Transcriptional Mapping and Nucleotide

Vol. 263, No 35, Issue of December 15, pp. 18857-18863,1988 Printed in U.S.A.

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc

Transcriptional Mapping and Nucleotide Sequence of the Escherichia coli fepA-fesEnterobactin Region IDENTIFICATION OF A UNIQUE IRON-REGULATED BIDIRECTIONAL PROMOTER* (Received for publication, June 22,1988)

Gregg S. PettisS, Timothy J. Brickman$, andMark A. McIntosh7 From the Department of Microbiology, School of Medicine, University of Missouri-Columbia, Columbia, Missouri 65212

Theiron-controlled fepA and fes-entF transcripts cations to thepromoter of the pColV-K30-derived aerobactin from the Escherichia coli enterobactin gene complex operon (5) and to act as a corepressor with ferrous iron in are expressed divergently from a limited genetic re- vitro by specific binding to the promoter region of an aerogion, thereby suggesting the existence of asingle, pos- bactin-lac2 operon fusion construction (6). Further assesssibly overlapping promoter junction for these genes. ment of Fur’s repressor role in iron transport, as well as the Thenucleotidesequence of a 1,997-base pair HpaI delineation of other potential common and specific factors, fragment specific for this genetic region allowed for awaits similar characterization of the other genetic systems the identification of an 1,122-base pair open reading involved in the Fur-controlled regulon. frame as the previously uncharacterized fes gene. Its For the enterobactin gene cluster, this process has been product, Fes (- M , 42,573) plays an essential but as slowed by its physical size (encompassing approximately 22 yet ambiguous role in the release of ferric iron from kb’ at min 13) (7) and its complexity (14 defined complementhe ligand. An additional small open reading frame of 216 nucleotides (encoding a potential product ofcal- tation groups). Genetic and biochemical evidence has demculated M, 8,271) was also identified betweenfes and onstrated the presence of seven distinct biosynthesis ( e n t ) entF. A portion of the remaining nucleotide sequence functions required for the conversion of the intermediates, chorismate and L-serine, into the iron ligand (8-10). Transdefined a 320-base pair control region for both the fepAand fes-entF messages. Primer extension analyses port of ferric enterobactin through the cell envelope involves placed the major in vivo transcription initiation sites the products of at least five additional cistrons (fep) (11, 12), to within 18 nucleotides of one another, thereby re- the most intensely studied being the outer membrane receptor vealing a novel, extensively overlapping bidirectional gene, fepA (13, 14). A further component, designated fes (15) promoter as well as long dual leader transcripts. This is absolutely required for the intracellular release of iron from promoter region contains multiple overlapping nucleothe ligand; however,its mode of action is still unresolved. The tide stretches which show strong homology to the con- clockwise gene order, as the E. coli chromosome is usually sensus Fur repressor-binding sequence, forms of which represented, of entD, fepA, fes, entF, fepE, fepC, fepG, fepD, are found inall E. coli iron-regulated promoters char- fepB, entC, entE, entB, entA has been derived from recent acterized to date. reports and unpublished observations. At the counterclockwise terminus of the enterobactin gene cluster, earlier analysis (16) demonstrated the existence of an iron-regulated transcript that is initiated upstream of fes and encodes at least The limitation of metabolic iron during aerobic growth of the fes and entF genes. This diverges from an inducible fepA Escherichia coli signals the manufacture and secretion of its transcript whichwas found during earlier ent-lac chromonative siderophore, enterobactin (enterochelin). A respondent somal fusion studies (17). From this organization, it seemed transport system, necessary for recovery of the ferrated ligand likely that many, if not all, of the genes encoded in this from the environment is concomitantly induced, as are comsegment of the enterobactin cluster are included on one of ponents specific for alternative, often exogenously encoded two opposing iron-regulated transcripts that are initiatedfrom siderophore species (1). Fundamental to this collective re- a commonregion. To demonstrate this, the nucleotide sesponse is thecoordinate expression of some 30 genes arranged quence of this intercistronic junction has been determined individually and in groups dispersed throughout the chromofollowed by mapping of the relevant transcription start sites. some (2). A controlling element for this induction has been The data reveal an important bidirectional promoter which defined by the fur locus (3, 4). Its protein product has been contains elements showing homology to other iron-regulated purified and shown to bind in the presence of various divalent promoters. Also included in this report is the nucleotide and * This work was supported by Grant DMB8416017 from the Na- deduced amino acid sequences of the enterobactin fes gene. tional Science Foundation (to M. A. M.). 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 solelyto indicate this fact. The nucleotide sequencefs) reported in thispaper h been submitted to theGenBankTM/EMBL Data Bankwith accession numberfs) 504216. $ Recipient of Predoctoral Traineeship 5 T32 GM07494 from the National Institute of General Medical Sciences. § Predoctoral Trainee supported by National Institute of Allergy and Infectious Diseases Grant 1 T32 AI07276. 7 To whom correspondence should be addressed.

EXPERIMENTAL PROCEDURES

Bacterial Strains, Plasmids, and Enzymes-E. coli MC4160(16) was used as the source for RNA. The recombinant plasmids PITS21 and pITS311 have been described previously (16). The plasmid construct pITS513 was created by ligating the 1.6-kb SspI-BarnHI insert fragment of pITS311 into SmaI-BamHI cut pGEM-3Z (Promega The abbreviations used are: kb, kilobase; CAPS, 3-[cyclohexylamino]-1-propanesulfonic acid; bp, base pair; REP, repetitive extragenic palindromic.

18857

18858

Enterobactin fepA-fesBidirectional Promoter

Biotec, Madison, WI). [y3*P]ATP(4500 Ci/mmol) and [(Y-~'P]~ATPdescribed (26). The aqueous phase was re-extracted once with phenol/ (600 Ci/mmol) were from ICN Radiochemicals (Irvine, CA). Isopro- chloroform (50:50) saturated with 0.02 M sodium acetate, pH5.5, and pyl-0-D-galactosidewas purchased from Sigma and used at 57 pg/ml. twice with chloroform followedby precipitation of the RNA in 3 Five-bromo-4-chloro-3-indolyl-~-~-galactoside (X-gal, final concen- volumes of absolute ethanol at -70 "C for 30 min. Yield was detertration 40 pg/ml) was from Boehringer Mannheim. Ampicillin (50 mined spectrophotometrically by measuring the AZM) value for each pg/ml) was used to select for the presence of plasmids. sample. Upon collection by centrifugation, the RNA was stored in Genetic Techniques and DNA Sequence Analysis-Plasmids were lyophilized samples at -70 "C. isolated by the procedure of Birnboim and Doly (18). Purification Primer Extension Analysis-For kinase treatment of oligonucleowas completed by precipitation of double-stranded DNA in the pres- tides, approximately 12.5 pmol of primer was mixed with 30 pmol ence of 800 mM NaCl and 6.5% polyethylene glycol as recommended [Y-~'P]ATP and 20 units of polynucleotide kinase in 50 mM Tris, pH by Promega Biotec. All other standard genetic techniques were per- 8.8, 10 mM MgCl,, 5 mM dithiothreitol, 0.1 mM spermidine, 0.1 mM formed as described (19). EDTA for a total volume of 30 pl. After incubation at 37 "C for 60 For nucleotide sequence analysis of the pITS311 insert, the modi- min, EDTA was added to 20 mM,and theaqueous phase was extracted fied M13 phages M13mp18 and M13mp19 (20) were utilized as cloning once with phenol/chloroform (5050) andonce with chloroform. DNA vectors for various fragments generated by cutting at the strategic was precipitated by the addition of one-tenth volume 3 M sodium restriction endonuclease cleavage sites shown in Fig. 1. Transfection acetate, pH 5.2,20pgof E. coli tRNA and 2.5 volumes absolute of E. coli strain JMlOl (20) resulted in the production of the desired ethanol, followed by storage at -70 "C for 30 min. Primer extensions single strand recombinant templates. These were purified using the were done basically as described by Fouser and Friesen (27). Approxmethod outlined in the M13 Cloning and Sequencing Handbook by imately 250 fmol of primer was mixed with either 60 pg of MC4160 Amersham Corp. (Arlington Heights, IL), except that after phenol RNA or 30 pg of MC4160 (pITS21) RNA in 7.5 pl of reverse tranextraction, the template was re-extracted twice with phenol/chloro- scriptase buffer (27) with added RNasin (Promega Biotec, Madison, form (5050) andonce with chloroform prior to ethanol precipitation. WI) to 1 unit/pl and dithiothreitol to 5 mM. After hybridization at Nucleotide sequences were determined by the Sanger dideoxy chain 45 "C for 1 h, 3.5 pl of modified mixture R (400 p~ dATP, dTTP, termination technique (21) as modified for using the Sequenase" dGTP, and dCTP, 30 mM MgCl,, 3 mM dithiothreitol, 0.6 pg/ml enzyme (United States Biochemicals, Cleveland, OH). Duplicate re- actinomycin D) and 20 units ofAMV reverse transcriptase were actions, in one of which dITP was used in place of dGTP, resolved added and the reaction mixture incubated at 42'C for 1 h. The compression areas and insured that each base was read a minimum aqueous phase was extracted once with phenol/chloroform (5050) of two times. Oligonucleotide primers for sequence analysis were and once with chloroform followed by precipitation of nucleic acids synthesized by the University of Missouri DNA core facility on a in the presence of 300 mM sodium acetate and 3 volumes absolute model 380A DNA synthesizer (Applied Biosystems, Inc., Foster City, ethanol at -70 "C for 30 min. The dried pellet was resuspended in 7 CAI. Reactions were analyzed on 0.2-0.4-mmfield gradient, 8% pl of 200 pg/ml RNase A, incubated at 37 'C for 30 min, and then2.5 polyacrylamide sequence gels which were exposed to Kodak X-Omat p1 of sequence analysis dye was added prior to heating to 90 "C for 5 RP5 film at -70 "C for examination of bands. Nucleotide sequences min. Half of each reaction was subjected to electrophoresis on an 8% were interpreted with the aid of the Microgenie Sequence Analysis polyacrylamide gel simultaneously with a sequence ladder generated Program (22) distributed by Beckman Instruments, Inc. (Palo Alto, with the same primer and anM13 template containing the pITS311 insert fragment. CAI. Partial Amino Acid Sequence Determination of Fes and EntFConstruction of pITS357, a fes-lacZ gene fusion, by ligation of a 263RESULTS bp Sau3A fragment containing the fes promoter and translational Nucleotide Sequenceof the fes Gene-The plasmid pITS311 initiation site into the BamHI site of the plasmid vector pMC1403 allowed iron-regulated expression of a Fes-@-galactosidasehybrid (Fig. 1)contains a 2.0-kb HpaI insert which encodes an intact protein consisting of the amino-terminal nine amino acids of Fes functional Fes protein of M,44,000 (16) in sodium dodecyl fused to &galactosidase beginning a t its eighth residue.' The entF- sulfate-polyacrylamide gel electrophoresis and extends into lac2 gene fusion construction pITS312 was described previously (16). The host strain used for overproduction of hybrid proteins was I PITS513 I MC4146-60 (23). Expression of fusion proteins was induced in 100 ml of early log phase LB broth culturesby addition of 2,2'-dipyridyl PITS311 I to 200 p~ concentration. After 1 h of induction, cells were harvested I by centrifugation, washed once with cold 0.9% NaC1, resuspended in 2 ml of solubilization buffer, and boiled 5 min. Samples (500 pl) were FeS Ent F 1 P c loaded onto 1.5-mm thick, 7.5% polyacrylamide gels and separated H s P w N H by electrophoresis according to Laemmli (24) a t 50 V constant voltage. Protein electrotransfer and microsequence analysis was performed c e c essentially as described by Matsudaira (25). Briefly, proteins were * electroblotted onto Immobilon-P polyvinylidene difluoride transfer 2 1" membranes (Millipore Corp., Bedford, MA) in atransfer buffer containing 10 mM CAPS, 10% methanol (v/v), pH 11.0, for 30 min at 0.5 A. After identification by brief Coomassie Blue R-250 staining, protein bands were excised and the membrane centered in the cartridge block of the sequenator. Proteins were analyzed on an Applied 1 kb I Biosystems model 470 gas-phase sequenator using the program No. t FIG. 1. Sequence analysis strategy for the pITS311 insert. O3RPTH and thephenylthiohydantoin derivatives were separated by reverse-phase high pressure liquid chromatography over a Brownlee The plasmid pITS311 was constructed as previously described (16). C-18 column. Samples were subjected to at least 5 cycles of sequence Plasmid pITS513 is a Fes+ derivative created as detailed in the text. analysis. The confirmed amino-terminal residues of Fes and EntF Approximate coding regions for each gene are denoted by bored (represented by their bold single-letter amino acid equivalents) are regions: the wide closed end of each box refers to the 5' region of the displayed in Fig. 2, along with the remaining deduced residues perti- gene, the nurrow closed end designates the 3' end and an open end indicates that only a portion of the gene is included in the insert. nent to thisreport. RNA Isolation-RNA was isolated essentially by the method of Arrows designate direction and extent of Sanger dideoxy sequence Aiba et al. (26). Cells were grown in 80 ml of LB broth untilearly log determined from M13 recombinant clones created with the restriction phase. The addition of FeSO, to 20 p M or 2,2'-dipyridyl to 200 pM sites indicated above on the insert line drawing. Two exceptions are resulted in high-iron uersus low-iron conditions, respectively. These indicated numerically; arrow 1 refers to sequence obtained from a conditions have been shown to repress or induce, respectively, the 263-bp Sau3A fragment cloned into M13mp18 whereas arrow 2 indifes-entF operon (16). The term iron-limiting refers to cells grown in cates sequence acquired with a 400-bp BglI fragment also cloned into LB broth alone. After further growth for 1 h, cells were harvested M13mp18. Arrows with open circlesindicate sequence obtained with and RNA recovered by two rounds of hot phenol extraction, as specific oligonucleotideprimers and theentire pITS311 insert region cloned into either M13mp18 or M13mp19 as template. Sites for (H) HpaI, (S)SspI, ( P ) PuuI, and ( N ) NruI are shown. 'T. Fleming, unpublished results.

3

- ..

-

t_

-

Enterobactin fepA-fes Bidirectional Promoter

18859

50

100

150 200 TTCTGCTCGGCGGCGGTAACGACAATAGTATCGTCATGTGAAACAGGAGTATCGGTCGGCTCTTGTGCCTGCGCTACCCCATAAATCCCCAGATTGACCA 300

250

ACAAGGCCAGGGAATGAATCTTCTTGTTCATTGTTTTATTCCTGCATTTTTGCCACGAATTGC~CTGTCGGGCATGGTCGTCATC~CACGACGCATCC FopA 350 400 CGCTACCGCGAAAACCTTTGATCCTGAAAGACACGCAGTGCAGTTGGTTAATTAATGTCCGCGCTTCCCACGGCGCGCCATTACGCTATTGCAAATGCAA ssp1 450 500 ATAGTTATCAATAATATTATCAATATATATTTCTGCAATC~TG~TTGCACAGTAAACATGGGGTTATGGTGTG~CGGCGTT~GTAGGAAGTGAG

Fes

550

-

600

AGCTGGTGGCAGTCGAAACATGGCCCGGAATGGCAGCGTCTG~TGACGAAATGTTTGAGGTCACTTTCTGGTGGCGTGATCCCC~GGTTCTG~G~T

M

F

B

V

T

F

W

W

R

D

P

Q

G

S

E

E

650 700 ACTCGACGATAAAGCGCGTATGGGTCTACATCACTGGTGTGACCGATCACCATCAGAACAGCCAGCCCCAGTCGATGCAGCG~TTGCAGGCACTAACGT Y S T I K R V W V Y I T G V T D H H Q N S Q P Q S M Q R I A G T N V 750 PVUI CTGGCAGTGGACGACACAACTCAATGCC~CTGGCGCGGCAGCTACTGCTTTATTCCCACCGAACGCGATGACATTTTTTCTGTACCATCCCCCGATCGC W Q W T T Q L N A N W R G S Y C F I P T E R D D I F S V P S P D R 850 900 CTCGAATTGCGCGAAGGCTGGCGAAAACTATTATTACCCCAGGCGATAGCCGATCCGCTGAACCTACAAAGCTGG~GGCGGGCGAGGGCACGCTGTTTCTG L E L R E G W R K L L P Q A I A D P L N L Q S W K G G R G H A V S 950

1000

CACTCGAAATGCCGCAAGCGCCTCTGCAACCGGGATGGGATTGTCCGCAAGCGCCAGAAATACCTGCCAAAGA~TTATCTGG~GTGAACGGTTGAA A L E M P Q A P L Q P G W D C P Q A P E I P A K E I I W K S E R L K

1100

1050

AAAGTCACGGCGTGTATGGATTTTTACCACCGGCGATGCAACAGCAG~G~CGCCCGCTGGCAGTTTTGCTCGATGGCGAATTTTGGGCGCAAAGTATG K S R R V W I F T T G D A T A E E R P L A V L L D G E F W A Q S M

1150

1200

CCCGTCTGGCCAGTGCTGACTTCGCTGACCCATCGTCAGCAACTTCCTCCCGCCGTGTATGTGTTGATCGACGCTATCGACACCACGCACCGCGCCCACG

P

V

W

P

V

L

T

S

L

T

H

R

Q

Q

L

P P 1250

A

V

Y

V

L

I

D

A

I

D

T T PVUI

H

R A H PvuI 1300

“

AACTGCCGTGTAATGCGGATTTCTGGCTCGCAGTACAGCAGCTATTGCCCCTTTTAGCGATCGTGCCGATCGCACCGT

E

L

P

C

N

A

D

F

W

L

A

V

Q

Q

E

L

L P 1350

L

V

K

A

I

A

P

F

S

D

R

A

D

R

T

V

1400 GGTTGCCGGGCAGAGTTTTGGTGGGCTTTCCGCGCTGTATGCCGGACTGCACTGGCCTGAACGCTTTGGCTGTGTATT~GCCAGTCAGGATCGTACTGG

V

A

G

Q

S

F

G

G

L

S

A

L

Y

A

G

L

H

W

P

E

R

F

G

C

V

L

S

Q

S

G

S

Y

W

1450 1500 T G G C C G C A T C G G G G C G G G C A G C A A G A G G G C G T G T T G G C G G G T A

W

P H R G G Q Q E G V L L E K L K A G E V S A E G L R I V L E A G NruI 1550 1600 TTCGCGAGCCGATGATCATGCGGGCCAATCAGGCGCTGTATGCGCAATTACACCCCAT~GAATCCATTTTCTGGCGTCAGGTTGACGGCGGACATGA I R E P M I M R A N Q A L Y A Q L H P I K E S I F W R Q V D G G H D 1650 ORF 1 1700 TGCGCTTTGTTGGCGCGGTGGCTTGATGCAGGGGCTAATCGACCTCTGGC~CCACTTTTCCATGACAGGAGTTGAATAT~GCATT~AGT~T~~~TT~G A L C W R G G L M Q G L I D L W Q P L F H D R S M A F S N P F 1750 1800 ATGATCCGCAGGGAGCGTTTTACATATTGCGCAATGCGCGCAGGGGCAATTCAGTCTGTGGCCGC~C~TGCGTCTTACCGGCAGGCTGGGACATTGTGTG D D P Q G A F Y I L R N A Q G Q F S L W P Q Q C V L P A G W D I v c 1850 Ent F 19% TCAGCCGCAGTCACAGGCGTCCTGCCAGCAGTGGCTGGAAGCCCACTGGCGTACTCTGACACCGACGAATTTTACCCAGTTGCAGGAGGCACAATGAG Q P Q S Q A S C Q Q W L E A H W R T L T P T N F T Q L Q E A Q M

S

1950 IipaI’ AGCATTTACCTTTGGTCGCCGCACAGCCCGGCATCTGGATGGCAG~CTGTCAGAATTACCCTCCGCCTGGAGCGTGGCGCATTACGTTGAGTT

Q

H

L

P

L

V

A

A

Q

P

G

I

W

M

A

E

K

L

S

E

L

P

S

A

W

S

V

A

H

Y

V

E

FIG. 2. Nucleotide sequence of the 1997-bp HpaI insert fragment of pITS311. Nucleotides are numbered starting with the 3’ half of the HpaI site in fepA. The startof the fepA coding region refers to theinitiation site on the opposite strand proposed previously (28). The single-lettercode for each amino acid of Fes, ORFl, and the beginning of EntF is centered below its corresponding codon. ORFl refers to the open reading frame which spans the intercistronic region between fes and en@. Amino acid residues for Fes and EntF confirmed by microsequence analysis are indicated by bold letters. The common names for pertinent restriction endonucleases are underlined above their respective palindromic recognition sequences. ‘HpaI, 3’ half of the HpaI restriction site generated by cleavage with this enzyme; Hpal’, 5’ half of the HpaI restriction site generated by cleavage with this enzyme.

the coding regions of both the fepA and entF genes (16, 28). A subclone designated pITS513 retains bothFes’ activity and production of the full-length Fes protein (data not shown), thereby localizing this approximate 1.2-kb gene as shown in Fig. 1. To precisely define the relevant end points for each cistron, the nucleotide sequence of the pITS311 insert was

determined (Fig. 2) by the Sanger dideoxy chain termination method with the strategy outlined in Fig. 1.For the fepA gene, its corresponding initiation point on the lower strand not shown is indicated at base position 231 (Fig. 2) and had been proposed from earlier nucleotide sequence analyses which concentrated on the fepA gene (28). For the portion of the

18860

Enterobactin fepA-fes Bidirectional Promoter

fepA coding region included within the pITS311 insert, no differences weredetected between the sequence now presented and that same regionshownpreviously. The singleopen reading frame of sufficient size to encode the Fes protein exists between base positions 458 and 1673. The ATG initiation codon starting at nucleotide 552 allows for an 1,122-bp open reading frame which encodesa protein of calculated M , 42,573, in good agreement with the aforementioned size estimate of Fes. The purine-rich region immediately upstream of this start site shows similarity but not striking homology to the consensus Shine-Dalgarno ribosome-binding sequence (5'-AGGAGG)(29).This initiation point was therefore verified by sequence determination of the amino-terminal 5 residues from a purified Fes-@-galactosidasehybrid protein which was produced by a fusion clone (pITS357) consisting of a 263bp Sau3A fragment encoding the first nine amino acids of Fes ligated into the BamHI site of the vector pMC1403 (30). Codon usage for fes (data not shown) resembled that for other E. coli genes whose proteins are notstrongly expressed (31),since the optimal codons representing the most abundant tRNA species wereoften not the preferred choice in this gene. The lack of a characteristic signal sequence for Fes is in accordance with all previous protein expression studies of the product (23, 32) and with its presumed cytoplasmic role in ferric enterobactin processing. Its distribution (36) of hydrophobic residues (data not shown), as well as charged amino acids, was fairly even; no membrane spanning helices were evident, although Fes may have a peripheral association with the cytoplasmic membrane, as predicted by the method of Klein et al. (33). In a deductive effort to better define the precise catalytic function of Fes, its primary sequence was assessed for structural relatedness to other known sequences by an automated homology search of the SWISS-PROT Release 6 data bank (Intelligenetics, Inc., Geneva, Switzerland); however, no significant homologies weredetected. Identification of ORFl-The final 493 bp (base positions 1505-1997) of Fig. 2 were shown previously along with the proposed entF start site at nucleotide 1894 (16). Verification of the displayed amino-terminal residues of EntF, like Fes, was accomplished by sequence determination of the first 12 residues of the purified EntF-@-galactosidasehybrid protein encoded by the fusion plasmid pITS312 (16). Definition of the precise gene boundaries for fes and entF allowed for the identification of an additional open reading frame (designated ORFl in Fig. 2) between these genes. ORFl begins at base position 1679, three nucleotides downstream from the opal stop codon of fes. It is preceded seven bases upstream within the fes gene by the potential ribosome-binding sequence, 5'AGGAG and spans the entireintercistronic region such that its terminationsignal TGA overlaps the ATG initiation codon for entF (base positions 1894-1897). It potentially encodes a 72-residue polypeptide with a calculated M , of 8271; attempts to visualize this product bygel electrophoresis havebeen unsuccessful. The transposon insertion M49 isolated previously onpITS312 (16) was shown by restriction mapping and complementation tests toreside between fes and entF-lac2 on that plasmid. Preliminary characterization of a strain containing the M49 mutation recombined into the chromosome suggests that siderophore production (or secretion) and transport are reduced but not eliminated.3An apparent reduction of enterobactin synthesis may beexplained by polarity effects on entF expression, as was noted by quantitative lacZ expression for pITS312 (16). The transport defect may result from decreased expression of the fepE gene, located downstream of G . Pettis, unpublished results.

1

2

3

4

5

G A T C

5'

FIG. 3. Primer extension of the fepA transcript. A 32Pendlabeled oligonucleotide primer corresponding to base positions 210 through 228 in Fig. 2 was annealed to total cellular RNA and extended using dNTPs and reverse transcriptase. The DNA products were separated by electrophoresis on an 8% polyacrylamide gel simultaneously with the Sanger dideoxy chain termination ladder produced with the identical primer and the fepA antisense strand of the pITS311 insert cloned into M13mp19 as the template. Bands were examined by autoradiography a t -70 "C for 9.5 h. RNA was extracted from MC4160 cells grown under high-iron ( l a n e I ) , iron-limiting ( l a n e 2), and low-iron ( l a n e 3) conditions and from MC4160 (pITS21) grown under high-iron ( l a n e 4 ) and low-iron ( l a n e 5 ) conditions. That portion of the sequence ladder pertinent to the transcription start sites is listed along with its complement. The major (heavy arrow) and minor (thin arrow) in vivo initiation points corresponding to the extended products seen are indicated. A reverse transcriptase pause site for this extension reaction is denoted by the asterisk.

entF and shown to be transcribed in the same direction: suggesting linkage to thefes-entF operon. From these initial data, a potential role for ORFl in the enterobactin system cannot be determined. Transcript Mapping for fepA and fes-entF Messages-As defined by the fepA and fes coding sequences (Fig. 2), there exists a 320-bp intercistronic region which is likely to house not only both promoters, but also certain iron-responsive elements for each transcript (16, 17). As an initial characterization, primer extension analysis was used to localize each promoter. Strain MC4160 was grown in the presence or absence of the recombinant plasmid pITS21 (which contains the complete entD, fepA, fes, and entF loci) (16), and total cellular RNA was extracted and hybridized to appropriate 5' end-labeled oligonucleotide primers. For the fepA transcript, extension from a primer corresponding to base positions 210228 in Fig. 2 resulted in the appearance of multiple bands in two distinct size ranges (Fig. 3). The lower,more intense doublet (marked with an asterisk) is interpreted as resulting from pausing of reverse transcriptase in a GC-rich region purportedly involved in a potential stem-and-loop secondary structure with a calculated free energy of -13.4 kcal (34). In conjunction with this view, the sequenceimmediately upstream of this pause area lacks homology to consensus E. coli promoter sequences (35). The upper region consists of several consecutivelyspaced bands, the twomost intense having comigrated with a T residue doublet, which corresponds to transcription start sites at 2 A residues on the opposite strand. The more 5' start appears to be the major in vivo initiation site. For the fes-entF transcript, a primer complementary to base positions 616 through 598 (in Fig. 2) was used to obtain the products shown in Fig.4, panel A . The lower band, B. A. Ozenberger, unpublished results.

Enterobactin fepA-fes A

G A T C

1

B

1

2

3

Promoter Bidirectional 4

18861

5

fes

4

A T

FIG.4. Primer extension of the fes-entF transcript. Panel A, an oligonucleotide complementary to base positions 616 through 598 in Fig. 2 was end-labeled with 32Pand used in extension reactionsas described in the legend for Fig. 3. The extendedproducts were separated by electrophoresis on an 8%polyacrylamide gel alongside the Sanger dideoxy sequence ladder generated with the same primer and template consisting of the fes antisense strand of the pITS311 insert cloned into M13mp18. Bands were examined by autoradiography a t -70 "C for 19 h. That portion of the ladder pertinent tosizing of the extended products is given along with its complement. The major (heavy arrow) and minor (thin arrow) i n vivo transcription initiation sitesare indicated. At this exposure extended productswere only evident for RNA isolated from MC4160 (pITS21) cells grown under low-iron conditions ( l a n e I). Panel B, same experiment as shown in panel A except that autoradiography was for 2.5 days a t -70 "C inthe presence of intensifying screens. RNA was from MC4160 (pITS21) grown under low-iron ( l a n e I ) or high-iron ( l a n e 2) conditions or from MC4160 cells grown under low-iron (lane 3), iron-limiting ( l a n e 4 ) , or high-iron ( l a n e 5)conditions. Arrows indicate iron-regulated bands in lanes 2and 3.

corresponding to the T residue start site, appears to be the preferred initiation point. The relative locations of the major ( T I ) and minor (T2)start sites for the fepA and fes-entF transcripts are shown in Fig. 5 . For both messages, the appearance of the identical respective bands using RNA from low-iron grown cells lacking PITS21 (Figs. 3 and 4, panel B ) ensures that these are the truechromosomal initiation points, at least under low-iron conditions. The increase in stable message under low-iron versus high-iron conditions was only evident using RNA fromplasmid-containing cells since bands were not evident even upon extended autoradiography for the experiments shown using RNA from MC4160 grown under high-iron or iron-limiting conditions. Taking into account minor differences in experimental conditions, a comparison of the relative amounts of corresponding extended products in the two separate reactions (Figs. 3 and 4) reveals that much more stable fepA transcript accumulates intracellularly than does fes-entF message. The major transcription initiation sites are separated by 18 bp (Fig. 5 ) , and each is preceded by strongly consensus -10 and -35 promoter determinants (35). A central AT-specific palindrome, 5"ATAATATTAT is present between these two start points. Within this core region, the proposed -10 sequences, both 5'-TAATAT, overlap one anotherand are centered 10 and 11bp, respectively,upstream from the major fes-entF and fepA +1sites. That this central region contains multiple potential -10 sequences is perhaps more than coincidental with the apparent indiscriminate transcription initiation observed inboth directions. The described -10 sequences are preceded 17 and 15bp, respectively, upstream by potential -35 regions each containing the most conserved

FIG. 5. Bidirectional promoter region located between the fepA and fes genes. The DNA sequence of the 320-bp intercistronic region between the fepA and fes genes (corresponding to base positions 232-551of Fig. 2) is listed in groups of 10 nucleotide base pairs. Shown flanking this sequence are the 5'-coding regions for fepA and fes, as denoted by the small open arrows and by the standard threeletter code for their respective amino-terminal aminoacid sequences. The direction of transcription for each message is indicated by the large filled arrows. Single base transcription start sites for each gene are boxed and referred to as TI or T2 where T1 is the major and T2 the minor in vivo initiation point. Potential -10 and -35 promoter determinants are indicated as such and were identified based on comparison to theconsensus E. coli promoter (35). The shaded, boxed regions resemble the consensus Fur repressor binding region (5).

fepA-fes FIG. 6. Potential iron boxes in the bidirectional promoter region.The consensus Fur-binding sequence (5) is shown with the most conserved nucleotides underlined. Four potential enterobactin iron boxes (shown highlighted in Fig. 5) are aligned with regard to the conserved base positions first and then to the overall consensus sequence. Each isnamed according to theantisense strand on which it appears. The numbers in parentheses refer to theposition of that end point (5'or 3') inrelation to themajor transcription start site (designated +1 in thiscase) for the relevant message.

base positions (35). Owing to the compactness of this promoter junction, these -35 sequences are each complementary to a portion of the opposing message. Completing the intercistronic region shown in Fig. 5 are long leader transcripts in both directions. No significant open reading frames with recognizable translational signals were detected in either. However, several potential stem-and-loop secondary structures were observed(data not shown); interestingly, the most energetically favoredones were confined to thefepA leader. The entire 1997-bp pITS311 insert sequence was searched for homology to theconsensus Fur repressor-bindingsequence ("ironbox") ( 5 ) . Taking into account the mostconserved bases, significant homology was found only in four overlapping regions (shownshaded in Fig. 5 and also listed in Fig. 6) of the bidirectional promoter. This arrangement suggests the possibility that multiple Fur interactions are involved in the iron-regulated expression of these transcripts. DISCUSSION

Subcloning and nucleotide sequence analysis specificfor the enterobactin region separating fepA and entF has led to the determination of the fes gene sequence. It has long been considered that the fes function isrequiredfor the final hydrolysis of ferric enterobactin upon its entrance into the cytoplasm of the cell, whereupon iron is necessarily made

Enterobactin fepA-fesBidirectional Promoter

18862

St" I -

Ent Dg

Fep A

A-

GCCGGATG.CG.CEGT.'GAACSCTTATCCGGCCTAC

0-

G C C T G A T G . C G ~ ~ T ~ ~ . ~ ~ A C ~ T T A T C A G G C C T C . C

Consensus

G C C ~ G A T G . C G ~ C G $ ..

. . .~ C G $ C T T A T C ~ G G C C T A C

FIG. 7. Spatial organization of the region 3' to fepA. The line drawing corresponds to the nucleotide sequence of the fepA antisense strand and approximates a region beginning within the fepA gene (designated FepA), extending through the two REP sequences (indicated by the arrows and denoted A and B ) and including a single open reading frameof over 600 bp (designated EntD*) which may correspondto the previously unidentified entD locus. The sequence to the left of the StuI site has been presented previously (28). The sequence to the right of this site was obtained with sequence analysis protocols similar to those described in the text?The exact spacing of the intercistronic region, as shown, is given in bp. Below the line drawing, the nucleotide sequences of the A and B copies are aligned above the consensus REP sequence (47). Arrows denote dyad symmetries. In their natural positions (as shown in the line drawing), these palindromes may form a potential secondary structure having a calculated free energy of -36.6 kcal as determined by the method of Tinoco et al. (34).

available for metabolic use (37-39). This enzymatic activity isolated from the soluble portion of cell extracts was given the trivial name enterochelin esterase (37); it was shown to be active in converting the cyclic triester form of either the ferrated or nonferrated forms of enterobactin to themonomer, 2,3-dihydroxybenzoylserine(15,40). Questions arose concerning the functional nature of Fes when certain synthetic structural analogues of enterobactin possessing backbones devoid of ester linkages could promote the growth of E. coli, including some that inhibited ferric enterobactin uptakeby competitive receptor binding and showed a potential dependence for the fes gene function (41-43). However, recent Mossbauer spectroscopy data demonstrated alternative intracellular fatesfor the natural versus the synthetic ligand (44). From this evidence it seems likely that upon initial internalization, ferric enterobactin is processed via an exquisitely specific pathway that is dependent on Fes activity. Still thereremain questions regarding the mechanism of iron removal. It has been demonstrated, for instance, that extracts of Bacillis subtilis are capable of causing non-hydrolytic reductive removal of iron from enterobactin analogues (45). Further understanding of the enzymatic role of Fes will require more direct studyof the purified protein. To addressthis, the plasmid construct pITS513, described in this report, has been utilized to overproduce (via T7 RNA polymerase) Fes for subsequent isolat i ~ nThe . ~ purified product can now be studied for its activity on both enterobactin and its syntheticanalogues. A genetic scheme for the observed coordinate regulation of the enterobactin system is now emerging. An important bidirectional promoter has been identified between the fepA and fes genes. Divergent, iron-regulated transcripts initiate 18 bp apart from within this control region. The clockwise transcript includes the fes gene, a 216-bp open reading frame of undetermined function, andthe entF gene. Downstream of this cistron is located the transportfunction encoded by fepE (12). The decreased transport activity presumably caused by polarity effects from the M49 insertion into ORFl between fes and entF and by the M1 insertion in entF (12), coupled with preliminary sequence dataandtranscript expression studies: suggest that fepE expression is controlled at least in part by this same large transcript. The other divergent message is known to encode the fepA gene. The biosynthetic locus entD has been localized downproduct has yet to be identified stream of fepA (23,32), but its and its direction of transcription is unknown. It was noted T. J. Brickman, unpublished results. M. McIntosh, unpublished results.

during RNA studies in this report that within a given RNA population, the amount of stable fepA transcript farexceeded the available fes-entF stable message. Many potential factors specific for expression of these messages could contribute to the observed result. These include competition for incoming RNA polymerase molecules by the overlapping promoters, attenuation-type effects, and potentialdisproportionate influence by ancillary factors such as Fur. However, selective degradative mechanisms may also influence the ratio of these messages. In this respect a region located approximately 100 bp 3' to fepA, beginning within a previously predicted (28) transcription terminator-like area, has now been identified (Fig. 7) as containing two copies of the repetitive extragenic palindromic (REP) sequence (46) in inverted orientation relative to each other. The REPsequence is a strongly conserved, remarkably ubiquitous feature of many E . coli transcriptional units. It occurs singly or in multiple arrays always with each copy in inverted orientation with respect to itsneighbor so as to maximize the number and extent of potential stem-andloop secondary structure formations (47, 48). Among its potential functions, the REP sequence has been shown directly to increase message stability of upstream genes, most likely by secondary structure formation and subsequent steric inhibition of 3'-5' exonucleases (49, 50). Interestingly, the two copies of REP identified downstream of fepA are immediately followedby a single open reading frame of over600 b ~ . ~ Should this open reading frame correspond to theentD gene, we would investigate the very interesting possibility that both fepA and entD are included on the iron-regulated transcript initiated upstream of fepA and that these REP sequences contribute to the observed stability of the 5' region of this transcript. A bidirectional control region has also been located between the divergent fepB and entC genes and is likely to influence expression of many, if not all, of the remaining enterobactin genes.8 Characterization of this control region suggests that it is quite distinct from the fepA-fes promoter junction. Several divergent, closely spaced promoters have been characterized for prokaryotes (51). Those of bacterial origin almost always involve an autoregulatory component that modulates expression of its own gene as well as those of the opposing transcript. An exception to this is found in the E . coli biotin anabolic operon. A bidirectional control region located between the genes bioA and bioB contains divergent promoters S. Armstrong, G. Pettis, L. Forrester, and M. McIntosh, manuscript in preparation. B. A. Ozenberger and T. J. Brickman, unpublished results.

Enterobactin fepA-fes Bidirectional Promoter which partially overlap (52). Amid these resides an operator for interaction with the tram-acting regulator of this system that lies a considerable distance away (7).Mutational analysis confirmed that, as a direct result of the compactness of this region, single base changes could have significant as well as distinct effects on both transcripts (52, 53). The fepA-fes promoter junction shares many similarities with its biotin counterpart. At present, none of the ent genes has demonstrated an additional autoregulatory function. Indeed, only the fur gene has been shown to mediate expression of this system in tram (3). Potential iron boxes within the overlapping promoters suggest a coordinate regulatory scenario reminiscent of the biotin system. A method has been devised that would allow for simultaneous monitoring of the potentially variable effects on divergent message expression resulting from single base regulatory mutations. A fusion of the structural p h A gene, encoding alkaline phosphatase, to fepA resulted in iron-regulated production of a hybrid protein with phosphatase activity? A plasmid was constructed containing the bidirectional promoter flanked by the fepA-phA fusion in one direction and fes and an entF-lac2 construct in the ~ t h e r Mutations .~ within the fepA-fes intercistronic junction located on this vector are now being assessed for their phenotypic effects on the quantitative expression of these enzymatically active hybrid proteins. Acknowledgments-We thank D. Pintel and K. Clemens for technical advice and helpful discussions during the course of these studies. We are grateful to J. Forrester for computer analysis of the fes sequence. We also thank J. Neuner for preparation of the manuscript.

REFERENCES 1. Neilands, J. B. (1981) Annu. Reu. Nutr. 1 , 27-46 2. Braun, V. (1985) Trends Biochem. Sci. 10,76-77 3. Hantke, K. (1981) Mol. Gen. Genet. 182,288-292 4. Bagg, A., and Neilands, J. B. (1985) J. Bacteriol. 161,450-453 5. de Lorenzo, V., Wee, S., Herrero, M., and Neilands, J. B. (1987) J.Bacteriol. 169,2624-2630 6. Bagg, A., and Neilands, J. B. (1987) Biochemistry 26,5471-5477 7. Bachmann, B. J. (1983) Microbiol. Reu. 47, 180-230 8. Young, I. G., Langman, L., Luke, R. K. J., and Gibson, F. (1971) J. Bacteriol. 106,51-57 9. Luke, R. K. J., and Gibson, F. (1971) J. Bacteriol. 107, 557-562 10. Woodrow, G. C., Young, I. G., and Gibson, F. (1975) J.Bacteriol. 124,1-6 11. Pierce, J. R., and Earhart, C. F. (1986) J. Bacteriol. 1 6 6 , 930936 12. Ozenberger, B. A., Nahlik, M. S., and McIntosh, M. A. (1987) J. Bacteriol. 169,3638-3646 13. Pugsley, A. P., and Reeves, P. (1976) J. Bacteriol. 1 2 6 , 10521062 14. Hancock, R. E.W., Hantke, K., and Braun, V. (1976) J. Bacteriol. 127, 1370-1375 15. Langman, L.,Young, I. G., Frost, G. E., Rosenberg, H., and Gibson, F. (1972) J. Bacteriol. 1 1 2 , 1142-1149 16. Pettis, G. S., and McIntosh, M. A. (1987) J. Bacteriol. 169,41544162 17. Fleming, T. P., Nahlik, M. S., and McIntosh, M. A. (1983) J. Bacteriol. 1 5 6 , 1171-1177

18863

18. Birnboim, H. C., and Doly, J. (1979) Nucleic Acids Res. 7 , 15131523 19. Maniatis, T.,Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning:A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 20. Norrander, J., Kempe, T., and Messing, J. (1983) Gene (Amst.) 26,101-106 21. Sanger, F., Nicklen, S., and Coulson, A.R. (1977) Proc. Natl. Acad. Sci. U. 5'. A . 7 4 , 5463-5467 22. Queen, C., and Korn, L. J. (1984) Nucleic Acids Res. 12,581-599 23. Fleming, T.P., Nahlik, M. S., Neilands, J. B., and McIntosh, M. A. (1985) Gene (Amst.)34,47-54 24. Laemmli, U. K. (1970) Nature 227,680-685 25. Matsudaira, P. (1987) J. Biol. Chem. 2 6 2 , 10035-10038 S., and de Crombrugghe, B. (1981) J. Biol. 26. Aiba,H.,Adhya, Chem. 256, 11905-11910 27. Fouser, L., and Friesen, J. (1986) Cell 4 5 , 81-93 28. Lundrigan, M. D., and Kadner, R. J. (1986) J. Biol. Chem. 2 6 1 , 10797-10801 29. Shine, J.. and Dalparno. - , L. (1974) . . Proc. Natl. Acad. Sci. U. S. A . 71,.1342-1346 30. Casadaban. M. J.. Chou. J. J.. and Cohen. S. N. (1980) J. Bacteriol: 143,971-980 31. Ikemura, T.(1981) J.Mol. Biol. 151, 389-409 32. Coderre, P. E., and Earhart, C. F. (1984) FEMS Microbiol. Lett. 25,111-116 33. Klein, P., Kanehisa, M., and Delisi, C. (1985) Biochim. Biophys. Acta 815,468-476 34. Tinoco, I., Borer, P. N., Dender, B., Levine, M. D., Uhlenbeck. 0. C:, Crothers, D. M., and Gralla, J. (1973) Nut. New Biol: 246,40-41 35. Hawley, D. K., and McClure, W. R. (1983) Nucleic Acids Res. 1 1 , 2237-2253 36. Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biol. 57, 105-132 37. O'Brien, I. G., Cox,G. B., and Gibson, F. (1971) Biochim. Biophys. Acta 237,537-549 38. Bryce, G. F., Weller, R., and Brot, N. (1971) Biochem. Biophys. Res. Commun. 4 2 , 8 7 1 4 7 9 39. Porra, R. J., Langman, L., Young, 1. G., and Gibson, F. (1972) Arch. Biochem. Biophys. 153,74-78 40. Greenwood, K. T.,and Luke, R. K. J. (1978) Biochim. Biophys. Acta 525,209-218 41. Hollifield, W.C., and Neilands, J. B. (1978) Biochemistry 1 7 , 1922-1928 42. Heidinger, S., Braun, V., Pecoraro, V. L., and Raymond, K. N. (1983) J. Bacteriol. 1 5 3 , 109-115 43. Ecker, D. J., Mantzanke, B. F., and Raymond, K. N. (1986) J. Bacteriol. 167,666-673 44. Mantzanke, B. F., Ecker, D. J., Yang, T.-S., Huynh, B. H., Muller, G., and Raymond, K. N. (1986) J.Bacteriol. 167,674-680 45. Lodge, J. S., Gaines, C. G., Arceneaux, J. E. L., and Byers, B. R. (1980) Biochern. Biophys. Res. Commun. 97,1291-1295 46. Higgins, C. F., Ames, G. F.-L., Barnes, W. M., Clement, J. M., and Hofnung, M. (1982) Nature 2 9 8 , 760-762 47. Stern, M. J., Ames, G. F.-L., Smith, N. H., Robinson, E. C., and Higgins, C. F. (1984) Cell 37, 1015-1026 48. Gilson, E., Clement, J. M., Brutlag, D., and Hofnung, M. (1984) EMBO J. 3,1417-1421 49. Newbury, S. F., Smith, N. H., Robinson, E. C., Hiles, I. D., and Higgins, C. F. (1987) Cell 4 8 , 297-310 50. Newbury, S. F., Smith, N. H., and Higgins, C. F. (1987) Cell 5 1 , 1131-1143 51. McClure, W. R. (1985) Annu. Reu. Biochem. 54, 171-204 52. Otsuka, A., and Abelson, J. (1978) Nature 2 7 6 , 689-694 53. Barker, D. F., Kuhn, J., and Campbell, A. M. (1981) Gene (Amst.) 13,89-102 I

.