Suppressor Mutations in rpoA Suggest that OmpR ControlsTranscription by Direct Interaction with the
a Subunit of RNA PolymeraseJAMES M. SLAUCH,t FRANK D. RUSSO, AND THOMAS J. SILHAVY*
Department of Molecular Biology, Lewis Thomas Laboratory,Princeton University, Princeton, New Jersey 08544-1014
Received 9 July 1991/Accepted 30 September 1991
We have isolated mutations in rpoA, the gene encoding the a subunit of RNA polymerase, that specificallyaffect transcriptional control by OmpR and EnvZ, the two-component regulatory system that controls poringene expression in Escherichia coli. Characterization of these mutations and a previously isolated rpoA allelesuggests that both positive and negative regulation of porin gene transcription involves a direct interactionbetween OmpR and RNA polymerase through the a subunit. Several of the rpoA mutations cluster in thecarboxy-terminal portion of the a protein, further suggesting that it is this domain of a that is involved ininteraction with OmpR and perhaps other transcriptional regulators as well.
Several different mechanisms have been proposed toexplain how regulatory proteins function to catalyze tran-scription initiation (1). A common feature of many of thesemodels is a direct physical interaction between the activatorprotein and the transcription apparatus. Evidence support-ing such an interaction has been presented for severalprokaryotic regulatory proteins. CRP, the cyclic AMP(cAMP) receptor protein, is an activator of many genes andoperons and controls expression of lac, for example, byfacilitating the binding of RNA polymerase (23). cI, Arepressor, regulates its own expression by enhancing opencomplex formation at PRM (17, 34). In this case as well,binding studies suggest a direct interaction (18). NR1 (NtrC)is the effector of the two-component regulatory system thatstimulates expression of the genes of the nitrogen regulon. Itactivates the glnAp2 promoter, for example, via a mecha-nism that involves DNA looping (40). A complete under-standing of transcriptional regulation requires a more de-tailed description of these activator-RNA polymeraseinteractions. In the case of cI, genetic studies suggest aregion of the repressor molecule that is involved in contact-ing RNA polymerase(6). However, in all cases, little isknown about the oppos-ing contact region of the polymer-ase. We have sought to address this question in our studiesof the two-component system that regulates porin geneexpression in Escherichia coli.The two-component regulatory system, OmpR and EnvZ,
controls the differential expression of the porin genes, ompFand ompC, in response to medium osmolarity. EnvZ is areceptor kinase that functions as a sensor of osmolarity andcommunicates this information to the effector, OmpR (anNR1 homolog), by a mechanism involving phosphorylationand dephosphorylation. OmpR-P controls transcription ini-tiation by binding to sites upstream of the appropriatepromoters (19). We have shown previously that OmpRworks in a positive fashion at the ompC promoter and bothpositively and negatively at the ompF promoter (37, 38). Inmedia of low osmolarity, levels of OmpR-P are low and
ompF expression is activated. In media of high osmolarity,levels of OmpR-P increase, ompC expression is activated,and ompF expression is repressed (11). Mutations in envZthat result in increased or decreased OmpR-P affect poringene expression accordingly (2, 31).
Several envZ alleles, e.g., envZ473 or envZII, causepleiotropic transcriptional defects. These alleles confer adominant OmpF-, OmpC-constitutive phenotype (37). Inaddition, they hinder the expression of several genes whichare not members of the porin regulon (e.g., malT and phoA),thus conferring pleiotropic xr and PhoA- phenotypes (9, 44).Genetic and biochemical analyses of these envZ mutantsshow that the phenotypes are not the result of a "rogue"kinase but rather are the consequence of high intracellularlevels of OmpR-P (2, 31, 36). Apparently, high levels ofOmpR-P can interfere with the activity of RNA polymeraseat certain promoters. Support for this prediction comes fromsuppressor analysis. Garrett and Silhavy (12) isolated extra-genic suppressors of envZ473, termed sez, that alleviate, atthe transcriptional level, all of the dominant negative phe-notypes conferred by envZ473. These mutations were tenta-tively mapped to rpoA, the gene for the a subunit of RNApolymerase. Matsuyama and Mizushima (26) have also dis-covered a mutation that maps in rpoA and prevents thesuppression of the pleiotropic envZ allele, envZJI, by theompR suppressor allele, ompR77 (25).The rpoA mutations described above suggest that OmpR-P
might interact with the a subunits ofRNA polymerase in theprocess of regulating porin gene expression. However, all ofthese rpoA mutations confer a phenotype only in the pres-ence of a pleiotropic envZ allele (12, 26). Therefore, it isquestionable whether this putative interaction betweenOmpR-P and the polymerase reflects the normal situation oris of significance only in strains carrying the pleiotropic envZalleles which cause abnormally high levels of OmpR-P. Inorder to distinguish between these two possibilities, we haveisolated and characterized mutations in rpoA that affect theregulation of the porin genes in a wild-type, i.e., ompR+envZ+, background. Our results suggest that regulatoryproteins such as OmpR may communicate directly withRNA polymerase via an interaction with the a subunit.
a Unless otherwise noted, strains were created in this study.
MATERIALS AND METHODSBacteria and bacteriophages. Parent bacterial strains and
bacteriophages are listed in Table 1. Mutant derivatives ofstrains listed in Table 1 were constructed simultaneously andare isogenic except for the mutant allele. Standard microbi-ological techniques were used for generalized transduction
and bacterial growth (35). Transformation of plasmid DNAwas carried out as previously described (35).Media, reagents, and enzymes. Most growth media were
prepared as previously described (35). M63 glucose minimalmedium (35) was prepared with added Casamino Acids(Difco) at 1 mg/ml. Glycerol-MOPS (morpholinepropane-sulfonic acid) alone and with added 15% filter-sterilizedsucrose was prepared as described by Neidhardt et al. (28)with 0.4% glycerol, 0.1 mg of thiamine, and 1 mg ofCasamino Acids per ml. The inducing medium for alkalinephosphatase assays was MOPS minimal medium with 0.4%glycerol made as described previously (28) except thatK2HPO4 was added to 132 ,uM. All restriction enzymes werepurchased from New England BioLabs. Reagents for DNAsequence analysis were obtained from United States Bio-chemical Corp.
Biochemical assays. Assays of alkaline phosphatase and,-galactosidase were performed with a microtiter assay asdescribed previously (38). The formula used to calculate theP-galactosidase activity was as follows:
(2 103) (A420min) (pmol of ONP. cm/4500 ml) (0.25 mlI0.78 cm)(A60) (ml of cell suspension added to reaction)
This calculation yields an activity that is given as (units perA6w. milliliter of cell suspension) -103, where units = ,umolof o-nitrophenol (ONP) formed per minute. Alkaline phos-phatase assays, adapted from the work of Brickman andBeckwith (5), were performed in an analogous fashion. Cellswere grown to stationary phase in low-phosphate glycerol-MOPS, centrifuged, and resuspended in 1 M Tris, pH 8.Activity is given as (units per A60- milliliter of cell suspen-sion) -103, where units = micromoles of para-nitrophenolformed per minute.
Localized mutagenesis of rpoA. Cultures of JMS3400(zhc-3::TnJO) were mutagenized with N-methyl-N'-nitro-N-nitrosoguanidine by a standard protocol and grown over-night in LB (35). Plvir lysates were grown on the mutagen-ized cultures and used to transduce either MH513 or MH225to tetracycline resistance on lactose-tetrazolium medium(35) containing 8 ,ug of tetracycline per ml.
Construction of pJS115 and pJS116. The plasmid pNO2530(3) was restricted with AatII, which cuts in pBR322 se-quences, and ApaI, which cuts 300 bp downstream from thetermination codon of rpoA. The protruding 3' ends of theDNA were digested with T4 DNA polymerase, and theplasmid was recircularized by ligation (32). The plasmidpJS116 was constructed from pJS115 by restricting theplasmid with HindIII and recircularizing the large fragment,deleting the translational start site and approximately 70% ofthe rpoA gene. All constructs were verified by restrictionanalysis. In addition, pJS115 was shown to complement anrpoA109 strain for P2 growth.PCR amplification and DNA sequence analysis. Mutational
changes were determined by polymerase chain reaction(PCR) amplification of a 1.1-kbp fragment, which includedthe entire rpoA structural gene, followed by DNA sequenceanalysis of the amplified fragment. The primers used for thePCR had the following sequences: 5'-CCgGATCcTCGAGCTTTACTCCAAG-3' and 5'-CCACTCTagaGATGGCGCATGACC-3'. These sequences correspond to the sense strandfrom base pairs 156 to 180 and the antisense strand from basepairs 1259 to 1236, respectively, of the published sequence ofthe operon containing rpoA, with changes (denoted bylowercase letters) to introduce restriction sites near the endsof the amplified fragment. PCR was carried out in a buffer
a The strains used were as follows: for cross la and b, JMS4546 (donor) and JMS72 (recipient); for cross 2, JMS4546 (donor) and JMS75 (recipient); and forcross 3, JMS76 (donor) and JMS4566 (recipient).
b SpC, spectinomycin.
containing 2.0 mM Mg2e for 30 cycles of 1 min at 94°C, 2 minat 55°C, and 2.5 min at 72°C.The sequence of the entire double-stranded amplified frag-
ment containing each mutation was determined directly by avariation of a published protocol (21). In addition to the twoprimers which were also used for amplification, the followingfour additional internal primers were used for sequencing:5'-CAGGATCGGATCGAACTCTG-3' (rpoA948), 5'-GTAGAAGCCGGCACATAACC-3' (rpoA659), 5'-CGCCTTCTTTGGTGCTGTAC-3' (rpoA409), and 5'-CCACTATATCGGTGATCTGG-3' (rpoA1033 rev). Details of the procedureused to amplify and sequence these mutations will be de-scribed separately.
Arabinose sensitivity assays. Sensitivity to arabinose wasquantitated on the basis of inhibition of growth. A lawn ofcells was plated on minimal glycerol agar in 2.5 ml of F topagar (35). Sterile filter disks (7-mm diameter) were thenplaced on the lawn, and 10-1ul aliquots of 0.1, 0.5, and 1%arabinose were placed on individual disks. Plates wereincubated overnight at 37°C, and the zone of inhibition wasmeasured.
Nucleotide sequence accession number. The sequence ofthe operon containing rpoA has been assigned GenBankaccession number J01685 XOO/66.
RESULTS
Rationale and mutant isolation. If OmpR regulates RNApolymerase activity by direct interaction with the a subunit,then it should be possible to isolate mutations in rpoA thataffect this interaction and alter porin gene expression. Inorder to test this hypothesis, we sought mutations thatdecreased either ompF or ompC expression in an otherwisewild-type background. This procedure differs from previousscreens that yielded mutations in rpoA because it does notinvolve pleiotropic envZ alleles.We performed localized mutagenesis of the region of the
chromosome containing rpoA and screened for mutationsthat decreased the ,B-galactosidase activity produced from an
ompF'-IacZ+ or an ompC'-lacZ+ operon fusion in anompR+ envZ+ background. Strain JMS3400 contains a TnJOinsertion that is approximately 65% linked to rpoA by P1transduction. Four independent cultures of this strain weremutagenized with nitrosoguanidine and used as donors totransduce either MH513, which contains an ompF'-lacZ+fusion, or MH225, which contains an ompC'-lacZ+ fusion,to tetracycline resistance on lactose-tetrazolium agar. Trans-ductants exhibiting decreased Lac activity were chosen forfurther analysis. Of 40,600 transductants in the ompF'-IacZ+fusion strain MH513, four mutants with decreased Lacactivity were isolated. Three of the four mutants came fromthe same mutagenized P1 lysate but were isolated in separatetransductions. In the case of the transductions into theompC'-IacZ+ fusion strain MH225, 44,500 transductantswere screened, and two independent mutants were isolatedand kept for further analysis. The mutations (termed dip,decrease in porin expression) that confer decreased Lacactivity to the ompF'-lacZ+ fusion and the ompC'-lacZ+fusion were all approximately 65% linked to the TnJO, thecorrect linkage for mutations in rpoA.The dip mutations are alleles of rpoA. To establish that the
dip mutations did actually map to the vicinity of rpoA, wecarried out more precise genetic mapping by using dip-54 asa representative allele. The dip-54 mutation and the linkedTnJO were mapped with respect to aroE and spc (rpsE) bythree-factor cross. The mutant strain was used to transduceJMS72 (aroE spcR ompF'-lacZ+) to aroE+, and transduc-tants were scored for resistance to spectinomycin and tetra-cycline and for decreased Lac activity (Table 2). This crossconfirmed that the zhc-3::TnJO was the most distal of thethree markers to aroE. This also allowed us to map thedip-54 allele with respect to the zhc-3::TnJO and spc byreciprocal three-factor cross by using the TnlO as the se-lected marker and scoring for the spc and dip alleles. Theresults of these three-factor crosses show unambiguouslythat the gene order is aroE dip-54 rpsE zhc-3: :TnlO (Table 2).To localize the dip alleles further, we directly sequenced
the rpoA gene from each mutant after amplification by PCR.
FIG. 1. Diagram of the cx subunit protein sequence showing changes caused by the various 1poA mutations with corresponding codonchanges. The rpoA109 mutation results in a change of Leu to His at amino acid 290 (9a). The rpoA341 allele, although it has not beensequenced, has been genetically mapped to the carboxy-terminal side of amino acid 160 (30). The sequence of rpoA77 (26) is not known.
In each case, the entire rpoA gene was sequenced, and allmutations were confirmed by sequencing the oppositestrand. The results of this analysis for the dip-50, dip-52,dip-53, dip-54, and sez-85 (12) alleles are shown in Fig. 1.With the exception of dip-S0, each of these mutations resultsin a single base pair substitution in the rpoA gene. We willrefer to them hereafter as rpoA alleles (dip-52 = rpoA52,etc.). The dip-S0 mutant contains two changes, both ofwhich are in rpoA, and we will refer to these changes asrpoAS0-1 and rpoAS0-2 (Fig. 1). We conclude that in eachcase presented, the Dip phenotype is the result of a mutationin rpoA.The rpoA alleles affect the porin regulon specifically. It is
not difficult to imagine mutations in the gene for the exsubunit of RNA polymerase that cause transcriptional de-fects. However, the mutations we sought are predicted toaffect interactions with OmpR specifically and would, there-fore, be distinguished from other a mutants by the restrictionof such transcriptional defects to promoters affected byOmpR.We reasoned that if the rpoA mutants were generally
defective in transcription, there would be a growth defectunder many, if not all, conditions. Initially we tested for suchgrowth defects by streaking the mutants on minimal mediasupplemented with glycerol, glucose, or maltose. Indeed,the two mutant alleles isolated by screening for decreasedompC expression caused a growth defect on minimal media,and these mutants were not characterized further. On theother hand, no growth defects were observed for the rpoA85(sez), rpoA50, rpoA52, rpoA53, or rpoA54 mutants. Weconfirmed this observation quantitatively by determininggrowth curves for JMS4540 (rpoA+) and mutant derivatives.These strains contain an insertion mutation in the ompF geneas a result of the lacZ fusion construct. Thus, all of thestrains are OmpF-, precluding any effects on growth ratethat might be observed because of decreased production ofOmpF porin in the rpoA mutants. The observed doublingtime of these mutants in both glucose and glycerol M63minimal media is indistinguishable from the isogenic rpoAtstrain. These results suggest that these rpoA mutations donot cause a generalized transcription defect.The cx subunit of RNA polymerase has previously been
implicated in promoter activation by other positive tran-scriptional activators. We were interested in examiningwhether the newly isolated rpoA mutations behaved likeother alleles that had been isolated to affect transcriptionalactivation. One such allele is rpoA109, which interferes withactivation of the late operon of bacteriophage P2 (41);suppressors of this mutation map to the gene encoding the
positive activator of this operon (10, 41). We found thatrpoA109 has no apparent effect on porin regulation. Con-versely, none of the rpoA mutations examined in this studyhave any apparent effect on the growth of P2, as determinedby examining P2 plaques on lawns of the various rpoAstrains. Clearly, the newly isolated rpoA mutations do notbehave like rpoA109.Another allele isolated to affect positive regulation is
rpoA341, which was originally termed phs and has a dra-matic effect on the expression of mnelAB, cysA, and the araoperon (13). The fact that the rpoA alleles isolated in thisstudy grow on minimal medium shows that they are notdefective in cysA expression. Also, these mutants are capa-ble of utilizing melibiose as the sole carbon source, suggest-ing that expression of the melAB locus is normal. We testedwhether the rpoA mutations affect ara expression by assay-ing sensitivity to arabinose. The strains used in this studycarry the araD139 allele and are sensitive to arabinose as aresult of the toxic accumulation of a metabolic intermediate.Any decrease in expression of the arabinose operon shoulddecrease this sensitivity. We assayed arabinose sensitivitythrough a disk assay analogous to antibiotic-sensitivity as-says. These assays showed no difference in the arabinosesensitivity of the wild-type and various rpoA strains, sug-gesting that expression of the arabinose operon is unaffectedin the rpoA mutants (strain JMS4640.0 and mutant deriva-tives). In addition to providing further evidence against ageneral transcriptional defect, the fact that ara expression isnormal indicates that the newly isolated rpoA mutations donot, in fact, behave like the previously isolated rpoA341allele. The newly isolated rpoA mutations, therefore, form aseparate class from those represented by either rpoA109 orrpoA341.A further possibility is that the rpoA mutations decrease
ompF expression indirectly by altering OmpR production.To test this hypothesis, we monitored the ,B-galactosidaseactivity produced from an ompR'-'lacZ protein fusion car-ried on a X-specialized-transducing phage, XpMLB954 (20;see below). 1-Galactosidase assays were performed withisogenic ompR+ envZ4 strains lysogenic for XpMLB954 andcontaining the various rpoA alleles (JMS 4650.0 and mutantderivatives). The rpoA alleles had no significant effect on the3-galactosidase activity produced from the ompR'-'lacZ
fusion compared with the rpoA+ control [28.1 + 0.9 (units/A600 milliliter of cell suspension) 1031. These resultsindicate that the decreased expression of ompF observed inthe rpoA strains is not due to the decreased production ofOmpR. Again, this result provides further evidence that therpoA mutations do not cause a geneerlized decrease in
a Cells were grown to mid-log phase in M63 glucose minimal medium withadded Casamino Acids. Activity of p-galactosidase was determined as de-scribed in Materials and Methods and is given as (units/A6w. milliliter of cellsuspension) 103, where units = micromoles of ONP formed per minute. Theaverage P-galactosidase activity and standard deviation for triplicate assays
on two colonies are shown. The strains used were JMS4430, JMS4420, andmutant derivatives.
transcription. Rather, their effects are restricted to specificpromoters, and these promoters are different from thoseaffected by rpoA alleles isolated in previous screens.
The defect in ompF expression caused by the rpoA alleles isOmpR dependent. If the effect on ompF transcription ob-served in the rpoA mutants is the result of decreasedinteraction between RNA polymerase and OmpR, the tran-scriptional regulator of ompF, then the transcriptional defectshould be dependent on the presence of OmpR. In otherwords, there should be no effect on the basal, OmpR-independent transcription from the ompF promoter. In orderto test this prediction, we measured the ,B-galactosidaseactivity produced from an ompF'-lacZ' operon fusion in thepresence of the various rpoA alleles in an ompR null back-ground. Transcription from the ompF promoter is very lowin the absence of OmpR. To maximize the measurable,-galactosidase activity, we used the 'P(ompF'-lacZ+)248Ufusion carried on a X-specialized-transducing phage (29). The
3-galactosidase activity produced from this particular fusionis relatively high and is measurable even in the absence ofOmpR (Table 3). As expected, the expression of the ompF'-lacZ+ fusion is decreased by the presence of the rpoA allelesin an ompR+ background. However, there is no effect on theexpression of the fusion in an ompR null background. Inother words, the decrease in expression from the ompFpromoter observed in the rpoA mutants is dependent on thepresence of OmpR.These results are in contrast to those obtained with
mutations that alter the ompF promoter. These mutationscause a decrease in ompF expression by direct effect on
RNA polymerase. In this case, a similar decrease in tran-scription was observed regardless of the presence or absenceof OmpR, even when the decrease in expression caused bythe promoter mutation was only 50% (43). These resultsargue that even small OmpR-independent effects would havebeen detected if they had been present in the experimentabove.The rpoA mutations alter regulation of ompF by OmpR. We
have shown previously (37) that OmpR can alter ompFexpression either positively or negatively, depending on
environmental conditions. To quantitate the effects of therpoA mutations on the regulation of ompF expression, we
assayed the P-galactosidase activity produced from our
standard ompF'-lacZ+ operon fusion, '$(ompF'-lacZ+)16-13(15), under conditions of low and high osmolarity (Table 4).The various rpoA alleles affect ompF transcription to various
TABLE 4. Effects of rpoA mutations on expression of anompF'-IacZ' fusion in an ompR' env'Z' background
13-Galactosidase activity in medium'rpoA - Fluctuationallele Without With 15% ratio'
" Cells were grown to mid-log phasc in glycerol-MOPS medium with addedCasamino Acids with and without 15% sucrose. Activity of P-galactosidasewas deterinined as described in Materials and Methods and is given as(units/A6w* milliliter of cell suspension) 103, where units = micromoles ofONP formed per minute. The average P-galactosidase activity and standarddeviation for duplicate assays of four colonies are shown. The strains usedwere JMS4540 and mutant derivatives.
b Fluctuation ratio = ,13-galactosidase activity in low osmolarity/p-galactosi-dase activity in high osmolarity.
degrees, with the least effect (19% decrease) caused byrpoA85 (sez) in high osmolarity and the greatest effect (60%decrease) caused by rpoA54 in low osmolarity. Note that inmost cases, the decrease in transcription is observed in bothlow and high osmolarity. In the case of rpoA53, however, thedefect is significantly greater in high-osmolarity medium.The significance of this variability of effect will be discussedbelow.We also characterized the effect of the rpoA alleles on
ompF transcription in the presence of various ompR andenvZ mutations (Table 5). The mutation ompR472 is termedan ompR2 mutation and is phenotypically OmpF constitu-tive, OmpC- (15, 37). As expected, the absolute ,B-galactosi-dase activity produced from the ompF'-lacZ' fusion ishigher in the ompR472 strain than in the wild-type parent.The rpoA mutations decrease this activity in a mannersimilar to that seen with the ompR+ strain.
In contrast, the rpoA mutations have no significant effecton the phenotype conferred by an omnpR3 allele, whichcauses repression of OmpF and constitutive expression ofOmpC (37). The mutation ompR107 causes reduced expres-sion of P-galactosidase from the ompF'-lacZ' fusion even inthe presence of the rpoA mutations, including the sez allelerpoA85 (Table 5).
TABLE 5. Effects of rpoA mutations on expression of anompF'-lacZ+ fusion in various ompR and enisZ mutants
3-Galactosidase actmvity' with ompR or envZ allelerpoAallele ompRens onpR472 ornpR107 en vZ473
a Cells were grown to mid-log phase in M63 glucose minimal medium withadded Casamino Acids. Activity of j-galactosidase was determined as de-scribed in Materials and Methods and is given as (units/Awo milliliter of cellsuspension)- 103, where units = micromoles of ONP formed per minute. Theaverage P-galactosidase activity and standard deviation for duplicate assays oftwo colonies are shown. The strains used were JMS4540, JMS4470, JMS4460.JMS4480, and mutant derivatives.
a Strains used were JMS4640.0, JMS4650.0, JMS4641.0, JMS4651.0, JMS4644.0, JMS4654.0, and mutant derivatives.bProduction ofOmpF (F) and OmpC (C) was monitored by testing sensitivity to the OmpF-specific phage, K20, and the OmpC-specific phage, SS4. + indicates
sensltivity to phage killing, and indikates resistance to phage killing.
We wondered whether the interpretation that the rpoAmutations do not suppress ompR107 might actually be toosimplistic. Perhaps the OmpR107 protein is incapable ofactivating the ompF promoter. It is possible, then, that rpoAmutations suppress the dominant negative OmpF- functionof OmpR107, but because of the defect in activation ofompF, there is no apparent change in the phenotype. If thisis indeed the case, then the provision of an additional OmpRprotein to activate ompF expression while an rpoA mutationsuppresses the dominant OmpF- phenotype of ompR107should lead to restoration of expression.To perform this experiment, we chose the previously
characterized ompR allele, the (:(ompR'-'1acZ)1I(Hyb) genefusion, to provide an additional OmpR protein capable ofactivating ompF transcription. The hybrid protein specifiedby this fusion contains all but the last amino acid of OmpRfused in frame to the carboxy-terminal fragment of P-galac-tosidase (4). The fusion protein is bifunctional, possessingboth P-galactosidase activity and OmpR activity. In partic-ular, it confers an OmpF-constitutive, OmpC- phenotype (4,36, 37). The various rpoA alleles were transduced intomerodiploids in which the 4)(ompR'-'lacZ)lJ(Hyb) fusion,carried on a specialized transducing phage, XpMLB954 (20),is in trans to ompR+, the ompR null mutation ompR101, orompR107. Production of OmpF and OmpC in these strainswas monitored by scoring sensitivity to phage that use theseproteins as receptors. First, we can show (Table 6) that therpoA mutations do not interfere with the activation of ompFby the 'F(ompR'-'IacZ)JJ(Hyb) allele as measured by the sen-sitivity to the OmpF-specific phage, K20 [O(ompR'-'lacZ)J1(Hyb) ompR10 merodiploids]. However, the ompR1074(ompR'-'1acZIJ(Hyb) merodiploid strains remain resistantto K20 in the presence of the various rpoA alleles, con-firming that these mutations do not suppress the dominantOmpF- phenotype conferred by ompR107.
Finally, we examined the effect of the rpoA mutations onthe expression of ompF in an envZ473 background. Thepleiotropic envZ473 allele confers an OmpF-, OmpC-con-stitutive phenotype, similar to the phenotype conferred byompR3 mutations. As shown in Table 5, rpoA85 (sez)partially suppresses the OmpF- phenotype conferred byenvZ473. Indeed, this is one of the criteria by which thisallele was isolated (12). The other rpoA alleles (dip), isolatedindependently of any pleiotropic envZ alleles, show variousphenotypes with respect to suppression of the OmpF-phenotype of envZ473. For example, rpoA50 has no effect on
expression of the ompF'-lacZ+ fusion in the envZ473 strain.On the other hand, rpoA52 and rpoA54 both suppress theenvZ473 phenotype, although not as strongly as doesrpoA85. Curiously, rpoA53 actually enhances the ompFexpression defect caused by envZ473.The results presented in this section show that the rpoA
mutations affect the regulation of ompF expression. Thevarious rpoA alleles have different effects, depending on theconditions tested. However, the differences in behaviorbetween alleles under one set of conditions cannot easily beused to predict how they will behave under another.The rpoA mutations affect the pleiotropic phenotypes of
envZ473. The rpoA85 (sez) allele was isolated as a suppressorof both the OmpF- and the pleiotropic phenotypes ofenvZ473 (12). We examined the effects of the rpoA allelesisolated in this study on the pleiotropic phenotypes con-ferred by envZ473 in several ways.
First, we assayed the alkaline phosphatase activity pro-duced in envZ+ and envZ473 strains containing the variousrpoA alleles after induction in low-phosphate medium (Table7). In the envZ+ background, the various rpoA alleles havevirtually no effect on the production of PhoA, providinganother example of the specificity of the rpoA mutations. Inthe envZ473 rpoA+ strain, phoA expression is dramaticallydecreased, indicative of the pleiotropic effects conferred byenvZ473 functioning through OmpR (36). As shown previ-
TABLE 7. Effects of rpoA mutations on the PhoA- phenotypeconferred by envZ473
a Cells were grown approximately 12 h in glycerol-MOPS medium contain-ing 132 IuM K2HPO4. Activity of alkaline phosphatase was determined asdescribed in Materials and Methods and is given as (units/A60. milliliters ofcell suspension) * 1, where units = micromoles of para-nitrophenol formedper minute. The average alkaline phosphatase activity and standard deviationfor duplicate assays of three colonies are shown. The strains used wereJMS4480, JMS4540, and mutant derivatives.
Cells were grown to mid-log phase in glycerol-MOPS medium with addedCasamino Acids with and without 15% sucrose. Activity of ,-galactosidasewas determined as described in Materials and Methods and is given as(units/Awo milliliter of cell suspension) 103, where units = micromoles ofONP formed per minute. The average P-galactosidase activity and standarddeviation for duplicate assays of four colonies are shown. The strains usedwere JMS4530 and mutant derivatives.
b Fluctuation ratio = P-galactosidase activity in high osmolarity/,B-galac-tosidase activity in low osmolarity.
ously (12), rpoA85 (sez) partially suppresses the effect ofenvZ473 on phoA expression. The rpoA alleles isolated inthis study show a variety of effects on the PhoA- phenotypeof envZ473. For example, rpoA53 partially suppresses thePhoA- phenotype but not as well as does rpoA85. BothrpoA52 and rpoA54 suppress the PhoA- phenotype ofenvZ473 better than rpoA85, whereas the allele rpoA50 doesnot have an effect. Note that there is no direct correlationbetween the suppression of the OmpF- phenotype (Table 5)and the suppression of the PhoA- phenotype of envZ473 bythe various rpoA alleles. None of the mutants produce asignificant amount of PhoA when grown in phosphate-richmedium (data not shown), indicating that the rpoA allelessuppress the PhoA- phenotype of envZ473 by preventingthe action of the pleiotropic envZ allele, not by causingconstitutive expression of phoA.The envZ473 allele also decreases the expression of malT,
the positive activator of the maltose regulon (9), therebyconferring a decreased ability to metabolize maltose. Wemonitored suppression of the envZ473 Mal- phenotype bythe various rpoA alleles through the ability of the variousstrains to utilize maltose on maltose-tetrazolium agar. Thevarious rpoA alleles suppress the Mal- phenotype ofenvZ473 in the same manner and to approximately the samedegree that they suppress the PhoA- phenotype. Similarresults are obtained by scoring sensitivity to phage X, thusmonitoring production of another protein of the maltoseregulon, LamB. These results indicate that those rpoAalleles that suppress the pleiotropic phenotypes conferred byenvZ473 suppress many, if not all, such phenotypes toapproximately the same degree.The rpoA mutations affect the regulation ofompC by OmpR.
The effect of the various rpoA mutations on the expressionof ompC was quantitated by measuring the P-galactosidaseactivity produced from an ompC'-lacZ+ operon fusion.Table 8 shows the results of 3-galactosidase assays per-formed with ompR+ envZ+ strains containing the variousrpoA alleles grown in media with low and high osmolarity.Under these conditions, the rpoA85 allele has no effect onompC expression. The alleles isolated in this study dodecrease ompC expression but not to the extent that theydecrease expression of ompF. This is especially the casewith high osmolarity, in which the greatest effect is a 28%decrease in 3-galactosidase activity caused by rpoA53.
TABLE 9. Effects of rpoA mutations on expression of anompC'-lacZ+ fusion in various ompR and envZ mutants
13-Galactosidase activity" with ompR and envZ allelerpoAallele ompRfZ ompR472 ompR107 envZ473
" Cells were grown to mid-log phase in M63 glucose minimal medium withadded Casamino Acids. Activity of 3-galactosidase was determined as de-scribed in Materials and Methods and is given as (units/A600 milliliter of cellsuspenision) 103, where units = micromoles of ONP formed per minute. Theaverage activity and stanidar-d deviation for at least duplicate assays of twocolonies are shown. The strains used were JMS4530, JMS4410, JMS4490,JMS4440 and mutant derivatives.
The effect of the rpoA alleles on the expression of ompC invarious ompR and envZ mutant backgrounds is shown inTable 9. While the rpoA alleles cause a reproducible de-crease in 3-galactosidase production from an ompC'-IacZ+fusion in the ompR+ envZ+ strains, there is no such effect onompC expression in the presence of ompR472 (OmpF con-stitutive, OmpC--), with the exception that rpoA54 showssome decrease. Several of the rpoA mutations confer a slightincrease in ,-galactosidase activity in the ompRI07 back-ground (OmpF, OmpC constitutive). However, in theenvZ473 background, several of the rpoA alleles causedecreased ,B-galactosidase production from the ompC'-lacZ+ fusion, with rpoA54 showing the greatest effect (a 50%decrease).The results presented in this section demonstrate that in
addition to their effects on ompF, the newly isolated rpoAmutations also affect the regulation of ompC. Again, theeffects of the various rpoA alleles are different, depending onconditions, and again, there is no straightforward way topredict the behavior of a given allele under one set ofconditions based on its behavior under another.
Overproduction of rpoA alters porin gene regulation. Inorder to gain further insight into the mechanism by which thevarious rpoA mutations affect porin gene expression, weattempted to perform diploid analysis by providing an rpoA +gene in trans. The pBR322 derivative, pJS115, was con-structed from pNO2530 (3) and contains the rpoA gene underthe transcriptional control of the lac promoter. Other DNAfrom the a operon is limited to approximately 300 bp on boththe 5' end and the 3' end of rpoA. As a control for the diploidanalysis, pJS115 was cleaved with HindlII and recircular-ized, deleting from the polylinker in lac through approxi-mately 70% of the rpoA gene to yield pJS116.The above plasmids were transformed into both ompR+
envZ473 strains and ompR' envZ+ strains containing thevarious rpoA alleles. In the former case, we were looking forrestoration of the envZ473 phenotype by rpoA + in thepresence of rpoA alleles that otherwise suppress envZ473.Unexpectedly, we found that introduction of the rpoA'plasmid, pJS115, itself suppresses the phenotypes conferredby envZ473, even in the absence of other rpoA suppressoralleles. Moreover, in an ompR+ envZ+ background, intro-duction of pJS115 decreases expression of ompF. Introduc-tion of the control plasmid pJS116 does not confer thesephenotypes. It should be noted that the rpoA+ plasmid has
no apparent effect on the growth properties of the cell.Moreover, the observed effects are both positive and nega-tive. In other words, it does not appear that overproductionof ax is simply interfering with transcription. One tantalizingpossibility is that high levels of cx interfere directly with thefunction of OmpR.
DISCUSSION
We have isolated and characterized mutations in rpoA, thegene encoding the cx subunit of RNA polymerase, that affectthe transcriptional regulation of genes controlled by OmpRand EnvZ. These rpoA mutations do not confer a generaltranscription defect, nor are they analogous to other rpoAalleles isolated previously. We suggest that these new rpoAalleles affect a specific interaction between OmpR and the axsubunit of RNA polymerase.There are several possible ways in which rpoA mutations
could cause the changes we observe in transcription. Per-haps the rpoA mutations simply decrease the ability ofOmpR to interact with RNA polymerase in a general faviion.Since OmpR works both positively and negatively, a de-creased interaction with the polymerase could cause either adecrease or an increase in transcription. However, this typeof general interaction mutation should affect all of thephenotypes mediated by OmpR and EnvZ. Most of the rpoAmutations characterized here do not fit this category. Forexample, the porin phenotypes conferred by ompR107 andenvZ473 are very similar. Both mutations alter OmpR, thuspreventing transcription of ompF (36, 37). Several of therpoA alleles suppress the OmpF- phenotype conferred byenvZ473 but do not suppress the OmpF- phenotype con-ferred by ompR107, even when ompF can be activated by anindependent mechanism. A general interaction mutationshould affect both of these phenotypes to the same degree.These results suggest that the rpoA mutations affect theinteraction between OmpR and the polymerase in a moreprecise fashion.We note further that the rpoA mutations we have de-
scribed have very different patterns of behavior under thevarious sets of conditions tested. Figure 2 summarizes theeffects of each of the rpoA alleles on ompF and ompCexpression in backgrounds of wild-type ompR and envZ,ompR107, and envZ473 and on phoA expression in anenvZ473 background. Examination of this graph reveals thatthe different rpoA alleles have distinct effects on thesevarious phenotypes, as evidenced by differences in theoverall pattern shown for each allele. This variation isparticularly striking when we compare the effects on thephenotypes conferred by envZ473. For example, rpoA85(sez) was isolated as suppressing the negative phenotypes ofenvZ473 but does not suppress the OmpF- phenotype ofompR107. In addition, the other rpoA alleles behave quitedifferently from rpoA85 in terms of suppression of theOmpF- and PhoA- phenotypes of envZ473. The rpoA50allele has no effect on either phenotype, whereas the rpoA52and rpoA54 alleles, which have only minor effects on theOmpF- phenotype, suppress the pleiotropic PhoA- pheno-type better than does rpoA85. The fact that the phenotype ofa particular rpoA allele in a given background cannot bepredicted from the phenotype of the same allele in a differentbackground suggests that the rpoA alleles are behaving in aspecific manner.The sequence analysis of the mutations rpoA52, rpoA54,
and rpoA85 is especially striking. Each changes one of twoadjacent proline residues in the carboxy-terminal region of
A B C D E F G A B C D E F GFIG. 2. Allele-specific pattern of rpoA mutations. Data from
Tables 5, 7, and 9 are presented as the maximum activity (%) for agiven promoter. For example, maximum activity from the ompC'-lacZ+ fusion is seen in the ompR+ envZ473 background (Table 9).This value is considered 100%. (A) ompC'-lacZ+ in ompR+ envZ+;(B) ompC'-lacZ+ in ompR107 envZ+; (C) ompC'-lacZ' in ompR+envZ473; (D) phoA in ompR+ envZ473; (E) ompF'-1acZ+ in ompR+envZ+; (F) ompF'-IacZ+ in ompR107 envZ+; (G) ompF'-lacZ+ inompR+ envZ473.
the a subunit. Changing either proline to a serine residueresults in similar phenotypes (rpoA52 and rpoA54), whereaschanging the second proline to leucine instead results inrather striking differences in the phenotype (rpoA85 andrpoA54; Fig. 2). This extreme specificity supports our con-tention that OmpR interacts directly with RNA polymerasethrough the a. subunit to control the transcriptional activityat the various promoters.Both of the previously isolated rpoA mutations, rpoA109
and rpoA341, result in changes in the carboxy-terminal halfof the ot subunit (Fig. 1) (9a, 30), a property shared by themutations characterized here, with two exceptions. TherpoA50 allele results in two changes in the a subunit, one atamino acid 28 (Leu to Phe) and one at amino acid 240 (Pro toSer). It is not known which of these changes is responsiblefor the observed phenotypes. However, the Leu to Phechange is conservative and, therefore, we suspect that thePro to Ser change is the significant one. The rpoA53 muta-tion changes Gly to Ser at position 3. However, the aminoacid change would seem minor, and we suspect that thismutation may exert its effect at the level of the mRNA byincreasing the efficiency of translation initiation; such aproposal is consistent with current views (for a review, seereference 27). Thus, it is possible that rpoA53 actuallyresults in an overproduction of a. Indeed, the phenotypeconferred by this mutation is similar to that conferred by aplasmid that increases the rpoA copy number (pJS115).Although we have characterized only a small number of
rpoA mutations that affect transcriptional regulation, most ofthese alter amino acid residues in the carboxy-terminalportion of the a protein. Accordingly, it is tempting tosuggest that it is this domain of the ot protein that isresponsible for interacting with the various transcriptional
VOL. 173, 1991 rpoA MUTATIONS AFFECT TRANSCRIPTIONAL CONTROL BY OmpR 7509
regulators. Results presented by Lombardo et al. in theaccompanying paper (22) strengthen this view.
All of these results are consistent with the hypothesis thatthe a subunit of RNA polymerase is involved in the regula-tion of gene expression by specifically interacting withvarious transcriptional regulators. In the case of OmpR, theresults of this interaction are both positive and negative.This implies that the negative regulation of ompF, and in thecase of envZ473, phoA and malT, is not the result of sterichindrance, the classic model for negative regulation, butrather the result of an active interaction that preventstranscription. Consistent with this interpretation is the factthat suppression of this negative regulation by rpoA85 isrecessive (12). If the mutant polymerase were simply over-coming steric hindrance by OmpR, it should be able to do sowhether wild-type polymerase is present and, thus, themutation should be dominant. This is clearly not the case.Indeed, Straney and Crothers (39) have provided evidencethat the lac repressor, the original paradigm for negativeregulation, does not function by causing steric hindrance butactively binds both the polymerase and operator site toprevent transcription initiation.The differential effects observed with the various rpoA
alleles are dependent both on the ompR and envZ back-ground and on the promoter in question. If these differencestruly reflect differences in interactions with OmpR, then wewould propose that there are subtle, but distinct, differencesin the various interactions of OmpR and the polymerase atdifferent promoters and under different conditions. Thesedifferences could be the result of distinct forms or confor-mations of OmpR or the result of slightly altered placementof OmpR and the polymerase at the various promoters orboth. Alternatively, the DNA could act as an allostericregulator of OmpR that slightly alters the function of theprotein, depending on which site OmpR binds (1, 24, 42).
This type of subtle and specific interaction between RNApolymerase and a transcriptional regulator is seemingly incontrast to the observations of Bushman and Ptashne, whopropose that activation is simply the result of "apposing anacidic patch to a single part of RNA polymerase" (7).Although the part of RNA polymerase is unspecified, theinteractions seem to be relatively nonspecific. For example,Bushman and Ptashne (7) were successful in converting therepressor protein X Cro into an activator simply by placingacidic residues in the DNA-binding helix-turn-helix regioncorresponding to the region in X repressor known to beinvolved in activation of transcription. Of course, these twomodels for transcriptional regulation are not mutually exclu-sive, and both mechanisms could certainly be invokeddepending on the needs of the cell.
ACKNOWLEDGMENTS
We thank Liya Shi for excellent technical assistance. We thankConrad Halling for the gift of rpoA109 and P2 phage and G. Christieand D. Ayers for communicating results prior to publication.
F.D.R. was supported by a predoctoral fellowship from theNational Science Foundation. This research was supported byPublic Health Service grant GM35791 to T.J.S.
ADDENDUM IN PROOF
K. Igarashi and A. Ishihama (Cell 65:1015-1022, 1991)have recently provided in vitro data consistent with thehypothesis that the carboxy-terminal portion of a is involvedin interaction with transcriptional regulators.
REFERENCES1. Adhya, S., and S. Garges. 1990. Positive control. J. Biol. Chem.
265:10797-10800.2. Aiba, H., F. Nakasai, S. Mizushima, and T. Mizuno. 1989.
Evidence for the physiological importance of the phosphotrans-fer between the two regulatory components, EnvZ and OmpR,in osmoregulation in Escherichia coli. J. Biol. Chem. 264:14090-14094.
3. Bedwell, D. M., and M. Nomura. 1986. Feedback regulation ofRNA polymerase subunit synthesis after conditional overpro-duction of RNA polymerase in Escherichia coli. Mol. Gen.Genet. 204:17-23.
4. Berman, M. L., and D. E. Jackson. 1984. Selection of lac genefusions in vivo: ompR-lacZ fusions that define a functionaldomain of the ompR gene product. J. Bacteriol. 159:750-756.
5. Brickman, E., and J. Beckwith. 1975. Analysis of the regulationof Escherichia coli alkaline phosphatase synthesis using dele-tions and 4)80 transducing phages. J. Mol. Biol. 96:307-316.
6. Bushman, F., C. Shang, and M. Ptashne. 1989. A single glutamicacid residue plays a key role in the transcriptional activationfunction of lambda repressor. Cell 58:1163-1171.
7. Bushman, F. D., and M. Ptashne. 1988. Turning X Cro into atranscriptional activator. Cell 54:191-197.
8. Casadaban, M. J. 1976. Transposition and fusion of the lacgenes to selected promoters in Escherichia coli using bacterio-phage lambda and mu. J. Mol. Biol. 104:541-555.
9. Case, C. C., B. Bukau, S. Granett, M. Villarejo, and W. Boos.1986. Contrasting mechanisms of envZ control of mal and phoregulon genes in Escherichia coli. J. Bacteriol. 166:706-712.
9a.Christie, G., and D. Ayers. Personal communication.10. Christie, G. E., E. Haggard-Ljungquist, R. Feiwell, and R.
Calendar. 1986. Regulation of bacteriophage P2 late-geneexpression: the ogr gene. Proc. Natl. Acad. Sci. USA 83:3238-3242.
11. Forst, S., J. Delgado, A. Rampersaud, and M. Inouye. 1990. Invivo phosphorylation of OmpR, the transcription activator ofthe ompF and ompC genes in Escherichia coli. J. Bacteriol.172:3473-3477.
12. Garrett, S., and T. J. Silhavy. 1987. Isolation of mutations in thea operon of Escherichia coli that suppress the transcriptionaldefect conferred by a mutation in the porin regulatory geneenvZ. J. Bacteriol. 169:1379-1385.
13. Giffard, P. M., and I. R. Booth. 1988. The rpoA341 allele ofEscherichia coli specifically impairs the transcription of a groupof positively-regulated operons. Mol. Gen. Genet. 214:148-152.
14. Hall, M. N., and T. J. Silhavy. 1979. Transcriptional regulationof Escherichia coli K-12 major outer membrane protein lb. J.Bacteriol. 140:342-350.
15. Hall, M. N., and T. J. Silhavy. 1981. The ompB locus and theregulation of the major outer membrane porin proteins ofEscherichia coli K12. J. Mol. Biol. 146:23-43.
16. Hancock, R. E. W., and P. Reeves. 1975. Bacteriophage resis-tance in Escherichia coli K-12: general pattern of resistance. J.Bacteriol. 121:983-993.
17. Hawley, D. K., and W. R. McClure. 1982. Mechanism ofactivation of transcription initiation from the XPRM. J. Mol. Biol.157:493-525.
18. Hwang, J., and G. Gussin. 1988. Interactions between Esche-richia coli RNA polymerase and lambda repressor mutations inPRM affect repression of PR. J. Mol. Biol. 200:735-739.
19. Igo, M. M., J. M. Slauch, and T. J. Silhavy. 1990. Signaltransduction in bacteria: kinases that control gene expression.New Biologist 2:5-9.
20. Jackson, D. E. 1984. M. S. Thesis. Hood College, Frederick,Md.
21. Kretz, K. A., G. S. Carson, and J. S. O'Brien. 1989. Directsequencing from low-melt agarose with Sequenase. NucleicAcids Res. 17:5864.
22. Lombardo, M.-J., D. Bagga, and C. G. Miller. 1991. Mutationsin rpoA affect expression of anaerobically regulated genes inSalmonella typhimurium. J. Bacteriol. 173:7511-7518.
23. Malan, T. P., A. Kolb, H. Buc, and W. R. McClure. 1984.Mechanism of CRP-cAMP activation of lac operon transcription
initiation activation of the P1 promoter. J. Mol. Biol. 180:881-909.
24. Martin, K., L. Huo, and R. F. Schleif. 1986. The DNA loopmodel for ara repression: AraC protein occupies the proposedloop sites in vivo and repression-negative mutations lie in thesesame sites. Proc. Natl. Acad. Sci. USA 83:3654-3658.
25. Matsuyama, S., T. Mizuno, and S. Mizushima. 1986. Interactionbetween two regulatory proteins in osmoregulatory expressionof ompF and ompC genes in Escherichia coli: a novel ompRmutation suppresses pleiotropic defects caused by an envZmutation. J. Bacteriol. 168:1309-1314.
26. Matsuyama, S., and S. Mizushima. 1987. Novel rpoA mutationthat interferes with the function of OmpR and EnvZ, positiveregulators of the ompF and ompC genes that code for outer-membrane proteins in Escherichia coli K-12. J. Mol. Biol.195:847-853.
27. McCarthy, J. E. G., and C. Gualerzi. 1990. Translational controlof prokaryotic gene expression. Trends Genet. 6:78-85.
28. Neidhardt, F. C., P. L. Bloch, and D. F. Smith. 1974. Culturemediumn for enterobacteria. J. Bacteriol. 119:736-747.
29. Ostrow, K. S., T. J. Silhavy, and S. Garrett. 1986. cis-actingsites required for osmoregulation of ompF expression in Esch-erichia coli K-12. J. Bacteriol. 168:1165-1171.
30. Rowland, G. C., P. M. Giffard, and I. R. Booth. 1985. phs locusof Escherichia coli, a mutation causing pleiotropic lesions inmetabolism, is an rpoA allele. J. Bacteriol. 164:972-975.
31. Russo, F. D., and T. J. Silhavy. EnvZ controls the concentrationof phosphorylated OmpR to mediate osmoregulation of theporin genes. J. Mol. Biol., in press.
32. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.
33. Schnaltman, C. A., and G. A. McDonald. 1984. Regulation ofouter membrane protein synthesis in Escherichia coli K-12:deletion of ompC affects expression of the ompF protein. J.
Bacteriol. 159:555-563.34. Shih, M., and G. Gussin. 1983. Differential effects of mutations
on discrete steps in transcription initiation at the X PRE pro-moter. Cell 34:941-949.
35. Silhavy, T. J., M. L. Berman, and L. W. Enquist. 1984.Experiments with gene fusions. Cold Spring Harbor Labora-tory, Cold Spring Harbor, N.Y.
36. Slauch, J. M., S. Garrett, D. E. Jackson, and T. J. Silhavy. 1988.EnvZ functions through OmpR to control porin gene expressionin Escherichia coli K-12. J. Bacteriol. 170:439-41.
37. Slauch, J. M., and T. J. Silhavy. 1989. Genetic analysis of theswitch that controls porin gene expression in Escherichia coliK-12. J. Mol. Biol. 210:281-292.
38. Slauch, J. M., and T. J. Silhavy. 1991. cis-acting ompF muta-tions that result in OmpR-dependent constitutive expression. J.Bacteriol. 173:4039-4048.
39. Straney, S. B., and D. M. Crothers. 1987. Lac repressor is atransient gene-activating protein. Cell 51:699-707.
40. Su, W., S. Porter, S. Kustu, and H. Echols. 1990. DNA-loopingand enhancer activity: association between DNA-bound NtrCactivator and RNA polymerase at the bacterial gInA promoter.Proc. Natl. Acad. Sci. USA 87:5504-5508.
41. Sunshine, M. G., and B. Sauer. 1975. A bacterial mutationblocking P2 phage late gene expression. Proc. Natl. Acad. Sci.USA 72:2770-2774.
42. Tan, S., and T. J. Richmond. 1990. DNA binding-inducedconformational change of the yeast transcriptional activatorPRTF. Cell 62:367-377.
43. Taylor, R. K., S. Garrett, E. Sodergren, and T. J. Silhavy. 1985.Mutations that define the promoter of ompF, a gene specifyinga major outer membrane porin protein. J. Bacteriol. 162:1054-1060.
44. Wandersman, C., F. Moreno, and M. Schwartz. 1980. Pleio-tropic mutations rendering Escherichia coli K-12 resistant tobacteriophage TP1. J. Bacteriol. 143:1374-1383.
