The FecI Extracytoplasmic-Function Sigma Factor of Escherichia coli Interacts with the β′ Subunit of RNA Polymerase

Abstract

Transcription of the ferric citrate transport system of Escherichia coli K-12 is mediated by the extracytoplasmic-function (ECF) sigma factor FecI, which is activated by ferric citrate in the growth medium. By using a bacterial two-hybrid system, it was shown in vivo that FecI binds to the β′ subunit of RNA polymerase. The inactive mutant protein FecI(K155E) displayed reduced binding to β′, and small deletions along the entire FecI protein led to total impairment of β′ binding. In vitro, FecI was retained on Ni²⁺-nitrilotriacetic acid agarose loaded with a His-tagged β′_1-313 fragment and coeluted with β′_1-313. Binding of FecI to β′ and β′_1-313 was enhanced by FecR_1-85, which represents the cytoplasmic portion of the FecR protein that transmits the inducing signal across the cytoplasmic membrane. Interaction of FecR with FecI was demonstrated by showing that isolated FecR inhibited degradation of FecI by trypsin. This is the first demonstration of binding of an ECF sigma factor of the FecI type to the β′ subunit of RNA polymerase and of binding being enhanced by the protein that activates the ECF sigma factor.

The ferric citrate transport system of Escherichia coli K-12 transports iron into cells. Binding of ferric citrate to the FecA outer membrane protein initiates a signal transduction mechanism that induces transcription of the fecABCDE transport genes. In the FecA crystal structure, (Fe³⁺-citrate)₂ is bound to FecA (12). In addition, FecA transports (Fe³⁺-citrate)₂ across the outer membrane (6, 18, 25). FecA with bound ferric citrate transmits the signal to FecR (2), an inner membrane protein with the N-terminal domain (residues 1 to 84) located in the cytoplasm and the C-terminal domain (residues 101 to 317) located in the periplasm (46). Residues 85 to 100 form a transmembrane segment through which the signal is transmitted across the cytoplasmic membrane. FecR interacts in the cytoplasm with the FecI sigma factor and activates FecI, which in turn directs the RNA polymerase core enzyme to the fecA promoter and initiates transcription of the fecABCDE transport genes (1, 10). Transcription of the fecIR regulatory genes located upstream of fecA and of the fecABCDE genes is regulated by the intracellular concentration of iron (2, 36). Under iron-replete conditions, the Fur repressor protein loaded with Fe²⁺ binds to the fecI and fecA promoters and represses transcription. When iron is limiting, FecA, FecI, and FecR are synthesized, and (Fe³⁺-citrate)₂ induces transcription of the fecABCDE genes. Under noninducing conditions, there is enough FecA to initiate the signaling cascade.

In previous studies, we determined in vitro and in vivo binding of FecA to FecR and of FecR to FecI. In the derived signal transduction model, the N terminus of FecA interacts with the C-terminal domain of FecR, and the N terminus of FecR binds to region 4 of FecI (11, 29, 39). A mutant FecA lacking the N-terminal domain transports ferric citrate but shows no induction activity (25). With FecR_1-85, the fec transport genes are constitutively transcribed, and point mutations in FecR_1-85 fail to activate FecI (34, 39, 46).

FecI belongs to the extracytoplasmic-function (ECF) sigma factors, which represent a subgroup of the σ⁷⁰ family (14, 20, 26, 31, 47). The RNA polymerase of E. coli is a multisubunit enzyme and consists of two functional forms, the core enzyme and the holoenzyme. The core enzyme is formed by the subunits β′, β, and α₂ and performs transcription elongation. One out of seven different σ subunits of E. coli binds to the core enzyme and forms the holoenzyme (28), which specifically recognizes promoters and initiates transcription. Of all these sigma factors, σ⁷⁰ and its binding to RNA polymerase and parts of the core enzyme have been characterized in the most detail (7). Recent studies have revealed a tight binding of σ⁷⁰ to β′_260-309 and to several additional but weaker binding sites of β and β′ (3, 4, 5, 15, 24, 27). Determination of the crystal structures of the holoenzymes of Thermus aquaticus and Thermus thermophilus and the crystal structure of the initiating form of the enzyme of T. aquaticus have confirmed these findings (32, 33, 45). The major interface between σ and the RNA polymerase core enzyme is formed between the β′ coiled coil and the 2.2 helical region of σ. Since FecI belongs to the σ⁷⁰ group but displays unusual features of activation by FecR, we studied the interaction of FecI with β′ and report, to our knowledge for the first time, binding of an ECF sigma factor to the β′ subunit and a β′ fragment.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The E. coli strains and plasmids used in this study are listed in Table 1. Cells were grown in tryptone-yeast extract (TY) medium or nutrient broth (NB) as previously described (30). When required, antibiotics were added to media at the following concentrations: ampicillin, 50 μg/ml; chloramphenicol, 40 μg/ml; and tetracycline, 12 μg/ml.

TABLE 1.

E. coli strains and plasmids

Strain or plasmid	Description	Reference or source
Strains
DH5α	endA1 hsdR17(rK− mK+) supE44 thi1 recA1 gyrA relA1 Δ(lacZYA-argF) U169 φ80ΔlacZM15	16
AB2847	aroB tsx malT thi	17
41/2	AB2847 cir fepA fhuA412	17
ZI418	araD139 ΔlacU169 rpsL 150 relA1 flbB5301 deoC1 ptsF25 rbsR thi aroB fecB::Mud1 (Ap lac)	44
BL21 (DE3)	F−hsdS galT; phage T7 polymerase under lacUV5 control	40
SU202	lexA71::Tn5 sulA211 sulA(op408/op+)::lacZ Δ(lacIPOZYA)169/F′lacIqlacZΔM15::Tn9	9
Plasmids
pHSG576	pSC101 derivative, Cmr	42
pBCKS+	ColE1 Cmr	Stratagene
pSV66	pHSG576 fecI fecR fecA	18
pT7-7	Phage T7 promoter, ori (ColE1) Apr	41
pMS604	ori ColE1 TetrlexA1-87-fos zipper	9
pDP804	ori p15A AmprlexA1-87408-jun zipper	9
pET19b	Phage T7 promoter; N-terminal His tag; ori (ColE1) Apr	Novagen
pHisβ′	pET19b rpoC	This study
pHisβ′313	pET19b rpoC1-313	This study
pAA73	pT7-7 fecI	1
pAA70	pT7-7 fecR	48
pR85	pT7-7 fecR1-85	This study
pHCβ′313	pBCKS+rpoC1-313	This study
pMMO43	pHSG576 fecI(E141A) fecR	35
pMMO44	pHSG576 fecI(K145E) fecR	35
pMMO45	pHSG576 fecI(K155E) fecR	35
pSM173	pMS604 lexA1-87-fecI	10
pSM71	pMS604 lexA1-87-fecI(E141A)	This study
pSM72	pMS604 lexA1-87-fecI(K145E)	This study
pSM73	pMS604 lexA1-87-fecI(K155E)	This study
pSM123	pMS604 lexA1-87-fecI fecR1-85	This study
pSM124	pMS604 lexA1-87-fecI fecR	This study
pSM112	pMS604 lexA1-87-fecR1-85	This study
pSM115	pMS604 lexA1-87-fecR	This study
pSM120	pDP804 lexA1-87-rpoC	This study
pSM84	pDP804 lexA1-87408-fecR	29
pSM85	pDP804 lexA1-87408-fecR1-85	29
pSM20	pMS604 lexA1-87-fecI(Δ9-32)	29
pSM21	pMS604 lexA1-87-fecI(Δ32-55)	29
pSM22	pMS604 lexA1-87-fecI(Δ52-67)	29
pSM23	pMS604 lexA1-87-fecI(Δ65-83)	29
pSM24	pMS604 lexA1-87-fecI(Δ79-114)	29
pSM25	pMS604 lexA1-87-fecI(Δ114-134)	29
pSM26	pMS604 lexA1-87-fecI1-132	29

Construction of plasmids.

rpoC (the gene encoding β′) and truncated rpoC fragments were synthesized by PCR. Plasmid pSM120 was obtained by using the primers rpoC_XhoI_for (5′-CGGGAGCCTCGAGGTGAAAGATTTATTAAAG-3′) and rpoC_Sa_rev (5′-GCGCTCAACCATTTTCTTGAGCTCTTAAATGGTGGTAGCAAGACC-3′). The resulting rpoC fragment was digested with XhoI and SacI and cloned into XhoI- and SacI-restricted pMS604. Plasmid pHisβ′ was created by using the primers rpoCNdeI_for (5′-CGGGAGCAAACATATGAAAGATTTATTAAAGTTTC-3′) and rpoC1407_BamHI (5′-GCGGATTAAGGATCCCTCGTTATCAGAACCG-3′). Plasmid pHisβ′313 was constructed by using the primers rpoCNdeI_for (5′-CGGGAGCAAACATATGAAAGATTTATTAAAGTTTC-3′) and rpoC_313-BamHI (5′-GAACCGGTGGATCCTTAACCGCGACGACCG-3′). Each of the rpoC fragments was cloned into pET19b (Novagen) restricted with NdeI and BamHI. Plasmid pHCβ′313 was created by using b′PromSacI_for (5′-GTATCAACATCGCCGCGGAAGACGAGTAATTC-3′) and b′313BamHI_rev (5′-GTTAGAACCGGTGGATCCTTAACCGCGACGACCG-3′), and the rpoC fragment was cloned into SacI- and BamHI-restricted pBCKS⁺. For construction of plasmids pSM120, pHisβ′, pHisβ′313, and pHCβ′313, the chromosomal DNA of strain E. coli DH5α was used as the DNA template.

fecI mutations in region 4 were synthesized by PCR with the primers FecIBstEII (5′-GATGCAGGTGACCATGTCTGACCGCGCC-3′) and FecIPstI (5′-GGTTAACACTGCAGCGAAAGCAGAAACGC-3′), and pMMO43, pMMO44, and pMMO45 were used as the DNA templates. The resulting fecI fragments were digested with BstEII and PstI and ligated into BstEII- and PstI-cleaved pSM604, yielding plasmids pSM71, pSM72, and pSM73, respectively.

fecIR_1-85 was amplified by PCR with the primers FecIBstEII (5′-GATGCAGGTGACCATGTCTGACCGCGCC-3′) and FecR85_PstI (5′-GCCGAGCAACTGCAGTTATCCTTTCATC-3′), and plasmid pSV66 was used as the DNA template. The fecIR_1-85 fragment was cloned into BstEII- and PstI-digested pMS604, resulting in plasmid pSM123.

fecIR was synthesized by PCR with primers FecI_XhoI (5′-CGCGAAAGCCAACTCGAGACCCTACAAC-3′) and FecR317_XhoI (5′-GAATTACTCGAGTTACAGTGGTGAAATG-3′), and plasmid pSV66 was used as the DNA template. The fecIR fragment was digested with XhoI and ligated into XhoI-restricted pSM173, resulting in plasmid pSM124.

Plasmid pSM112 was constructed by PCR amplification with primers FecR_BstEII (5′-GTTCCGTCTGGAGTGGTGACCATGAATCCTTTGTTAACC-3′) and FecR85_XhoI (5′-GCCGAGCAACTCGAGTAATCATTTCATTTCATCACGTGACC-3′). Plasmid pSV66 was used as the DNA template. The BstEII/XhoI fecR_1-85 PCR fragment was ligated into BstEII- and XhoI-cleaved pMS604.

Plasmid pSM115 was amplified with primers FecR_BstEII (5′-GTTCCGTCTGGAGTGGTGACCATGAATCCTTTGTTAACC-3′) and FecR_XhoI (5′-GAAATAAGAATTACTCGAGTTACAGTGGTGAAATGTTTATC-3′), and plasmid pSV66 was used as the DNA template. The fecR fragment was digested with BstEII and XhoI and cloned into BstEII- and XhoI-cleaved pMS604.

To obtain plasmid pR85, the NdeI/HindIII fragment of fecR_1-85 from plasmid pSM10 was cloned in the vector pT7-7, cleaved with NdeI and HindIII.

Recombinant DNA techniques.

Standard techniques (37) or the protocols of the suppliers were used for the isolation of plasmid DNAs, digestion with restriction endonucleases, ligation, transformation, and agarose gel electrophoresis. PCR amplification was carried out with Taq polymerase (Qiagen, Hilden, Germany) under standard conditions. DNA was initially denatured by heating to 94°C for 3 min. This was followed by 30 cycles consisting of denaturation at 94°C for 1 min, annealing at 54°C for 2 min, and extension at 72°C for 5 min. All DNA constructs were sequenced by the dideoxy chain termination method (38) on an A.L.F. DNA sequencer (Pharmacia Biotech, Freiburg, Germany) with the AutoRead Sequencing kit (Pharmacia Biotech).

Expression and purification of proteins.

Plasmids were introduced into E. coli BL21(DE3) by transformation for high-level expression. Cells were grown at 37°C in 200 ml of TY medium with either 50 μg of ampicillin/ml or 40 μg of chloramphenicol/ml. The cultures were grown to an optical density at 578 nm of 0.6 to 0.8, and then 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added. After further incubation for 2.5 h, cells were harvested by centrifugation and suspended in binding buffer (20 mM Tris-HCl, 0.5 M NaCl, 10 mM imidazole [pH 7.9]). Crude cell extracts were obtained by sonicating the cells three times for 60 s each. The truncated β′_1-313 fragment and FecR_1-85 did not form inclusion bodies.

FecR inclusion bodies were solubilized by incubation for 1 h at 25°C in binding buffer supplemented with 6 M urea. Undissolved material was removed by centrifugation (30,000 × g for 20 min), and the supernatant fractions were dialyzed against binding buffer supplemented with 1 M urea. Precipitated material was removed by centrifugation (30,000 × g for 20 min).

FecI inclusion bodies were purified as described previously (1). FecI was solubilized from inclusion bodies by incubation for 1 h at 25°C in binding buffer supplemented with 2 mg of N-dodecyl-N,N-dimethyl-3-ammonio-1-propanesulfonate per ml. Undissolved cell debris was removed by centrifugation (30,000 × g for 30 min).

His₁₀-tagged β′ inclusion bodies were solubilized by a procedure similar to that of Arthur and Burgess (4). Briefly, the inclusion bodies were solubilized in buffer C (20 mM Tris-HCl [pH 7.9], 0.5 M NaCl, 5 mM imidazole, 0.1% [vol/vol] Tween 20, and 10% glycerol) with 8 M urea. The protein solution was loaded onto Ni²⁺-nitrilotriacetic acid (NTA) resin. The column was washed with 10 bed volumes of buffer C containing 8 M urea, followed by 10 bed volumes of buffer C (without urea) to allow refolding. The bound proteins were eluted with 2 bed volumes of buffer C containing 250 mM imidazole.

Binding assays.

The solution containing β′_1-313 was loaded onto an Ni²⁺-NTA agarose column previously equilibrated with 10 bed volumes of binding buffer. Solutions containing FecI, FecR_1-85, or FecI and FecR_1-85 were applied to the column. After two wash steps with 10 bed volumes of wash buffer (20 mM Tris-HCl [pH 7.9], 0.5 M NaCl, 20 mM imidazole), bound fusion proteins were eluted with 2 bed volumes of elution buffer (20 mM Tris-HCl [pH 7.9], 0.5 M NaCl, 250 mM imidazole). Samples from the flowthrough, wash, and elution fractions were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Western blotting.

FecI was detected by Western blot analysis with anti-FecI antibodies. After electrophoresis, the proteins were electroblotted onto nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany). Blots were blocked overnight in 3% bovine serine albumin in TNT buffer (20 mM Tris-HCl [pH 7.5], 500 mM NaCl, 0.05% Tween 20), probed with the anti-FecI antibodies, washed with TNT buffer, and incubated with anti-rabbit immunoglobulin G conjugated with alkaline phosphatase (Sigma-Aldrich Chemie GmbH, Deisenhofen, Germany). The blots were developed with NBT-BCIP (nitroblue tetrazolium-5-bromo-4-chloro-3-indolylphosphate) (Serva, Heidelberg, Germany).

Trypsin cleavage.

Different amounts of trypsin (0.12, 0.25, 0.5, and 1 μg) were added to 43 μg of purified FecI and 49 μg of purified FecR. Digestion was carried out in 20 mM Tris-HCl (pH 8.0) at 25°C for 20 min, after which the samples were analyzed by SDS-PAGE.

Determination of β-galactosidase activity.

β-Galactosidase activity was determined as described by Miller (30) and Giacomini et al. (13). To determine the induction level, cells were grown in NB either with no additions or supplemented with 50 μM 2,2′-dipyridyl or 1 mM citrate. For the LexA-based repression system, cells were grown in TY medium supplemented with 1 mM IPTG.

RESULTS

FecI interacts with the β′ subunit of RNA polymerase.

The bacterial LexA-based two-hybrid system was used to examine in vivo heterodimerization between β′ and FecI. LexA is a transcriptional repressor that binds as a homodimer to the sulA promoter. It consists of an N-terminal DNA binding domain and a C-terminal dimerization domain. To determine heterodimerization, the C-terminal domain can be replaced by dimerization domains of other proteins. To prevent homodimerization of LexA hybrid proteins, the promoter of sulA is mutated such that wild-type LexA binds to one site and mutated LexA408 binds to the other site. Dimerization of the hybrid proteins was assessed by repression of chromosomal P_sulA::lacZ transcription of the reporter strain E. coli SU202. The LexA_1-87408 DNA binding domain was fused to the N terminus of β′, and the LexA_1-87 DNA binding domain was fused to the N terminus of FecI.

E. coli SU202 was transformed with plasmids carrying the lexA_1-87-fecI and lexA_1-87408-rpoC fusion genes. LexA_1-87-FecI combined with LexA_1-87408-Fos-Zipper served as a negative control. Compared to LexA_1-87-FecI/LexA_1-87408-Fos-Zipper, LexA_1-87-FecI/LexA_1-87408-β′ repressed P_sulA::lacZ transcription to an extent that β-galactosidase activity was reduced to 37%, which indicated heterodimer formation of the two proteins (Table 2).

TABLE 2.

Binding of wild-type and mutated FecI to the β′ subunit

Plasmids	Proteinsa	β-Galactosidase activity (Miller units)b
pSM173, pDP804	LexA1-87WT-FecI, LexA1-87408-Jun-zipper	243
pSM173, pSM120	LexA1-87WT-FecI, LexA1-87408-β′	92
pSM73, pSM120	LexA1-87WT-FecI(K155E), LexA1-87408-β′	165
pSM71, pSM120	LexA1-87WT-FecI(E141A), LexA1-87408-β′	115
pSM72, pSM120	LexA1-87WT-FecI(K145E), LexA1-87408-β′	88
pSM20, pSM120	LexA1-87WT-FecI(Δ9-32), LexA1-87408-β′	232
pSM21, pSM120	LexA1-87WT-FecI(Δ32-55), LexA1-87408-β′	265
pSM22, pSM120	LexA1-87WT-FecI(Δ52-67), LexA1-87408-β′	189
pSM23, pSM120	LexA1-87WT-FecI(Δ65-83), LexA1-87408-β′	224
pSM24, pSM120	LexA1-87WT-FecI(Δ79-114), LexA1-87408-β′	216
pSM25, pSM120	LexA1-87WT-FecI(Δ114-134), LexA1-87408-β′	269
pSM26, pSM120	LexA1-87WT-FecI(Δ133-173), LexA1-87408-β′	170

^aWT, wild type

^bDetermined by using the bacterial two-hybrid LexA-based system in E. coli SU202 sulA-lacZ.

Inactive FecI mutants display weak binding to β′.

To examine a correlation between FecI activity and FecI binding to β′, previously isolated FecI point and deletion mutants (29, 35) were tested for binding to β′. The K155E amino acid replacement in the predicted second helix of FecI reduced fecB-lacZ transcription to 10% of wild-type activity, whereas the E141A and K145E replacements in the first helix did not alter ferric citrate-induced fecB-lacZ transcription (35). The FecI deletion mutants showed no fecB-lacZ transcription (29). The FecI mutant proteins were fused to LexA_1-87, and β′ was fused to LexA_1-87408. FecI(K155E)-β′ repressed sulA-lacZ transcription only to 51% of wild-type FecI, whereas FecI(E141A)-β′ repressed to 84% and FecI(K145E)-β′ repressed to 100%. The FecI deletion mutants showed no or low repression activity (Table 2). The data demonstrate a correlation between FecI sigma factor activity and binding of FecI to β′.

FecR_1-85 enhances binding of FecI to β′.

FecR activates FecI in response to (Fe³⁺-citrate)₂ in the growth medium (34). The cytoplasmic FecR_1-85 fragment induces FecI-mediated fecA and fecB transcription constitutively in the absence of ferric citrate (34, 39). To examine the effect of FecR_1-85 on the binding of FecI to β′, fecR_1-85 was cloned downstream of lexA_1-87-fecI. In the presence of FecR_1-85, sulA-lacZ transcription by LexA_1-87-FecI combined with LexA_1-87408 β′ was more strongly repressed than in the absence or FecR_1-85 (51 versus 90 β-galactosidase units) (Table 3). In contrast, repression of sulA-lacZ in the presence of complete FecR was lower than that in the absence of FecR, which suggests that FecR in cells uninduced by ferric citrate inhibits binding of FecI to β′. Induction by ferric citrate could not be measured, since E. coli SU202, in which sulA-lacZ transcription was determined, for unknown reasons did not respond to ferric citrate.

TABLE 3.

Binding of wild-type FecI to the β′ subunit in the presence of FecR_1-85 and FecR and of mutated FecI to FecR_1-85

Plasmids	Proteinsa	β-Galactosidase activity (Miller units)b
pSM173, pSM120	LexA1-87WT-FecI, LexA1-87408-β′	90
pSM123, pSM120	LexA1-87WT-FecI, LexA1-87408-β′, FecR1-85	51
pSM124, pSM120	LexA1-87WT-FecI, LexA1-87408-β′, FecR	125
pSM112, pSM120	LexA1-87WT-FecR1-85, LexA1-87408-β′	223
pSM115, pSM120	LexA1-87WT-FecR, LexA1-87408-β′	242
pSM173, pSM85	LexA1-87WT-FecI, LexA1-87408-FecR1-85	38
pSM73, pSM85	LexA1-87WT-FecI(K155E), LexA1-87408-FecR1-85	49
pSM71, pSM85	LexA1-87WT-FecI(E141A), LexA1-87408-FecR1-85	50
pSM72, pSM85	LexA1-87WT-FecI(K145E), LexA1-87408-FecR1-85	51

^aWT, wild type.

^bDetermined by using the bacterial two-hybrid LexA-based system in E. coli SU202 sulA-lacZ.

To examine whether FecR_1-85 binds to β′ and affects binding of FecI to β′, FecR_1-85 was fused to wild-type LexA_1-87 and combined with LexA_1-87408-β′. The assay revealed no binding of FecR_1-85 to β′ (Table 3). The same result was obtained with LexA_1-87-FecR containing complete FecR combined with LexA_1-87408-β′ which did not repress sulA-lacZ transcription (Table 3).

The specificity of the reduction of FecI(K155E) and FecI(E141A) binding to β′ was tested by combining the LexA-FecI missense hybrid proteins with the LexA-FecR_1-85 hybrid protein. The mutations in FecI only slightly affected interaction with FecR_1-85 mutants, as the comparison to wild-type FecI demonstrates (Table 3). Therefore, the reduced sigma factor activity of FecI(K155E) is caused by weak binding to β′ and not by an impaired interaction with FecR.

Binding of FecI to (His)₁₀-β′_1-313 on an Ni²⁺-NTA agarose column.

In vitro Ni²⁺-NTA coimmobilization assays were used to confirm and extend the results obtained in vivo with the two-hybrid system. β′ and a C-terminally truncated β′ fragment, consisting of residues 1 to 313, were each fused to His₁₀ purification tags. The binding of FecI to (His)₁₀-β′ could not be tested since (His)₁₀-β′ did not bind to the Ni²⁺-NTA agarose column, possibly because (His)₁₀-β′ was solubilized from inclusion bodies in binding buffer supplemented with 8 M urea. In contrast, (His)₁₀-β′_1-313 was soluble in the binding buffer without urea and bound to the column. FecI solubilized from inclusion bodies in binding buffer with N-dodecyl-N,N-dimethyl-3-ammonio-1-propanesulfonate was active, as shown by fecA promoter DNA band shift assays with FecI and RNA polymerase (1).

FecI was applied to the column loaded with (His)₁₀-β′_1-313. The column was washed, and the bound proteins were eluted with imidazole. If the truncated β′_1-313 contained the interaction domain for FecI, it would bind FecI, and FecI would then coelute with (His)₁₀-β′_1-313. After a fraction of FecI eluted in the flowthrough and in the first wash from the column, the FecI protein coeluted with β′_1-313 (Fig. 1A). Since the FecI band was barely detectable in the elution fraction after staining, FecI was identified with a FecI antiserum. As shown in Fig. 1D, FecI clearly coeluted with (His)₁₀-β′_1-313.

FIG. 1.

(A to C) Binding of FecI to β′_1-313 (A), of FecI and FecR_1-85 to β′_1-313 (B), and of FecR_1-85 to β′_1-313 (C). The assay was carried out on an Ni²⁺-NTA column, to which (His)₁₀-β′_1-313 was bound. After the column was washed, the proteins were eluted and the fractions were analyzed by SDS-PAGE (15% polyacrylamide). Proteins in the gel were stained with Serva blue. (D) FecI was identified by Western blotting with FecI antiserum.

The in vivo two-hybrid system revealed enhancement of FecI binding to β′ by FecR_1-85. To support this finding, binding of FecI to β′_1-313 in the presence of FecR_1-85 was tested. FecI and FecR_1-85 were mixed and then applied together to the column loaded with (His)₁₀-β′_1-313. Small amounts of FecI and FecR_1-85 were detected in the flowthrough and the first wash (Fig. 1B). FecR_1-85 coeluted with FecI and (His)₁₀-β′_1-313, and more FecI interacted with β′_1-313 in the presence than in the absence of FecR_1-85. FecR_1-85 alone did not bind to (His)₁₀-β′_1-313 fixed to Ni²⁺-NTA agarose (Fig. 1C). These results demonstrate that FecI interacts specifically with the C-terminally truncated β′_1-313-fragment and indicate that FecR_1-85 enhances binding of FecI to (His)₁₀-β′_1-313.

FecR stabilizes FecI.

FecI in the presence and absence of FecR was digested with trypsin to determine whether FecR binding to FecI alters trypsin sensitivity of FecI by covering trypsin-sensitive sites or by induction of a conformational change in FecI. Purified FecR and FecI were treated with increasing amounts of trypsin. The samples were analyzed by SDS-PAGE, followed by immunoblotting with polyclonal antibodies directed against FecI. One microgram of trypsin per sample caused complete cleavage of FecI in the absence of FecR (Fig. 2, lane 5). Less FecI was degraded in the presence of FecR. FecR was stable at a trypsin concentration of 1 μg per sample (data not shown).

FIG. 2.

Degradation of FecI by trypsin in the presence and absence of FecR. FecI (43 μg) and FecR (49 μg) were incubated for 20 min at 25°C with increasing amounts of trypsin: 0 μg (lanes 1), 0.12 μg (lanes 2), 0.25 μg (lanes 3), 0.5 μg (lanes 4), and 1 μg (lanes 5). FecI was identified by Western blotting with a FecI antiserum.

To eliminate the possibility that addition of FecR increased the amount of protein beyond the level degraded by trypsin under the conditions used, a twofold-higher concentration of FecI was treated with trypsin. The results of Western blot analysis were the same as when smaller amounts of FecI were used (data not shown).

Excess β′_1-313 reduces transcription initiation.

If β′_1-313 interacts with FecI, a surplus of β′_1-313 should reduce transcription initiation of the fec transport genes. In an in vivo titration experiment, transcription was measured in E. coli ZI418, which carries a chromosomal fecB-lacZ fusion. E. coli ZI418 was transformed with pHCβ′313, comprising the gene encoding β′_1-313 cloned on the high-copy-number vector pBCKS⁺. Cells were grown in NB supplemented with 1 mM citrate for induction. The β-galactosidase activity decreased from 143 Miller units with the vector alone to 65 U with pHCβ′313. To rule out that β′_1-313 inactivated β-galactosidase, E. coli 41/2, which contains a lacZ gene, was transformed with pHCβ′313. Cells were grown in NB supplemented with 1 mM IPTG for induction. The β-galactosidase activity was 699 Miller units with the strain alone, 680 U with the vector, and 658 U with pHCβ′313. These data suggest that binding of β′_1-313 to FecI inhibits the transcription of the fec transport genes by competition with FecI binding to β′ of RNA polymerase.

DISCUSSION

The bacterial two-hybrid system revealed in vivo binding of FecI to the β′ subunit of RNA polymerase. β-Galactosidase activity was reduced to 30% of the unrepressed level when the LexA-FecI and the LexA-β′ hybrid proteins were combined. Interaction of FecI with β′ was also shown by competition experiments in which overexpressed β′_1-313 reduced ferric citrate-induced transcription of fecB-lacZ to less than half the level in cells lacking plasmid-encoded β′_1-313. Specificity of the interaction was tested with FecI point and deletion mutants. In mutant FecI(K155E), mutated in the second predicted helix of region 4.2 (29), fecB-lacZ transcription was reduced to 10% of the wild-type level in ferric citrate-induced cells (35); as a LexA hybrid protein in combination with LexA-β′, it repressed transcription of sulA-lacZ only to 30% of the wild-type FecI level. FecI(E141A) and FecI(K145E), which showed 91 and 99% of FecI wild-type fecB-lacZ transcription, respectively (35), repressed sulA-lacZ transcription to 78 and 104% of the wild-type FecI level, respectively. All of the FecI deletion mutants that are inactive as sigma factors (29) failed to interact with β′. Within the accuracy of these experiments, FecI binding to β′ reflects the activities of the FecI sigma factor. These findings are consistent with our previously published data showing that purified FecI causes band shifting of a DNA fragment consisting of the fecA −35 and −10 promoter fragment only when RNA polymerase core enzyme is added. FecI or RNA polymerase core enzyme alone did not retard the electrophoretic mobility of the DNA fragment (1).

The in vivo data for binding of FecI to β′ were supported by the in vitro data. In these experiments, the β′_1-313 fragment was used, since it remained soluble after disruption of the cells whereas β′ could be isolated only as an inclusion body. His-tagged β′_1-313 bound to Ni²⁺-NTA agarose retained FecI on the column, and FecI was coeluted with His-tagged β′_1-313.

In vivo binding of FecI to β′ is enhanced by FecR_1-85.

In a sense, FecR_1-85 mimics activated FecR since it causes FecI-meditated constitutive transcription of the fecABCDE transport genes, in contrast to FecR, which requires the signal from (Fe³⁺-citrate)₂-bound FecA to activate FecI (34, 39). FecR in uninduced cells decreased interaction of FecI with β′. Enhancement of FecI interaction with β′ by FecR in ferric citrate-induced cells was not studied, since the E. coli SU202 test strain did not respond to ferric citrate.

FecR_1-85 also increased in vitro binding of FecI to His-tagged β′_1-313 on Ni²⁺-NTA agarose. FecR_1-85 was coeluted together with His-tagged β′_1-313 and FecI from the Ni²⁺-NTA agarose, which indicated that it stayed bound to the FecI-β′ complex on the column during the washing procedure. This finding opens the possibility that in cells FecR_1-85 stays bound to FecI when FecI interacts with the RNA polymerase core enzyme. Only a fraction of FecI solubilized by detergent treatment of FecI inclusion bodies is active, as has been shown previously by DNA band shift experiments (1). In addition, sigma factors are very sensitive to proteolysis because they are most likely in an unstructured form (21); this could explain why sigma factors alone cannot be crystallized, whereas cocrystallization with the holoenzyme is possible (32, 33, 45). The sigma factor FecI might also exist in an unstructured form, and binding of FecR_1-85 could cause formation of an ordered FecI structure. Degradation of FecI by trypsin and inhibition of trypsin degradation by FecR support this conclusion.

Two crystal structures of the holoenzyme of bacterial RNA polymerase have been identified, one at 2.6-Å resolution (T. thermophilus) (45) and one at 4-Å resolution (T. aquaticus) (32, 33). These RNA polymerases show a high degree of similarity in sequence and structure, which implies a similar functional mechanisms (21). Regions 2 to 4 of the sigma factors bind to the surface of the RNA polymerase core enzyme along the upper half of the active-site cleft. Our findings that deletions from region 2 to 4 prevent FecI from interacting with β′ and that FecI(K155E), with a low activity as a sigma factor, displayed low binding to β′ suggest that FecI occupies a position on the RNA polymerase core enzyme similar to that of σ⁷⁰. The crystal structures further reveal that the subregions of the sigma factors are organized into globular domains and are well separated from each other. Therefore, FecR may bind to region 4 of FecI (11, 29) without disturbing binding of FecI to the surface of the RNA polymerase core enzyme.

The regulation of several sigma factors and their cognate anti-sigma factors has been characterized (8, 19, 20, 22, 23, 43). Upon receiving a stimulus from the environment, the sigma factor is released and binds to RNA polymerase to stimulate transcription. The binding of FecR to region 4 of FecI concurs with the binding of anti-sigma factors to binding sites on sigma factors. However, in contrast to anti-sigma factors that inhibit the cognate sigma factors, FecR is required for FecI sigma factor activity. In the absence of FecR, the activity of FecI is very low. FecI is converted to an active sigma factor in response to ferric citrate in the growth medium. The binding of (Fe³⁺-citrate)₂ to FecA causes a structural change in FecA, which is seen in the crystal structure of FecA occupied with (Fe³⁺-citrate)₂ compared to unoccupied FecA (12). Large movements are observed in the region exposed to the periplasm to which the segment that is essential for induction (residues 1 to 79) is linked (25) and which mediates interaction with FecR (11). The structure of this FecA segment is not resolved in the crystal analysis, presumably because it is flexible. The initial signal is a structural change in FecA. The altered FecA structure interacts with the periplasmic portion of FecR (11), which probably causes a structural change in FecR. FecR transmits the signal across the cytoplasmic membrane. The N-terminal cytoplasmic segment of FecR binds to region 4 of FecI (11, 29) and converts FecI to an active sigma factor.

Two models are proposed for FecI activation by FecR. In the first model, FecR causes a conformational change in FecI that enhances binding of FecI to RNA polymerase core enzyme. This model is supported by data obtained here with FecR_1-85, which is a soluble cytoplasmic form of FecR and reflects activated FecR. RNA polymerase holoenzyme then binds to the fecA promoter and initiates transcription of the fecABCDE transport genes. This model includes the possibility that activated as well as inactivated FecI is bound to FecR, and therefore, FecI and the RNA polymerase core enzyme are bound via FecR to the cytoplasmic membrane while transcription is initiated. This notion is supported by the finding that FecR_1-85 is found along with FecI bound to His-tagged β′ on Ni²⁺-NTA agarose. In the second model, FecR binds FecI, keeps FecI in solution, and protects FecI from protease degradation, as suggested by the trypsin degradation data. In the absence of a signal, FecR inhibits FecI sigma factor activity. After receiving the signal from ferric citrate-loaded FecA, FecR changes conformation and in turn may change the conformation of FecI, and FecI dissociates from FecR and directs the RNA polymerase core enzyme to the fecA promoter.