|Home | About | Journals | Submit | Contact Us | Français|
Colicin E2-tolerant (known as Cet2) Escherichia coli K-12 mutants overproduce an inner membrane protein, CreD, which is believed to cause the Cet2 phenotype. Here, we show that overproduction of CreD in a Cet2 strain results from hyperactivation of the CreBC two-component regulator, but CreD overproduction is not responsible for the Cet2 phenotype. Through microarray analysis and gene knockout and overexpression studies, we show that overexpression of another CreBC-regulated gene, yieJ (also known as cbrC), causes the Cet2 phenotype.
Colicins are protein antibiotics that have various modes of action. They are usually encoded on plasmids and, in many cases, alongside genes encoding colicin immunity factors, which protect colicin-producing cells from the colicin they produce. Of the enzymatic (E) colicins, some carry nuclease activity, including colicin E2, colicin E9, and colicin E3. These three proteins bind to susceptible cells via the surface protein BtuB (the vitamin B12 importer) and, through a series of events that are poorly understood, cross the cell envelope to enter the cytoplasm, where they degrade nucleic acids: colicins E2 and E9 target DNA; colicin E3 targets rRNA (11).
Cells can readily become tolerant of E colicins. Mutants usually have lost either the colicin receptor or some protein involved in colicin import. Loss-of-function mutations in btuB confer tolerance of high levels of colicins E2, E9, and E3. Almost 40 years ago, Escherichia coli mutants having a colicin E2-tolerant (Cet2) phenotype were identified. The Cet2 phenotype confers tolerance of colicins E2 and E9 only, while cells remain susceptible to colicin E3, and BtuB is intact (8, 9). Cet2 mutants were shown to overproduce an inner membrane protein (26), and the cet2 mutation was found to be dominant in trans and mapped at 99.9 min on the E. coli chromosome (8, 9). Using the Cet2 mutant RB208 as a source of genomic DNA, a clone able to transform E. coli cells to a Cet2 phenotype was identified. Since this clone carried a gene predicted to encode an inner membrane protein with properties identical to those overproduced in Cet2 mutants, the gene was named cet (15).
The cet gene is the last gene in the four-gene cre locus, so cet is also known as creD. The other genes in this locus are creA (hypothetical open reading frame [ORF]); creB, encoding a response regulator; and creC, encoding a sensor kinase. CreB and CreC form a classical two-component regulatory system, and we recently showed that CreBC are activated upon fermentation of glucose in minimal medium or during aerobic growth on minimal medium containing fermentation products, such as pyruvate, lactate, or acetate, as the sole carbon and energy source (10). CreBC controls the expression of a number of genes (the Cre regulon), some of which encode metabolic functions but several of which are hypothetical. One of the most tightly controlled Cre regulon genes is creD (5).
We have previously shown that the Cet2 strain RB208 has a point mutation in creC but that creD itself is wild type (5). Since the RB208 genomic clone capable of transforming cells to a Cet2 phenotype carries the whole cre locus, not just creD (15), our hypothesis is that the Cet2 phenotype of the transformant was due to a trans-dominant mutation in the cloned creC mutant allele activating one or more Cre regulon genes and that the Cet2 phenotype may or may not be caused by overexpression of creD. The aims of the experiments described in this paper were to test our hypothesis that the Cet2 phenotype is caused by activating mutations in CreBC and to definitively identify the Cre regulon gene that encodes the colicin E2 tolerance (Cet) protein.
All E. coli strains used in this study were obtained from the E. coli Genetic Stock Center. The wild-type E. coli strain was MG1655 (F− λ−) (CGSC 7740). Disruption of MG1655 chromosomal genes was performed either de novo or by using P1 transduction to replace the wild-type chromosomal gene with a disrupted version marked with a kanamycin resistance gene cassette from the appropriate mutant strain in the Keio collection (7, 13) (Table (Table1).1). Strains were routinely cultured at 37°C in LB broth or on LB agar (Oxoid Ltd., Basingstoke, United Kingdom). M9 minimal salts medium was used in some cases and was prepared using a base of 6 g/liter Na2HPO4, 3 g/liter KH2PO4, 1 g/liter NH4Cl, and 0.5 g/liter NaCl in water. Liquid cultures were grown with vigorous aeration (150 rpm) in conical flasks with foam bungs, where the culture occupied one-fifth of the total flask volume. All chemicals were obtained from Sigma-Aldrich Ltd. (Poole, United Kingdom).
MG1655 was transformed to chloramphenicol resistance (25 μg/ml) using plasmid pUB5962, carrying the cepH β-lactamase gene from Aeromonas hydrophila (3). Spontaneous MG1655(pUB5962) mutants were selected on LB agar containing 1 μg/ml (8 times the MIC) of cefotaxime (CTX) and 25 μg/ml chloramphenicol. The MIC of cefotaxime against the mutants was determined with a broth dilution method using LB broth. Derivatives of the CTX-resistant mutants that had spontaneously lost pUB5962 were selected following serial passage in LB broth containing no antibiotics. Loss of pUB5962 was confirmed by replica plating onto LB agar containing chloramphenicol (25 μg/ml) and PCR for cepH according to the method previously described (4), using the cepH internal primers previously used for reverse transcriptase PCR (RT-PCR) (5).
For preparation of P1 lysates, 5 ml LB-Ca2+ medium (LB broth supplemented with 2 mM CaCl2) was inoculated with 0.5 ml overnight LB broth culture of the donor Keio collection mutant (selected using 25 μg/ml kanamycin) and incubated at 37°C with aeration for 4 h. Each donor culture (0.1 ml) was then added to 0.1 ml of a 10−3, 10−4, or 10−5 dilution (diluted in LB-Ca2+) of stock P1 lysate in tubes containing 1 ml molten LB agar with 0.2% (wt/vol) glucose and 2 mM CaCl2 that were being kept at 45°C. Two milliliters of LB-Ca2+ broth was then added to each, and the mixture was vortexed and poured on top of set LB agar prior to incubation overnight at 37°C to reveal phage plaques. Two milliliters of LB-Ca2+ was then used to break up the soft upper agar layer containing these plaques, the pieces were added to 1 ml chloroform, the sample was homogenized on ice until it was smooth, and the sludge was centrifuged at 5,000 × g for 15 min (4°C). The clear supernatant containing P1 lysate was then stored at 4°C until it was required.
For transduction, 5 ml LB-Ca2+ was inoculated with 0.1 ml of an overnight culture of the recipient strain (CTX6) and incubated at 37°C with aeration for 4 h. Cells were harvested by centrifugation at 5,000 × g for 15 min at 4°C and resuspended in 0.5 ml LB-Ca2+. A 20% (vol/vol) dilution of the appropriate P1 lysate (in LB-Ca2+) was added to the cells, which were then incubated at 37°C without shaking for 20 min. M9 minimal salts medium (1 ml) was then added to the cells before they were harvested by centrifugation at 5,000 × g for 10 min (4°C). The cells were then resuspended in 4 ml M9 minimal salts medium before being pelleted as before. Following two further washes in 4 ml M9 minimal salts medium, the cells were resuspended in 2 ml LB broth, incubated at 37°C for 1 h with shaking, and harvested by centrifugation at 5,000 × g for 10 min (4°C). Each pellet was then resuspended in 0.2 ml M9 minimal salts medium, and the entire cell suspension was plated onto LB agar supplemented with kanamycin (25 μg/ml) and incubated at 37°C overnight. Transductants were checked by PCR in comparison with the wild type using primers flanking the gene of interest. The primers were identical to those used for checking the disruptions in the Keio collection mutants (7).
De novo disruption and deletion of chromosomal genes in E. coli MG1655 or CTX6 were performed using the PCR-mediated method of Datsenko and Wanner (13). Disruption and deletion of creB, creC, and creD were performed and checked using the primers previously described (10). Amplification of the disruption construct for lacZ was performed using the folowing primers: lacZ_F_KO, 5′-GGAATTGTGAGCGGATAACAATTTCACACAGGAAACAGCTGTGTAGGCTGGAGCTGCTTC-3′, and lacZ_R_KO, 5′-TTACGCGAAATACGGGCAGACATGGCCTGCCCGGTTATTACATATGAATATCCTCCTTAG-3′; the primers used for amplification of the disruption construct for yieJ were as follows: yieJ_F_KO, 5′-CTTTATCTTTGGGCTACTCAAAAGCAGACAGGATGTTTCTGTGTAGGCTGGAGCTGCTTC-3′, and yieJ_R_KO, 5′-GTGTGAATTACGCTCCGGCCTGTTCTCATTATTTAAATAACATATGAATATCCTCCTTAG-3′. To avoid possible polar effects and to delete the target gene, the resistance cassettes used to disrupt the genes were excised using the pCP20-encoded FLP recombinase as described elsewhere (13). Gene-specific primers used to check the deletion constructs were, for lacZ, lacZ_F_chk, 5′-GGCGCCCAATACGCAAACCG-3′, and lacZ_R_chk 5′-GAATAATAGCGAGAACAGAG-3′; for yieJ, they were yieJ_F_chk, 5′-GCTTCCTCGGAGTTGTTT-3′, and yieJ_R_chk, 5′-TGACAGCTACGTGACGAT-3′.
Estimation of Cre regulon gene expression levels was performed using the creD-lacZ reporter construct pUB6070, as previously validated and described (10). To allow these assays to be carried out with MG1655 and its derivatives, lacZ was first disrupted in each test strain, using the method described above.
An overnight LB broth culture of E. coli C600 carrying a colicin E2-encoding plasmid, ColE2-P9 (17), was diluted 1/10 with fresh broth. Using a sterile swab, a single streak of the diluted culture was applied down the center of a square LB agar plate. Following overnight incubation at 37°C, the lid was removed from the plate, which was overturned so that its surface was held 2 cm above a reservoir of chloroform for 15 min. This simultaneously killed the colicin E2-producing cells and released colicin E2 into the agar. Next, an overnight LB broth culture of each test strain was diluted 1/1,000 in fresh LB broth, and each was streaked once from left to right across the width of the plate at 90° to the colicin E2 streak. The plate was then incubated for a further 16 to 18 h at 37°C. Tolerance of colicin E2 was measured semiquantitatively by how close the test strain could grow to the colicin E2 streak relative to internal controls.
The method used to partially purify colicin E2 was adapted from Herschman and Helinski (18). In brief, a 500-ml LB broth culture of E. coli C600 carrying plasmid Col E2-E9 (17) was incubated overnight with shaking, and the cells were pelleted by centrifugation at 4,000 × g and 4°C for 10 min. The pellet was washed by resuspending the cells in 25 ml of 0.01 M potassium phosphate buffer, pH 7.0, containing 1 M NaCl, followed by recentrifugation as described above. The washed pellet was resuspended in 25 ml 0.01 M potassium phosphate buffer, pH 7.0, containing 1 M NaCl and 250 μl of EDTA-free protease inhibitor cocktail (Roche, West Sussex, United Kingdom), and the cells were lysed by sonication (four pulses of 20 s, separated by 20-s intervals on ice). The cell debris was pelleted by centrifugation at 5,000 × g and 4°C for 10 min, and ammonium sulfate was added to the supernatant to a final concentration of 20% (wt/vol). The mixture was stirred for 60 min at 4°C and clarified by centrifugation at 16,000 × g and 4°C for 10 min. Ammonium sulfate was added to the supernatant to give a total concentration of 40% (wt/vol), the mixture was stirred, and the precipitant was collected by centrifugation as before. The pellet was dissolved in 3 ml of 0.01 M potassium phosphate buffer, pH 7.0, and then clarified by centrifugation for 10 min at 48,000 × g and 4°C. The resulting colicin-containing supernatant was dialyzed against 1 liter of 0.01 M potassium phosphate buffer, pH 7.0, overnight at 4°C and clarified again by centrifugation for 10 min at 48,000 × g and 4°C. The colicin susceptibility was determined by adding doubling amounts of the colicin-containing supernatant to LB agar plates, followed by multispot inoculation of all test bacteria on the same plate using the standard BSAC agar dilution antibiotic susceptibility testing method (2). After incubation for 18 h at 37°C, the colicin susceptibility was recorded as the lowest dilution of colicin-containing supernatant required to totally inhibit growth of a test bacterium.
All strains used for microarray analysis were tested for unexpected gross mutations using comparative genomic hybridization as described elsewhere (19). Preparation of total RNA, labeling, hybridization, and analysis of array data were done as described previously (12), with the following modifications. RNA was extracted from three independent cultures (24 ml) of each strain grown to mid-logarithmic growth phase (A600 = 0.4). Each 24-ml culture was mixed with 48 ml of RNAprotect reagent (Qiagen Ltd.), and an RNeasy Midi kit was used to prepare total RNA according to the manufacturer's instructions (Qiagen Ltd.). Any contaminating DNA was removed from RNA samples while they were on the purification column, using RNase-free DNase (Qiagen Ltd.). Total RNA was reverse transcribed and labeled indirectly to give Cy3- and Cy5-labeled cDNAs, respectively, using the Amersham Cyscribe labeling kit, before being hybridized onto Corning Ultra GAPS glass slides printed with the 6,112 70-mer oligonucleotide Operon Array-Ready E. coli set 1.0 as described previously (12). The slides were hybridized using an Advalytix hybridization station, according to the manufacturer's instructions, and washed as described previously (12). The slides were scanned using an Axon 4100B scanner, and the data were analyzed using GenePix and Genespring software (Silicon Genetics) as described previously (12). Only genes differently expressed >3-fold with P values of <0.05 and within false-discovery rate (FDR) criteria (12) are reported in this study.
RT-PCR, including total RNA extraction and cDNA generation, was performed as described previously (10). A 1/1,000 dilution of cDNA was used for amplification of the 16S rRNA amplimer, whereas a 1/5 dilution of total cDNA was used to amplify yieI, yieJ, and creD sequences. Primers for 16S rRNA, yieI, and creD were as described previously (10). Amplification of yieJ was performed using primers yieJ_RT_F, 5′-AACCGTAGAGTGCGATTG-3′, and yieJ_RT_R, 5′-ACAGCGGAAGAGATAACC-3′. Following an initial denaturation at 95°C for 5 min, PCR (comprising sequential segments: 95°C for 30 s, 53°C for 30s, and 72°C for 1 min) was performed for 30 cycles. Changes in expression of yieI, yieJ, and creD were inferred from the intensity of the resultant PCR product under UV transillumination following agarose gel electrophoresis, where expression of the control 16S rRNA housekeeping gene was used as a cDNA loading control.
The yieJ coding sequence (including the stop codon) was amplified from MG1655 chromosomal DNA using primers yieJ_pBAD_F, 5′-CCATGGATGACTCAAAATATC-3′ (forward), and yieJ_pBAD_R, 5′-CCGGATGTGTGAATTACG-3′ (reverse). The yieJ amplimer was TOPO cloned in both the forward and reverse orientations into the pBAD vector (Invitrogen Ltd., Leek, Netherlands) according to the manufacturer's instructions to yield plasmids pUB6075 and pUB6076, respectively. Plasmid DNA was purified using the Qiagen midiprep kit according to the manufacturer's instructions (Qiagen Ltd., United Kingdom) and used to transform MG1655 to kanamycin resistance (25 μg/ml). Expression of yieJ from the pBAD vector was induced using arabinose (10 mM) as required.
We started this work with the hypothesis that Cet2 E. coli mutants have mutations that cause activation of CreBC, leading to overproduction of the colicin E2 tolerance protein, which is encoded by a CreBC-regulated (Cre regulon) gene. To test this hypothesis, we first set out to select CreBC hyperactive mutants using a method not involving colicin E2 and to test whether the mutants were tolerant of colicin E2.
Previous work in this laboratory showed that CreBC is able to activate transcription of cepH, a gene that encodes the class 1 cephalosporinase from A. hydrophila, when the gene is present in E. coli alongside upstream DNA sequence (5, 6). The reason for this is that the CreB response regulator targets genes having an upstream TTCACNNNNNNTTCAC (Cre tag) motif (where N is any base) (9). In A. hydrophila, cepH expression is controlled by the BlrAB two-component system. The signal sensor BlrB is 65% identical to CreC (21), and the response regulator, BlrA, is 63% identical to CreB (1, 21) and uses the same Cre tag motif to bind DNA and thus activate gene expression (6). Hence, cepH effectively has an upstream Cre tag and becomes part of the Cre regulon when cloned into E. coli (5, 6).
Overproduction of CepH confers cephalosporin resistance on E. coli (3). Thus, in order to obtain mutants with hyperactivation of CreBC, we selected cefotaxime-resistant (CTXr) mutants of E. coli MG1655 carrying a plasmid-borne cepH (pUB5962) (3). Mutation to CTXr (using CTX at 1 μg/ml for the selection of mutants) occurred at a frequency of 6 × 10−8. It was not possible to select CTXr mutants at a detectable frequency when the parent strain was MG1655ΔcreC or MG1655ΔcreB carrying pUB5962, consistent with activation of CreBC (either directly by mutation or indirectly, following mutation elsewhere) being the only way of obtaining CepH-hyperproducing mutants. The CTX MICs against 10 randomly selected CTXr mutants of MG1655(pUB5962) were determined to be 4 to 8 μg/ml. One representative mutant, CTX6(pUB5962), was chosen for further study. We selected CTX6(pUB5962) derivatives that had spontaneously lost pUB5962, as described in Materials and Methods, and CTX6 was found to have a point mutation causing an R247L substitution in the previously located (5, 23) CreC histidine kinase domain.
In order to test the effects of this creC mutation in CTX6, we employed a previously described reporter of creD promoter activity (plasmid pUB6070), which measures activation of Cre regulon gene expression (10). To do so, we made an unmarked lacZ deletion mutant of CTX6 and introduced pUB6070 by electroporation. As expected (10), Cre regulon reporter gene expression was found to be very low during growth of MG1655ΔlacZ(pUB6070) in LB broth. However, CTX6ΔlacZ(pUB6070) showed dramatic upregulation of Cre regulon reporter gene expression in this medium (Fig. (Fig.1).1). This upregulation was entirely dependent on creB and creC, since CTX6ΔlacZΔcreB(pUB6070) and CTX6ΔlacZΔcreC(pUB6070) expressed β-galactosidase at basal levels (Fig. (Fig.1).1). These data indicate that in CTX6, the R247L change found in CreC causes constitutive activation of CreBC signaling and Cre regulon gene expression.
We next tested the colicin E2 tolerance phenotypes of the strains and found that, as we had hypothesized would be the case, CTX6 is considerably more tolerant of colicin E2 than its parent strain, MG1655 (Fig. (Fig.2).2). Colicin E2 susceptibility similar to that of MG1655 was regained upon deletion of creB (Fig. (Fig.2)2) or creC (data not shown) in a CTX6 background, but deletion of creD did not affect colicin E2 tolerance in CTX6 (Fig. (Fig.2),2), so the previously identified cet gene does not actually encode a colicin E2 tolerance protein. These data were validated using agar dilution assays of colicin E2 susceptibility as described in Materials and Methods. MG1655, CTX6ΔcreB, and CTX6ΔcreC were all equally susceptible to colicin E2 at a maximum dilution that was four doublings (16-fold) lower than that of CTX6 and CTX6ΔcreD, which were equally tolerant. These data suggest that a Cre regulon gene (or genes) other than creD is responsible for CreBC-dependent colicin E2 tolerance.
In order to identify all Cre regulon genes and so focus on a possible colicin E2 tolerance protein(s), we employed global expression profiling of the CreC hyperactive mutant CTX6 versus CTX6ΔcreB. We used aerobic growth on M9 minimal medium containing 60 mM glycerol for this experiment, because it is a defined growth condition that does not activate CreBC (10), meaning that observed gene expression changes would be solely due to the Cet2 mutation. The expression-profiling studies revealed that a total of 9 genes were upregulated >3-fold (applying statistical analysis as described in Materials and Methods) in CTX6 relative to CTX6ΔcreB (Table (Table2).2). These genes included three previously identified (by RT-PCR ) Cre regulon genes, yieI (also known as cbrB ), creD, and yidS (also known as cbrA ), whose expression was upregulated more than 30-fold in the array experiment (Table (Table2).2). The genes found to be upregulated more than 10-fold were ynaI and mppA, predicted to be part of the same operon, and yieJ (also known as cbrC ), predicted to form an operon with the known Cre regulon gene, yieI (cbrB). The majority of the upregulated genes in CTX6 had unknown or putative functions (Table (Table22).
To identify the gene(s) responsible for colicin E2 tolerance, the 8 genes other than creD that were significantly upregulated in CTX6 relative to CTX6ΔcreB (Table (Table2)2) were systematically disrupted in CTX6 by P1 transduction of the kanamycin resistance cassette insertion from the Keio collection of mutants (7). The only gene whose disruption caused CTX6 to become more susceptible to colicin E2 was yieJ (Fig. (Fig.3).3). This observation was confirmed using quantitative colicin susceptibility assays: all the CTX6 disruption mutants had the same susceptibility as CTX6, except CTX6yieJ(Knr), which had 16-fold-increased susceptibility—the same susceptibility as MG1655, CTX6ΔcreB, and CTX6ΔcreB. To rule out the possibility that the effect observed with CTX6yieJ(Knr) was due to some other mutation transferred during the transduction process, a yieJ disruption mutant of CTX6 was made de novo and the kanamycin cassette was removed, effectively deleting yieJ, as described in Materials and Methods. CTX6ΔyieJ was found to be as susceptible to colicin as CTX6ΔcreB (Fig. (Fig.4).4). We also deleted yieJ in the parent strain, MG1655. This had no discernible effect on colicin E2 tolerance (data not shown), as would be expected given that positively regulated Cre regulon genes are not expressed at significant levels under growth conditions in which CreBC are inactive, e.g., growth on LB agar, as used for the colicin E2 susceptibility assays (5).
To validate the microarray data, RT-PCR was used to confirm upregulation of yieJ in CTX6 at a level similar to that seen for the known Cre regulon genes creD and yieI (Fig. (Fig.5).5). Expression of creD and yieI in CTX6 was unaffected by disruption of yieJ, so this gene does not confer colicin E2 tolerance by modulating CreBC activity or Cre regulon gene expression. The predicted 899-bp product was obtained from CTX6 total cDNA when PCR using the yieI forward and yieJ reverse RT-PCR primers was performed (not shown), confirming that yieI and yieJ are cotranscribed. However, disruption of yieI did not affect colicin E2 susceptibility (Fig. (Fig.3)3) or yieJ expression (Fig. (Fig.5)5) to any significant degree, meaning that there is no significant polar effect of yieI disruption on yieJ expression, despite their being part of the same operon.
To test whether overproduction of YieJ is sufficient to confer colicin E2 tolerance on an otherwise susceptible strain, we used the pBAD expression vector to overexpress yieJ in an MG1655 derivative in which yieJ had been deleted. The yieJ-expressing plasmid was called pUB6075. When arabinose was present in the medium (to activate yieJ expression), MG1655ΔyieJ(pUB6075) was considerably more tolerant of colicin E2 than MG1655ΔyieJ carrying plasmid pUB6076, i.e., pBAD with yieJ ligated in the opposite orientation (Fig. (Fig.66 A). By quantitative assay, yieJ overexpression using the arabinose induction method conferred an 8-fold (three doublings) increase in colicin tolerance. Increased colicin E2 tolerance in MG1655ΔyieJ(pUB6075) relative to MG1655ΔyieJ(pUB6076) was not observed in the absence of arabinose (Fig. (Fig.6B).6B). That overexpression of yieJ was achieved in MG1655ΔyieJ(pUB6075) upon arabinose induction but did not affect Cre regulon gene expression per se was demonstrated by RT-PCR (Fig. (Fig.77).
The creD gene is annotated in many online resources as encoding a protein responsible for tolerance of colicin E2. The source of this annotation is a paper that linked three observations: (i) that colicin E2 tolerant (Cet2) mutants such as RB208 overproduce an inner membrane protein with a molecular mass of approximately 45 kDa; (ii) that a genomic clone from RB208 capable of transforming a wild-type strain to Cet2 carries a gene predicted to encode a membrane protein of approximately 49 kDa; (iii) that RB208 and the Cet2 transformant both overexpress this putative membrane protein gene. The conclusion was that RB208 carries a mutation in this gene, or nearby, that causes the gene to be overexpressed and that this overexpression confers a Cet2 phenotype, so the gene was named cet (15). Following identification of cet as the fourth gene in the cre operon, it was given its alternative name, creD (5).
While these three observations (15) are entirely reproducible, we now know that RB208 has a wild-type creD sequence, but its creC has a point mutation conferring a Thr264Ser change immediately next to the predicted catalytic histidine of CreC (5). Furthermore, the RB208 genomic clone conferring a Cet2 phenotype in trans also carries this hyperactive creC mutant allele (15). Indeed, we showed in this paper that selecting a CreC hyperactive mutant by means not involving colicin E2 exposure generates a Cet2 mutant. We can go even further and say that while Cet2 strains do overproduce CreD, the gene disruption experiments reported above show that CreD overproduction does not cause the Cet2 phenotype. Accordingly, we suggest that any mention of colicin E2 tolerance should be removed from the annotation of creD in the various databases. Its true role remains to be found. We would also propose adding to the database entries for CreC that activation of this protein by mutation causes colicin E2 tolerance.
After discounting a role for CreD, we have provided definitive evidence that overproduction of YieJ in E. coli facilitates tolerance of colicin E2. We are confident that yieJ can be directly assigned to the Cre regulon because it shares with yieI a promoter containing a Cre tag sequence (5), which is the binding site of CreB (10), and its expression is upregulated in a CreB-dependent manner upon activation of CreC (Table (Table2).2). In fact, it has been stated previously, though without presenting any evidence, that yieJ is part of the Cre regulon, giving rise to its alternative name, cbrC, for “CreBC regulated” (28). This alternative name does not communicate a function, so it does not conform to the conventions of naming E. coli genes originally described by Demerec and colleagues (14) and recently reaffirmed by a consortium of E. coli researchers (25). We are now closer to a function for yieJ, but until the mechanism by which it causes colicin E2 tolerance is determined, we propose to keep the name yieJ. Recently, a screen for mutants that affect the colicin susceptibility of E. coli was published, but yieJ was not found to be among them (27). The explanation for this is that the screen used E. coli growing on complex media, where CreBC are not active, and so positively regulated Cre regulon genes were not expressed (5, 10).
Another recent report (20) has shown that when the arcA response regulator gene was deleted in a creC510 mutant encoding a hyperactive CreC (5), there was a greater effect on metabolism than if arcA was deleted in a creC wild-type strain. This highlights the possibility that there is some as-yet-undefined interaction between CreBC and ArcAB in vivo. It was interesting, therefore, to find that five known Cre regulon genes were not overexpressed in our CreC hyperactive mutant, CTX6, during growth under the conditions chosen for gene expression profiling. These other Cre regulon genes (ackA, pta, talA, radC, and trgB) were identified using RT-PCR following activation of CreBC by growing cells anaerobically on glucose minimal medium (5). However, the gene expression-profiling studies reported here deliberately used growth conditions that do not activate CreBC (aerobic growth on glycerol minimal medium). One would expect that metabolic regulators other than CreBC, including ArcAB, would have their activities altered upon switching between aerobic and anaerobic growth conditions. Thus, our observation fits with the idea that simply activating CreC through mutation does not reveal the full extent of the Cre regulon. Instead, the complete Cre regulon is seen only under conditions where CreC is activated and the activity of ArcA (and possibly other regulators) is also modified, perhaps releasing a block on activation of gene expression by CreB. Indeed, in spite of the small number and uncertain functions of the Cre regulon genes seen overexpressed in the Cet2 mutant, CTX6, in our current study on the cause of colicin E2 tolerance, the importance of CreBC as a metabolic regulator has been elegantly underlined in a recent analysis of metabolic flux (21).
We are aware of three other global-expression studies that have also reported perturbations in Cre regulon gene expression. The first involved treatment of cells with the antibiotic 4,5-dihydroxy-2-cyclopenten-1-one (24). The second involved activation of the envelope stress two-component system BaeSR (22). The third involved deletion of the biofilm/motility regulator MqsR (16). In all three cases, the only previously assigned Cre regulon genes found to be significantly upregulated were creD, yieI, and yidS. In two out of three, yieJ was also upregulated. Taken together with our own microarray studies of a Cet2 mutant, this suggests that these four genes make up the core Cre regulon that becomes activated in the absence of metabolic changes that also affect other regulatory systems. The role of this core regulon in responding to these various signals is not known, but the link with colicin E2 tolerance makes us speculate that one role of the Cre regulon might be to cause modification of the envelope, since previously identified colicin E2 tolerance mechanisms are associated with reducing the ability of colicin E2 to cross the envelope (11).
This work was supported by grants BB/C514266 to M.B.A. at Bristol University and EGA16107 and JIF13209 to Birmingham University from the United Kingdom Biotechnology and Biological Sciences Research Council (BBSRC).
Published ahead of print on 23 April 2010.