Analysis of the Ferric Citrate Transport Gene Promoter of Escherichia coli

Abstract

FecI, an extracytoplasmic-function σ factor, is required for initiation of transcription of the ferric citrate transport genes. A mutational analysis of the fecA promoter revealed that the nonconserved −10 region and a downstream regulatory element are important for fecA promoter activity. However, nucleotide substitutions in the well-conserved −35 region also have an effect on the fecA promoter activity. Titration of FecI suggests that the FecI-RNA polymerase holoenzyme does not bind strongly to the downstream regulatory element, which is therefore probably involved in a subsequent step of transcription initiation.

In Escherichia coli, transcription of the ferric citrate transport genes fecABCDE is controlled by a signal transduction mechanism that starts at the cell surface. (Fe³⁺ citrate)₂ binds to the outer membrane protein FecA and without further transport into the cell induces transcription of the fec transport genes (9, 14). The signal transmitted by FecA loaded with (Fe³⁺ citrate)₂ across the outer membrane is transmitted across the cytoplasmic membrane by FecR, a transmembrane protein (21, 28) that interacts with FecA in the periplasm and with the FecI extracytoplasmic-function (ECF) σ factor in the cytoplasm (7, 8, 18, 24). FecR enables FecI to bind to the RNA polymerase core enzyme; this complex then binds to the fecA promoter to initiate transcription of the fec transport genes (1, 2). The only promoter known to be recognized by FecI is that of the fecABCDE operon; no other σ factor of E. coli is endowed with such a narrow promoter specificity (17).

ECF σ factors belong to a subfamily of the σ⁷⁰ class, based on their sequence conservation and function across bacterial species (3, 10, 16, 20). Comparisons of sequences indicate that the genomes of Caulobacter crescentus, Pseudomonas aeruginosa, Nitrosomonas europaea, and Streptomyces coelicolor are particularly rich in ECF σ factors and contain 13, 19, 22, and 50 predicted ECF σ factors, respectively. σ factors share four conserved regions which can be further subdivided. Region 4.2 recognizes the −35 element, and region 2.4 recognizes the −10 element of promoter DNA. A bacterial promoter consists of at least about 60 bp spanning the positions −40 to +20 in relation to the +1 start site of transcription (12). A comparison of ECF σ factors reveals that the −35 sequence and the spacing but not the sequence between the −35 and −10 regions are well conserved (Table 1) (4, 5, 10, 11, 19, 20, 23, 29). The −10 sequences show less homology (Table 1), which is reflected by the low homology of regions 2.4 in the ECF σ factors (16). The characteristic feature of region 2.4 of σ⁷⁰, a set of hydrophobic residues that form an amphipatic α-helix, is not present in the corresponding region of ECF σ factors (16). The diversity of the −10 sequences and region 2.4 probably accounts for the coexistence of multiple members of the ECF σ-factor subfamily in the same species.

TABLE 1.

Promoter sequences recognized by ECF σ factors

ECF σ factor	Nucleotide sequence		Reference
	−35 element	−10 element
FecI	GAAAAT	TGTCCT
PA2468	GAAAATa	TGTCGGa	25
PA3899	GATATTa	TTTTTCa	25
SigR	GGGAAT	GTT	10
SigB	GGgaacb	cGTTab	10
AlgU/RpoE	GAAC	TCT	29
CarQ	GAAAC	CGT	29
SigW	TGAAAC	CGTA	29
SigX	TGTAACT	CGAC	29
PvdS	TAAAT	CGT	29

^aAlignment of sequences of the putative pa2466 and pa3901 promoters with that of the fecA promoter. The amino acid sequences of the PA2468 and PA3899 σ factors in the region 4.2 are very similar to that of FecI.

^bLowercase letters indicate sites of lower homology.

Previously, it was shown that transcription of the fec transport genes starts at nucleotide 2741 (7) of the fec sequence (22), resulting in a polycistronic mRNA (7). Binding of the FecI RNA polymerase holoenzyme to a 75-bp DNA fragment (from positions −61 to +13) was demonstrated by DNA band shift experiments (2). The smallest plasmid-encoded DNA fragment studied that inhibited chromosomal fec transport gene expression by binding the FecI RNA polymerase extended from positions −82 to +62. In vitro transcription by FecI RNA polymerase was shown with a 650-bp promoter DNA fragment (2). Footprinting scans revealed that the Fur repressor protein loaded with Mn²⁺ (less prone to oxidation than Fe²⁺) covered 38 nucleotides (from positions −38 to −1) of the coding strand of the fecA promoter in which the Fur consensus sequence was localized (positions −36 to −17) (7). In addition, preliminary evidence was obtained for the involvement of the promoter downstream region for fecA transcription (2).

Since the ECF promoters deviate in several respects from the σ⁷⁰ promoters, we examined the contribution of the −35, −10, and the downstream region for transcription of the fecA transport gene. Using site-directed mutagenesis, we constructed single-nucleotide changes in the fecA promoter (Fig. 1). To analyze the effects of the nucleotide substitutions, the mutated and the wild-type fecA promoters were fused to the gfp reporter gene. Plasmid pGFPA′ is a derivative of pFPV25 that contains a promoterless gfpmut3 gene (26) fused to the fecA promoter. Each construct was transferred into the fec deletion mutant E. coli AA93 transformed with plasmid pSV66 fecIRA, and the relative fluorescence was assayed in nutrient broth (NB) medium in the presence and absence of ferric citrate. The fluorescence was measured with the FL600 fluorescence reader (Biotek, Bad Friedrichshall, Germany) at an excitation wavelength of 485 nm and an emission wavelength of 530 nm.

FIG. 1.

Nucleotide sequence of the fecA promoter analyzed. The −35 and −10 elements are underlined, and the sites of nucleotide substitutions are in bold letters.

Point mutations in the −35 region affect fecA promoter activity.

All mutated fecA promoters with nucleotide substitutions in the −35 element (−35 to −30) were less active than the fecA wild-type promoter, but none decreased fecA promoter activity more than threefold (Table 2). Mutations of the most conserved nucleotides A₋₃₄ and A₋₃₃ reduced the relative fluorescence no more than to 81 and 69%, respectively, of wild-type activity. Nucleotide substitutions next to the −35 element from positions −40 to −36 had a small effect on the activity of the fecA promoter, which shows that the putative −35 element with the conserved sequence GAAAAT begins at position G₋₃₅. Since none of the single mutations abolished fecA promoter activity, double mutations were introduced at a few sites. Combination of the single mutation G₋₃₅T (57% activity) with T₋₃₀G (41%) or A₋₃₄C (81%) reduced the activity to 32 and 29%, respectively. The double mutant C₋₃₂ G₋₃₀ was also less active (21%) than the C₋₃₂ (43%) and G₋₃₀ (41%) single mutants (Table 2). It appears that nucleotide substitutions in the −35 region are tolerated to some extent, which is not unexpected since different ECF σ factors bind to similar −35 regions. A comparable result was obtained with the sigX promoter of the SigX σ factor (13). However, the finding that the crtL promoter of Myxococcus xanthus to which the CarQ ECF σfactor binds is critically dependent on a pentanucleotide sequence centered at the −31 position (19) indicates differences in the structural requirements of ECF σ factor promoters. Upstream deletions in the fecA promoter extending to nucleotides −28 and −19 reduced β-galactosidase activity of a plasmid-encoded fecA-lacZ operon fusion to 23 and 25% of wild-type activity, respectively, whereas a deletion covering the entire promoter region to +1 completely abolished fecA-lacZ transcription (1). These data demonstrate that the −35 region can be partially replaced by other nucleotide sequences. This finding is consistent with the lack of complete promoter inactivation by the point mutations.

TABLE 2.

Site-directed mutations introduced at the −35 element (−40 to −25), the −10 element (−18 to −5), and the downstream regulatory element^a

Plasmid	Sequences of the fecA promoter and the mutant fecA promoters	Relative fluorescence units (% of wild-type)
		NB + 50 μM dipyridyl	NB + 1 mM citrate
	−35 element
pGFPA′	GTAAGGAAAATAATTC	403 (2)	17,170 (100)
pGFP40	TTAAGGAAAATAATTC	363 (2)	13,471 (78)
pGFP39	GGAAGGAAAATAATTC	247 (1)	10,816 (61)
pGFP03	GTCAGGAAAATAATTC	669 (4)	7,555 (44)
pGFP37	GTACGGAAAATAATTC	391 (2)	16,101 (93)
pGFP36	GTAATGAAAATAATTC	958 (6)	13,661 (79)
pGFP06	GTAAGTAAAATAATTC	704 (4)	9,787 (57)
pGFP07	TTAAGGCAAATAATTC	1,287 (8)	13,907 (81)
pGFP08	TTAAGGACAATAATTC	1,013 (6)	11,847 (69)
pGFP09	GTAAGGAACATAATTC	635 (4)	7,383 (43)
pGFP10	GTAAGGAAACTAATTC	446 (3)	6,353 (37)
pGFP11	GTAAGGAAAAGAATTC	481 (3)	7,039 (41)
pGFP29C	GTAAGGAAAATCATTC	369 (2)	6,122 (35)
pGFP388	GTAAGGAAAATAGTTC	351 (2)	6,705 (39)
pGFP14	GTAAGGAAAATAAGTC	601 (4)	9,272 (54)
pGFP26	GTAAGGAAAATAATGC	293 (2)	6,745 (39)
pGFP25	GTAAGGAAAATAATTA	1,071 (6)	8,035 (46)
pGFP3530	GTAAGTAAAAGAATTC	482 (3)	5,736 (32)
pGFP3534	GTAAGTCAAATAATTC	287 (2)	5,199 (29)
pGFP3230	GTAAGGAACAGAATTC	369 (2)	3,764 (21)
pGFP3231	GTAAGGAACCTAATTC	262 (2)	6,633 (37)
	−10 element
pGFPA′	CGATTGTCCTTTTT	400 (2)	17,926 (100)
pGFP19	AGATTGTCCTTTTT	244 (1)	16,935 (94)
pGFP18	CTATTGTCCTTTTT	211 (1)	12,934 (72)
pGFP17	CGCTTGTCCTTTTT	244 (1)	13,639 (76)
pGFP16	CGAGTGTCCTTTTT	348 (2)	7,147 (39)
pGFP27	CGATGGTCCTTTTT	430 (2)	3,764 (21)
pGFP28	CGATTTTCCTTTTT	36 (0)	179 (1)
pGFP28C	CGATTCTCCTTTTT	235 (1)	8,097 (45)
pGFP28A	CGATTATCCTTTTT	525 (3)	8,502 (47)
pGFP29	CGATTGGCCTTTTT	322 (2)	2,151 (12)
pGFP30	CGATTGTACTTTTT	502 (3)	4,660 (26)
pGFP31	CGATTGTCATTTTT	54 (0)	717 (4)
pGFP32	CGATTGTCCGTTTT	72 (0)	1,255 (7)
pGFP8G	CGATTGTCCTTGTT	186 (1)	11,228 (62)
pGFP7G	CGATTGTCCTTTGT	208 (1)	5,563 (31)
pGFP6G	CGATTGTCCTTTTG	470 (3)	8,869 (49)
	Downstream regulatory element
pGFP41	CTCTCGTTCGACTCATAGCT	1,470 (8)	15,954 (89)
pGFP44	TTCCCGTTCGACTCATAGCT	1,308 (7)	14,162 (79)
pGFP48	TTCTCGTCCGACTCATAGCT	968 (5)	11,831 (66)
pGFP49	TTCTCGTTTGACTCATAGCT	89 (0)	896 (5)
pGFP50	TTCTCGTTCTACTCATAGCT	36 (0)	179 (1)
pGFP50C	TTCTCGTTCCACTCATAGCT	153 (1)	9,793 (54)
pGFP50A	TTCTCGTTCAACTCATAGCT	211 (1)	220 (1)
pGFP51	TTCTCGTTCGCCTCATAGCT	107 (0)	1,254 (7)
pGFP57	TTCTCGTTCGACTCATCGCT	609 (3)	8,066 (45)
pGFP60	TTCTCGTTCGACTCATAGCC	753 (4)	10,397 (58)

^aThe plasmids carry gfp fused to either the wild-type or mutated fecA promoter. Values (in Miller units) were determined for E. coli AA93 (Δfec) pSV66 (fecIRA) transformed with plasmids carrying the wild-type or mutant promoter fused to gfp. The bold letters indicate the changed nucleotides. The first base of each sequence given for the downstream regulatory element is at position +1.

Point mutations in the −10 region strongly reduce fecA promoter activity.

To determine the −10 element and its boundaries, single-nucleotide substitutions were introduced from −18 to −5. The mutated promoters were analyzed as described above for the −35 region. Mutations between positions −14 and −9 greatly reduced the activity of the fecA promoter (Table 2). Mutant fecA promoters with the nucleotide substitutions G₋₁₃T, C₋₁₀A, and T₋₉G almost completely abolished the activity of the fecA promoter, whereas mutations at C₋₁₂ strongly reduced fecA promoter activity. The G₋₁₃T substitution was replaced by the less extreme C and A substitutions. These mutants displayed a higher fecA promoter activity than did T₋₁₃. The mutations that completely inactivated the fecA promoter activity all had extreme nucleotide changes. The high fecA-gfp transcription rates of the −17 to −15 mutants define well the 5′ boundary of the −10 element, in contrast to the lower transcription of the −6 to −4 mutants (Table 2), which define less clearly the 3′ boundary. The data strongly indicate that the −10 region is essential for the activity of the fecA promoter and agree with data for SigX promoters and the crt promoter indicating that single-nucleotide replacements also abolished promoter activity.

Point mutations in the region downstream of the fecA transcription start site strongly reduce transcription.

The fecA downstream promoter element plays a critical role, as mutations C₊₉T and A₊₁₁C strongly reduced competition of a fecA promoter DNA fragment with the corresponding wild-type fecA promoter fragment for FecI-induced fecA-lacZ transcription. In addition, the T₊₇G mutation strongly reduced FecI-mediated binding of RNA polymerase to a 95-bp fecA DNA promoter fragment (2). In this study the level of fecA transcription was not determined. To further support this finding, we constructed by PCR mutagenesis single-nucleotide substitutions between positions +1 and +20. The promoter activity of the mutated downstream fragments was monitored by measuring gfp expression as indicated above. Mutants with the nucleotide substitutions C₊₉T, G₊₁₀T, and A₊₁₁C strongly reduced the relative fluorescence, resulting in 5, 1, and 7%, respectively, of wild-type promoter activity (Table 2). The mutation G₊₁₀A abolished fecA-gfp transcription as strongly as did G₊₁₀T (1% activity), whereas transcription by the G₊₁₀C promoter was 54% of wild-type activity (Table 2). Mutations at positions T₊₈, T₊₁₆, and T₊₂₀ affected fecA promoter activity much less, ranging between 45 and 66%.

DNA downstream of the transcription start site was shown to be important for other ECF σ factor-controlled promoters. DNA from nucleotides +50 to +120 of the crtL promoter is required for promoter activity (19). The DNA downstream of crtL contains an enhancer-like element that remains active when displaced to a site upstream of the promoter. The ECF σ factor PvdS is required for transcription of the pyoverdin synthesis genes of P. aeruginosa. For pvdF transcription the smallest fragment retaining promoter activity extended from nucleotides −91 to +34, indicating that a sequence element in the downstream region is required for maximal expression from this promoter (27). However, DNA downstream from the pvdE +1 site to nucleotide +195 did not increase the PvdS-dependent pvdE promoter activity (29). A study of the PvdS-dependent pvdA promoter finds that DNA downstream of the transcription start site is required for promoter activity in P. aeruginosa (15).

In order to relate promoter activity to promoter binding of FecI RNA polymerase, in vivo titration experiments were carried out. An excess of plasmid-encoded wild-type fecA promoter DNA binds the FecI-RNA polymerase complex, resulting in lesser availability of the holoenzyme for transcription of the chromosomal fecB-lacZ operon fusion in E. coli ZI418 (1). Mutations in the −10, −35, and downstream regions were examined. The wild-type fecA promoter and mutant fecA promoters fused to gfp were cloned into the high-copy-number vector pBCSK⁺. The β-galactosidase activity of E. coli ZI418 transformed with pBCSK⁺ and grown in the presence of the inducer (Fe³⁺ citrate)₂ amounted to 325 Miller units (Table 3). β-Galactosidase activity was reduced to 192 Miller units in cells with plasmid-encoded wild-type fecA promoter. Even though the plasmid-encoded fecA promoter did not completely reduce expression of chromosomal fecB-lacZ, this reduction was taken as 100%, in relation to which the reduction caused by the mutated fecA promoters was determined. The G₊₁₀T mutation reduced β-galactosidase activity only by 63% (Table 3). The T₋₁₂G mutation reduced β-galactosidase activity by 14%, while the A₋₃₁C mutation reduced β-galactosidase activity by 60%. The mutation in the −10 region most strongly impaired binding of the FecI RNA polymerase complex to the plasmid-encoded fecA promoter.

TABLE 3.

FecI-binding and induction activity of the wild-type and the mutated fecA promoter regions^a

Plasmid	Base changes	β-Galactosidase activity (percent FecI-binding)		Relative fluorescence (percent induction)
		NB medium	NB medium + 1 mM citrate	NB medium	NB medium + 1 mM citrate
pBCSK+		108	325 (0)	181	224
pHCA	Wild type	94	192 (100)	258	2,636 (100)
pHC10	A−31C	105	246 (60)	255	1324 (50)
pHC29	T−12G	138	307 (14)	178	909 (34)
pHC50	G+10 T	125	241 (63)	160	234 (9)

^aValues (in Miller units) were determined for E. coli ZI418 fecB-lacZ transformed with high-copy-number plasmids carrying the fecA wild-type promoter or fecA mutant promoter fused to gfp.

Activity of the wild-type and the mutated fecA promoter fragments on the high-copy-number plasmids in ZI418 was determined by gfp expression (Table 3). Despite the relatively high FecI RNA polymerase binding ability of the G₊₁₀T promoter mutant (63%), the induction activity was low (9%) (Table 3). The relative activities of the promoter mutants on the high-copy-number plasmids were higher (Table 3) than on the low-copy-number plasmids (Table 2). However, the absolute fluorescence values listed in Table 3 are lower than those listed in Table 2, which may be caused by the reduced availability of FecI RNA polymerase due to its binding to the chromosomal fecA promoter of ZI418, which is not present in the AA93 fec deletion mutant used for the data presented in Table 2. Induction levels by (Fe³⁺ citrate)₂ via the wild-type promoter were comparable (10-fold) in the two E. coli strains.

The results described here confirm and extend previous findings on the importance of the downstream region of the fecA promoter for fec transport gene transcription (2). Since binding of the FecI-RNA polymerase complex to the downstream region did not play a major role, the downstream region may be involved in a subsequent step of transcription initiation. For the initiation process in E. coli the closed RNA polymerase holoenzyme (RPc) turns into an open complex conformation. DNase I footprint assays showed that several E. coli promoters exist in two forms; one form displays protection of the promoter from −55 to −5 (RPc₁) against DNase I, and the other displays it from −55 to + 20 (RPc₂) (6). Although it is not clear how the conformational changes of RPc₁ to RPc₂ occur, it seems that the downstream DNA is placed into its binding site (12).

Mutations in the fec-specific −10 region strongly reduce FecI polymerase binding and fecA transcription. FecI RNA polymerase specifically binds to the −10 fecA promoter region, and for this reason mutations in the −10 region reduce promoter activity. This may apply to other ECF σ factors for which different −10 regions exist. In contrast, the fecA −35 region tolerates nucleotide replacements.