Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Predicting Inter-Species Cross-Talk in Two-Component Signalling Systems

  • Sonja Pawelczyk,

    Affiliations Department of Biochemistry, Oxford Centre for Integrative Systems Biology, University of Oxford, Oxford, United Kingdom, Department of Biochemistry, University of Oxford, Oxford, United Kingdom

  • Kathryn A. Scott,

    Affiliations Department of Biochemistry, Oxford Centre for Integrative Systems Biology, University of Oxford, Oxford, United Kingdom, Department of Biochemistry, University of Oxford, Oxford, United Kingdom

  • Rebecca Hamer,

    Affiliations Department of Biochemistry, Oxford Centre for Integrative Systems Biology, University of Oxford, Oxford, United Kingdom, Department of Statistics, University of Oxford, Oxford, United Kingdom

  • Gareth Blades,

    Affiliations Department of Biochemistry, Oxford Centre for Integrative Systems Biology, University of Oxford, Oxford, United Kingdom, Department of Biochemistry, University of Oxford, Oxford, United Kingdom

  • Charlotte M. Deane,

    Affiliations Department of Biochemistry, Oxford Centre for Integrative Systems Biology, University of Oxford, Oxford, United Kingdom, Department of Statistics, University of Oxford, Oxford, United Kingdom

  • George H. Wadhams

    george.wadhams@bioch.ox.ac.uk

    Affiliations Department of Biochemistry, Oxford Centre for Integrative Systems Biology, University of Oxford, Oxford, United Kingdom, Department of Biochemistry, University of Oxford, Oxford, United Kingdom

Predicting Inter-Species Cross-Talk in Two-Component Signalling Systems

  • Sonja Pawelczyk, 
  • Kathryn A. Scott, 
  • Rebecca Hamer, 
  • Gareth Blades, 
  • Charlotte M. Deane, 
  • George H. Wadhams
PLOS
x

Abstract

Phosphosignalling pathways are an attractive option for the synthetic biologist looking for a wide repertoire of modular components from which to build. We demonstrate that two-component systems can be used in synthetic biology. However, their potential is limited by the fact that host cells contain many of their own phosphosignalling pathways and these may interact with, and cross-talk to, the introduced synthetic components. In this paper we also demonstrate a simple bioinformatic tool that can help predict whether interspecies cross-talk between introduced and native two-component signalling pathways will occur and show both in vitro and in vivo that the predicted interactions do take place. The ability to predict potential cross-talk prior to designing and constructing novel pathways or choosing a host organism is essential for the promise that phosphosignalling components hold for synthetic biology to be realised.

Introduction

Synthetic Biology aims at designing new or redesigning existing biological circuits for a particular purpose [1][2]. In most cases this is achieved by transplanting biological components (DNA or proteins) from one organism into a non-native host to implement the new or redesigned circuit within that host cell. An implicit condition for this is that the introduced components should function either independently to, or at least interact minimally and in a well characterised way with, the native host systems.

Many of the early circuits which have been created have used transcriptional activators and repressors to control gene expression in Escherichia coli [3][4]. However, the small number of these parts which are available for the synthetic biologist to use has forced many to look outside these traditional components and to start to try to implement pathways designed around phosphorylation based signalling, either using MAPK pathways in eukaryotic cells [5] or two-component signalling pathways in bacteria [6]. A potential limitation of these approaches is that the host cells contain many of their own phosphosignalling pathways and it is currently not possible to determine in advance whether any of these will interact with, and cross-talk to, the introduced synthetic components. Some cross-talk has been shown in the MAPK pathways for osmolarity and pheromone sensing in Saccharomyces cerevisiae where both pathways share a MAPKKK [7][8]. Inter-species cross-talk in bacterial systems has been shown from Enterococcus faecium to Escherichia coli, its sensor kinase VanS can activate the E. coli response regulator PhoB [9]. A recent study by Antunes et al. has shown possible cross-talk across kingdoms where heterologous expression of PhoB and OmpR from E. coli in Arabidopsis are phosphorylated by endogenous cytokinin-mediated HK-signaling components [10].

Here we focus on bacterial two-component systems (TCS) and develop and validate a method which allows potential cross-talk to be predicted prior to the design and implementation of a synthetic circuit in a particular host cell type. TCS are ubiquitous in bacteria and lower eukaryotes [11]. They are the main signal transduction pathways in these species and how an organism can coordinate the activity of so many highly related signalling systems is the focus of much attention [12]. TCS generally consist of a sensor histidine protein kinase (HPK) and a response regulator (RR). Classically HPKs contain a periplasmic sensory region, a transmembrane region and a cytoplasmic signalling region. The cytoplasmic part of the kinase consists of a HAMP linker domain (present in Histidine kinases, Adenyl cyclases, Methyl-accepting proteins and Phosphatases), a dimerisation and histidine phosphoaccepting domain (DHp) and a catatlytic domain (CA). The RR usually consists of a phosphyl receiver (REC) domain and a regulatory DNA binding domain.

In a canonical TCS the histidine kinase senses environmental changes which trigger autophosphorylation and phosphotransfer to the response regulator. This leads to a conformational change in the response regulator allowing binding to a specific region of DNA and regulation of transcription [13][14].

Despite the fact that a single bacterium can contain many tens or hundreds of TCS, they generally show a high degree of specificity with a single HPK phosphorylating only its partner RR [15]. Many studies have tried to dissect the specificity of these interactions to determine how the fidelity of signal transmission in maintained and unwanted cross-talk prevented when the HPK and RR families share such significant sequence and structural similarity [16]. It is generally thought that the majority, if not all, of the specificity information is contained within the DHp domain of the HPK and the REC domain of the RR [17].

Interestingly, cross-talk between non-native and host TCS components has been observed previously. When the vancomycin resistance TCS (VanS-VanR) from Enterococcus faecilis BM4147 was introduced into E. coli it was found to cross-talk to the native PhoR-PhoB system, but only when either the VanS or PhoR kinases were deleted [9]. In this case cross-talk was attributed to the elimination of the phosphatase activity usually associated with the cognate bifunctional HPK. Indeed, the presence of a bifunctional kinase/phosphatase for its cognate RR may be a general mechanism through which cross-talk is reduced in vivo. The presence of specific phosphatases may also offset any inappropriate phosphorylation of the RR by small molecule phosphodonors such as acetyl-phosphate [18]. In vitro, histidine kinases have been shown to phosphorylate a wide range of RRs, however many studies have shown that the HPK demonstrates a distinct kinetic preference for its natural in vivo RR [19].

One of the most intensively studied TCS is the EnvZ/OmpR system of E. coli. It is involved in sensing osmolarity changes in the environment and regulating the transcription of the outer membrane porins OmpC and OmpF [20] and consists of the osmosensing histidine kinase EnvZ and the response regulator OmpR. Under conditions of high osmolarity, EnvZ undergoes autophosphorylation on histidine 243 [21] and this phosphoryl group is then transferred to aspartic acid 55 on OmpR (to give OmpR∼P) [14]. Under low osmolarity conditions EnvZ acts as a phosphatase, converting OmpR∼P to OmpR. Phosphorylation of OmpR results in a conformational change, allowing it to bind to the promoter regions of either ompC or ompF, depending on the osmotic level of the cell’s environment. Under conditions of high osmolarity there are high concentrations of OmpR∼P and transcription of ompC is preferred whereas under low osmolarity conditions there are low levels of OmR∼P and the transcription of ompF is favoured. [22]. In early attempts to redesign this system, the periplasmic ligand binding domain of the chemoreceptor Tar was fused with the cytoplasmic kinase/phosphatase domain of EnvZ, allowing the system to sense aspartate [23][25]. The resulting chimeric receptor, Taz, was able to activate OmpR in response to the extracellular levels of aspartate. In this paper we develop and validate a simple tool which assigns the likelihood of a given HPK interacting with particular RRs. We show that it can predict the presence of inter-species cross-talk when the non-native EnvZ/OmpR E. coli TCS components are introduced into a Rhodobacter sphaeroides host cell.

Results

We have developed a simple analysis tool for the prediction of inter-species cross-talk between components of bacterial two-component systems. We created two non-redundant databases containing the DHp domains of the HPKs and the REC domains of the RRs from sequenced bacterial genomes respectively. The sequences were extracted from the Pfam database [26] and clustered using the program TribeMCL [27]. This resulted in 227 clusters for the DHp domains, each containing up to 1193 sequences from 466 species. For the REC domains of the RRs, 433 different clusters were created, containing up to 1191 sequences from 472 species. To predict interactions between members of the different clusters we used a statistical method to assign each of the DHp and REC domains to operons [28]. Using the same definition as the MiST2 database [29] DHp and REC domain containing proteins encoded in the same operon were considered to be probable interaction partners if there were no more than two intervening genes. From these data sets, we generated a probability table showing the likelihood that DHp domains from a particular cluster would interact with REC domains from other clusters. (Table S4).

As there is little experimental data on inter-species cross-talk, we validated our method by predicting intra-species partners for orphan HPKs. As these genes are orphans, ie not encoded in an operon with a RR, they will not have contributed to our interaction dataset. The Caulobacter crescentus orphans are the best studied experimentally and hence provide an ideal set to validate our method. We also compare our results to those obtained with two other recent methods for predicting partners for orphan kinases, those of Burger and van Nimwegen [19] and Procaccini et al. [20].

C. crescentus contains eight orphan HPKs. Table 1 shows the predicted RRs for each orphan HPK using our clustering method, the Procaccini method and the Burger method. Experimentally validated RRs are highlighted in green, experimentally disproved interactions in red and untested partners in black [19], [30]. Overall, the three methods gave similar results with our approach predicting a partner for seven of the orphans including four of those which have been experimentally validated. We did not predict the experimentally determined RR CenR as a partner for CenK, although as stated by Procaccini et al [20] CenK generates very weak scores with their method, indicating that these pairings are not yet well described by any of the current models. Interestingly, all three methods predict a different RR for the HPK CC_0586 and further experiments to determine which, if any, of these are correct would be informative.

thumbnail
Table 1. Comparison of response regulator predictions for Caulobacter crecentus histidine kinases.

https://doi.org/10.1371/journal.pone.0037737.t001

Prediction of Interspecies Cross-talk

We have chosen to introduce the well-characterised EnvZ/OmpR TCS from E. coli into an alternate host species. The host species chosen was Rhodobacter sphaeroides WS8N [31], a well-characterised model organism whose genomic DNA can be easily manipulated. We used our clustering approach to predict which of the native HPKs may cross-talk with the E. coli EnvZ/OmpR TCS. EnvZ appears in HPK cluster 11 (Figure S1). Interestingly, one of the R. sphaeroides HPKs, RSP0203, also appears in this cluster suggesting that it is a potential candidate for cross-talk (Table S2). The sequences of both proteins show no homology in their periplasmic domains, however the sequence similarity between the cytoplasmic regions (amino acid 163–450 for EnvZ, 161–493 for RSP0203) containing the HAMP, DHp and CA domains of EnvZ from E. coli and RSP0203 from R. sphaeroides is 36% according to the BLAST database [32].

RSP0203 is one of five R. sphaeroides orphan kinases, those HPKs whose corresponding RR is not apparent from genomic context. From our analysis of the interactions between clusters we would predict that kinases found in cluster 11 are most likely to interact with RRs in cluster 13 (Table S4). Interestingly, not only does OmpR appear in RR cluster 13, but also one of the R. sphaeroides orphan RRs, RSP1138, also appears in this cluster. We therefore predicted that the orphan R. sphaeroides HPK RSP0203 would phosphorylate the orphan RR RSP1138 and that cross-talk may occur between the HPK/RR pairing from R. sphaeroides and the EnvZ/OmpR system from E. coli.

According to the Pfam database, RSP0203 is a 439 amino acid protein containing a HAMP domain, a HisK A phosphoacceptor domain and a HATPase c domain and is therefore a classical HPK, similar in its domain structure to E. coli’s EnvZ. However, its putative signal sensing periplasmic domain shows no similarity to that of EnvZ or to any other protein in the database, has no identifiable domains and the whole protein is classified as an integral membrane sensor signal transduction histidine kinase in the BLAST database [33].

According to the Pfam database [26], RSP1138 is a 232 amino acid protein that contains an N-terminal CheY-like response regulator receiver domain and a C-terminal DNA binding domain in a helix-turn-helix motif. Both RSP0203 and RSP1138 are located on chromosome 1 in the genome of R. sphaeroides in position 1918334 bp (RSP0203) and 2901588 bp (RSP1138).

In vitro Analysis of Interspecies Cross-talk

The purified soluble cytoplasmic domains of the histidine protein kinases E. coli EnvZ and R. sphaeroides RSP0203 were analysed for their ability to phosphotransfer in vitro to the response regulators OmpR and RSP1138 (Figure 1). Both EnvZ and RSP0203 autophosphorylate and these assays clearly showed that phosphotransfer not only took place from EnvZ to its native response regulator OmpR and from RSP0203 to RSP1138, but also from the R. sphaeroides RSP0203 to OmpR and from E. coli EnvZ to R. sphaeroides RSP1138. The phosphotransfer assays also confirmed the observations of Yoshida et al. [34], that the kinase activity of EnvZ is dominant over its phosphatase activity for OmpR, and demonstrated that the same applies for RSP1138, under these conditions. The precise kinetics of phosphotransfer to the different response regulators may vary, as shown by the differing levels of phosphorylated EnvZ remaining at each timepoint (Figures 1A and 1B). The amount of phosphorylated kinase and RR present at each timepoint is dependent on the rate of phosphotransfer between kinase and RR and on the RR autodephosphorylation rate. One explanation for the difference in the result for OmpR and RSP1138 is that both phosphotransfer to and autodephosphorylation of the RR are enhanced for the cognate pair. However, determining the precise kinetics of these reactions is beyond the scope of this investigation.

thumbnail
Figure 1. Phosphorylation assays of EnvZ/OmpR and RSP0203/RSP1138.

Both histidine kinases were autophophorylated and tested for phosphotransfer to the response regulators after different time intervals. A Phosphotransfer from EnvZ to OmpR. B Phosphotransfer from EnvZ to RSP1138. C Phosphotransfer from RSP0203 to OmpR. D Phosphotransfer from RSP0203 to RSP1138.

https://doi.org/10.1371/journal.pone.0037737.g001

In vivo Analysis of Cross-talk

To determine whether RSP0203 can interact in vivo with the non-native response regulator OmpR from E. coli we designed a testbed where we introduced parts of the EnvZ/OmpR TCS along with a reporter construct into R. sphaeroides WS8N. The gene encoding CheA1, a chemotaxis protein not expressed under the chosen cultivation conditions [35], was replaced by the E. coli promoter of ompC translationally fused to the yellow fluorescence reporter gene yfp to generate strain JPA1800.

To test whether E. coli OmpR can interact with the native R. sphaeroides RNA polymerase and activate transcription, R. sphaeroides WS8N and JPA1800 were transformed with the inducible expression vector pIND4 [36] expressing either OmpR (pINDOmpR) or the non-phosphorylatable OmpRD55A (pINDD55A). Cells were grown under photoheterotrophic conditions and their fluorescence output data were measured by fluorometry (Table 2). In the absence of OmpR, no fluorescence output could be detected from JPA1800, demonstrating that no R. sphaeroides RR was able to bind and to induce expression from the promoter region of ompC in this reporter strain. The introduction of OmpR caused a significant increase in the fluorescence output, showing that the E. coli protein could not only bind to the promoter region of ompC in JPA1800 but also interact with the R. sphaeroides RNA polymerase. The absence of the cognate E. coli kinase EnvZ in JPA1800 suggested that a kinase from R. sphaeroides was responsible for the phosphorylation of OmpR in this strain, as only OmpR∼P is predicted to undergo the conformational change allowing it to bind to the promoter region of ompC. To verify that OmpR∼P was responsible for the activation of yfp transcription in JPA1800, we transformed this reporter strain with pINDD55A. This mutation of OmpR, the exchange of aspartic acid in position 55 to alanine, prevents its phosphorylation and hence activation [37]. The fluorescence output of the cells expressing OmpRD55A was comparable to wild-type (WS8N) level, confirming that only OmpR∼P was responsible for transcription activation of the reporter gene yfp.

Further analysis of which kinase is responsible for the phosphorylation of OmpR in vivo required the introduction of the DNA sequence coding for OmpR into the R. sphaeroides genome. The gene encoding TlpC, a non-essential protein involved the chemotaxis signalling pathway [38], was replaced by that encoding the response regulator OmpR from E. coli in JPA1800 to generate JPA1802. The gene encoding TlpC was chosen for replacement as intracellular levels of expressed TlpC in R. sphaeroides are similar to expressed OmpR levels in E. coli [39] (M Gould, personal communication). The fluorescence output of JPA1802 clearly showed that OmpR was expressed in this strain, that it bound to the promoter region of ompC and that it initiated transcription of the reporter gene yfp (Figure 2).

thumbnail
Figure 2. Fluorescence output data of R. sphaeroides and its derived mutants.

Cells were grown under photosynthetic conditions for 72 h and analysed in a plate reader (Tecan, Austria). Transcription from the plasmids pINDEnvZ and pIND0203 was induced with 10 µM IPTG.

https://doi.org/10.1371/journal.pone.0037737.g002

These data (Figure 2, Table 3) showed that the E. coli OmpR must have been phosphorylated even in the absence of its cognate kinase (EnvZ) in R. sphaeroides, and from our previous analysis we predicted that the host kinase RSP0203 could be the source of this phosphorylation. The deletion of rsp0203 from the genome of JPA1802, forming JPA1806, resulted in a marked reduction in the fluorescence reporter output compared to JPA1802 (Figure 2). This strongly suggested that RSP0203 is the major histidine kinase responsible for the phosphorylation of OmpR in R. sphaeroides.

To confirm that RSP0203 was the major kinase phosphorylating OmpR in R. sphaeroides, RSP0203 was overexpressed from pIND4 and complementation of the RSP0203 deletion in JPA1806 restored wild-type levels of fluorescence (Figure 2). Interestingly, the overexpression of RSP0203 from pIND4 in JPA1802 did not increase the observed fluorescence signal. This suggests that sufficient OmpR∼P has already been produced in JPA1802 to saturate the single ompC promoter in this strain. The deletion of rsp0203 did not completely abolish the fluorescence output of the reporter strain JPA1806, which could be explained either if OmpR is also being phosphorylated by a small molecule phosphodonor such as acetyl phosphate [18] or by another unidentified HPK.

The gene encoding the native histidine kinase of OmpR, EnvZ, was also cloned into pIND4. This plasmid, pINDEnvZ, was expressed in the R. sphaeroides strains JPA1802 and JPA1806. This resulted in an obvious reduction of the fluorescence output of the reporter strains JPA1802 and JPA1806 (Figure 2), suggesting that the phosphatase activity of EnvZ dominated over its kinase activity under these conditions in R. sphaeroides and reduced transcription from the ompC promoter to close to wild-type levels [22].

Discussion

Synthetic biologists are constantly looking to increase the available parts lists from which they can design and construct novel pathways and phosphosignalling systems. Both bacterial two-component systems and eukaryotic MAPK pathways provide a plethora of opportunities. However, with the increased diversity of parts comes the problem of the number of potential host systems with which the synthetic systems could interact. The ability to predict potential cross-talk prior to designing and constructing novel pathways or choosing a host organism is essential for the promise that phosphosignalling components hold for synthetic biology to be realised.

In this study, we demonstrate that it is possible to introduce parts of TCS into non-host strains for use as synthetic biology components. We show that the a non-host transcription factor, in this case OmpR from the γ-subgroup bacterium E. coli, is capable of binding the α-subgroup bacterial host’s (R. sphaeroides) RNA polymerase to activate transcription of a reporter construct. We also demonstrate that a host kinase phosphorylated this non-native RR in vivo and hence that the potential issue of cross-talk is a real problem with the use of these systems.

Most, if not all, of the specificity of HPK/RR interactions is encoded in the DHp and REC domains [19]. We therefore extracted the sequences of these domains from available protein sequences and clustered them by their sequence similarity. Using no additional information other than genome context we calculated the frequency with which HPKs of a particular cluster interacted with members of RR clusters across 466 species. Using our prediction tool, we identified a native R. sphaeroides kinase (RSP0203) as the probable cause of cross-talk to OmpR. The in vitro analysis of cross-talk between EnvZ/OmpR and RSP0203 demonstrated that predicted interactions between proteins within the clusters generated by our analysis do occur as we showed that both these kinases from the same cluster phosphotransfer to the response regulator OmpR.

Deletion of rsp0203 from the JPA1802 reporter strain reduced the fluorescence expression levels, indicating in vivo that the predicted kinase RSP0203 was responsible for significant phosphorylation of the non-native RR OmpR. When the E. coli bifunctional kinase EnvZ was introduced into the R. sphaeroides reporter strains on a plasmid, a clear reduction of the fluorescence output was detectable, suggesting that the phosphatase activity of the native cognate partner protein exceeded the kinase activity of the predicted host’s cross-talking histidine kinase RSP0203. Unfortunately, experimental techniques for quantifying the specific kinase/phosphatase kinetics in vivo are not currently available, however further investigation of this observation are ongoing. The prediction of possible cross-talk between TCS is essential for our ability to utilise non-native biological components in the pursuit of synthetic biology.

A synthetic biologist proposing to use TCS proteins will have to consider the problem of cross-talk from host proteins. Our predictive tool can easily be used to provide guidance as to which (if any) of the host proteins would cross-talk to the synthetic components prior to the construction of synthetic strains. Assuming that both the bacterium from which the non-native components will be taken and the host are already contained within our database then simply searching the HPK or RR data tables for the GI number of the protein to be introduced will provide the cluster to which it has been assigned. These clusters can then be sorted by species and checked for the presence of any potential cross-talking host proteins. Should the protein to be introduced come from a bacterium which is not contained within our database, then these DHp or REC domain sequence can be compared against our dataset using a local BLAST search to find the HPK or RR cluster containing the most similar sequences (Figure S1, Methods S1, Tables S1, S2, S3 and S4) and a similar procedure followed.

Finally, our predictive tool could be used by researchers looking for the probable interaction partners of orphan HPKs and RRs. The probability table provides information about the most likely interaction clusters for a given orphan HPK or RR and may guide the search for the cognate pairings for these proteins. As was demonstrated by our in vitro phosphotransfer data, the orphan HKP RSP0203 was shown to phosphotransfer to the orphan RR RSP1138 as predicted from their cluster locations.

Our method also performed well in predicting the RRs for the orphan kinases of C. crescentus. Our predictions were in good general agreement with the experimental data [16], [30] and with predictions obtained by two other methods, those of Burger and van Nimwegen [19] and Procaccini et al [20]. The Burger method utilises a Bayesian approach to match orphan HPKs to RRs whereas the Procaccini method uses Direct-Coupling Analysis to extract residue-residue contacts and generates a scoring function to aid in the prediction of interaction partners. All three methods are well suited to the study of bacterial two component systems. For more general protein interaction predictions, a distinct advantage of the Burger method is that it can be applied where no large set of interaction partners is known, whereas the Procaccini method provides a direct scoring function as its output. As with the Procaccini method, ours requires a large dataset and also assumes a one to one relationship between HPK and RR which whilst generally true for TCS may not always be the case. The clear advantage of our method is simplicity, using no more information than the sequences and genome context information to generate the clusters and that it provides a simple analysis that predicts inter-species cross-talk across a wide range of potential host organism. For all prediction models the scarcity of experimental data somewhat hinders their validation. However, from the existing data in the literature [16], [30] and the analysis of the representatives of the clusters tested here our method appears to provide a good guide for researchers working in this area. Further validation of the method using members of other clusters is currently underway. Whilst most predictions were the same regardless of the method used, in the case of CC_0586 the three methods all predicted different RRs. This suggests that the ability for researchers to apply a combination of different methods for orphan protein interaction predictions will undoubtedly aid the experimental design in this area.

In summary, our results demonstrate the importance of considering and being able to predict how introduced synthetic biological components will interact with those of the native host cell. We have provided a simple bioinformatic tool to allow synthetic biologists to predict whether interspecies cross-talk between introduced and native two-component signalling pathways will occur and shown both in vitro and in vivo that the predicted interactions do take place. We have also shown that our prediction tool can aid in the identification of response regulators for orphan kinases.

Materials and Methods

Bioinformatics

Histidine kinases and response regulators.

A set of bacterial protein sequences from 504 non-redundant species were downloaded (February 2009) from the NCBI (ftp://ftp.ncbi.nih.gov/genomes/Bacteria, multiple strains of the same species were not included). HMMER version 3.0 ([40], http://hmmer.janelia.org) was used to find response regulator REC domains and DHp domains of histidine kinases. The PFAM hidden markov models for REC domains (PF00512; ([26]) and for DHp domains (PF00512) were employed.

16742 proteins were found to have at least 1 REC domain, where the e value of the hit was below the default inclusion threshold of 0.01. The sequences of all REC domains were extracted from these proteins. As some proteins had more than one REC domain, this gave 17448 sequences. 9881 proteins were found to have at least 1 DHp domain, where the e value of the hit was below the default inclusion threshold of 0.01. The sequences of the DHp domains were extracted from these proteins. As some proteins had more than one DHp domain, this resulted in 9906 sequences. All heuristic filters were turned off to ensure maximum sensitivity.

Clustering.

BLAST version 2.2.23 [33], [41] was used to compare all 9881 HK DHp domains to each other. DHp sequences were then clustered using TRIBE-MCL [27]. The same was done for the 17448 REC domain sequences. The I value for the clustering was 5.

Molecular Biology

Construction of R. sphaeroides mutant strains.

For all strains, plasmids and primers used in this study, please see Table 3 and Table 4. A transcriptional fusion of yfp (pEYFP-N1, Clontech) to the promoter region (300 bp upstream of ompC) of ompC from E. coli W3110 [42] flanked by the 500 bp up- and down-stream region of cheA1 (EcoRI, SalI) was cloned into pK18mobsacB [43], linearised with EcoRI and SalI, to create pK18ompCyfp. For the second mutation E. coli W3110 ompR (SalI, XbaI) flanked by the 500 bp up- and downstream region of the R. sphaeroides tlpC was cloned into pK18mobsacB, linearised with SalI and XbaI, resulting in pK18ompR. The constructs were sequenced to ensure that the regions were in frame and contained no errors. The mutations were introduced into the genome of Rhodobacter sphaeroides WS8N [31] by allelic exchange as described previously [43][44]. The correct insertions were confirmed by Southern Blot. This resulted in the R. sphaeroides mutants JPA1800 (cheA1::PompC-yfp) and JPA1802 (cheA1::PompC-yfp, tlpC::ompR).

For the deletion of RSP0203 the 600 bp up- and down-stream regions of rsp0203 were amplified by PCR including suitable restriction sites (MfeI, SalI) and were spliced together using overlap-extension PCR. The resulting fragment was cloned into MfeI and SalI linearised pK18mobsacB to create pK18RSP0203. The mutations were introduced into the genome of Rhodobacter sphaeroides WS8N and JPA1802 as described above. This resulted in the strains JPA1803 and JPA1806.The correct insertions were confirmed by Southern Blot.

Construction of plasmids.

For the expression of EnvZ and RSP0203 in R. sphaeroides, the full length sequences of E. coli W3110 EnvZ and R. sphaeroides RSP0203 were cloned into the expression vector pIND4, resulting in pIndEnvZ and pIND0203. These plasmids were introduced to R. sphaeroides as previously described [36].

The full length sequences of wild-type E.coli ompR and wild-type R. sphaeroides rsp1138 were amplified by PCR and cloned into linearised pQE60 (Qiagen) for the expression of HIS-tagged proteins.

The coding cytoplasmic sequences of R. sphaeroides rsp0203 and E.coli envZ were amplified by PCR and cloned into the expression vector pGEX (GE Healthcare).

Purification of HIS/GST-tagged proteins.

The HIS- and GST-tagged proteins were overexpressed and purified as described previously [45][46]. Cells containing the appropriate expression plasmids were grown in 2YT medium with antibiotics as appropriate to an OD600 of 0.8 at 37°C. Expression was induced by the addition of 0.1 mM IPTG for 20 h at 18°C. After induction, cells were harvested by centrifugation and resuspended in lysis buffer. Cells were lysed and filtered through a 0.45-µm (pore-size) syringe filter. The filtered supernatant was applied to a Ni-nitrilotriacetic acid agarose column (Qiagen), for HIS-tagged proteins, or a glutathione sepharose column (GE Healthcare), for GST-tagged proteins, equilibrated with lysis buffer. The column was washed lysis buffer prior to elution of the protein in lysis buffer supplemented with 250 mM imidazole (HIS-tagged proteins) or 10 mM reduced glutathione (GST-tagged proteins) (Figure S2).

Fluorescence assay.

Cells were plated on solid LB containing naldixic acid and kanamycin, if needed. The plates were incubated at 30°C for 48 h. Single colonies were transferred into succinate medium and incubated at 30°C for 48 h in the light to reach stationary phase.

To obtain the fluorescence (wavelength YFP: excitation 508 nm, emission 540 nm; wavelength CFP: excitation 430 nm, emission 480 nm; gain: 99; flashes: 25) and optical density (OD; absorbance: 700 nm) readings from the cells 100 µl of stationary phase cultures were applied to each well of a black clear bottom 96-well plate (Corning) and measured in a plate reader (Tecan, Infinite M200). To calculate the fluorescence intensity per cell, the fluorescence output data were divided by the OD.

Phosphotransfer assays.

Phosphotransfer assays of RSP0203 and EnvZ to OmpR were performed at 20°C in TGMNKD buffer (50 mM Tris HCl, 10% (v/v) glycerol, 5 mM MgCl2, 150 mM NaCl, 50 mM KCl, 1 mM DTT, pH 8.0) as previously described [47]. Reactions were initiated by the addition of 0.5 mM [γ-32P] ATP (specific activity 14.8 GBq mmol-1; PerkinElmer). The ATP-dependent phosphorylation of 5 µM RSP0203 and EnvZ was allowed to proceed for 30 min and then the phosphotransfer reactions were initiated by the addition of 20 µM response regulator. After 30 s, 60 s, 120 s, 240 s and 600 s a 10 µl aliquot of the reaction mixture was taken, quenched immediately in 10 µl of 3 x SDS-PAGE loading dye (7.5% (w/v) SDS, 90 mM EDTA, 37.5 mM Tris HCl, 37.5% glycerol, 3% (v/v) β-mercaptoethanol, pH 6.8) and analysed by SDS-PAGE (15% (w/v) acrylamide)).

To minimize hydrolysis of phosphoproteins, electrophoresis was performed at 4°C, with a run time of 90 minutes. Gels were exposed to phosphor screens (Kodak) and analyzed using an SF-phosphorimager with ImageQuant software (version 5.0, Molecular Dynamics) [48].

Supporting Information

Figure S1.

Flow diagram explaining the finding of potential interaction partners for orphan proteins. The diagram shows how a potential cognate partner for the orphan kinase RSP0203 from R. sphaeroides was found by applying our described method.

https://doi.org/10.1371/journal.pone.0037737.s001

(DOC)

Figure S2.

Coomassie-stained protein gel. Protein gel (RunBlue 4–20% precast gel, expedeon) of the purified GST-tagged cytoplasmic region of RSP0203 (56 kDa), the purified GST-tagged cytoplasmic region of EnvZ (57 kDa), HIS-tagged RSP1138 (28 kDA)and HIS-tagged OmpR (27 kDa).

https://doi.org/10.1371/journal.pone.0037737.s002

(DOC)

Methods S1.

Finding potential interaction partners.

https://doi.org/10.1371/journal.pone.0037737.s007

(DOC)

Author Contributions

Conceived and designed the experiments: SP KAS GHW. Performed the experiments: SP KAS GB. Analyzed the data: SP GHW. Contributed reagents/materials/analysis tools: RH CMD. Wrote the paper: SP GHW.

References

  1. 1. Bashor CJ, Horwitz AA, Peisajovich SG, Lim WA (2010) Rewiring cells: synthetic biology as a tool to interrogate the organizational principles of living systems. Annual Review of Biophysics 39: 515–537.CJ BashorAA HorwitzSG PeisajovichWA Lim2010Rewiring cells: synthetic biology as a tool to interrogate the organizational principles of living systems.Annual Review of Biophysics39515537
  2. 2. Lim WA (2010) Designing customized cell signalling circuits. Nat Rev Mol Cell Biol 11: 393–403.WA Lim2010Designing customized cell signalling circuits.Nat Rev Mol Cell Biol11393403
  3. 3. Jacob F, Monod J (1961) Genetic regulatory mechanisms in the synthesis of proteins. Journal of Molecular Biology 3: 318–356.F. JacobJ. Monod1961Genetic regulatory mechanisms in the synthesis of proteins.Journal of Molecular Biology3318356
  4. 4. Jacob F, Monod J (1961) On Regulation of gene activity. Cold Spring Harbor Symposia on Quantitative Biology 26: 193-&. F. JacobJ. Monod1961On Regulation of gene activity.Cold Spring Harbor Symposia on Quantitative Biology 26193-&
  5. 5. Bashor CJ, Helman NC, Yan S, Lim WA (2008) Using engineered scaffold interactions to reshape MAP kinase pathway signaling dynamics. Science 319: 1539–1543.CJ BashorNC HelmanS. YanWA Lim2008Using engineered scaffold interactions to reshape MAP kinase pathway signaling dynamics.Science31915391543
  6. 6. Ninfa AJ, Selinsky S, Perry N, Atkins S, Xiu Song Q, et al. (2007) Using two-component systems and other bacterial regulatory factors for the fabrication of synthetic genetic devices. In: Melvin I. Simon BRC, Alexandrine C, editors. Methods in Enzymology: Academic Press. pp. 488–512.AJ NinfaS. SelinskyN. PerryS. AtkinsQ. Xiu Song2007Using two-component systems and other bacterial regulatory factors for the fabrication of synthetic genetic devices.BRC Melvin I. SimonC. AlexandrineMethods in Enzymology: Academic Press488512
  7. 7. McClean MN, Mody A, Broach JR, Ramanathan S (2007) Cross-talk and decision making in MAP kinase pathways. Nat Genet 39: 409–414.MN McCleanA. ModyJR BroachS. Ramanathan2007Cross-talk and decision making in MAP kinase pathways.Nat Genet39409414
  8. 8. Haruo S (2010) Regulation of cross-talk in yeast MAPK signaling pathways. Current Opinion in Microbiology 13: 677–683.S. Haruo2010Regulation of cross-talk in yeast MAPK signaling pathways.Current Opinion in Microbiology13677683
  9. 9. Fisher SL, Jiang W, Wanner BL, Walsh CT (1995) Cross-talk between the histidine protein kinase VanS and the response regulator PhoB. Journal of Biological Chemistry 270: 23143–23149.SL FisherW. JiangBL WannerCT Walsh1995Cross-talk between the histidine protein kinase VanS and the response regulator PhoB.Journal of Biological Chemistry2702314323149
  10. 10. Antunes MS, Morey KJ, Tewari-Singh N, Bowen TA, Smith JJ, et al. (2009) Engineering key components in a synthetic eukaryotic signal transduction pathway. Mol Syst Biol 5: MS AntunesKJ MoreyN. Tewari-SinghTA BowenJJ Smith2009Engineering key components in a synthetic eukaryotic signal transduction pathway.Mol Syst Biol5
  11. 11. Wuichet K, Cantwell BJ, Zhulin IB (2010) Evolution and phyletic distribution of two-component signal transduction systems. Current Opinion in Microbiology 13: 219–225.K. WuichetBJ CantwellIB Zhulin2010Evolution and phyletic distribution of two-component signal transduction systems.Current Opinion in Microbiology13219225
  12. 12. Laub MT, Goulian M (2007) Specificity in two-component signal transduction pathways. Annual Review of Genetics 41: 121–145.MT LaubM. Goulian2007Specificity in two-component signal transduction pathways.Annual Review of Genetics41121145
  13. 13. Szurmant H, Hoch JA (2010) Interaction fidelity in two-component signaling. Current Opinion in Microbiology 13: 190–197.H. SzurmantJA Hoch2010Interaction fidelity in two-component signaling.Current Opinion in Microbiology13190197
  14. 14. Mattison K, Oropeza R, Byers N, Kenney LJ (2002) A phosphorylation site mutant of OmpR reveals different binding conformations at ompF and ompC. Journal of Molecular Biology 315: 497–511.K. MattisonR. OropezaN. ByersLJ Kenney2002A phosphorylation site mutant of OmpR reveals different binding conformations at ompF and ompC.Journal of Molecular Biology315497511
  15. 15. Verhamme DT, Arents JC, Postma PW, Crielaard W, Hellingwerf KJ (2002) Investigation of in vivo cross-talk between key two-component systems of Escherichia coli. Microbiology 148: 69–78.DT VerhammeJC ArentsPW PostmaW. CrielaardKJ Hellingwerf2002Investigation of in vivo cross-talk between key two-component systems of Escherichia coli.Microbiology1486978
  16. 16. Skerker JMP, Melanie S, Perchuk BS, Biondi EG, Laub MT (2005) T Two-component signal transduction ppathways regulating growth and cell cycle progression in a bacterium: A system-level analysis. PLoS Biol 3: e334.JMP SkerkerS. MelanieBS PerchukEG BiondiMT Laub2005T Two-component signal transduction ppathways regulating growth and cell cycle progression in a bacterium: A system-level analysis.PLoS Biol3e334
  17. 17. Gao R, Stock AM (2009) Biological insights from structures of two-component proteins. Annual Review of Microbiology 63: 133–154.R. GaoAM Stock2009Biological insights from structures of two-component proteins.Annual Review of Microbiology63133154
  18. 18. McCleary WR, Stock JB (1994) Acetyl phosphate and the activation of two-component response regulators. Journal of Biological Chemistry 269: 31567–31572.WR McClearyJB Stock1994Acetyl phosphate and the activation of two-component response regulators.Journal of Biological Chemistry2693156731572
  19. 19. Skerker JM, Perchuk BS, Siryaporn A, Lubin EA, Ashenberg O, et al. (2008) Rewiring the specificity of two-component signal transduction systems. Cell 133: 1043–1054.JM SkerkerBS PerchukA. SiryapornEA LubinO. Ashenberg2008Rewiring the specificity of two-component signal transduction systems.Cell13310431054
  20. 20. Stock JB NA, Stock AM (1989) Protein phosphorylation and regulation of adaptive responses in bacteria. Microbiol Rev 53: 450–490.NA Stock JBAM Stock1989Protein phosphorylation and regulation of adaptive responses in bacteria.Microbiol Rev53450490
  21. 21. Forst SA, Roberts DL (1994) Signal transduction by the EnvZ-OmpR phosphotransfer system in bacteria. Research in Microbiology 145: 363–373.SA ForstDL Roberts1994Signal transduction by the EnvZ-OmpR phosphotransfer system in bacteria.Research in Microbiology145363373
  22. 22. Russo FD, Silhavy TJ (1991) EnvZ controls the concentration of phosphorylated OmpR to mediate osmoregulation of the porin genes. Journal of Molecular Biology 222: 567–580.FD RussoTJ Silhavy1991EnvZ controls the concentration of phosphorylated OmpR to mediate osmoregulation of the porin genes.Journal of Molecular Biology222567580
  23. 23. Utsumi R, Brissette R, Rampersaud A, Forst S, Oosawa K, et al. (1989) Activation of bacterial porin gene expression by a chimeric signal transducer in response to aspartate. Science 245: 1246–1249.R. UtsumiR. BrissetteA. RampersaudS. ForstK. Oosawa1989Activation of bacterial porin gene expression by a chimeric signal transducer in response to aspartate.Science24512461249
  24. 24. Jin T, Inouye M (1993) Ligand binding to the receptor domain regulates the ratio of kinase to phosphatase activities of the signaling domain of the hybrid Escherichia coli transmembrane receptor, Taz1. Journal of Molecular Biology 232: 484–492.T. JinM. Inouye1993Ligand binding to the receptor domain regulates the ratio of kinase to phosphatase activities of the signaling domain of the hybrid Escherichia coli transmembrane receptor, Taz1.Journal of Molecular Biology232484492
  25. 25. Yoshida T, Phadtare S, Inouye M (2007) The design and development of Tar/EnvZ chimeric receptors. In: Melvin I. Simon BRC, Alexandrine C, editors. Methods in Enzymology: Academic Press. pp. 166–183.T. YoshidaS. PhadtareM. Inouye2007The design and development of Tar/EnvZ chimeric receptors.BRC Melvin I. SimonC. AlexandrineMethods in Enzymology: Academic Press166183
  26. 26. Finn RD, Mistry J, Tate J, Coggill P, Heger A, et al. (2010) The Pfam protein families database. Nucleic Acids Research 38: D211–D222.RD FinnJ. MistryJ. TateP. CoggillA. Heger2010The Pfam protein families database.Nucleic Acids Research38D211D222
  27. 27. Enright AJ, Van Dongen S, Ouzounis CA (2002) An efficient algorithm for large-scale detection of protein families. Nucleic Acids Research 30: 1575–1584.AJ EnrightS. Van DongenCA Ouzounis2002An efficient algorithm for large-scale detection of protein families.Nucleic Acids Research3015751584
  28. 28. Hamer R, Chen P-Y, Armitage J, Reinert G, Deane C (2010) Deciphering chemotaxis pathways using cross species comparisons. BMC Systems Biology 4: 3.R. HamerP-Y ChenJ. ArmitageG. ReinertC. Deane2010Deciphering chemotaxis pathways using cross species comparisons.BMC Systems Biology43
  29. 29. Ulrich LE, Zhulin IB (2010) The MiST2 database: a comprehensive genomics resource on microbial signal transduction. Nucl Acids Res 38: D401–407.LE UlrichIB Zhulin2010The MiST2 database: a comprehensive genomics resource on microbial signal transduction.Nucl Acids Res38D401407
  30. 30. Ohta N, Newton A (2003) The core dimerization domains of histidine kinases contain recognition specificity for the cognate response regulator. Journal of Bacteriology 185: 4424–4431.N. OhtaA. Newton2003The core dimerization domains of histidine kinases contain recognition specificity for the cognate response regulator.Journal of Bacteriology18544244431
  31. 31. Porter SL, Wilkinson DA, Byles ED, Wadhams GH, Taylor S, et al. (2011) Genome sequence of Rhodobacter sphaeroides strain WS8N. The Journal of Bacteriology. pp. JB.05257–05211.SL PorterDA WilkinsonED BylesGH WadhamsS. Taylor2011Genome sequence of Rhodobacter sphaeroides strain WS8N.The Journal of BacteriologyJB.0525705211
  32. 32. Tatusova TA, Madden TL (1999) BLAST 2 Sequences, a new tool for comparing protein and nucleotide sequences. FEMS Microbiol Lett 174: 247–250.TA TatusovaTL Madden1999BLAST 2 Sequences, a new tool for comparing protein and nucleotide sequences.FEMS Microbiol Lett174247250
  33. 33. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. Journal of Molecular Biology 215: 403–410.SF AltschulW. GishW. MillerEW MyersDJ Lipman1990Basic local alignment search tool.Journal of Molecular Biology215403410
  34. 34. Yoshida T, Qin L, Inouye M (2002) Formation of the stoichiometric complex of EnvZ, a histidine kinase, with its response regulator, OmpR. Molecular Microbiology 46: 1273–1282.T. YoshidaL. QinM. Inouye2002Formation of the stoichiometric complex of EnvZ, a histidine kinase, with its response regulator, OmpR.Molecular Microbiology4612731282
  35. 35. Shah DSH, Porter SL, Harris DC, Wadhams GH, Hamblin PA, et al. (2000) Identification of a fourth cheY gene in Rhodobacter sphaeroides and interspecies interaction within the bacterial chemotaxis signal transduction pathway. Molecular Microbiology 35: 101–112.DSH ShahSL PorterDC HarrisGH WadhamsPA Hamblin2000Identification of a fourth cheY gene in Rhodobacter sphaeroides and interspecies interaction within the bacterial chemotaxis signal transduction pathway.Molecular Microbiology35101112
  36. 36. Ind AC, Porter SL, Brown MT, Byles ED, de Beyer JA, et al. (2009) Inducible-expression plasmid for Rhodobacter sphaeroides and Paracoccus denitrificans. Appl Environ Microbiol 75: 6613–6615.AC IndSL PorterMT BrownED BylesJA de Beyer2009Inducible-expression plasmid for Rhodobacter sphaeroides and Paracoccus denitrificans.Appl Environ Microbiol7566136615
  37. 37. Delgado J, Forst S, Harlocker S, Inouye M (1993) Identification of a phosphorylation site and functional analysis of conserved aspartic acid residues of OmpR, a transcriptional activator for ompF and ompC in Escherichia coli. Molecular Microbiology 10: 1037–1047.J. DelgadoS. ForstS. HarlockerM. Inouye1993Identification of a phosphorylation site and functional analysis of conserved aspartic acid residues of OmpR, a transcriptional activator for ompF and ompC in Escherichia coli.Molecular Microbiology1010371047
  38. 38. Wadhams GH, Martin AC, Porter SL, Maddock JR, Mantotta JC, et al. (2002) TlpC, a novel chemotaxis protein in Rhodobacter sphaeroides, localizes to a discrete region in the cytoplasm. Molecular Microbiology 46: 1211–1221.GH WadhamsAC MartinSL PorterJR MaddockJC Mantotta2002TlpC, a novel chemotaxis protein in Rhodobacter sphaeroides, localizes to a discrete region in the cytoplasm.Molecular Microbiology4612111221
  39. 39. Pratt L, STJ (1995) Two-component signal transduction. eds Hoch J, Silhavy TJ American Society for Microbiology Press, Washington, D. C. pp. 105–107.L. PrattSTJ1995Two-component signal transduction.eds Hoch J, Silhavy TJ American Society for Microbiology Press, Washington, D. C105107
  40. 40. Eddy SR (1998) Profile hidden Markov models. Bioinformatics 14: 755–763.SR Eddy1998Profile hidden Markov models.Bioinformatics14755763
  41. 41. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, et al. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Research 25: 3389–3402.SF AltschulTL MaddenAA SchäfferJ. ZhangZ. Zhang1997Gapped BLAST and PSI-BLAST: a new generation of protein database search programs.Nucleic Acids Research2533893402
  42. 42. Bachman B (1972) Pedigrees of some mutant strains of Escherichia coli K-12. Bacteriol Rev 36: 252–557.B. Bachman1972Pedigrees of some mutant strains of Escherichia coli K-12.Bacteriol Rev36252557
  43. 43. Schäfer A TA, Jäger W, Kalinowski J, Thierbach G, Pühler A (1994) Small mobilizable multipurpose cloning vectors derived from the Escherichia coli plasmids pK18 and pK19 - Selection of defined deletions in the chromosome of Corynebacterium glutamicum. Gene 145: 69–73.TA Schäfer AW. JägerJ. KalinowskiG. ThierbachA. Pühler1994Small mobilizable multipurpose cloning vectors derived from the Escherichia coli plasmids pK18 and pK19 - Selection of defined deletions in the chromosome of Corynebacterium glutamicum.Gene1456973
  44. 44. Penfold RJ, Pemberton JM (1992) An improved suicide vector for construction of chromosomal insertion mutations in bacteria. Gene 118: 145–146.RJ PenfoldJM Pemberton1992An improved suicide vector for construction of chromosomal insertion mutations in bacteria.Gene118145146
  45. 45. Martin AC, Wadhams GH, Armitage JP (2001) The roles of the multiple CheW and CheA homologues in chemotaxis and in chemoreceptor localization in Rhodobacter sphaeroides. Molecular Microbiology 40: 1261–1272.AC MartinGH WadhamsJP Armitage2001The roles of the multiple CheW and CheA homologues in chemotaxis and in chemoreceptor localization in Rhodobacter sphaeroides.Molecular Microbiology4012611272
  46. 46. Porter SL, Wadhams GH, Armitage JP (2008) Rhodobacter sphaeroides: complexity in chemotactic signalling. Trends in Microbiology 16: 251–260.SL PorterGH WadhamsJP Armitage2008Rhodobacter sphaeroides: complexity in chemotactic signalling.Trends in Microbiology16251260
  47. 47. Scott KA, Porter SL, Bagg EAL, Hamer R, Hill JL, et al. (2010) Specificity of localization and phosphotransfer in the CheA proteins of Rhodobacter sphaeroides. Molecular Microbiology 76: 318–330.KA ScottSL PorterEAL BaggR. HamerJL Hill2010Specificity of localization and phosphotransfer in the CheA proteins of Rhodobacter sphaeroides.Molecular Microbiology76318330
  48. 48. Porter SL, Armitage JP (2002) Phosphotransfer in Rhodobacter sphaeroides chemotaxis. Journal of Molecular Biology 324: 35–45.SL PorterJP Armitage2002Phosphotransfer in Rhodobacter sphaeroides chemotaxis.Journal of Molecular Biology3243545