Interaction at a distance between multiple operators controls the

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 267, No. 9, Issue of March 25, pp. 6114-6121 1992

Printed in ir.S.A.

Interaction at a Distance between Multiple Operators Controls the Adjacent, Divergently Transcribed gZpTQ-gZpACB Operons of Escherichia coli K- 12”

(Received for publication, May 2, 1991)

Timothy J. LarsonS, John S. Cantwell, and Ali T. van Loo-Bhattacharya From the Department of Biochemistry and Nutrition, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061 -0308

The glp regulon of Escherichia coli encodes the proteins required for utilization of en-glycerol 3-phosphate and its precursors. Transcription of the divergently transcribed glpTQ and glpACB operons is ini- tiated at sites separated by 132 base pairs (bp) of DNA. These operons are controlled negatively by glp repressor and positively by the CAMP-CAMP receptor protein (CRP) complex. The locations of the binding sites for the glp repressor and for CAMP. CRP in the control regions of these operons were determined by DNase I footprinting. Binding of the glp repressor protected the region -32 to -51 (OT) in the glpTQ promoter, which was also the binding site for cAMP.CRP. Four repressor binding sites (-41 to -60 (OA1), -9 to -28 (0.42)s +12 to -8 (OA~) , and +52 to +33 (OA4)) and two CAMP-CRP binding sites (+11 to -11 and -30 to -51) were found in the glpACB promoter region. Compari- son of the sequences of the repressor binding sites found in the glpTQ-glpACB control region with those operators previously described in the glpD operon al- lowed formulation of a consensus operator sequence which was the palindrome 5’-WATGTTCGWTAWC- GAACATW-3‘ (W is A or T). The role of each operator was assessed by measuring repression in constructs where individual operators were altered by site-directed mutagenesis. Alteration of OT did not signifi- cantly decrease repression of either operon. Each of the glpACB operators contributed to repression of both operons. These results suggest involvement of glpACB operator(s) in control of glpTQ expression perhaps via formation of a repression loop. Evidence supporting this hypothesis was obtained by measuring the degree of repression of the glpTQ promoter in constructs containing 6- or 10-bp insertions between the glpTQ and glpACB operators. A 6-bp insertion located within 0.42 or between OT and OA1 eliminated repression of the glpTQ promoter, whereas significant repression was maintained in the case of a 10-bp insertion within 0 ~ 2 .

The genes of the glp regulon of Escherichia coli encode the proteins needed for the utilization of glycerol, sn-glycerol 3-

* This work was supported by National Science Foundation Grant DMB-8696072. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$To whom correspondence and reprint requests should be ad- dressed. Tel.: 703-231-7060.

phosphate (glycerol-P),’ and glycerophosphodiesters (Fig. 1; Lin, 1976, 1987). These genes are arranged in five different operons positioned at three different locations on the chro- mosome of E. coli. The glpTQ operon encodes glycerol-P permease (Eiglmeier et al., 1987) and glycerophosphodiester- ase (Tommassen et al., 1991). This operon is located near minute 49 on the linkage map and is transcribed in the counterclockwise direction. The glpACB operon encodes the subunits of the anaerobic glycerol-P dehydrogenase (Cole et al., 1988). This operon is directly adjacent to and is transcribed divergently from the glpTQ operon. The glpFK operon encodes glycerol diffusion facilitator and glycerol kinase (Lup- ski et al., 1990; Sweet et al., 1990). The operon is located near minute 88 and is transcribed in the counterclockwise direction. The glpD gene encodes aerobic glycerol-P dehydrogenase (Austin and Larson, 1991). This gene is located near minute 75 and is transcribed in the clockwise direction, divergently from glpEGR. The glpR gene encodes the glp repressor, and glpE andglpC encode protein of unknown function (Schweizer et al., 1986; Schweizer and Larson, 1987).

The glp operons are controlled transcriptionally in response to three types of environmental stimuli (Lin, 1987). Specific negative regulation is mediated by the glp repressor. Its affinity for operator sites on DNA is decreased in the presence of glycerol-P, the inducer for the regulon. Results of studies employing a strain harboring a thermolabile repressor indicated that the order of sensitivity of the glp operons to control by the glp repressor is glpD > glpT > glpK (Freedberg and Lin, 1973). The glp operons are also subject to catabolite repression during growth in the presence of glucose, presum- ably because transcription is positively influenced by the CAMP-CAMP receptor protein (CRP) complex. The order of sensitivity of the operons to catabolite repression is opposite that observed for control by glp repressor (Lin, 1987).

Expression of the operons encoding the glycerol-P dehydrogenases is sensitive to the repiratory state of the cell. Anaer- obic glycerol-P dehydrogenase activity is maximal during anaerobic growth with fumarate present as electron acceptor. Under these conditions, the FNR protein has a positive influence on the transcription of genes involved in anaerobic respiration, including the glpACB operon (Kuritzkes et al., 1984; Iuchi et al., 1990). Aerobic glycerol-P dehydrogenase activity is maximal under well oxygenated growth conditions. Anaerobiosis results in repression of the glpD operon by the arcA/arcB-encoded two-component regulatory system (Iuchi et al., 1990).

The abbreviations used are: glycerol-P, sn-glycerol 3-phosphate; kb, kilobase pair(s); bp, base pair(s); CRP, cyclic AMP receptor protein; GlpR, glp repressor protein; FNR, anaerobic activator protein.

6114

Control ofglpTQ-glpACB Gene Expression in E. coli 6115

FIG. 1. Metabolism of glycerol-P and its precursors in E. coli. The genetic symbols are italicized. Inhibition of glycerol kinase by fructose-1,6-bis- phosphate (FDP) and by enzyme IIIG" of the phosphoenolpyruvate phospho- transferase system is indicated by the dashed arrows. Abbreviations: G3P, glycerol-P; G3POR, glycerophosphodiesters; GAP, glyceraldehyde 3-phosphate; DHAP, dihydroxyacetone phosphate.

Glycerol

G3P

G3POR

OUTER MEMBRANE -\

CYTOPLASMIC MEMBRANE

Enzyme 1 1 1 e Glycerol -

Glycerol ATP

fg lp.4 CB)

P h o s p h o - diesterose

G 3 P s y n t h a s e

Phospholipid (glpQ) biosynfhesis

G3POR CYTOPLASM

PERIPLASM /

The glp repressor has been overproduced and purified to near homogeneity in an active form which binds both inducer and operator DNA (Larson et al., 1987). The repressor is tetrameric under native conditions and contains four identical 30-kDa subunits. The repressor binding sites in the control region for the glpD operon have been identified by DNase I footprinting (Ye and Larson, 1988). The repressor binds to tandemly repeated operators of hyphenated dyad symmetry located just downstream from the start site of transcription. The present work was undertaken in order to elucidate molecular details of repressor-mediated regulation of the divergently transcribed glpTQ-glpACB operons.

EXPERIMENTAL PROCEDURES

Materiak-Restriction endonucleases and DNA ligase were obtained from Boehringer Mannheim Gmbh or New England Biolabs. Sequenase, Klenow fragment of DNA polymerase I, and the DNA sequencing kit were obtained from US. Biochemicals. Du Pont-New England Nuclear supplied [cx-~'S]~ATP and [cY-~'S]~GTP. Cyclic AMP, o-nitrophenyl j3-D-galactopyranoside, isopropylthio-P-D-galactopyranoside, p-nitrophenyl phosphate, 5-bromo-4-chloro-3-indolyl- (3-D-galactopyranoside (X-Gal), 5-bromo-4-chloro-3-indolyl phosphate (X-P), and sn-glycerol3-phosphate were from Sigma. Casamino acids and maltose were from Difco Laboratories. Oligonucleotides were synthesized on an Applied Biosystems Model 381A synthesizer and purified by OPC (Applied Biosystems) as recommended by the manufacturer. A phosphorylated EglII linker [d(pCAGATCTG)] was obtained from New England Biolabs. The glp repressor was purified to homogeneity as described previously (Larson et al., 1987). Cyclic AMP receptor protein (CRP) was purified to near homogeneity from strain N4830 (pCRP-1) (Harman et al., 1986). The conditions for induction of CRP synthesis are those described earlier for overpro- duction of the glp repressor (Larson et al., 1987). After induction at 42 "C for 5 h, a crude extract was prepared, and polyethyleneimine and ammonium sulfate fractionation were performed as described by Blazy and Ullmann (1986), followed by phosphocellulose chromatography (Larson et al., 1987). Fractions from the phosphocellulose column which contained the bulk of CRP activity were concentrated by ammonium sulfate precipitation. Chromatography on a Sephadex G-200 column (1.5 X 50 cm) developed with 0.05 M potassium phosphate (pH 7.5), 0.2 M KCI, 1 mM EDTA, 0.5 mM dithiothreitol, and 10% glycerol resulted in a preparation that was nearly homogeneous as assessed by sodium dodecyl sulfate-polyacrylamide gel electropho- resis. The specific activity of the final product was 20 pmol of CAMP bound per mg of protein. Protein concentrations were estimated using the method of Bradford (1976) with bovine serum albumin as stand- ard.

Bacterial Strains and Growth Media-LB medium (Miller, 1972)

MEDIUM

with 0.2% glucose and 100 pg/ml ampicillin was used for growth of bacteria prior to isolation of plasmid DNA. Strain JM109 [recAl endA1 gyrA96 thi hsdRl7 supE44 reLAl Aflac-proAE) IF' traD36 proAE' lacp lacZAMl5)l ( Yanisch-Perron et al., 1985) was used as host during construction of recombinant plasmids. Selection of trans- formants was on plates of LB medium supplemented with 100 pg/ml ampicillin, 40 pg/ml X-Gal, and 0.5 mM isopropyl-1-thio-P-D-galactopyranoside. Strain DH5aF' [F' ~8OdlacZAM25 recA1 e n d l gyrA96 thi-2 hsdRI7 supE44 reLA1 A(lacZ-argF)CJl69] (Liss, 1987) was used for propagation of recombinant M13 phage. Strain N4830 (F- his strR ilvA relA Ag(XcZ857 AEAM AHI)) (Adhya and Gottes- man, 1982) harboring pCRP-1 (Harman et al., 1986) was used as the source of CRP. Strain TL73 (MC4100 glpR2 recAl (Larson et al., 1982)) harboring either pSH58 (gZpR*) or pACYC184 (the vector for cloning of glpR+ in pSH58) was used to asses the affect of the glp repressor on various glpT-phoA or glpA-lac2 transcriptional fusions. Cells growing logarithmically on AB minimal medium (Clark and Maalse, 1967) supplemented with 0.2% maltose, 0.1% casamino acids, 30 pg/ml chloramphenicol, and 100 pg/ml ampicillin were used for these experiments. For aerobic growth, 5-ml cultures were incubated in tubes (18 X 150 mm) on a roller drum. For anaerobic growth, cells were grown in sealed tubes filled to the top starting with a 1 to 2% inoculum of a culture growing aerobically.

Isolation of DNA-For quick screening, plasmid DNA was isolated from 1.5-ml cultures using a rapid technique (Rodriguez and Tait, 1983). DNA was also isolated from 100-ml cultures using a protocol outlined by Promega. This DNA was used for sequencing reactions and for preparation of radiolabeled DNA. M13 single-stranded DNA was prepared by a procedure described in the Sequenase manual (US . Biochemicals).

Sequencing of DNA-Double-stranded plasmid or single-stranded M13 DNA was used as template for sequencing reactions carried out by the dideoxy chain termination technique (Sanger et al., 1977) using the Sequenase kit (US. Biochemicals).

Construction of Recombinant Plasmids-The 725-base pair SstII- HpaI fragment of pSH69 (Ehrmann et al., 1987) containing the control region for the glpACB-glpTQ operons was ligated into the vector pBluescript KS(+) (Stratagene) which had been cleaved with SstII and EcoRV. The resulting plasmid (pHH11) was used as the source of a 268-base pair HaeIII-NruI fragment containing the control region. This fragment was inserted into the SmaI site of pBluescript KS(+), and the resulting plasmid was named pATHN. Nucleotide sequencing was used to determine the orientation of the insert. TheglpACE operon was transcribed toward the adjacent EcoRI site of the vector, and the glpTQ operon was transcribed toward the EamHI site of the vector in pATHN. The 268-base pair HaeIII-NruI fragment of pHHl1 also was cloned in both orientations into the SmaI site of M13mp19 (Yanisch-Perron et al., 1985). Sequencing reactions prepared on single-stranded DNA was used for verification of the sequence of the control region and were co-electrophoresed with DNase I footprinting reactions to serve as size markers.

6116 Control of glpTQ-glpACB Gene Expression in E. coli A plasmid (pSH58) carrying the wild-type glpR gene was con-

structed by cloning of a 2.7-kb BglII fragment from pSH21 (Schweizer et al., 1985) into the BarnHI site of pACYC184 (Chang and Cohen, 1978).

DNase I Footprinting-DNase I footprinting was performed using the conditions previously described (Ye and Larson, 1988), with details provided in the legends to the figures. Two sets of experiments were done, in which the DNA fragment was radiolabeled at either end. End-labeled DNA was prepared by digestion of pATHN with either BarnHI or EcoRI followed by incubation with Sequenase in the presence of ( C Y - ~ ~ S I ~ A T P and [ ( U - ~ ~ S I ~ G T P (for BamHI-cleaved DNA) or [ C Y - ~ ~ S J ~ A T P (for EcoRI-cleaved DNA). A second digestion with the other enzyme followed by purification on an agarose gel resulted in 288-base pair fragments where the label was incorporated 11 and 12 bases away from the NruI site at the beginning of glpA, or, in the other case, 5 and 6 base pairs away from the Hue111 site near the beginning of glpT.

Oligonucleotide-directed Mutagenesis-The technique of Kunkel et al. (1987) was used for making specific alterations in the nucleotide sequence of the control region of the glpTQ-glpACB operons. Single- stranded M13 templates containing uracil were generated by three cycles of propagation in strain RZ1032 (dutl ungl). The mutagenic oligonucleotides used are described in Figs. 2 and 3.

Construction of Insertion and Operator Mutations-Insertion mutations in the glpTQ-glpACB control region of pATHN were con- structed as shown in Fig. 2. A 2-bp insertion at the unique BstBI site was generated by sequential treatment of pATHN with BstBI, Se- quenase in the presence of dNTPs, and T4 DNA ligase. The resulting plasmid (pATHN + 2) contained a unique NruI site in place of the original BstBI site. A 10-bp insertion at this position was generated by sequential treatment of pATHN + 2 with NruI, 8-bp BglII linkers, and T4 DNA ligase, BglII, and T4 DNA ligase. The resulting plasmid (pATHN + 10) contained a unique BglII site. A 4-bp deletion at the BglII site of pATHN + 10 was created using oligonucleotide-directed mutagenesis. The HindIII-BarnHI fragment of pATHN + 10 was cloned into M13mp19 for this purpose. The desired clone contained a unique SphI site and a 6-bp insertion relative to the wild type. The HindIII-BarnHI fragment was recloned to pBluescript (pATHN + 6).

Mutations altering each of the operators were generated using oligonucleotide-directed mutagenesis (Fig. 3). Where convenient, operators were altered with simultaneous creation of restriction sites to facilitate identification of the desired clones. A construct containing a 6-bp insertion between OT and 0 ~ 1 was generated using the same technique (AT + 6; Fig. 3). M13mp19 containing the HindIII-BarnHI fragment of pATHN served as the uracil-containing template in the mutagenesis reactions. The nucleotide sequence of the entire HindIII-

Wild Type -01

slph . . . GCATATTCGCTCATMTTCGAGTGAAACGTGATTTC . . . CGTATUGCGAGTATTAAGCTCACmCCACTAlVLG

ESwI 1 . . . GCATATTCGCTCATAATT CGMGTGIUACGTGATPTC . . . SlnT

CGTATAAGCGAGTATTAAGC TTTCACTTTGCACTAAAG

1. DNA polymerase

2. DNA ligase + dNTP's I

+ 2 nLph . . . GCATATTCGCTCATAATTCQCGAAAGTGAAACGTGATTTC . . .

CGTATMGCGAGTATT~GCOCTTTCACTTTGCACTAAAG

1. -1 2. W I I linker

+ DNA ligase d(CAGATCTG) I

W I I +10 nLph . . . G C A T A T T C G C T C A T A A T T C ~ A Q A ~ Q C G A A A G T G T T C . . . d.@

CGTATAAGCGAGTATTAAGCQTCTAQACQCTITCACTTTGCAC~AAAG 3'-GCGAGTATTI\P.GCGT----ACG~CA~GC-S Oliqo:

1. clone upper strand in U13 2. Directed mutagenesis using

above oligo (30-mer) I .I

g&A . . . GCATATTCGCTCATAATTCQCATQCGUAGTGAAACGTGATTTC . . . % k Z CGTATAAGCGAGTATTAAGCQT~CG~TCACTTTGCACT~G

FIG. 2. Construction of insertion mutations in the glpACB- glpTQ control region. The region surrounding the unique BstBI site of pATHN is shown.

BarnHI fragment was determined in each case to verify the presence of the desired changes.

To assess the effects of the above mutations on expression of the glpTQ and glpACB operons, the HindIII-BarnHI fragments from the wild type and 10 mutant derivatives were cloned into the promoter- probe vector pCB267 (Schneider and Beck, 1987). This vector contains promoterless, divergently arranged phoA and lac2 genes with an intervening multiple cloning site. The resulting plasmids were named pATCB, ~ATCBOTM, PATCBOA~M, ~ATCBOA~M, pATCBOA3M, ~ A T C B O A ~ M and ~ A T C B O A ~ P , pATCB + 6, pATCBOA2 + 2, pATCBOA2 + 6, and p A T c B 0 ~ 2 + 10. In each case, @-galactosidase expression was controlled by a glpA-lacZ transcriptional fusion, and alkaline phosphatase expression was controlled by a glpT-phoA transcriptional fusion. The structure of each plasmid was verified by restriction analysis, and the presence of the desired insert in each was confirmed by nucleotide sequence analysis of the entire HindIII-BarnHI insert using a primer complementary to the 5' end of the lac2 gene.

Enzyme Assays-@-Galactosidase activity was determined on cells permeabilized using chloroform and sodium dodecyl sulfate as described by Miller (1972). The absorbance at 420 nm was determined after removal of cells by centrifugation. Specific activities are expressed in Miller units and were determined by duplicate or triplicate assays on each batch of cells.

Alkaline phosphatase activity was determined using a modification of the method described by Schneider and Beck (1987). Cells were washed and resuspended in 1 M Tris-HC1 (pH 8) and then permeabilized using chloroform. Duplicate assays were then carried out at 25 "C in 1 ml of 1 M Tris-HC1, pH 8, containing 1 mM p-nitrophenyl phosphate. The change in absorbance at 410 nm was monitored continuously; values reported are those obtained where activity was linear with respect to time and amount of cells added. Specific activities are expressed as described by Schneider and Beck (1987).

RESULTS

Identification of the Binding Sites for the glp Repressor in theglpTQ-glpACB Control Region-The control region for the divergently transcribed glpTQ-glpACB operons has been lo- calized to a 272-bp region separating the two initiation codons (Eiglmeier et al., 1987; Cole et al., 1988). Most of this region is contained on a 268-bp NruI-Hue111 restriction fragment. This fragment was employed in DNase I footprinting studies in order to locate the binding sites for the glp repressor and CAMP-CRP. Separate experiments were done where the fragment was radiolabeled on either end so the distance between

Control of glpTQ-glpACB Gene Expression in E. coli 6117

the labeled end and the binding sites could be minimized, and the regions protected on either strand could be analyzed.

The results of DNase I footprinting analysis involving purified glp repressor and end-labeled DNA are shown in Fig. 4. Binding of repressor protected several regions from digestion by DNase I. The areas protected are indicated on the nucleotide sequence shown in Fig. 5. It is apparent that there are multiple binding sites for the repressor in the control region for the glpACB operon. The largest area of protection

A S C l n R C R P - .. . I 2 3 1

I C l o R C R P

FIG. 4. DNase I footprinting using GlpR and CRP. The size standards (labeled S) were generated from M13mp19 clones containing the NruI-HaeIII fragment in opposite orientations. The sizes of the lowest bands shown (C, in both cases) are 64 and 36 nucleotides for the standards shown in A and B, respectively. A, the top strand (Fig. 5) was labeled at the 3' end near the HaeIII site. GlpR, reactions were carried out using the following concentrations of glp repressor tetramers and analyzed in the indicated lanes: I , 0 nM; 2, 0.2 nM; 3, 0.4 nM; 4 and 8, 0.8 nM; 5, 1.6 nM; 6 and 9, 3.2 nM; 7, 6.4 nM. Reactions run in lanes 8 and 9 also contained 2 mM glycerol-P. The arrow indicates the position of the DNase I hypersensitive site. CRP, reactions were carried out using the following concentrations of CRP dimers and analyzed in the indicated lanes: 1, 0 nM; 2, 25 nM; 3, 50 nM; 4, 50 nM. Reactions analyzed in lanes 1-3 also contained 0.25 mM CAMP. B, the bottom strand (Fig. 5) was labeled at the 3' end near the NruI site. GlpR, reactions were carried out using the following concentrations of glp repressor tetramers and analyzed in the indicated lanes: 1 , 0 nM; 2, 0.2 nM; 3, 0.5 nM; 4, 1 nM; 5, 5 nM; 6, 20 nM. CRP, reactions were carried out using the following concentrations of CRP dimers and analyzed in the indicated lanes: I , . 0 nM; 2, 25 nM; 3, 50 nM; 4, 100 nM; 5, 50 nM. Reactions analyzed in lanes 1 through 4 also contained 0.25 mM CAMP.

(0A2 and 0 ~ 3 ) included the start point of transcription and the -10 region. Protection of a relatively small area in the glpTQ promoter region was observed (Figs. 4 and 5). This region coincided with the binding site for cAMP CRP. Bind- ing of the glp repressor to all of these areas was greatly decreased when the inducer (glycerol-P) was included in the footprinting reactions (Fig. 4A).

The footprinting data were used to estimate the relative binding affinity of the repressor for the various operator sites. Repressor had the highest apparent affinity for OA1, OT, and the tandem OA2-OA3 pair, and lower affinity for OA4. Protec- tion of the high affinity operators was complete a t a repressor concentration of 0.4 or 0.5 nM. Significant protection of On4 did not occur until 5 nM or 20 nM repressor was used in the footprinting analysis (Fig. 4).

Cleavage by DNase I of a region of DNA between 0 ~ 1 and 0A2 was enhanced by binding of repressor (Fig. 4A, arrow). These results, along with results to be presented below, indicate that the DNA in this region may be bent or looped when bound to repressor.

Identification of the Binding Sites for the CAMP. CRP Com- plex-Expression of the glpTQ operon is dependent upon the presence of the CAMP-CRP complex. A sequence similar to the consensus sequence for interaction with CAMP. CRP was previously identified (Eiglmeier et al., 1987). CRP binding to this region was demonstrated directly by performing DNase I footprinting experiments. The results, shown in Fig. 4 and summarized in Fig. 5, show that CRP does indeed bind to the proposed site. Binding was dependent upon the presence of CAMP. The CRP site was centered 41-42 bp upstream from the start point of transcription of the glpTQ operon.

Two additional binding sites for CRP were discovered in the promoter region for the glpACB operon using this analysis (Figs. 4 and 5). One was centered 40-41 bp upstream from the first start point of transcription of the glpACB (Eiglmeier et al., 1989). This site coincides with the region predicted to interact with FNR (Eiglmeier et al., 1989). It is not surprising that FNR and CRP bind to the same site, because these two related transcription factors recognize very similar DNA sequences (Zhang and Ebright, 1990). The other CRP site identified by footprinting is centered 2-3 bp downstream from the start point of transcription documented by Eiglmeier et al. (1989; Fig. 5). Binding of cAMP.CRP to both of these areas was abolished when cAMP was omitted from the footprinting reactions.

Binding of CRP to each of its sites resulted in enhanced sensitivity of the DNA to cleavage by DNase I within the CRP binding sites. This is especially true for the sites labeled CT and CAI (Fig. 4). The enhanced sensitivity is likely due to CRP-induced bending of the DNA (Kim et al., 1989; Schultz et al., 1991).

Comparison of the glp Operator Sites-Inspection of the sequences present in the regions protected by the glp repressor revealed that these sequences could be aligned with the previously identified tandem operators of the glpD operon (Ye and Larson, 1988). By comparison of all 14 half-sites, the consensus half-site 5'-WATGTTCGWT-3' (W = A or T) for repressor binding was derived (Fig. 6). The various operators matched the consensus operator sequence to varying degrees. I t is interesting to note that the operons encoding the dehydrogenases are controlled by tandem operators and that the operator that matches the consensus most closely overlaps the start point of transcription in both cases. A single operator was found in the glpTQ promoter region overlapping directly the CRP site. The glpTQ operator did not match the consensus operator as well as most of the other operators.

6118 Control of glpTQ-glpACB Gene Expression in E. coli

&QI 'A4 CGTCACTTGATTGCGAGTCGCGAGTTTTCATTGTTATCCC GCAGTGAACTAACGCTCAGCGCTCWGTAACAATAGGGA

D S S Q S D R T K H """""""""

<-" AT GAT A CG (OA4M) A A T T GC T (OA4P)

and CRP sites in the glpACB-glpTQ FIG. 5. Positions of the operator

control region. Regions protected by the glp repressor or CRP are indicated by dashes or doubb dashes, respectively. The -10 sequence and start point of transcription (+1) for each operon are underlined. The sequences of the insertions (+2, +6, and +lo) into the BstBI site (TTJCGAA) are indicated for the top strand. The sequence changes present in the operator mutations (OTM, AT +6, O A ~ M , OA2M, 0 ~ 3 M , O A ~ M , and 0 ~ 4 P ) are indicated for the bottom strand.

Operator sequence

CRP/O """"__==== """"_ """""""""- ,=E=== ==== =

T T T A T G A A T G T T T T C T C G C G G C W C T C A A G A A A C G G AAATACPTACAAAAGAATTGTAGCGCCGTTGAGTTCTTTGCC

-10 +1

============ 5============

I I TA TTG ( O F )

TA (AT + 6 )

-">

C A G G T T C T C T C A C T G M T C A G C C T G T T A A T C A T A A A T A A G G T C C A A G A G A G T G A C T T A G T C C G A C A A T T A G T A G T A T T C T T

m 1 1 1 M L S I F

0 Match with consensus

OD1 T A T G T T C G A T A A C G A A C A T T 100

OD2 T A T G A G C T T T A A C G A A A G T G 70

OA1 A A T G T T C A A A A T G A C G C A T G 65

OA2 A C T T T C G A A T T A T G A G C G A A 50

O A l T A T G C G C G A A A T C A A A C A A T 75

CAQ A A T G G T A A A A A A C G A A C T T C 70

OT T G T G T G C G G C A A T T C A C A T T 65

Frequency Of A6 A10TllG12Tg T8 Cg G, A7 Tg

5 2 5 6 4 Occurence: T5 T2 C2 T2 C, G A A T A

C 2 C A A C T 2 C G C

G G G G T

Half-s i te Consensus: 5"W A T G T T C G W T-3'

FIG, 6. Comparison of the glpD and glpACB-glpTQ operator sites. The operator sequences are written in the direction of transcription of the indicated gene. The gap in the center corresponds to the center of operator symmetry.

Effect of Operator Mutations on Repressor Control of glpTQ- glpACB Expression-To determine which of the operators identified by footprinting are important for regulation of the glpTQ-glpACB operons, constructs were generated where the five operators were individually altered. Mutations (OTM, OAlM, OAZM, O A ~ M , and OA4M) were designed so that the sequences resembled less closely the operator consensus, especially at positions located 4 and 7 bases from the center of operator symmetry? The specific alterations left the other promoter elements intact (-10 and CRP/FNR site). In a sixth construct, OA4 was altered so that it matched the consensus operator sequence perfectly (OA4P). DNA fragments harboring the wild type or mutant control regions were cloned into

These positions are critical for repressor-operator interaction (W. Oh and T. J. Larson, unpublished observations).

pCB267 (Schneider and Beck, 1987) such that glpTQ promoter activity could be monitored by measuring alkaline phosphatase activity, and glpACB promoter activity could be monitored by measuring ,&galactosidase activity. The resulting plasmids were introduced into two strains, one lacking glp repressor [TL73(pACYC184)] and the other [TL73(pSH58)] producing glp repressor from a multicopy plasmid. The effi- cacy of the glp repressor was assessed by comparing alkaline phosphatase or &galactosidase activities produced in the presence or absence of the repressor. The activities were measured after growth under aerobic or anaerobic conditions. Anaerobic growth conditions stimulated the glpACB and the glpTQ promoters by more than 10-fold and 5-fold, respectively.

The effects of the operator mutations on repressor-mediated regulation of the glpTQ and glpACB promoters are shown in the top portion of Table I. Alteration of OT (OTM) did not result in a significant decrease in repression of the glpTQ operon during aerobic growth. Repression of glpTQ was decreased %fold during anaerobic growth (Table I). OTM had little influence on repression of the glpACB operon under anaerobic conditions, but increased repression aerobically.

Alterations in the glpACB operators had a very significant effect on regulation of both promoters. The greatest decrease in repression of the glpTQ promoter was observed in the case of the OA1 mutation, where repression was almost eliminated. OA3 and OA4 were also important for regulation of glpTQ, however, as repression decreased relative to the wild type in the strains harboring the mutant constructs. Repression of glpTQ increased when OA4 was changed to match the consensus operator perfectly ( O A ~ P , Table I). Mutation of 0 ~ 2 decreased repression of glpTQ only slightly. The relative importance of the operators for regulation of glpTQ was 0 ~ 1

Results for regulation of the glpACB promoter involving the various constructs described above showed that each of the glpACB operators contributed to regulation by glp repressor. OA3 was clearly the most critical for regulation of the

> OA3 > 0 ~ 4 > 0 ~ 2 > OT.

Control of glpTQ-glpACB Gene Expression in E. coli 6119 TABLE I

Repressor control of divergent glpACB-glpTQ promoters in wild type and mutant constructs Aerobic growth Anaerobic growth

Construct' glpTQ promoter" glpACB promoterb glpTQ promoter" glpACB promoterb

glpR glpR+ Repressiond glpR glpR+ Repression glpR glpR+ Repression glpR glpR+ Repression

Wild type 3.50 0.11 31 (11) 32.3 1.12 29 (10) 14.3 0.55 26 (6) 412 2.5 166 (6) OTM 10.10 0.37 27 (5) 80.4 1.26 64 (5) 11.5 0.89 13 (3) 502 3.4 149 (6) 0 ~ l M 2.48 1.06 2 (2) 21.9 1.21 18 (2) 16.9 6.85 2 (3) 420 5.8 72 (3) 0 ~ 2 M 3.27 0.22 15 (5) 11.0 1.24 9 (5) 23.2 1.6 15 (2) 554 11.5 48 (2) 0 ~ 3 M 2.16 0.39 6 (5) 14.0 1.80 8 (4) 13.0 2.5 5 (2) 178 22.8 8 (2) 0 ~ 4 M 3.15 0.31 10 (4) 8.7 1.16 8 (4) 15.5 2.5 6 (2) 109 5.5 20 (2) 0 ~ 4 P 3.21 0.07 46 (4) 17.4 0.72 24 (4) 22.1 0.56 39 (3) 467 2.2 212 (3)

0 ~ 2 + 6 2.60 3.54 1 (5) 5.0 1.27 4 (4) 16.1 11.6 1 (2) 124 4.7 26 (2) 0 A 2 + 2 3.36 0.71 5 (6) 13.6 1.27 11 (5) 15.2 2.3 7 (2) 267 4.6 58 (2)

0A2+10 4.30 0.63 7 (5) 6.7 0.93 7 (4) 19.6 2.2 9 (2) 193 1.9 100 (2) AT+6 0.28 0.46 1 (4) 8.6 1.16 7 (4) 3.5 6.3 1 (4) 159 3.1 51 (4)

Promoter activity was monitored by measuring the specific activity of alkaline phosphatase. * Promoter activity was monitored by measuring the specific activity of 6-galactosidase. 'pCB267 containing the indicated control region was introduced into strain TL73(pACYC184) [glpR] or TL73

Repression is defined as the specific activity in the glpR strain divided by the specific activity in the glpR+ strain. Numbers in Parentheses indicate the number of independent determinations performed in order to determine

(pSH58) [glpR+I.

mean repression ratios.

glpACB operon particularly under anaerobic conditions. Dur- ing anaerobiosis, there was a striking increase in repression mediated by glp repressor. The relative importance of the operators for regulation of glpACB under anaerobic conditions was OA3 > 0 ~ 4 > 0A2 > 0~1. The same trend was observed under aerobic conditions, with OAl being least critical for repression (Table I).

Interaction between the Operators-The above results in- dicating that the glpTQ promoter is controlled by operators OAl, OA3, and OA4 is unexpected in light of the relatively long distance between these operator sites and the glpTQ promoter. OA1 is located -72 to -91 relative to the start point of transcription for glpTQ. This is farther upstream than usual for a negative control element influencing a a-70 promoter (Collado-Vides et al., 1991; Gralla, 1991). However, a higher order DNA structure, such as a repressor-mediated DNA loop, possibly between OA1 and OA3 or 0 ~ 4 , may influence expression of glpTQ over a larger distance. The following phasing experiments were designed to test this possibility.

First, the effects on repression of insertions of 2, 6, or 10 base pairs made at the unique BstBI site located in OA2 were tested. If repressor-mediated DNA looping between operators is important for repression, then insertion of 6 base pairs (one-half of a helical turn) between the operators involved should decrease repression, whereas insertion of 10 base pairs (one helical turn) should have less influence on repression. Disruption of OA2 should pot influence repression of glpTQ to a large extent, as this operator is least critical for control of glpTQ (see results for O A ~ M , above).

Insertion of 6 base pairs at the BstBI site totally eliminated repression of the glpTQ promoter by glp repressor ( 0 A 2 + 6; Table I). In fact, a small but reproducible stimulation by repressor of the 0 A 2 + 6 glpTQ promoter was observed during aerobic growth. Significant repression of the glpTQ promoter was retained in the case of the 2- or 10-base pair insertions (0.42 + 2 and 0 A 2 + 10; Table I). The insertion mutations had similar effects on repression of the glpACB promoter (Table I), except in this case repression was never totally eliminated, perhaps due to the presence of multiple operators. The results are consistent with a model for regulation in which repressor binds simultaneously to operator sites flank- ing the site of the insertions, with the intervening DNA

forming a loop. Because OT is apparently not critical for repressor control of the glpTQ promoter, a repression loop involving OA1 and 0 ~ 3 or 0 ~ 4 seems the most plausible. Operators 0 ~ 3 and 0 ~ 4 are separated from 0 ~ 1 by 4.95 and 8.76 helical turns, assuming 10.5 base pairs per helical turn. The distance and phasing are appropriate for loop formation between OA1 and either of the two most distant glpACB operators (Hochschild and Ptashne, 1986).

To provide further evidence for involvement of remote operators in the regulation of glpTQ, the effect on repression of a 6-base pair insertion between OA1 and OT was tested. It was found that this phasing mutation completely eliminated repression of glpTQ (AT + 6; Table I). This alteration also decreased repression of glpACB 3- to 4-fold. Although this insertion was made in the region between the promoter elements of the divergent operons, the insertion drastically reduced the strength of the glpTQ promoter and, to a lesser extent, the glpACB promoter (Table I). These results are suggestive of interaction between transcription factors bound at the divergent promoters. One possibility would be interaction between the two CRP sites. The fact that OTM increases glpACB promoter activity supports this idea. OTM changes the CRP site for glpTQ to match the CRP binding site consensus more closely and might be expected to cause an increase in apparent promoter strength.

DISCUSSION

With the completion of this work and that described in the accompanying paper (Weissenborn and Larson, 1992), the control regions of each of the glp operons have been charac- terized with regard to the binding sites for the glp repressor and CAMP. CRP. The common theme for the arrangement of the repressor binding sites is tandem repetition of the operator. This is especially noteworthy for the control of the operons encoding the aerobic and anaerobic glycerol-P dehydrogenases. Cooperative binding of repressor to adjacent sites, which is reminiscent of the binding of X repressor to its operators (Gussin et al., 1983), is likely to play an important role in the relatively tight control of these operons by the glp repressor. As has been pointed out by Lin (1976,1987), rapid shut down of the operons encoding the dehydrogenases would be desirable when glycerol-P in the medium is depleted,

6120 Control of glpTQ-glpACB Gene Expression in E. coli

Decreasing the levels of these enzymes would allow utilization of the remaining glycerol-P for the biosynthesis of membrane phospholipids.

Besides interaction of glp repressor at adjacent operators, it is clear that interactions between operators separated by greater distances are important for complete repression of the glpTQ and glpACB operons. The degree of repression of the glpTQ promoter was decreased, to differing extents, by mutations in each of the glpACB operators. Somewhat unexpected was the finding that changes in OT had little effect on repression of glpTQ, while changes in OA1 practically eliminated repression. Mutation of OA3 reduced repression of glpTQ 5- to 6-fold. Elimination of OA2, which is directly adjacent to 0 ~ 3 , reduced repression only 2-fold. 0 A 2 could exert a small effect on glpTQ due to its influence on binding of repressor to OA3. OA4, the operator most remote from glpTQ, also was required for full repression of the glpTQ promoter, especially during anaerobic growth.

One model which explains regulation of glpTQ from these remote operators invokes the existence of a repression loop. Evidence in support of such a model was obtained through analysis of the effects of phasing mutations and individual operator-constitutive mutations on repression of the divergent operons. Evidence for repressor-mediated DNA looping in the glpFK operon has also been obtained (Weissenborn and Larson, 1992). It is possible that tetrameric glp repressor (Larson et al., 1987) is able to bind two appropriately oriented operators simultaneously, with the intervening DNA forming a loop. If this is the case, regulation of the glp operons by glp repressor would resemble control of other catabolic operons such as lac, gal, deo, and ara by the respective repressors. In each of these instances, repressor binding to widely separated operators has been invoked to explain repression (Adhya, 1989; Gralla, 1989).

If OT is not critical for repression of glpTQ, what is the nature of the repression complex responsible for regulation? We believe the data are most consistent with a model in which a primary repression loop involving OA1 and 0 ~ 3 is formed. These operators are the most critical for regulation of glpTQ and are separated by five helical turns of DNA (assuming 10.5 base pairs per helical turn). A stable repression loop involving these two operators also would be responsible for the tight control of glpACB because the loop encompasses the entire promoter. Such a loop might also place 0 ~ 4 and OT in juxtaposition such that repression of glpTQ results. 0 ~ 4 and OT are positioned with precise symmetry relative to 0 ~ 3 and OAl, with operator centers of symmetry 40 base pairs removed from centers of the adjacent operators. An explanation must then be provided for the minimal influence of OTM on repression. One possibility is that the three base pair change con- verting OT to OTM was not severe enough to eliminate repression, especially if the primary repression loop involving 0 ~ 1 and OA3 is very stable and places 0 ~ 4 and OT in proper juxtaposition for simultaneous interaction with repressor. An- other explanation would be that repressor bound at 0 ~ 4 alone interferes with binding of CRP or polymerase at the glpTQ promoter due to the spatial proximity of the binding sites in the looped DNA. Thirdly, it is possible that repressor bound at one of the glpACB operators contacts CRP bound at OT or disrupts a positive interaction occurring between two CRP sites in the divergent promoters. The first possibility is fa- vored, because the phasing mutation A T + 6 decreased repression of glpACB, which indicates that an element on the glpTQ side of AT + 6 (most likely OT) is important for repression of glpACB. The observations that mutations affecting 0 ~ 4 decrease repression of glpTQ and that glp repressor footprints

OT in vitro also support the first possibility. Characterization of the influence of operator-constitutive

mutations on expression of the glpACB promoter supports the conclusion that each of the four operators is important for regulation of the glpACB operon. The effects of the mutations on expression of glpACB are most readily apparent during anaerobic growth, where much higher repression ratios are obtained. The differences in repression observed during anaerobiosis may be due to the increase in negative supercoiling of the DNA under these conditions (Dorman et al., 1988), which may favor formation of a repression complex. It is clear that the order of importance of the operators for regulation of glpACB is 0 ~ 3 > 0 ~ 4 > 0 A 2 > 0~1. Thus, OA3, which directly overlaps the -10 region of the glpACB promoter, is most critical for regulation of glpACB.

The spacing between FNR-binding site and the start of transcription of FNR-dependent promoters seems to be well conserved (Eiglmeier et al., 1989). The same is true for CRP- binding sites (Gaston et al., 1990; Ushida and Aiba, 1990). Therefore, it was somewhat surprising to find that the insertion mutations, which were generated at the unique BstBI site located between the -10 promoter element of glpACB and the CRP/FNR site, did not eliminate glpACB promoter activity. Walker and DeMoss (1991) and Bell et al. (1990) found that a variety of insertions (including 4, 6, 8, or more nucleotides) between the FNR site and the start of transcription abolished FNR-dependent anaerobic induction. The presence of glpACB promoter activity both anaerobically and aerobically in the +2, +6, and +10 mutant constructs suggests that the second CRP site, which overlaps the start point for transcription (as indicated in Fig. 5), might promote initiation of transcription from a downstream promoter. An appropriately spaced -10 element for this putative promoter may be the sequence TAAATG which is located in 0 ~ 4 . The corre- sponding -10 sequence in the 0 ~ 4 M situation would by TAGCTA. The sequence changes in 0 ~ 4 M would be expected to decrease promoter activity, which was observed (Table I). Although in vivo evidence for the function of the putative promoter has not been sought, it should be pointed out that the sequence of the downstream CRP/FNR site matches the consensus sequence for CRP and FNR binding more closely than that of the upstream CRP/FNR site.

Acknowledgments-The graduate students in the course Biochem- istry 5104 (Summer 1991) are gratefully acknowledged for the isolation and sequencing of three of the oligonucleotide-directed mutations (OJM, OTM, and AT + 6). We thank James G. Harman for the gift of pCRP-1, Christoph Beck for providing pCB267, Heidi J. Hoffmann for construction of pHH11, Herbert Schweizer for construction of pSH58, Deborah L. Weissenborn for preparation of Fig. 1, and Karin Eiglmeier and Stewart T. Cole for providing information regarding the start point of transcription of glpACB prior to publication.

REFERENCES

Adhya, S. (1989) Annu. Rev. Genet. 23, 227-250 Adhya, S., and Gottesman, M. (1982) Cell 29,939-944 Austin, D., and Larson, T. J. (1991) J. Bacteriol. 173 , 101-107 Bell, A. I., Cole, J. A., and Busby, S. J. W. (1990) Mol. Microbiol. 4,

Blazy, B., and Ullmann, A. (1986) J. Biol. Chem. 261,11645-11649 Bradford, M. M. (1976) Anal. Bwchem. 72 , 248-254 Chang, A. C. Y., and Cohen, S. N. (1978) J. Bacteriol. 134 , 1141-

Clark, D. J., and Maalbe, 0. (1967) J. Mol. Biol. 2 3 , 99-112 Cole, S. T., Eiglmeier, K., Ahmed, S., Honore, N., Elmes, L., Ander-

son, W. F., and Weiner, J. H. (1988) J. Bacteriol. 170,2448-2456 Collado-Vides, J., Magasanik, B., and Gralla, J. D. (1991) Microbiol.

Reu. 55,371-394 Dorman, C. J., Barr, G. C., NiBhriain, N., and Higgins, C. F. (1988) J. Bacteriol. 170 , 2816-2826

1753-1763

1156

Control of glpTQ-glpACB Gene Expression in E. coli 6121

Ehrmann, M., Boss, W., Ormseth, E., Schweizer, H., and Larson, T.

Eiglmeier, K., Boos, W., and Cole, S. T. (1987) Mol. Microbiol. 1 ,

Eiglmeier, K., Honore, N., Iuchi, S., Lin, E. C. C., and Cole, S. T.

Freedberg, W. B., and Lin, E. C. C. (1973) J. Bacteriol. 115,816-823 Gaston, K., Bell, A., Kolb, A., Buc, H., and Busby, S. (1990) Cell 6 2 ,

Gralla, J. D. (1989) Cell 57, 193-195 Gralla, J. D. (1991) Cell 6 6 , 415-418 Gussin, G. N., Johnson, A. D., Pabo, C. O., and Sauer, R. T. (1983)

in Lambda ZZ (Hendrix, R. W., Roberts, J. W., Stahl, F. W., and Weisberg, R. A., eds) pp. 93-121, Cold Spring Harbor Laboratory, NY

Harman, J. G., McKenney, K., and Peterkofsky, A. (1986) J. Biol. Chem. 261,16332-16339

Hochschild, A., and Ptashne, M. (1986) Cell 44,681-687 Iuchi, S., Cole, S. T., and Lin, E. C. C. (1990) J. Bacterwl. 1 7 2 , 179-

Kim, J., Zwieb, C., Wu, C., and Adhya, S. (1989) Gene (Amst.) 85,

Kunkel, T. A., Roberts, J. D., and Zakour, R. A. (1987) Methods

Kuritzkes, D. R., Zhang, X.-Y., and Lin, E. C. C. (1984) J. Bacteriol.

Larson, T. J., Schumacher, G., and Boss, W. (1982) J. Bacteriol. 152 ,

Larson, T. J., Ye, S., Weissenborn, D. L., Hoffmann, H. J., and

Lin, E. C. C. (1976) Annu. Rev. Microbwl. 30,535-578 Lin, E. C. C. (1977) in Escherichia coli and Salmonella typhimurium:

Cellular and Molecular Biology (Neidhardt, F. C., ed) Vol. 1, pp. 244-284, American Society for Microbiology, Washington, D. C.

J. (1987) J. Bacteriol. 169 , 526-532

251-258

(1989) Mol. Microbwl. 3,869-878

733-743

184

15-23

Enzymol. 154,368-382

157,591-598

1008-1021

Schweizer, H. (1987) J. Biol. Chem. 262,15869-15874

Liss, L. R. (1987) Focus 9 , 13 Lupski, J. R., Zhang, Y. H., Rieger, M., Minter, M., Hsu, B., Ooi, B.

G., Koeuth, T., and McCabe, E. R. B. (1990) J. Bacteriol. 172 ,

Miller, J. H. (1972) in Experiments in Molecular Genetics, pp. 352- 355, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Rodriguez, E. L., and Tait, R. C. (1983) Recombinant DNA Tech- niques: an Introduction, pp. 164-165, Addison-Wesley Publishing Co., Reading, MA

Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. 5'. A. 74,5463-5467

Schneider, K., and Beck, C. F. (1987) Methods Enzymol. 153, 452- 461

Schultz, S. C., Shields, G. C., and Steitz, T. A. (1991) Science 253 ,

Schweizer, H., and Larson, T. J. (1987) J. Bacterid. 169, 507-513 Schweizer, H., Boos, W., and Larson, T. J. (1985) J. Bacteriol. 161,

Schweizer, H., Sweet, G., and Larson, T. J. (1986) Mol. Gen. Genet.

Sweet, G., Gandor, C., Vogele, R., Wittekindt, N., Beuerle, J., Tm- niger, V., Lin, E. C. C., and Boos, W. (1990) J. Bacterid. 172,424- 430

Tommassen, J., Eiglmeier, K., Cole, S. T., Overduin, P., Larson, T. J., and Boos, W. (1991) Mol. h Gen. Genet. 226,321-327

Ushida, C., and Aiba, H. (1990) Nucleic Acids Res. 18,6325-6330 Walker, M. S., and DeMoss, J. A. (1991) Mol. Microbiol. 6, 353-360 Weissenborn, D. L., Wittekindt, N., and Larson, T. J. (1992) J. Biol.

Yanisch-Perron, C., Vieira, J., and Messing, J. (1985) Gene (Amst.)

Ye, S., and Larson, T. J. (1988) J. Bacteriol. 170, 4209-4215 Zhang, X., and Ebright, R. E. (1990) J. Biol. Chem. 265 , 12400-

6129-6134

1001-1007

563-566

202,488-492

Chem. 267,6122-6131

33,103-119

12403

Documents

Interaction at a distance between multiple operators controls the