein are required for autoregulation of araC in Escherichia coli (DNA loops/operators/repression/gene regulation/cooperative interactions) EILEEN P. HAMILTON AND NANCY LEE* Department of Biological Sciences, University of California, Santa Barbara, CA 93106 Communicated by John Carbon, November 9, 1987 (received for review September 24, 1987) ABSTRACT Three binding sites for AraC protein were shown to be required for the autoregulation of araC: arall, araOl, and araO2. Selective inactivation of AraC-binding sites on the DNA demonstrated that araOl and araO2 are required in vivo to produce repression of araC in the presence of arabinose, whereas arall and arw02 are required in its ab- sence. We found that the low-affinity site araO2 is essential for araC autoregulation; araO1 and aral! provide high-affinity AraC-binding sites, which allow cooperative binding at araO2. Profound effects on the ara8AD promoter and the araC promoter are produced by ligand-induced changes in AraC occupancy of functional sites on the DNA. We suggest that AraC exerts its multiplicity of controls through two alternative states of cooperative interactions with DNA and we illustrate this with a model. This model presents our interpretations of activation and repression of the araBAD operon and the autoregulation of the araC gene. PBAD I PC ara Pl Fes araC araBAD mRNA araC mRNA I +50 0 -50 -100 -150 -200 -250 -300 -350 ara1782 1[ 1l L] araO2 212 aral55 A T- TGGAC ------ G C TA (AraC Binding) Proteins that repress gene activity were believed for many years to act by binding to a single site (operator) within a promoter, blocking transcription initiation (1). More re- cently, additional operator sites at considerable distances from the promoter regions have been found to play a role in negative regulation, including the negative aspect of araBAD operon control [gal (2, 3), lac (4-6), ara (7-9), deo (10, 11), nrd (12)]. These multiple-repressor-binding sites are neces- sary for full repression of these operons. It has been pro- posed that the secondary operators serve to enhance repres- sor activity by stabilizing protein-DNA complexes through cooperative binding (13). The araC gene, which encodes the transcriptional activa- tor of the arabinose genes in Escherichia coli, is homeostat- ically autoregulated under inducing and noninducing condi- tions (14). In vitro studies initially suggested that araC is transcriptionally regulated by the competitive binding of AraC protein to a site (araOl) congruent with the RNA polymerase-binding site of the araC promoter (15, 16) (Fig. 1). Direct selection of cis-acting, autoregulation-minus mu- tants in an araC-lacZ fusion strain gave primarily "promot- er-up" mutations with increased affinities for polymerase rather than decreased binding of AraC (19). Without an araOl - mutant that shows selective loss of AraC binding while retaining the ability to bind polymerase, the in vivo role of araO, in araC autoregulation cannot be established. In addition to araOl, there are three other AraC protein- binding sites (aral, araI2, and araO2) located near the araC gene promoter (Fig. 1). araO2, which lies within the leader region of the araC gene, is essential for repression of the araBAD operon (7). Schleif and co-workers (7-9) have postulated that the cooperative binding of AraC molecules to TTGACA ----------------- TATAAT (Promoter) -110 -120 -130 -140 -150 AATCAATGTGGACTTTTCTGCCGTGATTATAGACACTTTTGTTACGCG TTAGTTACACCTGAAAAGACGGCACTAATATCTGTGAAAACAATGCGC 11 11 Lo T AC AG startpoint araC A TG TC transcription araD 259 FIG. 1. Controlling region of the araBAD and araC operons, drawn to scale. The araBAD (PBAD) and araC (Pc) promoters are shown along with the locations of the protein-binding sites. RNA polymerase (Pol)-binding sites, large open boxes; catabolite gene activation protein (CAP)-binding site, small open box; AraC protein- binding sites (12 for aral2, I1 for araIl, 01 for araOl, and 02 for araO2), black boxes. Numbering of base pairs is relative to the araBAD transcription start site at + 1. The positions of the araO2 and ara(112) deletions and the single base pair deleted in the araI2 mutant are also indicated. The nucleotide sequence of the wild-type araC promoter from -103 to -150 is shown, with the four base substitutions in the araOl - mutant indicated below. This mutant has a fifth substitution, a spontaneous A to T transversion at -104, which lies outside the consensus sequences. The locations of the AraC-binding consensus sequence (AraC Binding) (17) and the consensus promoter sequence (Promoter) (18) are shown in relation- ship to the mutations in the araOl - strain. araI and araO2 results in araBAD repression. We have recently shown (17) that the araI site is separable into two adjacent regions, each containing a 17-base-pair (bp) AraC- binding consensus. These regions, designated araI1 and araI2, differ greatly in their affinities for AraC in the absence of inducer. We proposed that the induction of araBAD by arabinose is caused by a switching of the cooperative AraC binding from arall/araO2 to aral1/araI2. This ligand- Abbreviations: X-Gal, 5-bromo4chloro-3-indolyl P-D-galactoside; Pc, araC promoter; PBAD, araBAD promoter. *To whom reprint requests should be addressed. 1749 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. I

Upload: doanmien

Post on 06-May-2018

216 views

Category:

Documents


2 download

TRANSCRIPT

Proc. Nati. Acad. Sci. USAVol. 85, pp. 1749-1753, March 1988Biochemistry

Three binding sites for AraC protein are required forautoregulation of araC in Escherichia coli

(DNA loops/operators/repression/gene regulation/cooperative interactions)

EILEEN P. HAMILTON AND NANCY LEE*Department of Biological Sciences, University of California, Santa Barbara, CA 93106

Communicated by John Carbon, November 9, 1987 (received for review September 24, 1987)

ABSTRACT Three binding sites for AraC protein wereshown to be required for the autoregulation of araC: arall,araOl, and araO2. Selective inactivation of AraC-binding siteson the DNA demonstrated that araOl and araO2 are requiredin vivo to produce repression of araC in the presence ofarabinose, whereas arall and arw02 are required in its ab-sence. We found that the low-affinity site araO2 is essential foraraC autoregulation; araO1 and aral! provide high-affinityAraC-binding sites, which allow cooperative binding at araO2.Profound effects on the ara8AD promoter and the araCpromoter are produced by ligand-induced changes in AraCoccupancy of functional sites on the DNA. We suggest thatAraC exerts its multiplicity of controls through two alternativestates of cooperative interactions with DNA and we illustratethis with a model. This model presents our interpretations ofactivation and repression of the araBAD operon and theautoregulation of the araC gene.

PBADI PC

ara Pl Fes araC

araBAD mRNA araC mRNA

I+50 0 -50 -100 -150 -200 -250 -300 -350

ara1782 1[ 1l L] araO2212aral55 A

T- TGGAC ------ GC TA (AraC Binding)

Proteins that repress gene activity were believed for manyyears to act by binding to a single site (operator) within apromoter, blocking transcription initiation (1). More re-cently, additional operator sites at considerable distancesfrom the promoter regions have been found to play a role innegative regulation, including the negative aspect ofaraBADoperon control [gal (2, 3), lac (4-6), ara (7-9), deo (10, 11),nrd (12)]. These multiple-repressor-binding sites are neces-sary for full repression of these operons. It has been pro-posed that the secondary operators serve to enhance repres-sor activity by stabilizing protein-DNA complexes throughcooperative binding (13).The araC gene, which encodes the transcriptional activa-

tor of the arabinose genes in Escherichia coli, is homeostat-ically autoregulated under inducing and noninducing condi-tions (14). In vitro studies initially suggested that araC istranscriptionally regulated by the competitive binding ofAraC protein to a site (araOl) congruent with the RNApolymerase-binding site of the araC promoter (15, 16) (Fig.1). Direct selection of cis-acting, autoregulation-minus mu-tants in an araC-lacZ fusion strain gave primarily "promot-er-up" mutations with increased affinities for polymeraserather than decreased binding of AraC (19). Without anaraOl - mutant that shows selective loss of AraC bindingwhile retaining the ability to bind polymerase, the in vivorole of araO, in araC autoregulation cannot be established.

In addition to araOl, there are three other AraC protein-binding sites (aral, araI2, and araO2) located near the araCgene promoter (Fig. 1). araO2, which lies within the leaderregion of the araC gene, is essential for repression of thearaBAD operon (7). Schleif and co-workers (7-9) havepostulated that the cooperative binding ofAraC molecules to

TTGACA ----------------- TATAAT (Promoter)

-110 -120 -130 -140 -150AATCAATGTGGACTTTTCTGCCGTGATTATAGACACTTTTGTTACGCGTTAGTTACACCTGAAAAGACGGCACTAATATCTGTGAAAACAATGCGC

11 11 LoT AC AG startpoint araCA TG TC transcription

araD 259

FIG. 1. Controlling region of the araBAD and araC operons,drawn to scale. The araBAD (PBAD) and araC (Pc) promoters areshown along with the locations of the protein-binding sites. RNApolymerase (Pol)-binding sites, large open boxes; catabolite geneactivation protein (CAP)-binding site, small open box; AraC protein-binding sites (12 for aral2, I1 for araIl, 01 for araOl, and 02 foraraO2), black boxes. Numbering of base pairs is relative to thearaBAD transcription start site at + 1. The positions of the araO2and ara(112) deletions and the single base pair deleted in the araI2mutant are also indicated. The nucleotide sequence of the wild-typearaC promoter from -103 to -150 is shown, with the four basesubstitutions in the araOl - mutant indicated below. This mutant hasa fifth substitution, a spontaneous A to T transversion at -104,which lies outside the consensus sequences. The locations of theAraC-binding consensus sequence (AraC Binding) (17) and theconsensus promoter sequence (Promoter) (18) are shown in relation-ship to the mutations in the araOl - strain.

araI and araO2 results in araBAD repression. We haverecently shown (17) that the araI site is separable into twoadjacent regions, each containing a 17-base-pair (bp) AraC-binding consensus. These regions, designated araI1 andaraI2, differ greatly in their affinities for AraC in the absenceof inducer. We proposed that the induction of araBAD byarabinose is caused by a switching of the cooperative AraCbinding from arall/araO2 to aral1/araI2. This ligand-

Abbreviations: X-Gal, 5-bromo4chloro-3-indolyl P-D-galactoside;Pc, araC promoter; PBAD, araBAD promoter.*To whom reprint requests should be addressed.

1749

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

I

1750 Biochemistry: Hamilton and Lee

induced transition allows the araBAD promoter to changefrom the repressed to the induced state, with a 1200-foldincrease in its transcriptional activity.The present study has been undertaken to determine (i) if

araO1 is involved in the in vivo autoregulation of araC and(ii) if the other AraC-binding sites (araI1, araI2, and araO2)play a significant role in araC autoregulation. We selectivelyinactivated these sites in a strain carrying an araC-lacZprotein fusion and measured the activities of the fusion geneunder inducing and noninducing conditions. A wild-typearaC gene was introduced by lysogenization with a Aparatransducing phage, so that only single copies of the ara geneswere present. We were surprised to find that three AraC-binding sites were involved in araC autoregulation. In thepresence of arabinose, araO2 and araO1 were required torepress araC, whereas, in the absence of inducer, araI1 andaraO2 [the same sites that were shown to produce repressionof araBAD (9)] were required. To our knowledge, no casehas previously been reported where the cooperative interac-tion between protein molecules bound to widely separatedDNA sites is absolutely required for repression. We presenta model describing the ligand-dependent states of occupancyof all four AraC-binding sites and their respective roles in theregulation of araC and araBAD expression.

MATERIALS AND METHODSMedia and General Methods. Media used included Luria-

Bertani medium (LB; ref. 10), mineral glycerol medium withor without 0.4% L-arabinose [per 100 ml: 0.7 g of K2HPO4,0.3 g of KH2PO4, 0.1 g of (NH4)2S04, 10 mg ofMgSO4-7H2O, 0.2 ml of glycerol, 4 mg of L-leucine, 0.4 mg ofthiamine, and MnCl2 to a final concentration of 20 ,M],MacConkey arabinose medium with or without 100 ,ug ofampicillin per ml (Difco MacConkey agar base with 1%L-arabinose), tryptone medium (per 100 ml: 1 g of Bactotryp-tone, 0.5 g of NaCl, and 0.4 mg of thiamine), and tryptonebottom agar (with 5-bromo-4-chloro-3-indolyl f-D-galac-toside (X-Gal) (tryptone medium with 1.5% Bactoagar and0.0028% X-Gal). DNA manipulations used in the construc-tions of bacterial strains and plasmids were as described(20).

Strain Constructions. The scheme for the construction ofstrains used in our studies is outlined in Fig. 2. AraC-bindingsites were inactivated by three different mutagenic proce-dures. Oligonucleotide-directed in vitro mutagenesis (21)was used to delete 20 bp (from - 264 to - 283, inclusive) ofthe araO2 site, including the entire AraC-binding consensussequence (17). This method was also used to create a mutantwith four base substitutions in the AraC-binding site araO1(Fig. 1). Digestion of ara DNA cut by BamHI (at - 44) bythe exonuclease BAL-31 resulted in the deletion of 76 bp(from - 7 to - 82, inclusive) containing aral1, araI2, and halfof the araBAD polymerase-binding site. We also used apreviously identified, spontaneous chromosomal mutation inaral2, which deletes the base pair at position -55 andeliminates AraC binding to araI2 in vitro (17).The plasmid pNL20 was used as an integration and rescue

plasmid. It contains ara DNA from 1816 to -44, includingthe entire araB-coding region and part of the araBADpromoter. Also contained on the plasmid are most of thelacZ gene (374-3455), the distal end of the araC gene plusabout 800 bp of downstream sequence (from -877 to- 2006), and pBR322 DNA from 2066 clockwise to 25. Fig. 2shows this plasmid after a Sau3A restriction fragment de-rived from ara (from - 44 to - 330) had been cloned into theBamHI site at the ara/lac junction [formed by joining theara BamHI site at -44 to the BamHI site at 374 in the lacZgene on the plasmid pMC1871 (Pharmacia P-L Biochemi-cals)]. This method of cloning was used to construct plas-

Modification of AraC-binding sites by:

Oligonucleotide directed site specific mutagenenesisaraO2 and araOl

BamHI restriction and Bal31 digestionara(I,12J-

Spontaneous chromosomal mutationaral2-

'I Purify restriction fragmentsclone into pNL20

araC'(1-877 to -2006)

! iacZ-1374 to 34551

Integrate into an araC,/promoter-null mutant

araO1

araO2

Red papillae atopwhite colonies on

MacConkey ara amp

Integrate intowild type strain

ara(1112)aral2

White papillae atopred colonies on

MacConkey ara amp

Cure Plasmids

Amps XGal+ Ara-

ResLysogenize with

XparaC' A/Haploid Pc/acZ(XparaC+)

NL 31-067NL 31-217NL 31-282NL 31-332NL 31-338

Amps XGal+ Ara

scue with pNL20Sequence

WT

araO2-araOl-

ara(l112)-aral2-

Haploid PClacZ(araC)NL 31-024NL 31-212NL 31-259NL 31-320NL 31-337

FIG. 2. Scheme used to construct strains for testing autoregula-tion.

mids carrying each AraC-binding site mutation and resultedin the fusion of the sixth codon of araC to the eighth codonof lacZ. Those plasmids that had a functional araBADpromoter (the ones carrying the araO1- and araO2- muta-tions) were put into NL20-185a, a haploid strain with a 3-bpdeletion (from -53 to -55) that rendered it araBAD -.Integration events were detected as red papillae on the whitecolonies on MacConkey arabinose/ampicillin agar. The plas-mids that carried a mutant araBAD [due to the araI2- andara(112) - mutations] were put into the ara+ strain NL2O-000. Integration events resulted in white papillae on the redcolonies on MacConkey arabinose/ampicillin plates. Allstrains containing integrated plasmids were purified, inocu-lated into LB, incubated overnight at 440C, diluted, andplated on MacConkey arabinose plates. Cells cured of theplasmid, but retaining the araC-lacZ fusions, were white onMacConkey arabinose, sensitive to ampicillin, and blue ontryptone plates containing X-Gal.To confirm the genotypes of these strains, the araC-4acZ

Proc. Natl. Acad. Sci. USA 85 (1988)

Proc. Natl. Acad. Sci. USA 85 (1988) 1751

haploids were transformed with pNL20 and grown in LBwith ampicillin with several transfers. Rescued promotersfrom the araO1- (NL31-259) and araO2- (NL31-212) strains(which are PBAD+) gave rare recombinant AraB + plasmids,which were easily detected after transformation of NL20-314, a recA - araB - (araB716, a deletion from 436 to 634)strain. Plasmid DNA from Ara' colonies was isolated; DNAcontaining the ara regulatory region was cloned into an M13phage and sequenced by the Sanger dideoxy method (22).Rare recombinant plasmids from the PBAD strains [NL31-337 araI2- and NL31-320 ara(II2)-] were detected as bluecolonies on X-Gal plates after transformation of NL20-272,which is recA and contains the araC766 deletion (from - 626to - 1698). The regulatory region DNA from these plasmidswas also cloned and sequenced.We lysogenized our haploid araC-lacZ fusion strains with

a AparaC' to make AraC' derivatives. Stable lysogenswere isolated, and the presence of AparaC+ prophages wasverified by their ability to produce high-titer lysates thatcomplemented araC766. These strains, NL31-217 (araO2 ),NL31-282 (araO1-), NL31-332 [ara(I1I2)-i, and NL31-338(araI2-), were shown to be single lysogens by their sensi-tivity to AcI9O cl7 (23).

RESULTSA Mutation of araO1 with Unimpaired Polymerase Binding.

The AraC-binding site araO1 was thought to be the operatorresponsible for araC autoregulation (15, 16). This assump-tion was based primarily on two facts: (i) araO1 overlaps theRNA polymerase-binding site of the araC promoter and (it)AraC protein and RNA polymerase compete for binding tothis site in the presence of arabinose in vitro. In vivoexperiments showing the effect of araO1 on autoregulationhave not been possible due to the lack of mutants. Weconstructed an araO1- mutant (araO1259) by site-directedmutagenesis. The mutant araO1259 contained five basesubstitutions, four of which lie within a 17-bp AraC-bindingconsensus (17). This mutation left intact the -35 and - 10hexanucleotides of the overlapping polymerase binding site,as shown in Fig. 1. DNase I footprinting (DNase protection)showed that there was a reduction by a factor of 8 in theaffinity of the mutant araO1 for AraC protein as comparedwith the wild type (Fig. 3). As indicated below, thesechanges appeared to have no effect on polymerase binding.

Effect of araOl on Autoregulation in Vivo. To detect theeffect of araO1259 on araC autoregulation, we put themutation in cis to a chromosomal araC-IacZ fusion. Thismutation, in the absence of an intact araC gene, had littleeffect on the araC promoter (Table 1, line 4), indicating thatthe base substitutions in araO1259 did not significantly alterpolymerase binding. We introduced into this strain a singlecopy of araC by way of a Apara. When the lysogen wastested for its degree of autoregulation, we found a largedisparity between the induced and noninduced cells (Table1, lines 5 and 6). Unlike the wild-type control, which showedrepression by AraC in the presence and absence of inducer(Table 1, lines 1-3), the araO1259 lysogen showed a com-plete loss of autoregulation in the presence of arabinose(Table 1, line 6), whereas the normal repression (by a factorof 10) was observed in the absence of the sugar (Table 1, line5). This unexpected finding suggested that the mechanism ofaraC autoregulation might be more complex than previouslyassumed. Ligand-induced alterations in the occupancy ofvarious AraC-binding sites have been demonstrated (17).The regulation of araC may also involve such changes, sincethe role of araO1 in autoregulation changes with the pres-ence of inducer. To test this possibility and to locate thesite(s) responsible for the repression of araC in the absence

araO 1-259Mutant

1 2 3 4 5 6 7 8Wild Type12345678

-~~~~~~~~~4

A t~~ ~ ~ ~~~~~~~5-5-60

_! w _--7~~~-0,_,4 w e

-90

--120

-130 Iaraol

*-140

"b

* -150

FIG. 3. DNase proteation experiment showing decreased araOlprotection in the ara01259 mutant. The binding mixtures (50 1ul)were as described (17). End-labeled DNA fragments were at 7.5 nM,L-arabinose was 33 nM, and AraC protein was at the concentrationsindicated: lanes 1 and 8, no AraC; lanes 2, 10 nM; lanes 3, 20 nM;lanes 4, 40 nM; lanes 5, 80 nM; lanes 6, 160 nM; and lanes 7, 320 nM.Nucleotide positions are indicated on the right. Arrows indicate thepositions of mutated bases in the ara01259 DNA. The promoterfragment containing the araO1259 mutation exhibited the samedegree of protection at 320 nM AraC (lane 7) as did the wild-typefragment at 40 nM AraC protein (lane 4). No difference in theprotection of araI was observed.

of arabinose, we tested the remaining nearby AraC-bindingsites for their involvement in araC autoregulation.Mutations in the Other AraC-Binding Sites. There are three

other AraC-binding sites near the transcriptional startpointof the araC gene: araO2, araI1, and araI2 (Fig. 1). Weselectively inactivated these sites and obtained araO2-(araO2212), araI2- (araIS5), and ara(112)- (araI782) deriv-atives of the araC-lacZ fusion strain. The araISS mutationprevents AraC binding at araI2 while retaining wild-typeaffinity for AraC at araI1 (17). AraO2212 was a deletion of 20bp of araO2 DNA that included the entire AraC-bindingconsensus sequence (17). The araI782 deletion left the CAPconsensus sequence intact but removed araI1, araI2, andpart of the araBAD polymerase-binding site (Fig. 1).Role of arall in araC Autoregulation. The araI site was

found to be an indispensable element in araC autoregulationonly in the absence of inducer. Inactivation of araI1 by theara(I12) - deletion affected autoregulation in the absence ofinducer (Table 1, line 8) but not in its presence (Table 1, line9), a situation that is the exact converse of that seen with thearaO1259 mutant. That it was araI1 not araI2 that wasinvolved with araC regulation was shown by the observationthat the araI2- mutation (araIS5) did not affect autoregula-tion (Table 1, lines 10-12).

Biochemistry: Hamilton and Lee

. -10011,1111,11im

-110

qwm

''.

1752 Biochemistry: Hamilton and Lee

Table 1. Effect of AraC-binding site mutations on araCautoregulation in vivo

i3-GalactosidaseStrain Mutation araC L-Arabinose activity31-024 None - + 132 ± 1331-067 + - 12.7 ± 0.531-067 + + 14.8 ± 1.631-259 araO1- - + 105 + 231-282 + - 12.5 ± 0.631-282 + + 132 + 2131-320 ara(II2) - 87 ± 231-332 + - 46.4 ± 2.931-332 + + 4.3 ± 0.132-337 araI2 - + 110 ± 531-338 + - 14.7 ± 0.331-338 + + 12.6 ± 0.431-212 araO2- - + 101 + 631-217 + - 116 ± 1131-217 + + 108 ± 6

The strains carrying the indicated mutations in cis to an ara-C-4acZ fusion were grown approximately eight generations inmineral glycerol medium with or without arabinose, as indicated.3-Galactosidase was then assayed as described by Miller (24) and

expressed in Miller units. Each value shown in the last columnrepresents the average of either four or six independent determina-tions. The somewhat reduced levels observed with the ara(112) -derivatives (lines 7-9) may be due to the removal by the deletion ofbases lying in the immediate neighborhood of the CAP consensussequence (25).

Role of araO2 in araC Autoregulation. The araO2 strainwas autoregulation-minus under inducing and noninducingconditions (Table 1, lines 13-15). When araO2 was removedby the araO2212 mutation, the araC' allele was unable torepress 83-galactosidase synthesis when the cells were grownin either arabinose or glycerol medium. Thus, three discreteAraC-binding sites are involved in maintaining the repressedstate of the araC gene. araO2 and araOl are both requiredfor autoregulation in the presence of inducer, whereas araO2and araIl are required when inducer is absent. At any giventime, two of the three sites are operative; the araO2 site mustremain intact for repression to occur, whereas araOl andaral1 alternate to maintain repression in the presence and inthe absence of inducer, respectively.

DISCUSSIONThe promoters of araC and araBAD are subject to transcrip-tional control by the AraC protein. When the inducer isabsent, araBAD and araC are repressed by AraC. Uponaddition of the inducer, AraC becomes an activator of thearaBAD operon while continuing to repress the araC gene.These highly selective and diversified actions seem to de-mand an unusual degree of functional complexity in theAraC protein.Our findings that there is a ligand-induced change in the

state of occupancy of AraC-binding sites near these promot-ers and that such changes have profound effects on PBAD(17) and Pc activities have led us to believe that AraC exertsthis multiplicity of control through alternate states of coop-erative binding to DNA. This simple model is shown in Fig.4. We postulate that, in the absence of inducer, a singleinteraction between ligand-free AraC molecules facilitatestheir cooperative binding to aral1 and araO2; this configura-tion produces repression of araBAD (7, 9) and autoregula-tion of araC. Upon the introduction of arabinose, AraCundergoes a ligand-induced conformational change that pre-cludes its cooperative binding to araI1 and araO2. Instead,two different interactions become established between fig-and-bound AraC molecules: the binding to araIl/araI2 leads

araB . 1 1i] araC

I, CAP 0, °

Pol

airaC rnRNA

I + L-arabinose

araB [__r L~gEv7CA l2121 CAfP 04 02Po1Pol

araBAD) mRNA

Pol

C.. ..RNAaraC mONA

araC

FIG. 4. Integrative model for araC autoregulation, araBADactivation, and repression. The cooperative interactions of AraCprotein bound to different sites are responsible for autoregulation inthe absence (Upper) and presence (Lower) of L-arabinose. Proteins:AraC protein, shaded boxes; catabolite gene activator protein, smallopen boxes; RNA polymerase, large open boxes. Protein-bindingsites are designated as in Fig. 1. Bold lines connecting AraC proteinboxes show cooperative interactions. Dashed polymerase boxessignify partial occupancy by RNA polymerase due to the autoregu-lation of araC.

to PBAD activation (17) and the binding to araOl/araO2restores araO2 occupancy and maintains araC repression.The above model accounts for the various phenotypes

observed after selective inactivations of the four AraC-binding sites. The loss of araI2 function prevented araBADactivation (17) without affecting autoregulation. Inactivationof araI abolished araBAD activation (ref. 8; N.L. and C.Francklyn, unpublished data), araBAD repression (9), aswell as araC autoregulation in the absence of arabinose butnot in its presence. Inactivation of araO1 eliminated araCautoregulation in the presence of inducer only and had noeffect on either the activation or the repression of araBAD(ref. 9; unpublished data). Removal of araO2 was accompa-nied by the loss of araBAD repression but not its activation(7), and the autoregulation of araC was completely abolishedby the araO2 deletion.The model is based on two key facts: (i) the DNA sites that

bind AraC have very different affinities for the protein (15,16) and (it) the binding of ligand induces a conformationalchange in AraC that alters the sites of cooperative binding onthe DNA (ref. 17; this paper).The araO2 site possesses a very low affinity for the AraC

protein (16) and does not bind AraC in vivo in the absence ofcooperativity (9). We propose that the role of araI1 andaraO1 in maintaining araC autoregulation is to provide thesites required for cooperative binding of AraC to the low-affinity site araO2. The alternate participation of these twohigh-affinity sites is compatible with in vivo binding data (9).It is not known whether AraC is bound to araOl in vivo inthe absence of inducer. This high-affinity site (16), even ifbound, has now been shown to have no effect on araCtranscription in glycerol medium. Congruency of protein-binding sites does not necessarily preclude simultaneousoccupancy (17, 26, 27).We believe that the conformation of AraC protein plays a

direct role in determining which cooperative interaction willoccur. The conformational change in AraC induced by thebinding of L-arabinose (28) is necessary for its cooperativebinding to araI and araI2, which produces araBAD activa-tion (18). Our experiments suggest that the araO1araO2interaction leading to araC repression also requires theligand-bound conformation of AraC protein.

Proc. Natl. Acad Sci. USA 85 (1988)

Proc. Natl. Acad. Sci. USA 85 (1988) 1753

The transfer of AraC binding from araI1 to araO1 upon theaddition of inducer must transiently disengage AraC boundat araO2, since the latter site is incapable of noncooperativebinding (9). This may account for the arabinose-inducedtransient derepression of the araC gene. It has been reportedthat araC expression increases in the first 15 min afterarabinose addition before autoregulation is reestablished andaraC expression returns to preinduction levels (29, 30). Wesuggest that the decrease in autoregulation following theaddition of arabinose represents a transient escape synthesiswhen araIl/araO2-mediated autoregulation is replaced bythat resulting from the araOl/araO2 interaction.The alternate states of AraC occupancy depicted in Fig. 4

represent a dynamic equilibrium, governed by the ligand-dependent changes in AraC conformation. There is consid-erable evidence, however, suggesting that the requirementfor arabinose in the allosteric transition of AraC is notabsolute and that a very small amount of AraC activator ispresent even in the uninduced cell (7, 31, 32). This activatoris responsible for the 12-fold stimulation of the araBADpromoter seen when repression is prevented by the elimina-tion of the araO2 site. We propose that the repression ofaraBAD exists in the uninduced cell because the araIl/araO2 interaction precludes the araIl/araI2 interaction. AnAraC molecule bound at araI1 is capable of entering intoassociative interactions with either araO2 or araI2, depend-ing on its conformational state. In the absence of inducer,AraC at araI1 is locked into a cooperative interaction witharaO2, preventing its participation in an araIl/araI2 associ-ation. We envision that these mutually exclusive cooperativeinteractions constitute the basis for araBAD repression,since this promoter has no affinity for RNA polymerasewhatsoever in the absence of AraC protein (33). This modelfor the mechanism of araBAD repression predicts that anymodification that favors the binding of AraC to aral1/araO2would enhance repression, and any modification that favorsthe binding of AraC to araIl/aral2 would decrease it. Inview of the proximity of araI2 and the polymerase-bindingsite at PBAD (17), it would not be surprising if a promotermutant that strengthened the polymerase interaction withDNA also favored the occupancy of araI2 by AraC, at theexpense of the aral1/araO2 interaction. This may explain thefinding that some mutations that reduced araBAD repressionmap within the RNA polymerase-binding domain (9).

It has been suggested that when a protein binds coopera-tively to widely separated sites, the intervening DNA formsa loop (see ref. 13 for a review). Looping has been incorpo-rated into the repression models of many operons (2-12).The repression of araBAD in the absence of inducer hasbeen postulated to involve the formation of a DNA loop (7,9, 17), since the phasing of the two sites involved inrepression, aral1 and araO2, is critical (7), an observationthat supports the idea of loop formation (34, 35). There is yetno experimental evidence suggesting that a DNA loop formsbetween araO1 and araO2. Examination of the araC leadersequence (36) shows that the center-to-center distance be-tween the AraC-binding consensus sequences (17) withinaraOl and araO2 is 158 bp, which represents an integralnumber of helical turns (15.0) in B-form DNA. The phasingof araO1 and araO2 therefore suggests the existence of asimilar DNA loop structure.How AraC occupancy generates repression of araC re-

mains to be elucidated. The binding of AraC could eitherdirectly block polymerase progress (37) or produce a DNAconformation unfavorable for polymerase entry at the araCpromoter. Further work is needed to determine the mecha-nism.

We have proposed that AraC protein exerts positive andnegative transcriptional regulation and mediates cellularresponse to inducer by two alternate states of DNA occu-pancy. Like the phage A cI protein, the mode of action ofAraC-i.e., whether positive or negative-is governed bythe DNA sites that are occupied (38). The widespreadoccurrence of multiple, and often widely separated, bindingsites on the DNA in different biological systems (13, 39, 40)suggests their importance in regulation. The ara systemprovides a model where a single protein interacts with itsfour cognate sites to produce a multiplicity of controls.

We thank Chris Francklyn and Dr. Gary Wilcox for helpfuldiscussions. This work was supported by National Institutes ofHealth Grants 5 RO GM14652-19 and 507 RR07099-20.

1. Reznikoff, W. S., Siegele, D. A., Cowing, D. & Gross, C. A. (1985)Annu. Rev. Genet. 19, 355-387.

2. Majumdar, A. & Adhya, S. (1984) Proc. Natl. Acad. Sci4 USA 81,6100-6104.

3. Kuhnke, O., Krause, A., Heibach, C., Gieske, U., Fritz, H.-J. &Ehring, R. (1986) EMBO J. 5, 167-173.

4. Besse, M., von Wilcken-Bergman, B. & Miller-Hill, B. (1986) EMBO J.5, 1377-1381.

5. Mossing, M. C. & Record, M. T., Jr. (1986) Science 233, 889-892.6. Eismann, E., von Wilcken-Bergman, B. & Muller-Hill, B. (1987) J. Mol.

Biol. 195, 949-952.7. Dunn, T. M., Hahn, S., Ogden, S. & Schleif, R. F. (1984) Proc. Nail.

Acad. Sci. USA 81, 5017-5020.8. Hahn, S., Dunn, T. & Schleif, R. (1984) J. Mol. Biol. 180, 61-72.9. Martin, K., Huo, L. & Schleif, R. F. (1986) Proc. Natl. Acad. Sci. USA

83, 3654-3658.10. Dandanell, G. & Hammer, K. (1985) EMBO J. 4, 3333-3338.11. Valentin-Hansen, P., Albrechtsen, B. & L0ve Larsen, J. E. (1986)

EMBO J, 5, 2015-2021.12. Tuggle, C. K. & Fuchs, J. A. (1986) EMBO J. 5, 1077-1085.13. Ptashne, M. (1986) Nature (London) 322, 697-701.14. Casedaban, M. J. (1976) J. Mol. Biol. 104, 557-566.15. Lee, N. L., Gielow, W. 0. & Wallace, R. G. (1981) Proc. Natl. Acad.

Sci. USA 78, 752-756.16. Hendrickson, W. & Schleif, R. F. (1984) J. Mol. Biol. 174, 611-628.17. Lee, N., Francklyn, C. & Hamilton, E. P. (1987) Proc. Natl. Acad. Sci.

USA 84, 8814-8818.18. Hawley, D. K. & McClure, W. R. (1983) Nucleic Acids Res. 8,

2237-2255.19. Lee, J.-H., Burke, K. & Wilcox, G. (1986) Gene 46, 113-121.20. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning:A

Laboratory Manual (Cold Spring Harbor Laboratory, Cold SpringHarbor, NY).

21. Zoller, M. J. & Smith, M. (1983) Methods Enzymol. 100, 468-500.22. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci.

USA 74, 5463-5467.23. Shimada, K., Weiberg, R. A. & Gottesman, M. E. (1973) J. Mol. Biol.

80, 297-314.24. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold Spring

Harbor Laboratory, Cold Spring Harbor, NY), pp. 352-355.25. Liu-Johnson, H.-N., Gartenberg, M. R. & Crothers, D. M. (1986) Cell

47, 995-1005.26. Hochschild, A., Irwin, N. & Ptashne, M. (1983) Cell 32, 319-325.27. Ho, Y.-S., Wulff, D. & Rosenberg, M. (1986) in Regulation of Gene

Expression, eds. Booth, I. & Higgins, C. (Cambridge Univ. Press,Cambridge, UK), pp. 79-103.

28. Wilcox, G. (1974) J. Biol. Chem. 249, 6892-6894.29. Ogden, S., Haggerty, D., Stoner, C. M., Kolodrubetz, D. & Schleif, R.

(1980) Proc. Natl. Acad. Sci. USA 77, 3346-3350.30. Hahn, S. & Schleif, R. (1983) J. Bacteriol. 155, 593-600.31. Englesberg, E., Squires, C. & Meronk, F., Jr. (1969) Proc. Natl. Acad.

Sci. USA 62, 1100-1107.32. Lichensteinj H. S., Hamilton, E. P. & Lee, N. (1987) J. Bacteriol. 169,

811-822.33. Englesberg, E. (1961) J. Bacteriol. 81, 996-1006.34. Hochschild, A. & Ptashne, M. (1986) Cell 44, 681-687.35. Krimer, H., Niem6ller, M., Amouyal, M., Revet, B., von Wilcken-

Bergmann, B. & M0ller-Hill, B. (1987) EMBO J. 6, 1481-1491.36. Miyada, C. G., Horwitz, A. H., Cass, L. G., Timko, J. & Wilcox, G.

(1980) Nucleic Acids Res. 22, 5267-5274.37. Deuschle, U., Gentz, R. & Bujard, H. (1986) Proc. Natl. Acad. Sci.

USA 83, 4134-4137.38. Ptashne, M. (1986) A Genetic Switch (Blackwell Scientific, Palo Alto,

CA).39. Dynan, W. S. & Tijian, R. (1985) Nature (London) 316, 774-778.40. Struhl, K. (1987) Cell 49, 295-297.

Biochemistry: Hamilton and Lee