INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Many bacterial plasmids carry genetic elements that confer upon the host a selective advantage. One of these is pColV-K30, a large low-copy-number plasmid which was isolated from a pathogenic strain of Escherichia coli (13, 33, 34). In addition to pathogenic traits such as serum resistance (2), aerobactin iron uptake (35), and enhanced epithelial adherence (8), pColV-K30 was found to carry the determinants for production and secretion of the peptide antibiotic colicin V (ColV), which is toxic for related strains of the family Enterobacteriaceae.

The ColV genes were characterized from a 9.4-kb subcloned fragment of plasmid pColV-K30 (12, 13). The four genes which are essential for ColV production, secretion, and host immunity, reside within 4.5 kb of this fragment and are arranged in two converging operons (12, 13). The genes encoding a secretion accessory protein, CvaA, which belongs to the membrane fusion protein family (10), and the transporter, CvaB, which belongs to the ATP binding cassette transporter family, are encoded in a single operon (13, 14). The genes encoding the immunity protein Cvi and toxin CvaC are encoded in the opposing operon (13, 14).

Unlike most colicin systems, the ColV secretion system does not appear to be inducible by an SOS response (17). Secretion of ColV was found to be regulated by iron via Fur (6, 14), similar to pathogenic traits such as iron uptake and -hemolysin secretion. Fur represses transcription of genes in the presence of iron by binding promoters that contain a Fur box (4, 9). However, results of initial studies suggested the possibility that the observed iron-dependent ColV activity was partially due to iron-dependent expression of the transporter and accessory proteins, CvaA and CvaB, allowing increased secretion of the toxin under iron-poor inducible conditions. In support of this, studies using translational Mudlac fusions indicated that cvaB and cvi were under iron control, in that they both exhibited increased -galactosidase activity with the addition of the iron chelator, dipyridyl (14). Transcriptional regulation of these operons by Fur was also suggested by previous sequence analysis that identified two potential Fur boxes, previously termed iron response elements; one was 179 bp upstream of the cvaA gene, and the other was 50 bp upstream of the cvi gene (14). These two Fur boxes presumably control transcription from the promoters for the converging cvaA-cvaB and cvi-cvaC operons, respectively.

This study shows that the expression of both Cvi and CvaA is iron regulated and identifies iron-regulated promoters of both the cvaA and cvi genes. Transcription of cvi occurs from the previously predicted promoter and associated Fur binding site. However, for cvaA, iron-dependent transcription does not occur from the previously suggested promoter and Fur box upstream of cvaA but instead occurs from a promoter that lies more than 320 bp upstream of the translational start codon and is associated with a new Fur binding site identified in this study. The transcriptional activity from this promoter can be induced up to 10-fold by iron depletion. However, sequences downstream of the transcriptional start apparently modulate such induction fivefold and down-regulate overall transcription approximately fourfold. The deletion of sequences encompassing a long stem-loop structure from the downstream sequences partially relieves this down-regulation. In addition, the modulating properties of these sequences were further substantiated for the cvi promoter, which is also regulated by iron. The results show that analogous placement of the cvaA downstream sequences modulates transcription from the cvi promoter in a manner similar to that observed for cvaA. Sequences including the predicted RNA secondary structure showed the greatest down-modulation. In summary, this study identified the promoters responsible for iron-dependent expression of both Cvi and CvaA and revealed that transcription of cvaA is more complex than anticipated.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and culture media. The E. coli strains used in this study were MC4100 (F lacU169 araD136 relA1 rpsL150 flbB5301 deoC7 ptsF25 thi-1) and DH5 [supE44 lacU169 (80 lacZM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1]. DH5 was used for initial transformations. Luria-Bertani medium (LB; 10 g of tryptone, 5 g of yeast extract, and 10 g of NaCl per liter) was used for growth of competent cells and growth of cells after transformation. TB (10 g of tryptone and 8 g of NaCl per liter) was used for all growth conditions unless otherwise noted, as is or with FeCl₃ or 2,2'-dipyridyl added at the concentrations indicated. Ampicillin and glucose were added at final concentrations of 100 µg/ml and 0.5%, respectively. Cultures were grown in TB-ampicillin-glucose with or without FeCl₃ until specific times during the growth cycle, either FeCl₃ or dipyridyl was added, and then growth resumed for an additional 40 min before harvesting.

Transformations and DNA manipulations. Standard procedures were used to transform E. coli and for DNA manipulations (32). Plasmids were isolated from Qiagen (Chatsworth, Calif.) gels. Restriction enzymes, phage T4 DNA ligase, and T4 polynucleotide kinase were used as described by the manufacturers. DNA fragments were purified from agarose gels by using the Qiaex gel extraction protocol as recommended by the manufacturer (Qiagen). Oligonucleotide primers (Table 1) were synthesized by the -cyanoethylphosphoramidite method with an ABI 381A or 392 DNA synthesizer (Applied Biosystems, Foster City, Calif.) in the Biology Core Facility of Georgia State University. PCR was performed with T4 DNA polymerase as described elsewhere (27).

Plasmids and plasmid constructions. pHK11 obtained from Roberto Kolter (Harvard Medical School, Boston, Mass.) was derived from pBR322 with a 9.4-kb fragment from pColV-K30 of which a 4.5-kb segment contains the cvaA, cvaB, cvi, and cvaC genes in two converging operons (12). Transcriptional fusions of cvaA sequences (ALZ-n) upstream of the lacZ gene were constructed by PCR using 5' primers with BamHI sites and 3' primers with a HindIII site cloned into the same restriction sites of the transcriptional fusion vector pQF50 (11). The cvi promoter was similarly cloned by PCR upstream of lacZ into the 5'-SalI-BamHI-3' site (ILZ-1). This construction was then used to place specific cvaA sequences immediately downstream of the cvi promoter in the BamHI-HindIII site, generating a cvi-cvaA hybrid transcriptional lacZ fusion. Table 1 lists all of the primers and summarizes the constructions; the cvaA, cvi, and cvi-cvaA hybrid constructions are depicted in Fig. 5. Most PCR fragments were generated by using pHK11 as the template. A SmaI-SmaI fragment and an AvaII-AvaII fragment of pHK11 were used for cvaA sequences to construct ILZ-1-4 and ILZ-1-6, respectively. The annealing temperature during PCR was 55°C. PCR products were ethanol precipitated and digested with BamHI and HindIII, then purified from agarose gels by using Qiaex, and cloned into the transcriptional lacZ fusion vector pQF50 (11) that was similarly digested and purified. Ligation was followed by transformation into E. coli DH5. Transformants were screened by plating onto LB-glucose supplemented with ampicillin (100 µg/ml) and 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal; 50 µg/ml). Plasmid insertions were confirmed by restriction enzyme analysis and then transformed into MC4100. Sequences were verified by using PRISM Ready Reaction DyeDeoxy terminator cycle sequencing kit (Applied Biosystems), carried out at the Georgia State University Biology Core Facility.

RNA isolation. MC4100/pHK11 cells were grown in 1 liter of TB-ampicillin-glucose. The optical density at 600 nm (OD₆₀₀) was monitored, and at mid-logarithmic (OD₆₀₀ = 0.4), late logarithmic (OD₆₀₀ = 0.8), and stationary (OD₆₀₀ = 1.0) growth phases, either 0.1 mM FeCl₃ or 0.1 mM 2,2'-dipyridyl was added to 100-ml aliquots. The cultures were grown for an additional 40 min to repress or induce iron/Fur-regulated transcription, after which the culture was chilled on ice and harvested at 10,000 × g for 5 min. RNA was purified as described previously (18).

Primer extension and DNA sequencing. For primer extension, 21 µg of each purified RNA sample was denatured with formaldehyde as described previously (18). Synthetic oligonucleotides A14 and A17, antisense to specific cvaA sequences, and ALZ-16 and ALZ-21, specific for cvi sequences (Table 1), were labeled with [-³²P]ATP (Amersham, Arlington Heights, Ill.) by using T4 polynucleotide kinase according to the manufacturer's specifications. The labeled primers were purified by using G-25 Micro-Select spin columns (5 Prime  3 Prime, Inc., Boulder, Colo.) and then used to prime the reverse transcription of the RNA, using murine myeloblastosis virus reverse transcriptase (New England Biolabs, Beverly, Mass.) or Superscript II reverse transcriptase (Gibco BRL, Gaithersburg, Md.). Primer extension products were purified by using a QIAquick nucleotide removal kit (Qiagen). These products were run adjacent to the corresponding DNA sequence ladder generated by primer A14, A17, A16, or A21 on a 6% Sequagel polyacrylamide gel (National Diagnostics, Atlanta, Ga.). DNA sequencing was performed with a Sequenase dideoxy sequencing kit and [-S³⁵]dATP. Radiograms were visualized with a phosphoimager (Fuji Medical Systems, Inc., Stamford, Conn.) or by exposure to Hyperfilm-MP (Amersham) or Biomax MR film (Eastman Kodak Co., Rochester, N.Y.).

S1 nuclease analysis. The same RNA samples from logarithmic phase cells that were purified and used as described above for primer extensions were subjected to S1 nuclease analysis. Primers A14 and A17, radiolabeled with [-³²P]ATP as described above, were used to construct a DNA probe for annealing to the cvaA transcript. Purified plasmid pHK11 was alkaline denatured and annealed with labeled primer A14 or A17. DNA probes specific for the cvaA promoter regions were synthesized with the Klenow fragment of DNA polymerase (Promega, Madison, Wis.) and restricted with EcoRV, located at 345. Probes with radioactivity of 50,000 cpm were used for annealing to 21 µg of RNA and then subjected to digestion with S1 nuclease (Ambion, Austin, Tex.) as described by the manufacturer. Undigested radiolabeled DNA was precipitated with ethanol and then run on a 6% polyacrylamide gel adjacent to Superscript II A14 or A17 primer-extended samples and A14 or A17 DNA sequence ladders.

Western blot analysis and quantitation of CvaA. Approximately 40 µg of total protein for each sample was used for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)/Western blotting performed as described elsewhere (16), using purified CvaA-specific antibodies (19) and alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin G as secondary antibodies. The immunoprobed proteins were detected by using 2,2'-di-p-nitrophenyl- 5,5'-diphenyl-D,D'[3,3'-dimethoxy-4,4'-diphenylene] ditetrazolium chloride and 5-bromo-4-chloro-3-indolyl phosphate as the substrate. CvaA bands detected by Western blotting were quantified with the PDI Image Analyzing System (Protein Databases Inc., Huntington Station, N.Y.).

Western blot analysis of Cvi. Antibodies to Cvi were generated by using a synthetic peptide of the C-terminal 26 amino acids predicted for Cvi (CYFVGDNYYSISDKIKRRSYENSDSKA) combined with Freund's adjuvant and inoculated into rabbits according to standard protocols (16). Serum produced specific reactions to Cvi protein in MC4100/pHK11 but not in MC4100/pLY11 with cvaA and cvaB only (data not shown) was collected postinoculation. Although nonspecific bands were minimal, the serum was treated twice with MC4100 acetone powder (16). The treated Cvi serum was diluted 1:2,000 for immunoblotting. Approximately 40 µg of total protein of cell samples of MC4100 alone or of MC4100/pHK11 repressed with 0.1 mM FeCl₃ or induced with 0.1 mM 2,2'-dipyridyl at an OD₆₀₀ of 0.4 for 40 min was used for SDS-PAGE/Western blotting performed as described previously (16). The immunoblot was treated with labeled secondary antibodies and developed as described for CvaA above.

-Galactosidase assays. MC4100 cells containing lacZ promoter fusion plasmids were grown overnight in 2 ml TB-ampicillin-glucose with 0.05 mM FeCl₃. Overnight cultures were diluted 1:20 to an OD₆₀₀ of approximately 0.1 in 3 ml TB-ampicillin-glucose with 0.05 mM FeCl₃ unless otherwise noted. At mid-logarithmic phase, (OD₆₀₀ = ~0.4), 1.5 ml of cells was induced with 0.1 mM 2,2'-dipyridyl. After 40 min of growth in either FeCl₃ or dipyridyl, cells were placed on ice, permeabilized, and then assayed in duplicate for -galactosidase activity as described elsewhere (26).

Chemicals and reagents. All chemicals and reagents were from Sigma (St. Louis, Mo.) except where indicated otherwise. All DNA-modifying enzymes were from Boehringer (Indianapolis, Ind.) unless noted otherwise.

Oligonucleotide synthesis. All oligonucleotides used for primer extensions and PCR for cloning were constructed in the Georgia State University Biology Department Core Facility.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Iron dependence of CvaA and Cvi production in immunoblots. Previous studies have shown that ColV production is under iron regulation. Presumably, promoters associated with Fur binding sites identified upstream of the cvaA and cvi genes of the two converging ColV operons are responsible for iron-regulated production of the ColV proteins (see Fig. 3A). To confirm that CvaA and Cvi are also iron regulated, immunoblot analyses were used to determine the levels of CvaA and Cvi produced under different iron conditions. CvaA antibodies were used for immunoblotting to determine the iron and growth phase dependence of CvaA synthesis in total cell samples collected at mid-logarithmic, late logarithmic, and stationary phases of growth after the addition of iron or dipyridyl. Dipyridyl enhances transcription by alleviating Fur repression, resulting in transcriptional expression of genes subject to regulation by Fur and/or iron. Immunoblotting and quantification of the CvaA bands showed that low levels of CvaA is produced throughout the growth cycle even in the presence of iron. This finding indicates that CvaA is not completely repressed by iron and is expressed constitutively at low levels in the presence of iron (Fig. 1A). Under conditions where iron is depleted by dipyridyl, CvaA production is increased in the log phases of growth. However, in the stationary phase, dipyridyl no longer induces production of CvaA and a basal level is exhibited (Fig. 1A). The stationary-phase levels of CvaA with dipyridyl are similar to the basal levels observed with iron present. The maximum induction of CvaA observed in the late logarithmic phase compared to levels with iron is about fivefold (Fig. 1A). Therefore, it appears that the production of CvaA is repressed by iron levels and induced by iron depletion except in the stationary phase of growth.

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Characterization of iron-dependent expression of Cvi and CvaA. Immunoblotting using purified anti-CvaA antibodies (A) or antibodies to a synthetic Cvi peptide (B) was performed with 40 µg of total protein of MC4100/pHK11 cells grown in TB-ampicillin-glucose and exposed to either 0.1 mM FeCl3 (Fe) or 0.1 mM 2,2'-dipyridyl for 40 min at mid-logarithmic (ML), late logarithmic (LL), and stationary (S) phases of growth. After immunoblotting, CvaA (A)- and Cvi (B)-specific bands were quantified by scanning densitometry. The data were normalized to 100% of the highest value obtained. (A) Quantitative immunoblot for CvaA. Lane 1 corresponds to MC4100 without plasmid so that no CvaA is produced. Lane 2 corresponds to MC4100/pHK11 cells grown for the duration in FeCl3, overnight with FeCl3, reinoculated into TB with FeCl3, and then collected at late log phase, indicating that this sample corresponds to the constitutive basal level of CvaA produced in the presence of FeCl3. Lanes 3 to 5 correspond to MC4100/pHK11 cell samples exposed to FeCl3 for 40 min; lanes 6 to 8 correspond to cell samples exposed to dipyridyl for 40 min. (B) Quantitative immunoblot for Cvi. Mid-log-phase cell samples of MC4100 without plasmid (lane 1) and MC4100/pHK11 exposed to FeCl3 (lane 2) or 2,2'-dipyridyl (Dip; lane 3) were used for immunoblotting with Cvi-specific peptide antibody.

Identification of the ColV promoters by primer extension analyses. To determine the promoter locations and transcriptional regulation of the cvaA and cvi genes, primer extension assays were performed on total RNA isolated from MC4100/pHK11 cells grown in TB containing either FeCl₃ or 2,2'-dipyridyl at time points corresponding to mid-logarithmic, late logarithmic, and stationary growth phases. These RNA samples correspond to the cell samples used for Fig. 1A. A promoter and Fur box, FB-I (see Fig. 3B), for iron-dependent transcription of cvi have been predicted from DNA sequence analysis (14). However, their location and iron dependence have not been verified. The transcriptional start for the cvi gene was mapped by using a primer (A21) antisense to the sequence immediately downstream of the translational start identified for cvaC (12, 13). The primer specific for cvaC was used since cvaC is downstream of cvi and presumably cotranscribed with cvi and because the two genes are separated by only 214 bp. Primer extensions revealed a single band which is present in all samples throughout the growth cycle and is reduced with iron and increased by iron depletion especially in the logarithmic phase (Fig. 2A). The band representing the transcriptional start for cvi was mapped more closely by using primer A16 immediately downstream of the cvi translational start in primer extensions of the mid-log-phase samples. Three closely spaced bands present in samples induced with dipyridyl could represent the transcriptional start for cvi (Fig. 2A, box). The strongest of the three bands was chosen as the +1 site, although all are close enough to utilize the same promoter elements. The promoter region for cvi is shown in Fig. 3C. Mapping the cvi transcriptional start from primer extensions shows that it is only 40 bp upstream of its translational start. The possible 10 and 35 promoter elements are indicated and correspond well to the +1 band (Fig. 3C). The 10 elements resides within the Fur binding site (FB-I) (Fig. 3B) identified previously for cvi (14).

View larger version (96K): [in this window] [in a new window]   FIG. 2.   Identification of putative transcriptional starts for the cvaA and cvi genes. Primer extension assays were performed on RNA isolated from MC4100/pHK11 grown in TB-ampicillin-glucose and exposed to either 0.1 mM FeCl3 (Fe) or 0.1 mM 2,2'-dipyridyl (Dip) for 40 min at mid logarithmic (lanes a), late logarithmic (lanes b), and stationary (lanes c) phases of growth. (A) Primer A21 was used to map the transcriptional start for cvi of RNA samples with Fe (lanes 1 to 3) or Dip (lanes 4 to 6). The inset is of primer extensions, using primer A16 of mid-log-phase RNA samples from cells grown with Fe (lane 1) or Dip (lane 2) to determine the sequence corresponding to the transcriptional start for cvi. (B) Primer A17 was used to map the transcriptional start for cvaA of RNA samples with Fe (lanes 1 to 3) or Dip (lanes 4 to 6). (C and D) Primer extensions using Superscript II reverse transcriptase and S1 nuclease analysis were performed on the same RNA samples from mid-log-phase cells as used for panels A and B. Primers A14 (C) and A17 (D) were used for primer extensions and to construct probes for S1 nuclease 5' mapping of cvaA RNA. Lanes 1 and 2 correspond to primer extension of RNA samples grown in iron and Dip; lanes 3 and 4 correspond to S1 nuclease analysis of the same RNA samples. Lane 5 is the probe.

View larger version (58K): [in this window] [in a new window]   FIG. 3.   (A) Genetic arrangement of the four genes required for the production, export, and immunity of ColV. Arrows indicate the direction of transcription for both operons, and the open boxes represent the open reading frames corresponding to the genes. The predicted number of amino acids (aa) residues for each gene product and approximate locations of promoters and sequences with Fur box homology are shown. (B) Alignment of the consensus Fur box with ColV Fur boxes. The previously identified Fur boxes for cvi (FB-I) and cvaA (FB-A2) are compared to the new Fur box identified here (FB-A1) and the Fur consensus (4, 9). The numbers in parentheses are sequence positions corresponding to the sequences in panels C and D. (C) Promoter region for cvi, showing the transcriptional start (+1), the predicted FB-I for cvi, the 10 and 35 promoter elements, and the translational start for cvi. (D) Promoter region for cvaA. Sequence upstream of and including the start codon for the cvaA gene is shown. The predicted 10 and 35 sequences for P1 are underlined, the +1 base is in bold-face, and FB-A1 and FB-A2 sequences are boxed. Locations of A14 and A17 primers used for primer extensions are complementary to the lightly shaded sequences. Two possible translational start sites for cvaA are underlined and labeled; the TTG is the newly identified translational start (19). The start codons for two potential leader peptides, L1 and L2, downstream of the transcriptional start are underlined. The newly sequenced region is numbered negative up to 150 relative to the published sequence (13).

Iron dependence of cvi transcription in a cvi promoter fusion to lacZ. To verify that the potential promoter identified by primer extension corresponds to the iron depletion-induced transcription of cvi, 90 bp of the cvi promoter region from bp 4387 to 4286 (Fig. 3C) was subcloned upstream of lacZ in a transcriptional fusion vector, yielding ILZ-1. -Galactosidase assays with MC4100/ILZ-1 grown with either iron or dipyridyl showed that transcription from the cvi promoter is induced by iron depletion fourfold over repressed levels with iron (Fig. 4A). However, the synthesis of Cvi induced with iron depletion compared to basal levels with iron in the log phase is 8.5-fold, more than double that observed for the Cvi transcriptional fusion, ILZ-1. This difference may indicate translational control.

View larger version (21K): [in this window] [in a new window]   FIG. 4.   Iron dependence of cvi and cvaA promoters in lacZ fusions. (A) cvi promoter activity. MC4100/ILZ-1 was grown to mid-log phase in TB with 0.05 mM FeCl3 and then for an additional 40 min in FeCl3 (filled column) or induced for 40 min with 0.1 mM 2,2'-dipyridyl (Dip2; shaded column). (B) cvaA promoter activity. MC4100/ALZ-1 was grown in TB-ampicillin-glucose and exposed to either 0.1 mM FeCl3 (Fe) or 0.1 mM 2,2'-dipyridyl for 40 min at mid-logarithmic (ML), late logarithmic (LL), and stationary (S) phases of growth. Cells were placed on ice after 40 min, and then 200-µl aliquots of cells were solubilized and assayed for -galactosidase activity as described in Materials and Methods.

Iron-dependent transcriptional activity in lacZ fusion with the entire cvaA promoter region. For comparison to primer extensions and immunoblots, samples of MC4100 cells transformed with ALZ-1, containing the entire 320-bp cvaA promoter region fused to lacZ, were grown under conditions identical to those used for primer extension and immunoblots and were assayed for -galactosidase activity. The ALZ-1 fusion was designed as transcriptional, since the 3' fusion point was upstream of the ATG translational start originally proposed for cvaA (13). However, evidence recently showed that translation of cvaA occurs from a TTG upstream of the originally proposed ATG translational start (19). Although this fusion is out of frame with translation, it could extend into coding regions of cvaA. Therefore, we constructed a similar fusion (ALZ-1') with the 3' end upstream of the TTG. The activities in the downstream ATG fusion are not significantly different from activities observed for the upstream TTG fusion when samples are grown under the same conditions (data not shown).

Transcription from P1 in lacZ fusions is modulated by downstream sequences. Except as noted in Fig. 4B, only transcriptional fusions upstream of TTG are included in the assays described below. Also in Fig. 4B, higher levels of activity with both iron and dipyridyl were achieved on samples grown in TB with no iron until the time when iron or dipyridyl was added. Because some induction occurs in TB with no iron or dipyridyl, this method gave higher and slightly variable levels of activity. Therefore, it was important to optimize conditions to achieve full repression so that induction occurred from the same fully repressed point. Thus, all other -galactosidase assays were performed only on log-phase cells grown under conditions which achieved optimal repression, in TB with 0.05 mM FeCl₃ for the duration, followed by induction of an aliquot of the repressed samples at mid-log phase with 0.1 mM 2,2'-dipyridyl.

View larger version (50K): [in this window] [in a new window]   FIG. 5.   Summary of cvaA, cvi, and cvi-cvaA promoter fusions and -galactosidase activities. The cvaA promoter P1 and the P2, P3, and P4 regions are depicted as differently shaded bars. The cvi promoter is represented by the bar with diagonal hash marks (D). The sequence fusion positions are indicated with the 5' fusion position above and left and the 3' position at the bottom and right side of the bar. The mutations of the start codons of potential leader peptides, L1 and/or L2 (Fig. 3D), from ATG to ACG are indicated above the bar in which each occurs. Single mutations of L1 or L2 are indicated as mL1 and mL2, respectively. A double mutation of both L1 and L2 is indicated as mL1/L2. Deletions within the P3 region are represented as spaces. -Galactosidase activities (Miller units) measured in the presence of FeCl3 (Fe) or dipyridyl (Dp), assayed as described in Materials and Methods, are listed at the right; 50 µl of cells was assayed for fusions with high activity (ALZ-5 and -7); 100 µl of cells was assayed for intermediate activity (ALZ-18, -21, and -23); 200 µl of cells was assayed for all other constructions. Volumes of cells for the assay were adjusted to 200 µl with TB. (A) -Galactosidase assays of cvaA P1 promoter fusions with or without downstream sequences included. (B) -Galactosidase assays comparing P3 deletions to fusions with P3 (ALZ-1') and with P1 or P1P2 only. (C) -Galactosidase assays of P1 total promoter region fusions with mutated potential leader start codons. (D) -Galactosidase assays for the cvi promoter fusion and the cvi-cvaA hybrid promoter fusions. Activities and standard deviations are averages from two to seven separate assays.

Secondary structure analysis. One possibility for the down-regulation resulting from the presence of sequences downstream of the P1 promoter is that it is due to secondary structures in the transcript. The primer extension bands could result from termination of reverse transcriptase as a result of secondary structures in the RNA transcripts. In fact, RNA secondary structures could be predicted within the cvaA RNA upstream of the artifactual primer extension bands mentioned above. One of these secondary structures occurs in the P3 region which was responsible for the down-regulation in transcriptional fusions and is designated the P3 stem-loop (Fig. 6A). Also possible is an extended secondary structure partly composed of the P3 stem-loop, encompassing the 3' end of the P2 and the entire P3 region; it is designated LP3 (Fig. 6B). Additional fusions were constructed to test the effects of the potential secondary structures.

View larger version (26K): [in this window] [in a new window]   FIG. 6.   RNA secondary structure predictions. Secondary structures within the RNA upstream of the cvaA translational start were predicted by using MacDNASIS (Hitachi Software Engineering Co., San Bruno, Calif.). (A) RNA secondary structure predicted within the P3 region downstream of the cvaA transcriptional start. (B) Alternative extended secondary structure for the P3 stem-loop in panel A (LP3). (C) Secondary structure predicted for the upper LP3 deletion, ALZ-18. Sequence positions are indicated by numbers at the beginning and end of the secondary structure.

Up-regulation of cvaA transcription in promoter fusions with deletions in the downstream regions. To ascertain the importance of the apparent reduction in activity caused by the presence of the P3 region, we constructed a series of deletions of the P3 region as shown in Fig. 5B. In ALZ-14, the upper part of the P3 stem-loop was deleted so that it was identical to ALZ-1' except for the deletion of bp 45 to 100. Interestingly, this was not sufficient to cause significant up-regulation of transcription from P1, and activities in this fusion were only slightly higher than those observed for ALZ-1' (Fig. 5A, and B). Therefore, other parts of the P3 region may be involved in down-regulation. A fusion with a deletion of all of the P3 stem-loop (ALZ-18) from bp 30 to 100 resulted in a 32 to 34% increase in expression compared to ALZ-1' (Fig. 5B). This increased expression is still not significant compared to the levels seen with ALZ-5 and ALZ-7 without P3 (Fig. 5A). The extended stem-loop LP3 is followed by a U-rich stretch (UUUUCCUU) (Fig. 6B) similar to that in some transcriptional terminators (34). Even though the upper portion of P3 was deleted in ALZ-14 and the entire upper stem-loop was deleted in ALZ-18, a lower stem-loop with a free energy of 8.2 kcal/mol could still form in this construction (Fig. 6C). Therefore, we constructed a fusion in which both the upper stem-loop and most of the lower stem-loop were deleted (ALZ-21). This construction had a deletion of bp 15 to 110 and significantly alleviated the down-regulation, resulting in an approximate 2.5-fold increase in transcription (ALZ-21) compared to ALZ-1' (Fig. 5A and B). Interestingly, with iron, the transcriptional levels in ALZ-21 (122 Miller units) were higher than those for ALZ-5 (P1 alone) (87 Miller units) but less than those for ALZ-7 (P1P2) (203 Miller units) (Fig. 5A). This finding suggests that with iron, the deletion in ALZ-21 may allow the enhancing potential of the P2 region to be partially exhibited. However, the induced levels for ALZ-21 were still less than those observed for ALZ-5 (P1 alone).

Transcriptional activity from P1 is modulated by mutations downstream of the transcriptional start site. The enhanced transcriptional activity in the P2 region and down-regulation in the P3 region were possibly due to translation of two potential leader peptides encoded downstream of the transcriptional start within the P2 region. The first potential leader, L1, is 10 amino acids, MNELCYFNIL, and the second, L2, is 29 amino acids, MSYVTLIFSDNNLNQLDYCHLINNDILSS. Mutation of the start codon of either leader from ATG to ACG does not significantly change -galactosidase activity (ALZ-19 and ALZ-20) (Fig. 5C). Interestingly, mutation of both potential leader peptides decreases the induced levels compared to ALZ-1' (Fig. 5C). It is not certain that translational attenuation is involved in the down-regulation. It is more likely that the reason for the down-regulation by the double mutation in the P2 region is an interference with the enhancing potential of this sequence. Thus, it is possible that sequences downstream of the potential Fur box and promoter influence transcription, transcription factors, or other regulators.

Transcription of cvi is modulated by placing specific cvaA sequences downstream of the cvi promoter. The modulating properties of the sequences downstream of the cvaA transcriptional start were tested further on a similarly regulated heterologous promoter. Thus, the cvi-lacZ transcriptional fusion, ILZ-1, was used to evaluate the effect of the cvaA downstream sequences on the iron-regulated cvi promoter. The P2, P3, and P4 regions of cvaA sequences between the cvaA promoter and gene were placed between the cvi promoter and lacZ gene of ILZ-1. The cvi-cvaA hybrid plasmids constructed are depicted in Fig. 5D. -Galactosidase assays with the cvi and cvi-cvaA hybrid transcriptional lacZ fusion plasmids showed fourfold induction by iron depletion in the cvi promoter fusion ILZ-1, similar to that observed for the cvaA promoter fusion with downstream sequences, ALZ-1' (3.9-fold) (Fig. 5A and D). The hybrid fusions revealed that including the cvaA P2 region behind the cvi promoter (ILZ-1-6) enhances transcription as it does for cvaA (ALZ-7) (Fig. 5A and D). For ILZ-1-6, the increase with P2 is similar for conditions with iron (40%) and dipyridyl (30%), whereas for ALZ-7 the increase with P2 is much greater with iron, which concurrently modulates the induction with dipyridyl from 10-fold (ALZ-5) to 5-fold (ALZ-7) (Fig. 5A and D). Including both downstream P2P3 regions (bp 56 to 115) behind the cvi promoter (ILZ-1-13) down-regulates transcription as it does following the cvaA promoter (ALZ-8) compared to the promoters, ILZ-1 and ALZ-5, respectively, alone (Fig. 5A and D). When all of the LP3 secondary structure and a following U-rich sequence are included (P2P3 and the following 19 bp of P4, bp 56 to 134) downstream of the cvi promoter (ILZ-1-27), the decrease is even greater. Comparison of cvi transcription in ILZ-1-27 to that ILZ-1-6 with P2 only shows that the down-regulation by including bp 48 to 134 is 3.6-fold for iron and 4.8-fold for dipyridyl (Fig. 5D). This is similar to comparison of the analogous constructions in cvaA; P1 with P234 (ALZ-1') compared to P1 with P2 only (ALZ-7) is also reduced 3.6-fold for iron and 4.8-fold for dipyridyl (Fig. 5A). When the entire downstream cvaA region (P234) is placed behind cvi (ILZ-1-28), the activity is similar to that in ILZ-1-27 without the remaining 3' sequences from bp 135 to 208 (Fig. 5D).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Historically, a virulence factor was considered a pathogenic trait only if it was directly involved in host injury. Therefore, the ColV trait was not considered pathogenic since the ColV genes are not directly required for pathogenicity (30). However, the definition of pathogenicity has been revised to include any trait which increases the proliferation or survival of a bacterium during the infection process (24). With this definition, the ColV trait could be considered a pathogenic trait since its presence aids in propagating the pathogenic ColV population by eliminating competing microorganisms (6). By maintaining the ColV population, it also facilitates proliferation of the pathogenic traits of the ColV plasmid. Furthermore, its importance to the infectious process is evident since it resides on the same plasmid and thus is closely linked to many other traits which do cause host injury. In this regard, it is significant that the ColV toxin was found to be regulated by iron and Fur, similar to many of the pathogenic genes to which it is closely linked. Earlier studies showed that the production of ColV is constitutively elevated in Fur mutants (6) and is induced in response to iron limitation (6, 14). However, these observations are solely based on the detection of ColV activity in the supernatant (6, 14). It has not been shown whether the ColV operons (Fig. 3A) are transcriptionally regulated by iron, and the operon promoters have not been identified. The cvaA gene is upstream of and overlapping cvaB and contained an identifiable consensus sequence for Fur-dependent regulation (FB-A2) (14). Therefore, a promoter upstream of the cvaA gene was presumably responsible for transcription of both genes. Though the previously identified Fur binding site for cvi was a close match to the consensus, that identified for cvaA was a weak match to the consensus (Fig. 3B). Therefore, there was some question as to whether cvaA was actually subject to Fur repression. This study set out to assess the iron dependence of Cvi and CvaA production and to characterize the predicted iron-responsive promoters of the cvaA and cvi genes of the ColV operons.

It was shown that production of both Cvi and CvaA is induced by iron depletion (Fig. 1A). It was also found that the transcription of cvaA is indeed iron regulated, but it begins from a promoter well upstream from that previously identified and lies very close to a sequence (FB-A1) which is a close match to the consensus Fur binding site (Fig. 3B). Results of three independent methods, primer extension, S1 nuclease protection, and transcriptional -galactosidase fusions, used to identify the cvaA promoter support the conclusion that transcription of cvaA occurs from one promoter, P1, located 320 bp upstream of its translational start. In contrast to cvaA, this study verified that the cvi promoter resides exactly where it was predicted, about 50 bp upstream of the cvi gene, within the previously identified Fur binding site (FB-I) (14). Transcription from both the cvi and cvaA promoters is highly induced in the logarithmic phase of growth by iron depletion.

Transcriptional promoter fusions for cvaA revealed that transcription from P1 alone is inducible 10-fold and considerably greater than transcription from P1 with downstream sequences present (ALZ-1' and ALZ-8) (Fig. 5A). It is interesting that including the P2 region with P1 enhances the transcriptional activity as well as modulating the relative induction to fivefold. This increase may be due to inherent flexibility in the AT-rich P2 region. Maximal promoter functioning is facilitated by contacts with DNA segments both upstream and downstream of the 10 and 35 promoter elements (7). In light of this finding, P2 flexibility immediately downstream of the promoter may contribute to enhanced DNA contacts with RNA polymerase, which could explain why transcription from the P1 promoter with the P2 region present is enhanced with both iron and dipyridyl. The potential flexibility in the P2 region might also interfere with the binding of Fur, accounting for the more than doubled transcription with iron when P2 is included.

We also addressed the possibility that transcription of cvaA is influenced by translation of potential leader peptides which might alter the formation of the LP3 secondary structure. However, mutations of the start codons of either leader did not result in any remarkable change in -galactosidase activities. Thus, it does not appear that translation of a leader peptide occurs or contributes to regulation of this system. Interestingly, a double mutation of both leader peptides decreased the induction. It is interesting that the double mutation changes two ATGs to ACGs. The effect of changing two bases from T to C in the enhancer region may affect the flexibility of the DNA, resulting in decreased transcription.

When included downstream of P1P2, the P3 region reduces transcription from P1 considerably. In reverse, the deletion or absence of sequences downstream of P1P2 results in fourfold-higher transcription from P1P2. This observation appears similar to those of early studies of the attenuation mechanisms for the histidine (his) and tryptophan (trp) operons, which showed increased expression of the structural genes with the deletion of a sequence similarly positioned between the promoter/operator and the structural genes (1, 20). As with the deletion of the P3 region, the deletion-dependent increases observed for his and trp are independent of their normal repressible nature; levels of both basal and derepressed transcription are increased. The deleted sequences for trp and his encode a rho-independent transcriptional terminator which controls the operon by attenuation. However, the up-regulation observed for the P3 deletion is much less than that observed for rho-independent terminators. The P3 region combined with upstream sequences forms a stable stem-loop structure, LP3. Although the AU-rich stem region of LP3 does not closely resemble a typical rho-independent terminator, which is much shorter and rich in CG base pairing, it is followed by a U-rich stretch, typical of most terminators (31). The deletion of a large portion of the LP3 sequence contributes to considerable derepression. Thus, this sequence is presumably responsible for the down-regulation.

This possibility was confirmed by testing the modulating properties of the cvaA sequences on the cvi promoter. The cvi-cvaA hybrid fusions confirm that the 114-bp P2 cvaA sequence (56 to 48) enhances transcription from the cvi promoter when it is placed downstream of it (Fig. 5D). The hybrid fusions also confirm that the cvaA sequences P2P3 plus 19 bp in P4 (bp 56 to 134) contribute to maximal down-modulation of transcription from cvi (ILZ-1-27). These sequences make up the entire LP3 secondary structure and following poly(U) sequence (Fig. 6B). Including the remaining secondary structure and the poly(U) sequence behind the cvi promoter results in even lower transcription than without the poly(U), an observation which supports a termination-like mechanism (Fig. 5A). However, without the poly(U) sequence, the cvaA sequences (56 to 115) down-regulate transcription considerably suggesting that the poly(U) enhances only the down-regulation. Alternatively, the secondary structure may interfere with translation. Increased secondary structure in the 5' untranslated regions of eukaryotic mRNA has been shown to decrease the efficiency of translation (23, 29). The same phenomena may be responsible for the down-regulation in this case since including potential secondary structure in the LP3 sequences contribute to considerable down-regulation. Secondary structures in both the 5' and 3' untranslated regions of RNA are known to affect transcript stability (3, 5, 22, 24).

In speculating on the physiological role of the down-regulation observed for the cvaA promoter, it is interesting to note the transcriptional activities of both the cvi and cvaA promoters in similar lacZ fusions. Fusions of the cvi promoter alone to lacZ are induced fourfold (up to 250 Miller units). Under the same conditions, these levels for cvi are similar to those for the cvaA promoter fusion with downstream sequences, which are induced fourfold (up to 220 Miller units). Without the downstream sequences, inducible transcription from the cvaA promoter is 916 Miller units, fourfold higher than for cvaA with downstream sequences or cvi alone. Therefore, the downstream sequences appear to regulate transcription of cvaA to levels similar to that of cvi. Evidence indicates that cvaB may be cotranscribed with cvaA, and cvaC also appears to be cotranscribed with cvi (unpublished observations). Therefore, the down-regulating properties of the cvaA downstream sequences appear to equalize transcription of the two operons. This feature may be important for optimal function of secretion. However, it is not certain whether differences in translational efficiency are present. It is also possible that transcriptional down-regulation of the cvaA-cvaB operon is a feature designed to protect the host cell. The protein sequence for CvaB predicts that it is an integral inner membrane protein with six membrane-spanning domains (14). Furthermore, evidence indicates that CvaB may be detrimental to the cell if overexpressed since attempts to overproduce it have failed (unpublished observations). This is not unusual for transmembrane secretory proteins. The SecY protein with 10 transmembrane domains is deleterious to E. coli cells when overproduced and/or unassociated with its complex partners (21). Thus, the down-regulating properties of the sequences downstream of the cvaA promoter may also be essential for cell survival by placing a limit on the amount of CvaB produced in addition to coordinating equivalent transcription of the genes involved in ColV production and secretion. Additionally, since there is the capacity for fourfold-elevated transcription from the cvaA promoter which is decreased under these conditions, there may be unknown conditions, such as severe stress, whereby transcription of the secretory genes must be significantly heightened, such that the need for rapid production outweighs preserving the safety of the cells from overproduction of the secretion genes. In this case, some unknown factor(s) may interfere with the down-regulation by the downstream sequences.

In all, this study identifies the two iron-regulated promoters of the ColV operons and reveals a unusual mechanism of regulation for fine-tuning transcription of a gene involved in secretion of the bacterial toxin ColV. The observation that transcription from the promoter alone is significantly higher than with the modulating downstream sequences suggests that there may be unknown mechanisms whereby the down-regulation is averted. The exact nature of the down-modulation is under investigation.