Use of a Two-Color Genetic Screen To Identify a Domain of the Global Regulator Lrp That Is Specifically Required for pap Phase Variation

Abstract

The global regulator Lrp plays a central role as both a repressor and an activator in Pap phase variation. Unlike most other members of the Lrp regulon such as ilvIH, activation of papBA transcription requires the coregulator PapI and is methylation dependent. We developed a two-color genetic screen to identify Lrp mutations that inhibit Pap phase variation but still activate ilvIH transcription, reasoning that such mutations might identify PapI binding or methylation-responsive domains. Amino acid substitutions in Lrp at position 126, 133, or 134 greatly reduced the rate of Pap switching from phase off to phase on but had much smaller effects on ilvIH transcription. In vitro analyses indicated that the T134A and E133G Lrp variants maintained affinities for pap and ilvIH DNAs similar to those of wild-type Lrp. In addition, both mutant Lrp’s were as responsive to PapI as wild-type Lrp, evidenced by an increase in affinity for pap Lrp binding sites 4, 5, and 6. Thus, in vitro analyses did not reveal the step(s) in Pap phase variation where these Lrp mutants were inhibited. In vivo analyses showed that both the T134A and E133G Lrp mutants activated transcription of a phase-on-locked pap derivative containing a mutation in Lrp binding site 3. Further studies indicated that the T134A Lrp mutant was blocked in a step in Pap phase variation that does not involve PapI. Our data suggest that these mutant Lrp’s are defective in a previously unidentified interaction required for the switch from the phase-off to the phase-on pap transcription state.

The global regulator leucine-responsive regulatory protein (Lrp) plays an essential and central role in regulating the phase variation of pyelonephritis-associated pili (Pap) in Escherichia coli (4). Pap expression is regulated in part by the binding of Lrp to two sets of DNA target sites located in the pap upstream regulatory region (11). The binding of Lrp at pap DNA target sites 1 to 3, which overlap the papBA pilin promoter, blocks transcription and Pap fimbrial expression, resulting in the off expression phase (Fig. 1A). Under conditions of high levels of cyclic AMP (cAMP), the PapI coregulatory protein is expressed from the divergent papI promoter. PapI causes an increase in the affinity of Lrp for pap DNA sites 4 and 5, located over 100 bp upstream from sites 1 to 3 (11). Transcription from the papBA promoter is activated by Lrp bound at sites 4 and 5 and by cAMP receptor protein (CRP), which binds about 30 bp upstream of pap site 4 (7) (Fig. 1A). Analysis of Lrp activation mutants has shown that binding of Lrp to pap DNA sites 4 and 5 is necessary but not sufficient for transcription activation (17). Lrp participates in activating papBA transcription, since lrp null mutants do not express Pap fimbriae (2, 3). Thus, Lrp may act as either a repressor or an inducer of transcription of the pap operon, depending on which set of pap DNA binding sites Lrp occupies (17).

FIG. 1.

Isolation and identification of mutations in lrp that specifically block transcription of the papBA operon. (A) Two-color genetic screen. The lacZ and phoA reporter genes were used to simultaneously detect transcription from ilvIH and papBA as described in Materials and Methods. Base pair locations of the regulatory regions of papBA and ilvIH are shown relative to the transcription start site at +1. Lrp binding sites are depicted as open rectangles (1, 18). The GATC-I and GATC-II sites are depicted as black boxes within Lrp binding sites 5 and 2, respectively. These GATC sites are substrates for Dam and are differentially methylated in Pap phase-on and -off states (3). (B) Functional map of Lrp. Amino acid regions of Lrp which are required for DNA binding, activation, and leucine responsiveness are based on an lrp mutational analysis by Platko and Calvo (13). Lrp mutations at amino acids 126, 133, or 134 that block PapI-dependent pap transcription are shown by arrows.

Pap phase variation is also regulated epigenetically via deoxyadenosine methylation of two pap DNA GATC sites designated GATC-I and GATC-II (16). GATC-I is located within Lrp binding site 5, whereas GATC-II is located within Lrp binding site 2 (Fig. 1A). Methylation of the pap DNA GATC-I site reduces the affinity of Lrp for sites 4 and 5 and blocks transition to the phase-on state. It is likely that this process serves to lock cells in the off state until DNA replication occurs and a hemimethylated GATC-I site is generated (12). In vitro binding analyses have shown that Lrp binds to pap regulatory DNA containing a hemimethylated GATC-I site with significantly higher affinity than to pap DNA containing a fully methylated GATC-I site (12). In contrast to the negative role that methylation of GATC-I plays in Pap phase variation, genetic evidence indicates that methylation of GATC-II is required for the transition from the off to the on state (3). It is possible that methylation of GATC-II reduces the affinity of Lrp for pap DNA sites 1 to 3, freeing the papBA promoter to bind RNA polymerase (3).

The pap operon is a member of the methylation-dependent fimbrial operons which are characterized by the following: first, they require Lrp for activation and repression but are not modulated by leucine; second, they contain GATC-box DNA regions which are contained within the Lrp binding sites and are substrates for deoxyadenosine methylase (Dam); and third, they express PapI or coregulatory proteins homologous to PapI. Nonfimbrial operons of the Lrp regulon such as ilvIH do not contain a PapI-like coregulator or GATC-box regulatory sequences, and they do not require Dam for transcription activation (5). To learn more about the roles that Lrp plays in coordinating Pap phase variation, we developed a two-color genetic screen to identify mutations in lrp that blocked Pap phase variation but had little to no effect on ilvIH transcription. Notably, these lrp mutants map within a region, between codons for amino acids 126 and 134 inclusive, which lies outside of the DNA binding and transcription activation regions of Lrp identified by Platko and Calvo (13). Our data suggest that these mutant Lrp’s are defective in a step in the switch from Pap phase off to phase on which has not yet been identified.

MATERIALS AND METHODS

Two-color genetic screen.

The two-color genetic screen was carried out with E. coli DL2561 (ΔphoA ΔlacZ lrp-402) containing a chromosomal copy of ilvIHp-lacZ and plasmid-borne papBAp-phoA (Table 1). This E. coli isolate was constructed as follows. First, the Mu dI1734 lac insertion within ilvIH in E. coli CV975 was transduced into E. coli mPh2phoA by phage P1 transduction (14). Second, a papBAp-phoA operon fusion was constructed in which a promoterless phoA gene was inserted downstream and tandem rrnB1 transcription terminators were inserted upstream of the papBA promoter. The E. coli K-12 phoA gene was amplified by PCR with oligonucleotides 5′-CACTCGTCGACCGGTTTTATTTCAGCCCCAG-3′ and 5′-CTCGGATCCACAGATCTGTACATGGAGAAATAAAGT-3′. The resulting DNA fragment was digested with restriction endonucleases BamHI and SalI and ligated to a plasmid vector pACYC184 derivative lacking most of tet (bp 2544 to 2820) to construct plasmid pDAL446. The 1.76-kb pap regulatory region, which includes the papBA promoter, was amplified by PCR with plasmid pDAL354 (11) and oligonucleotides 5′-GGAAGATCTCAGCGGATCAATTCCCAATTC-3′ and 5′-GGAAGATCTCGCATTTCCTGACCGACTGA-3′. The resulting pap DNA fragment contained tandem rrnB1 transcription terminators, derived from plasmid pRS551 (15), between the vector sequence and the papBA promoter. This pap DNA fragment was digested with BglII and ligated to the BglII-digested plasmid vector pDAL446 to construct the pap-phoA fusion vector pDAL454. Plasmid pDAL454 was transformed into E. coli mPh2phoA containing ilvIH-lacZ to construct the double-fusion isolate DL2561.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain, plasmid, or phage	Relevant characteristics	Source or reference
E. coli strains
XL1-Red	endA1 gyrA96 thr-1 hsdR17 supE44 relA1 lac mutD5 mutS mutT Tn10	Stratagene
DL842	MC4100 lrp-402a	2
DL844	MC4100 lrp-420a	2
DL852	DL842 papBAp-lacZYA	2
mPh2phoA	araD139 ΔlacU169 Δ(brnQ phoB) phoR rpsL phoA	C. Manoil and J. Beckwith
DL1742	MC4100 λ354-1 lysogen Φ(papBA1 ′lacZ)	This study
DL1784	MC4100 Δlrp	17
DL1911	DL1784 λ354-1 lysogen Φ(papBA1 ′lacZ)	3
DL2121	MC4100 λ354-13 lysogen Φ(papBA13 ′lacZ)	11
DL2162	DL2121 lrp-420	11
DL2176	DL844 λ354-13 ΔpapI lysogen, lrp-420	This study
DL2188	MC4100 λ354-13 ΔpapI lysogen	11
DL2560	mPh2phoA ilvIH::Mu dI1734 (pDAL454)	This study
DL2561	DL2560 lrp-402	This study
DL2582	DL2561 (pMBF1)	This study
DL2669	DL2561 (pMBF1lrp501) LrpE133G	This study
DL2693	DL2561 (pMBF1lrp503)b	This study
DL2725	DL852 (pMBF1)b	This study
DL2743	DL852 (pMBF1lrp501)b	This study
DL2746	DL852 (pMBF1lrp502)b	This study
DL2747	DL852 (pMBF1lrp503)	This study
CV975	F−ara thi Δ(lac pro) ilvIH::Mu dI1734	14
DL2760	DL842 ilvIH::Mu dI1734	This study
DL2761	DL2760 (pMBF1)	This study
DL2763	DL2760 (pMBF1lrp501)	This study
DL2766	DL2760 (pMBF1lrp502)	This study
DL2767	DL2760 (pMBF1lrp503)	This study
DL2162	DL2121 lrp-420	11
DL2917	DL2162 (pMBF1)	This study
DL2918	DL2162 (pMBF1lrp501)	This study
DL2921	DL2188 lrp-402	This study
DL2931	DL2921 (pMBF1)	This study
DL2932	DL2921 (pMBF1lrp501)	This study
DL3298	DL1911 (pMBF1)	This study
DL3299	DL1911 (pMBF1lrp501)	This study
DL3255	DL1784 (pMBF1lrp501)	This study
DL3300	DL1911 (pMBF1lrp502)	This study
DL3301	DL2162 (pMBF1lrp502)	This study
DL3302	DL2921 (pMBF1lrp502)	This study
DL3303	DL1742 lrp-402	This study
DL3304	DL3303 (pMBF1)	This study
DL3305	DL3303 (pMBF1lrp501)	This study
DL3306	DL3303 (pMBF1lrp502)	This study
Plasmids
pMBF1	pREG153lrp	2
pACYC184	p15A cam tet replicon	6
pDAL446	pACYC184phoA	This study
pDAL354	pRS550 Φ(papBA ′lacZ)	3
pDAL454	pACYC184 Φ(papBA ′phoA)	This study
Phages
λ354	λRS45-pDAL354 recombinant phage	3
λ354-1	λ354 with pap GATC-I mutated to GCTC	3
λ354-13	λ354 with bp 1161–1166 replaced by GCTAGC	11

^alrp-402 is the new designation for mbf-2 in which lrp contains an mini-Tn10 element inserted 199 bp from the translation start site, and lrp-420 is the new designation for mbf-20 (4). Both mutations block expression of Lrp based on complementation and immunoblot analyses (reference 4 and unpublished data). 

^blrp-501 contains an A-to-G mutation at bp 398 that causes a Glu-to-Gly alteration at codon 133 of Lrp. Similarly, lrp-502 contains an A-to-G mutation at bp 400, yielding Lrp^T134A, and lrp-503 contains a C-to-T mutation at bp 377, yielding Lrp^A126V. Plasmid pMBF1 contains the wild-type lrp gene (4). 

Mutagenesis of lrp.

The low-copy-number plasmid pMBF1 containing lrp (2) was mutagenized by passage through the mutagenic strain XL1-Red (Stratagene). DL2561 was transformed with mutagenized plasmid pMBF1 by electroporation (Electroporator II; Invitrogen). E. coli transformants were screened on M9 minimal medium containing 5-bromo-4-chloro-3-indolyl phosphate (X-Phos) and 6-chloro-3-indolyl-β-d-galactoside (Red-Gal) indicators (Research Organics). Red colonies containing lrp mutations that specifically reduced expression of papBA but not ilvIH were picked for further analysis. Plasmids were isolated by Qiaprep (Qiagen) and transformed back into DL2561. Strains which still maintained the mutant phenotype were chosen for further study. Sequencing of the mutations was performed at the sequencing facility at the University of Utah.

β-Galactosidase and phase variation assays.

β-Galactosidase (1) and phase variation analyses (3) were performed as previously described. Plating of 10⁵ E. coli DL2746 cells (expressing the T134A Lrp mutant [Lrp^T134A]), did not yield any blue (Lac⁺) colonies (see Table 2). Blue DL2746 colonies were obtained by picking blue sectors, present in some colonies, and streaking onto M9 minimal medium until nonsectored blue colonies were obtained. Such colonies were assumed to have arisen from a single phase-on cell (1) and were used to obtain the rate of switching from phase on to phase off (see Table 2). For E. coli containing ilvIH-lacZ and papBA-lacZ operon fusions, the level of transcription was assumed to be equal to the β-galactosidase activity. The standard deviations from the means of β-galactosidase activities were calculated with data from two separate colonies with three replicates each for Tables 2 to 5.

TABLE 2.

Effect of lrp mutations on Pap phase variation and ilvIH transcription

Strain designationa	Pap phase variation			ilvIH transcription
	Pap off-to-on switch rateb	Off-to-on switch rate ratio (mutant to wt)	Pap on-to-off switch rate	Level (Miller units)c	Ratio (mutant to wt)
DL2725 (wt Lrp)	4.83 × 10−4	1.0	3.7 × 10−2
DL852 (Lrp−)	NAd	NA	NA
DL2747 (LrpA126V)	1.86 × 10−5	0.04	3.8 × 10−2
DL2743 (LrpE133G)	2.51 × 10−5	0.05	3.6 × 10−2
DL2746 (LrpT134A)	<1 × 10−5e	<0.02	3.1 × 10−2
DL2761 (wt Lrp)				289 ± 4	1.0
DL2760 (Lrp−)				3 ± 1	0.01
DL2767 (LrpA126V)				72 ± 1	0.25
DL2763 (LrpE133G)				274 ± 3	0.95
DL2766 (LrpT134A)				63 ± 1	0.22

^aThe Lrp allele present in each E. coli isolate is shown in parentheses. wt, wild type. 

^bPap switch rates from phase off to phase on were calculated by plating phase-off bacteria onto M9 minimal medium containing glycerol and the Lac indicator 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal); counting the number of phase-on colonies (M), the total number of colonies (N), and the number of generations of growth (g); and using the formula M/N/g as described previously (1). The on-to-off rates were calculated as described above by plating phase-on bacteria and counting the number that switched to phase off. Both the off-to-on and on-to-off switch rates are expressed as the number of phase switches per cell per generation. 

^cβ-Galactosidase activity levels were used as a measure of ilvIH-lacZ transcript level. Means and standard deviations were calculated with data from two separate colonies with three replicates each. See reference 10 for Miller units. 

^dNA, not applicable since there is no transcription in the absence of Lrp. 

^ePlating of 10⁵ E. coli DL2746 cells did not yield any blue colonies. However, some colonies contained blue sectors, which were used to obtain blue colonies for determination of the phase-on-to-phase-off rate (see Materials and Methods). 

TABLE 5.

Analysis of wild-type and mutant Lrp’s in the pap Lrp binding site 3 mutant background^a

E. coli strainb (pap Lrp site 3)	Lrp phenotype	PapI phenotype	papBA transcript activity (Miller units)	% of papBA transcription
DL2917	wt Lrp	PapI+	5,085 ± 388	100
DL2162	Lrp−	PapI+	6 ± 2	0.1
DL2918	LrpE133G	PapI+	4,800 ± 593	94.4
DL3301	LrpT134A	PapI+	2,819 ± 221	55.4
DL2931	wt Lrp	PapI−	3,705 ± 128	100
DL2921	Lrp−	PapI−	6 ± 1	0.2
DL2932	LrpE133G	PapI−	4,283 ± 298	115.6
DL3302	LrpT134A	PapI−	1,292 ± 95	34.9

^aRefer to the footnotes to Table 4 for explanations of the column headings. 

^bAll E. coli strains contained the 6-bp substitution mutation at Lrp binding site 3 of pap (Table 1). 

Preparation of cell extracts and DNA probes.

Cell extracts were prepared by growing 200-ml bacterial cultures at 37°C to an optical density at 590 nm of 0.5 and centrifuging the cultures at 2,000 × g for 30 min at 4°C. Cells were washed two times in dialysis buffer (40 mM Tris HCl [pH 7.5], 0.1 mM EDTA, 1 mM dithiothreitol, 60 mM KCl) and resuspended in 5 ml of dialysis buffer. Next, 10 μl of 0.5 M EDTA and 5 μl of 50-mg/ml lysozyme were added to the cells, which were held on ice for 30 min prior to sonication in Falcon 2057 tubes in a cup horn sonicator at 4°C. Cell breakage was checked by phase-contrast microscopy. Cell debris was separated by centrifugation at 15,000 × g for 20 min at 4°C. Protein levels in the supernatant solution were quantitated by the Bradford assay (Bio-Rad), and extracts were flash frozen in liquid nitrogen and stored at −70°C.

Nonmethylated DNA probes used for gel retardation analysis were prepared by PCR as follows. A 167-bp pap DNA probe containing Lrp binding sites 1 to 3 spanning bp −118 to +49 was amplified with oligonucleotides 5′-TTTATCTGAGTACCCTCTTG-3′ and 5′-CCCTTCTGTCGGGCCCC-3′. A 166-bp pap DNA probe containing Lrp binding sites 4 to 6 spanning bp −278 to −112 was amplified with oligonucleotides 5′-CTCTATGTTTGCTTTATTTGTTC-3′ and 5′-AGATAAAAACATCATGGCAAA-3′. A 327-bp pap DNA probe containing Lrp binding sites 1 to 6 spanning bp −278 to +49 was amplified with oligonucleotides 5′-CTCTATGTTTGCTTTATTTGTTC-3′ and 5′-CCCTTCTGTCGGGCCCC-3′. The ilvIH probe (bp −311 to +2) containing all six Lrp binding sites was prepared as described previously (8). Oligonucleotides (17 pmol) were radiolabeled with 50 to 100 μCi of [γ-³²P]ATP (6,000 Ci/mmol) and polynucleotide kinase. Prior to the use of oligonucleotides for PCR amplification, buffer was changed to 10 mM Tris-Cl, pH 7.5, with Micro Bio-Spin 6 columns (Bio-Rad). Plasmids pDAL337 and pCV112 were used for amplifications of pap and ilvIH DNA sequences, respectively (8). Following PCR, DNA probes were gel purified on 5% polyacrylamide gels in Tris-borate-EDTA buffer (9). DNA was eluted from excised gel pieces by rotating the gels overnight in 0.6 ml of gel elution buffer (0.5 M ammonium acetate, 1 mM EDTA, 0.1% sodium dodecyl sulfate). DNA was precipitated by addition of 1 ml of 95% ethanol, collected by centrifugation, and washed two times in 70% ethanol.

Preparation of hemimethylated pap site 4 to 6 DNA was as follows. Radiolabeled oligonucleotides 5′-CTCTATGTTTGCTTTATTTGTTC-3′ and 5′-AGATAAAAACATCATGGCAAA-3′ were used to amplify pap sites 4 to 6 by PCR. Following the change of buffer to 10 mM Tris, pH 7.5, some of the pap DNA, labeled at both ends, was methylated in vitro with Dam (New England Biolabs). Methylation was assessed by resistance of the pap GATC-I site to cutting by MboI. Both fully methylated and nonmethylated pap DNAs were denatured by incubation in 0.3 N NaOH and separated on 8% acrylamide (0.24% bisacrylamide) strand-separation gels in Tris-borate-EDTA buffer as described previously (9). Electrophoresis was carried out at 350 V with water cooling until xylene cyanol indicator dye was near the bottom of the 20-cm-long gel. Under these conditions the two single-stranded pap DNAs were separated from each other and migrated more slowly than the double-stranded DNA. Each single-stranded, methylated pap DNA was eluted from the gel as described above and mixed with an equimolar amount of complementary nonmethylated pap DNA strand in annealing buffer (0.1 M NaCl, 10 mM Tris-Cl [pH 7.8], 1 mM EDTA). DNAs were heated to 95°C for 1 min, cooled 1°C per min to 25°C to anneal the strands, and analyzed on 8% acrylamide gels (described above). The hemimethylated state of pap DNAs was confirmed by resistance to digestion with DpnI and MboI, which cut fully methylated and nonmethylated GATC sites, respectively. The position of the methyl group (top versus bottom of a DNA strand) in the hemimethylated site 4 to 6 DNA analyzed in Fig. 2C was not determined.

FIG. 2.

Measurement of the affinities of wild-type and mutant Lrp’s for pap and ilvIH DNAs. The relative affinities of wild-type and mutant Lrp’s for papBA and ilvIH DNAs were analyzed by mobility shift analysis as described in Materials and Methods. Results obtained with wild-type Lrp are shown by circles, those obtained with Lrp^E133G are shown by squares, and those obtained with Lrp^T134A are shown by triangles. Measurements obtained in the absence of PapI are shown as open symbols, whereas those obtained in the presence of PapI (120 nM) are shown as filled symbols. The Lrp levels in extracts were equalized by immunoblot analysis and are shown as relative values. The percentage of DNA bound was calculated based on the level of free DNA remaining at each Lrp level. (A) Nonmethylated pap site 1 to 3 DNA (bp −118 to +49); (B) nonmethylated ilvIH DNA (bp −311 to +2); (C) hemimethylated pap site 4 to 6 DNA (bp −278 to −112). The base pair numbering is relative to the papBAp transcription start site (Fig. 1).

DNA mobility shift assay and immunoblotting.

DNA mobility shift assays were performed with high-ionic-strength gels as described previously (12) except that reaction mixtures (20 μl in dialysis buffer) contained NaCl (100 mM), poly(dI-dC) (2 μg), and acetylated bovine serum albumin (1 μg). Protein-DNA complexes were quantitated on a model GS-250 phosphorimager (Bio-Rad). PapI was purified by thrombin cleavage of a purified glutathione S-transferase–PapI fusion as described previously except that storage buffer was 50 mM Tris-Cl (pH 8.0)–0.15 M NaCl–2.5 mM CaCl₂–0.1% Triton X-100–10 mM dithiothreitol. PapI was used within 1 week (8). The Lrp protein levels in the extracts were quantitated by immunoblotting onto polyvinylidene difluoride membranes. Lrp was detected with rabbit polyclonal anti-Lrp antiserum followed by chemiluminescence detection (Super Signal; Pierce) and phosphorimager analysis (Bio-Rad).

RESULTS

Isolation of pap-specific lrp mutants by a two-color genetic screen.

The goal of this study was to identify the region(s) of Lrp that is specifically required for regulation of Pap phase variation. This goal was carried out by simultaneously detecting transcription from the ilvIH and papBA operons by a two-color genetic screen. Since the ilvIH operon lacks a PapI-like coregulator and is not controlled by the methylation state of its DNA, we reasoned that Lrp mutants showing reduced pap transcription but normal ilvIH transcription might have mutations within the PapI binding domain (8) or another Lrp region required specifically for pap gene regulation.

A two-color genetic screen was set up by transforming E. coli DL2561 (ΔphoA ΔlacZ lrp-402) carrying both papBAp-phoA and ilvIHp-lacZ operon fusions with a mutagenized pool of lrp-bearing plasmids (Materials and Methods). Transformants were plated on medium containing X-Phos (blue) and Red-Gal indicators (Fig. 1A). Colonies expressing both papBA and ilvIH operons were purple due to the blue and red colors of their respective indicators. E. coli containing a mutant lrp resulting in transcription activation of ilvIH but not papBA formed red colonies and was isolated at a frequency of 2 × 10⁻⁴. We also observed blue colonies representing mutations in lrp that inhibit activation of ilvIH but not pap and white colonies which contain an lrp mutation that disrupts activation of both operons. These blue and white mutants were not analyzed further.

Eleven colonies with a stable red color phenotype were further characterized by DNA sequence analysis. The strongest red colony phenotypes were due to changes in the codons for amino acids 126, 133, or 134 in Lrp (Fig. 1B). Quantitative analyses showed that mutant Lrp^A126V reduced the rate of switching of Pap from off to on by 25-fold and that Lrp^T134A reduced this rate by over 50-fold, yet both mutants displayed only a four- to fivefold reduction in ilvIH transcription (Table 2). Mutant Lrp^E133G reduced the rate of switching of Pap from off to on by 20-fold but maintained normal activation of ilvIH, showing that this mutant is specifically defective in pap transcription (Table 2). All three Lrp mutants displayed on-to-off switch frequencies similar to that of wild-type Lrp, indicating that the mechanism by which cells turn off Pap fimbrial expression is distinct from the mechanism by which they turn on fimbrial expression (see Discussion).

Characterization of the Lrp^T134A and Lrp^E133G mutants. (i) DNA binding and PapI response.

Previous work investigating the interactions of Lrp with ilvIH identified DNA binding, leucine response, and transcription activation regions of Lrp (13). To determine if the Lrp mutants identified here were phenotypically defective in any of these functions, we first measured the affinities of the Lrp mutants with both pap and ilvIH DNAs. Lrp levels in the extracts were measured by immunoblotting and equalized (Materials and Methods). DNA mobility shift analyses showed that Lrp^T134A and Lrp^E133G bound to ilvIH DNA with affinities similar to that of wild-type Lrp (Fig. 2B). Lrp appears to bind to the pap GATC-II region containing sites 1 to 3 in phase-off cells (3). Gel shift analysis showed that Lrp^T134A (Fig. 2A) and Lrp^E133G (not shown) bound to pap sites 1 to 3 with affinities similar to that of wild-type Lrp.

Based on genetic analyses, binding of Lrp at pap sites 4 and 5 is essential for transcription (3, 11) and is dependent upon the presence of the coregulator PapI (8). To test the hypothesis that the Lrp mutants are defective in switching from phase off to phase on due to a reduced affinity for pap sites 4 to 6, we first measured the levels of binding of Lrp to sites 4 to 6 in response to increasing amounts of PapI (Fig. 3A). A level of Lrp sufficient for a shift of about 15% of the DNA probe was used in this experiment. Addition of PapI enhanced the binding of both Lrp^E133G (Fig. 3A, lanes 4 to 10; quantitated in Fig. 3B) and wild-type Lrp (Fig. 3B) to pap DNA as evidenced by an increase in Lrp-pap DNA complexes. The effect of PapI on Lrp binding was specific, since Lrp-pap DNA complexes formed in the presence of PapI were competable by an excess of unlabeled pap site 4 to 6 DNA (Fig. 3A, lane 3). In addition, a shift in migration of the pap DNA probe was not observed with a cell extract from E. coli that does not express Lrp. These results indicate that Lrp was in the complex and that PapI does not bind specifically to pap DNA under these conditions (Fig. 3A, lane 2). An additional minor pap DNA complex which was due to specific binding of an unidentified protein in the extracts was also observed under these conditions (Fig. 3A).

FIG. 3.

PapI increases the affinities of wild-type and E133G mutant Lrp’s for pap DNA sites 4 to 6. (A) Mobility shift analysis. Mobility shifts were carried out with Lrp^E133G and ³²P-labeled pap site 4 to 6 DNA (40,000 cpm, 0.5 nM) as described in Materials and Methods. An Lrp^E133G level sufficient for a shift of about 15% of pap DNA probe was added to each sample (3 μg of DL3255 cell extract) except that the sample analyzed in lane 2 contained 3 μg of DL1784 (Lrp⁻) extract instead of Lrp^E133G extract. Various amounts of PapI were added to each sample as indicated below in a total volume of 20 μl. The positions of unbound and bound pap DNAs are shown at the left. Lane 1, 120 nM PapI; lane 2, 120 nM PapI (Lrp⁻ extract); lane 3, 120 nM PapI plus 125 nM unlabeled pap sites 4 to 6 (250-fold molar excess over the molar concentrations of labeled pap DNA sites); lane 4, no PapI addition; lane 5, 1.2 nM PapI; lane 6, 2.4 nM PapI; lane 7, 6 nM PapI; lane 8, 12 nM PapI; lane 9, 60 nM PapI; lane 10, 120 nM PapI. (B) Summary of mobility shift data. Data obtained with wild-type Lrp (circles) and Lrp^E133G (squares) are shown. Results are expressed as fractions of free pap DNA probe remaining compared with the level observed in the absence of PapI (Fig. 3A, lane 4).

Previous results have suggested that the switch from phase off to phase on occurs following DNA replication, when the GATC-I site is hemimethylated (3). It is therefore possible that the reduced abilities of the Lrp mutants to switch from phase off to phase on might be caused by a decrease in affinity for hemimethylated DNA. Alternatively, these Lrp mutants might be nonresponsive to PapI, which increases the affinity of Lrp for pap sites 4 to 6 by about fourfold (11) and is required for transition to the phase-on transcription state (12). These possibilities were tested by measuring the affinities of mutant and wild-type Lrps for both nonmethylated and hemimethylated pap sites 4 to 6 in the presence and absence of PapI. Our results showed that Lrp^E133G and wild-type Lrp bound to hemimethylated sites 4 to 6 with similar affinities (Fig. 2C). Addition of PapI resulted in approximately threefold increases in the affinities of both the mutant and wild-type Lrp’s for hemimethylated pap DNA (Fig. 2C). PapI also induced threefold increases in the affinities of Lrp^E133G and Lrp^T134A for nonmethylated pap sites 4 to 6, making their affinities similar to that of wild-type Lrp (data not shown).

Previous work indicated that although the affinity of wild-type Lrp for pap sites 4 to 6 was increased fourfold by addition of PapI in vitro, the affinity of wild-type Lrp for pap sites 1 to 3 was somewhat decreased (less than twofold) (11). These data raise the possibility that the mutant Lrp’s aberrantly respond to PapI by binding with higher affinities to pap sites 1 to 3, thus inhibiting the transition to the phase-on state. As shown in Fig. 2A, addition of PapI did not significantly affect the binding of wild-type Lrp to pap sites 1 to 3, in contrast to PapI’s stimulatory effect on binding of wild-type Lrp to pap sites 4 to 6 (Fig. 2C). Lrp^T134A behaved similarly to wild-type Lrp since its affinity for pap sites 1 to 3 was not significantly altered by addition of PapI (Fig. 2A). Together, these results indicate that the amino acid substitutions in Lrp^E133G or Lrp^T134A do not significantly affect the mutants’ affinities for pap sites 1 to 3 or sites 4 to 6. Moreover, the mutant Lrp’s retained the ability to respond to PapI, as was evidenced by an increase in affinity for pap sites 4 to 6 similar to the affinity of wild-type Lrp for these pap sites in the presence of PapI.

(ii) Leucine response.

Certain operons within the Lrp regulon are affected by leucine addition and are considered leucine responsive. The transcriptional activities of operons, such as ilvIH, which are positively regulated by Lrp are decreased by leucine. The addition of leucine reduces transcriptional activation of ilvIH 10-fold in E. coli containing wild-type Lrp and 16-fold in E. coli containing Lrp^E133G (Table 3). These data show that the leucine response region of Lrp^E133G remains fully functional.

TABLE 3.

Lrp^E133G is leucine responsive

Strain designa- tion	Relevant E. coli strain background	Leucine additiona	ilvIH transcrip- tional activity (Miller units)b	Leucine response (+ leucine/ − leucine ratio)
DL2760	Lrp−	−	6.4 ± 1	NAc
DL2761	Wild-type Lrp	−	273 ± 5	NA
DL2761	Wild-type Lrp	+	27.2 ± 1	0.1
DL2763	LrpE133G	−	232 ± 3	NA
DL2763	LrpE133G	+	13.5 ± 1	0.06

^aLeucine responsiveness was determined by measuring ilvIH-lacZ transcription in the presence of 100 μg of leucine per ml. 

^bMeans and standard deviations were calculated with data from two separate colonies with three replicates each. 

^cNA, not applicable. 

Analysis of the effects of Lrp^T134A and Lrp^E133G on pap transcription in vivo.

The in vitro analyses shown above indicated that the Lrp^T134A and Lrp^E133G mutants retained intact DNA binding and PapI responsiveness. In addition, Lrp^E133G was still responsive to leucine (Table 3). Because these in vitro assays did not provide a clue as to the possible defect(s) of the Lrp mutants, we explored Lrp function in vivo. Previously we showed that an A-to-C transversion mutation at GATC-I (GCTC-I) resulted in a phase-on-locked phenotype (3). Although this mutation does not disrupt binding of Lrp to pap sites 4 and 5, the ability of Dam to methylate the GATC-I site is precluded. Methylation of this site serves to keep cells in the off transcription state by blocking binding of Lrp in the presence of PapI (12). Further analysis of the GCTC-I mutant showed that in the absence of papI, cells remained locked in phase on with only about a twofold reduction in pap transcription, which was dependent on the presence of Lrp (Table 4). Thus, pap transcription from the GCTC-I mutant occurs in the absence of PapI.

TABLE 4.

Analysis of wild-type and mutant Lrp’s in the pap GCTC-I mutant background

E. coli straina	Lrp phenotypeb	PapI phenotypec	papBA transcript activity (Miller units)d	% of papBA transcriptione
DL3298	wt Lrp	PapI+	2,280 ± 126	100
DL1911	Lrp−	PapI+	3 ± 0	0.1
DL3299	LrpE133G	PapI+	1,584 ± 74	69.5
DL3300	LrpT134A	PapI+	37 ± 1	1.6
DL3304	wt Lrp	PapI−	1,152 ± 91	100
DL3303	Lrp−	PapI−	1 ± 0	0.1
DL3305	LrpE133G	PapI−	1,023 ± 29	88.8
DL3306	LrpT134A	PapI−	39 ± 2	3.4

^aAll strains contained a chromosomal papBAp-lacZ operon fusion in which the pap GATC-I site was mutated to a GCTC sequence (GCTC-I) (3). 

^bMutant and wild-type (wt) Lrp’s were present in trans on low-copy-number plasmid pMBF1 (2). 

^cPapI⁻ cells contained a deletion of the entire papI gene (11). 

^dThe β-galactosidase assay was run in triplicate for each of two separate colonies for every isolate, and the standard deviations from the means are shown. See reference 10 for Miller units. 

^ePercentages of papBA transcription relative to that of the PapI⁺ (top four results) or PapI⁻ (bottom four results) wild-type Lrp control are shown. 

Although wild-type Lrp activated pap transcription in the GCTC-I mutant to high levels in the presence or absence of PapI, the Lrp^T134A mutant was unable to do so. Using Lrp^T134A, we observed a 60-fold decrease in pap transcription compared with that in wild-type Lrp in PapI⁺ cells. A similar pap transcription level was observed in the absence of PapI (Table 4). Thus, Lrp^T134A is unable to activate pap transcription under conditions in which the PapI-dependent step(s) has been bypassed. These results indicate that Lrp^T134A is defective in a function that does not involve interaction with PapI and are consistent with the in vitro data described above indicating that PapI increases the affinity of Lrp^T134A for pap sites 4 to 6 in a manner similar to that of wild-type Lrp (Fig. 2).

Analysis of the effects of Lrp^E133G in the GCTC-I background showed that pap transcription was reduced less than twofold compared to that of wild-type Lrp, in the presence or absence of PapI (Table 4). Thus, under these conditions Lrp^E133G activates pap transcription to a level near that of wild-type Lrp.

Because the pap transcription phenotype of Lrp^T134A did not depend upon the presence of PapI in the GCTC-I background, we next explored the possibility that the step in Pap phase variation blocked in this mutant involves binding of Lrp to sites 1 to 3, which occurs in the absence of PapI (12). A substitution mutation within Lrp binding site 3 of pap DNA disrupts the cooperative binding of Lrp to sites 1 to 3 and results in a PapI-independent, phase-on-locked phenotype similar to that observed for the GCTC-I mutant analyzed above (11). As shown in Table 5, introduction of wild-type Lrp into the pap-13 mutant background (6 bp substitution in pap Lrp binding site 3) resulted in high level of papBA gene expression in the presence or absence of PapI that was almost totally dependent on Lrp. Similar high levels of gene expression were observed for Lrp^E133G in the presence and absence of PapI. Moreover, papBA expression in the pap Lrp binding site 3 mutant was reduced only 2-fold with Lrp^T134A (Table 5, papI⁺ background), in contrast to the 60-fold reduction in transcription observed with the GCTC-I mutant (Table 4). In the absence of papI, papBA transcription was reduced an additional twofold, consistent with the in vitro DNA binding results presented above indicating that Lrp^T134A is still PapI responsive (Table 4). These results indicate that Lrp^T134A and Lrp^E133G retain their transcription activation functions under conditions in which binding of Lrp to site 3 is inhibited.

DISCUSSION

The goal of this study was to identify a region(s) of the global regulator Lrp that is specifically required for Pap phase variation to help define the mechanisms by which Lrp responds to PapI and the pap DNA methylation state. We identified lrp mutations that differentially inhibited papBA transcription and ilvIH transcription. Our data show that mutations at codons for amino acids 126, 133, or 134 of Lrp greatly reduced the rate of switching of pap from off to on (20- to more than 50-fold) but had much smaller effects on activation of ilvIH (less than 4-fold [Table 2]). These Lrp mutations overlap a region of Lrp that is associated with leucine response and is adjacent to a transcription activation region (Fig. 1B) (13). In vivo analyses showed that Lrp^E133G remained responsive to leucine (Table 3). Moreover, Lrp^T134A and Lrp^E133G activated pap transcription to 55 and 94% of wild-type levels, respectively, in the pap Lrp binding site 3 mutant background (Table 5). Thus, these Lrp mutants retained the ability to activate pap transcription.

Our data indicate that the E133G and T134A Lrp mutants were not altered in their affinities for either ilvIH or pap DNA sequences, consistent with previous data showing that the DNA binding domain of Lrp spans amino acids 13 through 70, well beyond the area of the mutations identified here (13). Moreover, Lrp^E133G and Lrp^T134A maintained responsiveness to PapI, as evidenced by an increase in their affinities for both nonmethylated and hemimethylated pap DNA sites 4 to 6 (Fig. 2C and 3). These results were puzzling, since we expected to find alterations in either responses to PapI or the ability to bind hemimethylated DNA, both hallmarks of Pap phase variation that are not involved in ilvIH gene regulation (3–5, 11, 16).

Further in vivo analysis of the Lrp mutants was carried out with two types of phase-on-locked pap mutants which are both PapI independent. Using the GCTC-I background and Lrp^T134A, we found that pap transcription was reduced over 60-fold compared to that of wild-type Lrp (Table 4). These results are consistent with the >50-fold decrease in the rate of switching of pap from phase off to phase on measured with this Lrp mutant (Table 2). The Lrp^T134A mutant was unable to activate pap transcription under conditions in which wild-type Lrp activated transcription in the absence of PapI. These results indicate that Lrp^T134A is defective in a step in Pap phase variation that does not involve PapI. Lrp^E133G, in contrast, showed less than a 2-fold decrease in pap transcription in the GCTC-I background even though it displayed a 20-fold decrease in the rate of switching of pap from phase off to phase on compared with that of wild-type pap (Tables 2 and 4). The reason for this difference in the abilities of the Lrp mutants to activate pap transcription in the GCTC-I background is not clear, but it correlates with the severity of the Pap phase variation phenotype (Table 2).

In contrast to results obtained with the GCTC-1 background, Lrp^T134A activated pap transcription to near wild-type levels in the pap Lrp binding site 3 mutant background (Table 5). These results indicate that the Pap phase variation step blocked in the Lrp^T134A mutant is prior to activation of pap transcription. Possible locations for a block are where Lrp dissociates from pap DNA sites 1 to 3 or binds to sites 4 to 6. DNA mobility shift analyses indicated that the affinity of Lrp^T134A for pap DNA sites 1 to 3 and 4 to 6 appeared normal in vitro (Fig. 2A and data not shown). It is therefore likely that there is an additional factor(s) required for the Pap switch from phase off to phase on in vivo that we have not yet identified. Such a factor(s) might be identified by isolating second-site mutations that suppress the Lrp^T134A pap transcription phenotype.

Why is Lrp^T134A unable to activate pap transcription in the pap GCTC-I background but able to activate transcription of the pap Lrp binding site 3 mutant (Tables 4 and 5)? Previous results showed that the affinity of wild-type Lrp for pap-13 DNA (the pap site 3 DNA) decreased threefold for sites 2 and 3 but increased for sites 5, 6, and 1 in the absence of PapI (11). Thus, the phase on state of the pap site 3 mutant may differ from that of wild-type pap in that Lrp may be bound to pap sites 5, 6, and 1 in the former and sites 4 and 5 in the latter, based on in vitro analyses (14). In contrast, Lrp bound to the GCTC-I mutant DNA with an affinity similar to that of wild-type pap (5). Unlike the pap site 3 mutant, the pap GCTC-I mutant is locked in phase on because it cannot be methylated at GATC-I. Presumably, this methylation is required to reduce the affinity of Lrp for the GATC-I region and to maintain cells in the off state until sufficient PapI is present to facilitate binding of Lrp to pap sites 4 and 5. The processes involved in the transition from phase off to phase on with the GCTC-I mutant may thus more closely parallel those of wild-type pap than those of the site 3 mutant, even though both pap mutants are locked in phase on and PapI independent. This hypothesis is supported by additional data indicating that at 23°C the pap site 3 mutant remains locked in phase on but that transcription from the GCTC-I mutant, like that of wild-type pap, is shut off by the normal thermoregulatory mechanism (unpublished data). It is therefore possible that a critical step(s) in pap gene regulation is bypassed in the pap site 3 mutant, enabling Lrp^T134A to activate transcription under these conditions.

Previous results showed that the rate of switching of pap from phase off to phase on for cells grown in glucose is 35-fold lower than that for cells grown in glycerol, yet the on-to-off switch rates for cells grown on either one of the two carbon sources are similar (1). It is likely that the decrease in the rates of switching by glucose from off to on is the result of lowered PapI levels, which are under cAMP-catabolite activator protein control. These results suggest that the on-to-off switch is independent of PapI levels and occurs by a pathway distinct from the off-to-on switch. The data presented here support this hypothesis, since although the three Lrp mutations analyzed greatly reduced the rate of switching from off to on, they had little or no effect on the rate of switching from on to off, which occurs at a 100-fold higher rate (Table 2). Further work is needed to understand the mechanism by which Lrp dissociates from pap DNA sites 4 to 6 and how the transition to the off transcription phase occurs.

The two-color genetic screen described here is an extension of previous work on CRP by Zhou et al. (19). In their study, crp mutants specifically defective in activation of lac transcription but retaining DNA binding activity were isolated by simultaneously screening ribose (Rbs) and lactose (Lac) fermentation on tetrazolium indicator media. For that screen a lacUV5-O^CRP promoter in which CRP represses transcription was used. Thus, CRP activation mutants gave a Rbs⁻ Lac⁻ phenotype since they failed to activate rbs transcription and bind to the lacUV5-O^CRP promoter to repress lac transcription. The advantage of the approach used in the present study is that since each reporter gene yields a different colony color, mutations that affect only one operon can easily be detected. This method should be widely applicable for the identification of specific functional regions of global regulators.