Identification of ColR binding consensus and prediction of regulon of ColRS two-component system
© Kivistik et al; licensee BioMed Central Ltd. 2009
Received: 18 November 2008
Accepted: 16 May 2009
Published: 16 May 2009
Skip to main content
© Kivistik et al; licensee BioMed Central Ltd. 2009
Received: 18 November 2008
Accepted: 16 May 2009
Published: 16 May 2009
Conserved two-component system ColRS of Pseudomonas genus has been implicated in several unrelated phenotypes. For instance, deficiency of P. putida ColRS system results in lowered phenol tolerance, hindrance of transposition of Tn4652 and lysis of a subpopulation of glucose-grown bacteria. In order to discover molecular mechanisms behind these phenotypes, we focused here on identification of downstream components of ColRS signal transduction pathway.
First, highly similar ColR binding sites were mapped upstream of outer membrane protein-encoding oprQ and a putative methyltransferase-encoding PP0903. These two ColR binding sequences were used as an input in computational genome-wide screening for new potential ColR recognition boxes upstream of different genes in P. putida. Biological relevance of a set of in silico predicted ColR-binding sites was analysed in vivo by studying the effect of ColR on transcription from promoters carrying these sites. This analysis disclosed seven novel genes of which six were positively and one negatively regulated by ColR. Interestingly, all promoters tested responded more significantly to the over-expression than to the absence of ColR suggesting that either ColR is limiting or ColS-activating signal is low under the conditions applied. The binding sites of ColR in the promoters analysed were validated by gel mobility shift and/or DNase I footprinting assays. ColR binding consensus was defined according to seven ColR binding motifs mapped by DNase I protection assay and this consensus was used to predict minimal regulon of ColRS system.
Combined usage of experimental and computational approach enabled us to define the binding consensus for response regulator ColR and to discover several new ColR-regulated genes. For instance, genes of outer membrane lipid A 3-O-deacylase PagL and cytoplasmic membrane diacylglycerol kinase DgkA are the members of ColR regulon. Furthermore, over 40 genes were predicted to be putatively controlled by ColRS two-component system in P. putida. It is notable that many of ColR-regulated genes encode membrane-related products thus confirming the previously proposed role of ColRS system in regulation of membrane functionality.
Two-component signal systems are the main means for sensing the changing environment in a prokaryotic world . Typically, bacterial signal transduction systems consist of two components, a sensor histidine kinase and a response regulator. A specific compound or a physicochemical property of the environment acts as a signal triggering the activation of a membrane embedded sensor, which in turn autophosphorylates and thereafter passes the signal to a response regulator via phosphoryl group transfer . Phosphorylated response proteins mostly act as DNA binding transcription factors by activating or repressing the expression of target genes.
The number of genes for two-component proteins varies greatly between the genomes of sequenced bacteria, being for instance zero in case of Mycoplasma genitalium and 62 in a well-known model organism Escherichia coli . The abundance of two-component systems seems to correlate with environmental and pathogenic versatility of a bacterium. Pseudomonas bacteria that colonise different habitats such as soil, water, plants and animal tissues, possess many two-component signal systems to cope with various environments. For example, over a hundreds genes encoding two-component system proteins are present in the genome of Pseudomonas aeruginosa .
The ColRS two-component signal transduction system, intrinsic to Pseudomonas species, consists of a sensor kinase ColS and a response regulator ColR. colRS operon is well-conserved among the sequenced members of Pseudomonas genus  suggesting that ColRS system could be important to these bacteria. ColRS pathway was first characterised in P. fluorescens as a system involved in the ability of bacteria to competitively colonise plant roots . Our group observed a totally different function for ColRS system as transposition of Tn4652 was inhibited in phenol-starving colR- and colS-deficient P. putida . Additionally, recently, we demonstrated that colR-deficient P. putida is sensitive to phenol  and displays a serious defect on solid glucose medium where a subpopulation of the mutant lyses . ColRS system was also shown to be important in resistance of P. putida to divalent metal ions, especially Mn2+ . Importantly, the precise checkpoint of a response regulator ColR remained unclear in case of all these ColR-dependent phenotypes. However, two recent publications suggest that seemingly unrelated phenotypes of colR-deficient P. fluorescens and P. putida can most probably be explained by the compromised cell membranes of colR mutants. Namely, the first ColR-regulated genes that were identified in these two species, encoded different membrane functions. We demonstrated that oprQ and algD, which encode a porin protein and an exopolysaccharide alginate biosynthesis enzyme, respectively, are under the direct control of ColR in P. putida . Concomitantly, de Weert et al  reported for P. fluorescens that an operon downstream of colRS hypothetically coding for membrane associated proteins, methyltransferase (orf222) and lipopolysaccharide kinase (inaA/wapQ), is also regulated by ColR.
In spite of several recent studies disclosing the ColRS system as an important signal transduction pathway for pseudomonads, little is known about the downstream components of this signal system. Only two target genes of ColR have been identified in P. putida  and our unpublished data show that neither oprQ nor algD are involved in phenotypes characteristic to colR mutant. Thus, more ColR-regulated genes should be present in the genome of P. putida. In this study we aimed to determine the ColR binding consensus sequence and to use it in the screen of P. putida genome for the presence of new potential ColR binding sites, which would predict new ColR target genes and operons.
Despite the fact that ColR bound to the promoter DNA of algD in a gel mobility shift assay  we could not detect a ColR-binding site in this DNA by DNase I footprinting (data not shown). This suggests that the binding dynamics of response regulator ColR could be somewhat different if algD promoter is compared to the promoters of oprQ and PP0903. Most probably ColR recognition sequences in the algD promoter diverge from ColR consensus (see Discussion).
It is important to point out that one new ColR-regulated gene, PP0900, locates just upstream of colRS genes and is transcribed divergently from PP0901 encoding colR (Fig. 1a). Thus, the putative ColR-binding site affecting the expression of PP0900 is actually located between PP0900 and colR. Analysis of this particular promoter in the direction of colR (PP0901) did not reveal clear ColR-responsiveness under any conditions examined (Fig. 3). Therefore we conclude that ColR does not auto-regulate its own expression.
Seven sequences that were used for creating ColR binding consensus
ACGATAGTTCAGCCTTTT TTCACGTTTTTTTCAC AACGCAAAGCTTTGCA
conserved hypothetical protein
TGAAATGCCCACATAGGC TTCACACTTTTTTCAC AAGGACTGGCTCAGTA
TTACGCATTTACATTTAG TTCACTATTTTTTCAC CTTTGCCTCCATAGGC
GtrA family protein
GGGGAAAATTCACCGCAC TTGACGGTTTTTTCAC TGATCCTTCACGGACA
PAP2 family protein
GCATCGCATTATCAGCGG CTAAGATTTTTTTGAC GAAATATTGACAAGCT
secreted hemolysin-type calcium-binding bacteriocin, putative
GGTTATATCAACCTGCAT TTCACGACTTTTTGAC ATTTGTAATCGCACAC
outer membrane protein OprE3
TATCAGGTCAACGGGTGC CCAACGTTTTTTTCAC ATTGGCATGACCTGAT
conserved hypothetical protein
Two-component signal transduction systems reside at the top of regulatory cascades. Therefore, to decipher the role of a particular two-component pathway it is crucial to specify the downstream components of the cascade. This study focuses on identification of target genes of a Pseudomonas two-component system ColRS, thus serving as an indispensable step in the way of unravelling the mechanisms that trigger different phenotypes related to ColRS deficiency.
Previously, we have searched for ColR-regulated genes by using a promoter library created from total chromosomal DNA of P. putida . This screen disclosed only two ColR target genes, thus forcing us to try different approach to acquire more information about the putative regulon of ColRS system. We aimed to define a binding consensus for the transcription factor ColR and to search for similar sequences in the upstream regions of genes in the genome of P. putida. Seven experimentally verified ColR binding sites demonstrated a highly conserved ColR-binding motif with 8 fully conserved nucleotides in the 16-bp-long core binding box (Fig. 6). The binding motifs of OmpR subfamily response regulators, among whom ColR belongs http://www.pseudomonas.com, consist typically of direct repeats separated by four to five nucleotides . Careful inspection of ColR binding sites reveals that there are also two direct repeats. Although one of the repeats is much less conserved than the other, such structure of the binding site indicates that ColR binds to the DNA as a dimer. Notably, our data suggest that in addition to the highly conserved core sequence, ColR may also recognize and bind to less conserved sites. Namely, ColR regulates algD promoter in vivo and binds to this promoter in a phosphorylation-dependent manner according to gel mobility shift experiments . Yet, computational analysis could not find a good ColR binding motif and DNase I protection analysis did not locate ColR binding site in the promoter region of algD. Nevertheless, the sequence upstream of algD contains many T-rich tracks characteristic for ColR binding consensus. If we lowered the specificity of Virtual Footprint prediction by allowing three mismatches in the input consensus then four putative ColR binding sites were found in algD promoter region, which was previously shown to bind ColR in DNA shift assay (data not shown). Probably ColR binds to these less conserved and thereby with lower affinity sites, but DNA cleavage by DNase I destabilizes the ColR-DNA complex and hence the ColR protected area cannot be detected by DNase footprint assay. Analogously, the nucleoprotein complex between the ColR and the upstream region of PP1636 probably dissociates due to DNase I cleavage-caused destabilization.
Given that some promoter regions (e.g. those of PP1692 and PP2322/2323) contain putatively more than one ColR consensus-resembling site (Additional file 1) one may hypothesise that co-operative binding of ColR to several sites may be necessary for regulation of these ColR target promoters. In this connection it is interesting to note that in silico analysis of seven experimentally verified ColR-binding promoters, revealed more putative ColR boxes than identified by DNase footprint assay. Namely, promoter regions of oprQ, PP0900/0901 and PP2560/2561 contain additional less-conserved ColR boxes (data not shown). Further experiments should prove whether these additional sites really support ColR binding to these promoters.
In this study several new ColR-dependent promoters were identified. Intriguingly, the effect of ColR absence upon promoter activities was rather modest (PP0035, PP0900, PP0737, PP1636) or not detectable (PP0036, PP2560, PP3766) when bacteria grew on glucose medium (Fig. 3). The greatest effect was seen in case of PP0903 promoter whose activity was 10-fold down-regulated in a ColR-deficient P. putida (Fig. 1). However, the role of ColR in regulation of its target genes was confirmed by analysis of promoter activities under conditions of ColR over-expression. Indeed, most promoters tested were strongly influenced by ColR over-expression, especially when phenol was present in the growth medium (Fig. 1 and Fig. 3). This data indicate that the amount of transcriptionally active ColR is not sufficient enough to influence significantly most of its target promoters in wild-type P. putida in our assay conditions. Given that the phosphorylated form of ColR bound to its recognition sequences more avidly it is reasonable to conclude that actually the phosphorylated form of ColR is limiting in bacteria growing on glucose solid medium. Therefore, we consider that most probably the ColS-activating signal is low under our assay conditions. ColR over-expression could apparently mimic the conditions evoked by the signal and therefore the ColR-dependence of its target promoters was seen more clearly in the situation of ColR over-expression. This scenario is in good accordance with our previous data showing that over-expression of ColR can compensate the defect in signal transfer from ColS to ColR when the participation of ColRS system in regulation of Tn4652 transposition was examined . The signal sensed by ColS is not known so far. However, our observation that the impact of ColR on the expression of its target promoters was greater in the presence of phenol (Fig. 1 and Fig. 3) indicates that phenol-caused stress could be one of the conditions where ColRS signal transduction pathway is, at least partially, activated. This suggestion agrees with our previous results demonstrating the participation of ColRS system in phenol tolerance of P. putida .
Current study shows that ColR regulates directly genes locating upstream and downstream of its own gene. Like in P. fluorescens, ColR of P. putida also activates the downstream of colRS locating operon, that was suggested to be involved in fine-tuning of the outer membrane permeability . It is notable that ColR regulates several other outer membrane protein-encoding genes confirming the role of ColRS system in the regulation of membrane functionality [8, 9, 11]. In addition to our previous finding that outer membrane protein-encoding oprQ is negatively controlled by ColR , the current study revealed that ColR is a negative factor also for PP0737 which codes for a protein orthologous to lipid A 3-O-deacylase PagL in P. aeruginosa . The outer membrane-locating PagL modifies lipopolysaccharides by deacylation of lipid A at 3-O-position [18, 19]. PagL becomes important under specific conditions only, e.g., it will be needed for mutants deficient in aminoarabinose-modified lipid A to resist cationic antimicrobial peptides . In pathogenic Gram-negative bacteria the PagL-dependent deacylation of lipid A reduces the ability of lipid A to activate the Toll-like receptor 4 of the host, thus helping pathogens to avoid innate immune recognition . The role of PagL has not been studied in P. putida so far.
Our data show that besides affecting the composition of outer membrane, ColR also regulates other membrane compartments. For instance, ColR activated PP0900 and PP1636 coding for two cytoplasmic membrane-locating enzymes, putative type 2 phosphatidic acid phosphatase (PAP2) and diacylglycerol kinase DgkA, respectively. These two enzymes most probably affect the membrane lipid homeostasis as they reversely regulate the abundance of phosphatidic acid and diacylglycerol, the precursor of phospholipid synthesis and the by-product of the synthesis of membrane-derived oligosaccharides, respectively . Given that prediction of ColR regulon revealed another fatty acid and phospholipid metabolism related gene acpP (PP1915), coding for acyl carrier protein, as a putative ColR target gene, indeed, ColRS system may be involved in phospholipids homeostasis.
Since ColR is highly conserved among all Pseudomonas species  it was reasonable to presume that ColR-binding sites may also be similar in pseudomonads. Genome-wide prediction of potential ColR-binding sites in P. fluorescens PfO-1 disclosed seven genes that could be members of ColR regulon both in P. putida and P. fluorescens (see Additional file 2). Namely, ColR-binding box was found upstream of P. fluorescens PfO-1 genes orthologous to P. putida dgkA-1, colR/PP0900, PP0737, PP0903, PP1058, PP1692 and PP5152 (data not shown). Additional screening of P. aeruginosa PAO1, P. syringae tomato DC3000 and P. syringae phaseolicola 1448A revealed that all these organisms contain perfect ColR-recognition sites upstream of dgkA-1 orthologs. Therefore, it is highly possible that dgkA is a member of ColR regulon in all these pseudomonads.
Current study identified a 16-bp-long binding consensus of response regulator ColR ((T/C)(T/C)NA(C/G)NN(T/C)TTTTT(C/G)AC), which helped us to discover new genes controlled by ColRS two-component system in P. putida. Notably, several new ColR target genes (PP0035, PP0737, PP0900, PP1636 and PP2560) code for different membrane proteins supporting our previous assumption that the primary target of ColRS two-component system is the cell membrane. Regulon prediction suggests that ColR could regulate over 40 genes and many of them code for membrane-associated functions as well. However, it is not clear yet which of the target genes are responsible for specific ColR-related phenotypes such as lowered phenol tolerance, hindrance of transposition of Tn4652 and glucose-induced lysis of a subpopulation of the colR mutant [7–9]. Considering the number of ColR regulon genes it is highly possible that not one but several ColR target genes are involved in the formation of above-mentioned phenotypes. Given that ColRS two-component system is regulating several membrane-related genes, our further experiments are directed towards clarification the role of ColR target genes in the membrane functionality.
P. putida strain PaW85 , which is isogenic to fully sequenced KT2440 , its colR- deficient derivative PaWcolR  and PaWcolR derivative strain PaWRtaccolR capable of over-expression of ColR  were used in this study. E. coli strain DH5α (Invitrogen) was used for DNA manipulations. Bacteria were grown in Luria-Bertani (LB) medium  or in M9 minimal medium  containing 10 mM glucose. Concentration of phenol was 2.5 mM in minimal medium. When necessary, the growth medium was supplemented with ampicillin (100 μg ml-1) or chloramphenicol (20 μg ml-1) for Escherichia coli and with benzylpenicillin (1,500 μg ml-1), chloramphenicol (300 μg ml-1) or kanamycin (50 μg ml-1) for P. putida. X-gal (5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside) (75 μg ml-1) was added to the growth medium for visual evaluation of promoter activities. 0.5 mM IPTG (isopropyl β-D-1-thiogalactopyranoside) was used to induce Ptac promoter. P. putida was incubated at 30°C and E. coli at 37°C. E. coli and P. putida cells were electrotransformed according to the protocol of Sharma and Schimke .
For cloning of different promoter regions into promoter probe plasmid p9TTBlacZ  the PCR-amplified DNA fragments were used. Restriction site-containing oligonucleotides used in the process are listed in Additional file 3. Different promoter regions were amplified by PCR using the purified chromosomal DNA of P. putida PaW85 as a template. PCR fragments were restricted with the appropriate enzyme (Additional file 3) and cloned upstream of lacZ gene in the plasmid p9TTBlacZ. Orientation of a promoter fragment was verified by PCR.
All enzyme activities presented in this paper were measured from solid-medium-grown bacteria. Bacteria grown both on glucose or glucose plus 2.5 mM phenol containing M9 minimal medium were scraped off from the plates using toothpicks and suspended in M9 solution. For one suspension 24-hours-grown bacteria were collected from a sector comprising approximately one-twelfth of the Petri plate. β-galactosidase activity was assayed according to a previously described protocol .
ColR and N-terminally truncated ColS used in DNA gel mobility shift assay, were over-expressed and purified as His-tagged proteins by published protocol . Oligonucleotides used in PCR to generate DNA probes are listed in Additional file 3. Gel mobility shift assay was performed according to a previously described protocol .
DNA fragments for DNase I footprinting assay were amplified from the purified chromosomal DNA of P. putida PaW85 by PCR. Oligonucleotides used to generate DNA probes by PCR are listed in Additional file 3. One oligonucleotide was end-labelled by phosphorylation with [α-32P]-ATP and thus PCR reactions created products with specific labelling of one DNA strand. The labelled DNA fragments were purified by native 5% polyacrylamide gel electrophoresis, eluted (buffer containing 0.5 M NH4Ac, 10 mM MgAc, 1 mM EDTA and 0.1% SDS) and re-suspended in water. For the binding reaction, different amounts of purified P. putida his-tagged ColR protein (concentrations of ColR are specified in the text as they varied in case of different DNA probes) were combined with 30 000 c.p.m. of labelled DNA fragment, 25 mM Tris-HCl (pH 7.5), 10 mM MgCl2, 1 mM CaCl2, 0.1 mM EDTA, 50 mM KCl, 5 μg of BSA, 5 μg of salmon sperm DNA and 5% glycerol in a final volume of 100 μl. Reactions with different volume of proteins were equalized with the addition of appropriate amount of ColR storage buffer. To test the binding of phosphorylated ColR with the DNA probes, ColS was first autophosphorylated by incubation in the presence of 0.1 mM ATP in the reaction buffer for 15 min. After addition of ColR to the phosphorylated ColS and further incubation for 15 min, labelled DNA was added to the reaction mixture. ColR was allowed to bind to DNA during 20 min at room temperature before the start of digestion by DNase I (0.25 U, Fermentas) for 3 min. Reactions were stopped by the addition of 100 μl of a solution containing 0.1 M EDTA, 0.1% sodium dodecyl sulphate, 1.6 M ammonium acetate and 20 μg of sonicated salmon sperm DNA per ml. The footprinting reaction mixtures were subsequently extracted once with phenol and chloroform (1:1 v/v) and once with chloroform and, finally, the DNA was precipitated with ethanol. The DNA fragments were resuspended in 7 μl of sequence loading buffer (deionized formamide containing 10 mM EDTA, 0.3% bromophenol blue and 0.3% xylene cyanol) and loaded onto a 6.5% polyacrylamide gel that contained 8 M urea. DNA sequencing reactions were performed with a Sequenase version 2.0 kit (US Biochemicals) and were loaded on a sequencing gel as size markers. After the run, the gels were dried and exposed to a PhosphorImager screen (Amersham Biosciences).
Putative ColR binding sites in the genomes of P. putida KT2440,P. aeruginosa PAO1, P. fluorescens PfO-1, P. syringae tomato DC3000 and P. syringae phaseolicola 1448A were searched using two programs: the PredictRegulon server  and the Virtual Footprint server . Variations were made in the input sequence length and strand orientation. In Virtual Footprint predictions also different number of mismatches from the ColR binding consensus IUPAC code was allowed. Except for the parameters mentioned above, the programs were used with default settings. The new binding sites of P. putida ColR presented in this study were identified in a step-by-step process meaning that new predictions were made with every additional confirmed ColR binding site. Namely, first prediction with two input sequences (ColR sites in promoters of oprQ and PP0903) disclosed potential ColR binding sites in upstream regions of PP1636 and between divergently located PP0900 and PP0901 (colR). After experimental verification of ColR site between PP0900 and PP0901, the second round of prediction was performed with three input sequences resulting, for instance, in prediction of potential ColR sites upstream of PP0737 and PP0035. After verification of these sites, the third prediction was performed etc. Following such step-by-step process we were able to map seven ColR recognition sites, which were used as input in final prediction presented in Additional file 1.
We are grateful to Tiina Alamäe, Marta Putrinš, Heili Ilves and Hanna Hõrak for critically reading the manuscript. This work was supported by grants 6025 and 7829 from the Estonian Science Foundation to R.H., by funding of Targeted Financing Project TLOMR0031 from Estonian Ministry of Research and Education to M.K., and grant HHMI 55005614 from the Howard Hughes Medical Institute International Research Scholars Program to M.K.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.