Abstract
The pathogenesis of diarrhoeal disease due to human enterotoxigenic Escherichia coli absolutely requires the expression of fimbriae. The expression of CS1 fimbriae is positively regulated by the AraC-like protein Rns. AraC-like proteins are DNA-binding proteins that typically contain two helix–turn–helix (HTH) motifs. A program of pentapeptide insertion mutagenesis of the Rns protein was performed, and this revealed that both HTH motifs are required by Rns to positively regulate CS1 fimbrial gene expression. Intriguingly, a pentapeptide insertion after amino acid C102 reduced the ability of Rns to transactivate CS1 fimbrial expression. The structure of Rns in this vicinity (NACRS) was predicted to be disordered and thus might act as a flexible linker. This hypothesis was confirmed by deletion of this amino acid sequence from the Rns protein; a truncated protein that lacked this sequence was no longer functional. Strikingly, this sequence could be functionally substituted in vivo and in vitro by a flexible seven amino acid sequence from another E. coli AraC-like protein RhaS. Our data indicate that HTH motifs and a flexible sequence are required by Rns for maximal activation of fimbrial gene expression.
- EMSA, electrophoretic mobility shift assay
- ETEC, enterotoxigenic Escherichia coli
- HTH, helix–turn–helix
- MBP, maltose-binding protein
-
↵†Present address: NIBRT, Dublin City University, Glasnevin, Dublin 9, Ireland.
-
Five supplementary figures, showing expression and purification of His-tagged Rns and mutant derivatives and models of the Rns N- and C-termini [in 2D and 3D (PDB) format], are available with the online version of this paper.
Edited by: U. Dobrindt
INTRODUCTION
Enterotoxigenic Escherichia coli (ETEC) is a major cause of diarrhoeal disease in humans and animals (Qadri et al., 2005; Turner et al., 2006). ETEC pathogenesis is dependent on production of both a toxin (heat-labile toxin and/or heat-stable toxin) and fimbriae (Gaastra & Svennerholm, 1996). The Rns protein was originally identified as being required for the expression of CS1 and CS2 fimbriae of ETEC (Caron et al., 1989). Rns also positively autoregulates its gene expression (Froehlich et al., 1994) and activates the genes yiiS (unknown function) (Munson et al., 2002) and cexE, which encodes an uncharacterized extracytoplasmic protein (Pilonieta et al., 2007). In addition, Rns can act as a negative regulator, repressing the expression of NlpA (Bodero et al., 2007).
The coo operon (cooBACD) that encodes the CS1 fimbriae is repressed by H-NS and activated by Rns (Murphree et al., 1997). Rns activates the expression of the coo genes directly by binding to two sites upstream of the coo promoter. As site II overlaps the −35 hexamer, it has been suggested that Rns activates transcription from the coo promoter by forming direct contacts with RNA polymerase (Munson & Scott, 1999). A highly unconventional arrangement of binding sites are involved in Rns autoactivation. Rns binds to a site far upstream of the rns promoter and to two sites downstream of it (Munson & Scott, 2000). Downstream binding sites are usually associated with repressors of transcription, yet activation of the rns promoter in vivo is dependent on the upstream binding site and at least one of the downstream sites (Munson & Scott, 2000).
Rns is a member of the AraC family of transcriptional regulators (Caron et al., 1989), a family of proteins defined by a 100 aa region of homology that contains two predicted helix–turn–helix (HTH) DNA-binding motifs (Gallegos et al., 1997; Ibarra et al., 2007). Among AraC family members, there is a low level of sequence conservation at the first predicted HTH motif (HTH1). This may represent involvement of HTH1 in the recognition of the different target sites of each member (Gallegos et al., 1997). In contrast, the sequence of the second predicted HTH motif (HTH2) was found to be much less variable, prompting the suggestion that this motif may be involved in a common role other than DNA binding, such as forming contacts with RNA polymerase (Gallegos et al., 1997). Rns has two predicted HTH motifs in its C-terminal domain (Munson et al., 2001), so it is assumed that this area of Rns is involved in DNA binding, although this has not been verified previously.
Less is known about the role of the N-terminal domain of Rns. Rns does not appear to respond to an exogenous ligand (Basturea et al., 2008). Thus, while the possibility cannot be ruled out, it is unlikely that the N-terminal portion of Rns is an effector-binding domain. Additionally, it has been reported that the N terminus of Rns does not act as a dimerization domain (Basturea et al., 2008). However, the N-terminal domain has been found to be essential for transcriptional regulation and DNA binding by Rns (Basturea et al., 2008).
Because of its importance in regulation of fimbrial gene expression among human ETEC (Anantha et al., 2004; Bodero et al., 2008), mutagenesis of the rns gene was undertaken to define functionally relevant parts of the Rns protein.
METHODS
Bacterial strains and growth conditions.
Plasmids and E. coli strains used in this study are listed in Table 1⇓. Bacteria were cultured at 37 °C in Luria broth or on Luria agar. Where appropriate, antibiotics were used at the following concentrations: 50 μg carbenicillin ml−1, 50 μg kanamycin ml−1, 50 μg spectinomycin ml−1, 10 μg tetracycline ml−1 and 30 μg chloramphenicol ml−1.
Strains and plasmids used in this study
DNA manipulations.
DNA manipulations were performed according to standard protocols (Sambrook et al., 1989). Plasmid DNA was purified using the HiYield Plasmid kit (Real Biotech Corporation) according to the manufacturer's instructions. PCR was performed with Phusion High Fidelity DNA Polymerase (Finnzymes). Restriction endonucleases and Quick T4 DNA Ligase were purchased from New England Biolabs and used according to the manufacturer's instructions. DNA samples were purified by agarose gel electrophoresis followed by gel extraction with the Geneclean II kit (QBiogene) as described by the manufacturer. Standard methods were used for the transformation of plasmid DNA (Dower et al., 1988). DNA sequencing and synthesis of all oligonucleotide primers were performed by MWG-Biotech (Ebersberg, Germany). Site-directed mutagenesis was carried out with the QuikChange kit (Stratagene).
Pentapeptide scanning mutagenesis.
The GPS-LS linker-scanning system (NEB) was used according to the manufacturer's instructions to introduce transprimer insertions at random positions throughout the rns gene in the plasmid pSS2192. The resulting mutagenized plasmids were transformed into E. coli strain XL-1. The several thousand resultant colonies were pooled and grown overnight with antibiotic selection for both the target plasmid and the transprimer resistance markers. Plasmid DNA prepared from this overnight culture was digested with restriction endonuclease PmeI to remove all but 15 bp of the transprimer. The transprimer-free backbone was religated and transformed into E. coli strain XL-1 to amplify the library of plasmids containing 15 bp insertions. Plasmid DNA prepared from a pool of the resultant colonies was transformed into E. coli strain XL-1 containing the reporter plasmid pCooGFP. Bacteria harbouring mutant rns genes were discerned by observing their colony fluorescence under UV light. Colonies with altered fluorescence (relative to that seen with bacteria expressing the wild-type rns gene) were selected and the presence of a 15 bp insertion in their rns genes was verified by a PCR/restriction endonuclease screen. The precise sequence and location of the insertion was determined by DNA sequencing.
Western blotting.
Western blotting was performed as described previously (Fagan et al., 2008). Antisera were generated commercially by Convance. Anti-Rns sera were raised against a purified 6×His-tagged Rns (derived from pHisRns) by three successive immunizations with 500 μg protein. The serum was absorbed against lysates of E. coli XL-1 Blue harbouring pQE30 and used in Western blotting at a dilution of 1 : 1000.
RNA isolation and reverse transcriptase PCR.
Total RNA was isolated from mid-exponential-phase bacterial cultures using the RiboPure-Bacteria kit (Ambion). Contaminating genomic DNA was removed by DNase I treatment. RT-PCR was performed using 20 ng template total RNA and the Qiagen OneStep RT-PCR kit according to the manufacturer's instructions. The PCR products were visualized on a 2 % (w/v) agarose gel.
β-Galactosidase assay.
The β-galactosidase levels were measured by using the method of Miller (1992). Assays were performed in duplicate on samples that had been grown in duplicate. The data were expressed as the means of four measurements. The sds are indicated as error bars in each figure. Assays were performed independently at least twice.
Spectrofluorimetry.
GFP levels were measured from supernatant fractions of bacteria lysed with CelLytic B, a non-denaturing detergent (Sigma). Duplicate 100 μl aliquots of each cell lysate were dispensed to a black, flat-bottomed 96-well plate. Fluorescence levels were measured in a Wallac 1420 Victor2 multilabel counter set at an excitation wavelength of 485 nm and an emission wavelength of 535 nm. The values obtained were standardized with respect to optical density and expressed per ml culture. The final values were expressed as averages of all of the readings obtained for each strain. Assays were performed independently at least twice. A representative dataset is shown.
Maltose-binding protein (MBP) purification.
E. coli strain KS1000 harbouring plasmids pRare and pMRns5 (or its mutant derivatives) was used for overexpression and purification of MBP : : Rns fusions. Bacteria that had been induced for expression were harvested and resuspended in 1 ml sonication buffer [50 mM Tris/HCl pH 7.5, 10 % (v/v) sucrose, 100 mM NaCl, 1 mM EDTA] and then frozen at −80 °C for 2–3 h. The cell suspensions were thawed on ice and then lysed by sonication. The insoluble material was removed by centrifugation. The soluble supernatant fraction was collected and made up to a volume of 50 ml in column buffer (20 mM Tris/HCl, 200 mM NaCl, 1 mM EDTA, pH 7.4) prior to being loaded onto an amylose column. MBP fusion protein was purified using amylose resin (New England Biolabs) as described by the manufacturer.
Electrophoretic mobility shift assay (EMSA).
The binding of purified MBP : : Rns (or mutant derivatives) to the coo and rns promoter regions was analysed by EMSA using the LightShift Chemiluminescent EMSA kit (Pierce). To generate DNA probes for EMSAs, a 195 bp region of the coo promoter was amplified by PCR from plasmid pCooGFP-2 and a 434 bp region of the rns promoter was amplified by PCR from plasmid pSS2192. The primers in these PCRs had been biotinylated at their 5′-ends; therefore, they produced biotin-end-labelled PCR products. The PCR products were gel purified. Approximately 20 pg biotinylated DNA probe was incubated with a range of protein concentrations at 37 °C for 10 min in a reaction buffer containing 10 mM Tris (pH 7.5), 50 mM KCl, 1 mM dithiothreitol, 2 ng poly(dI.dC) μl−1, 0.1 μg BSA μl−1, 0.025 % (v/v) Nonidet P-40 and 6.5 % (v/v) glycerol. Binding reactions were separated on 5 % non-denaturing polyacrylamide gels at 100 V for 65 min and then transferred to a nylon membrane at 80 V for 1 h. After the transfer, DNA was cross-linked to the membrane at 120 mJ cm−2 using a UV-light cross-linker (Uvitec). The biotinylated DNA was detected with a streptavidin–horseradish peroxidase conjugate and a chemiluminescent substrate as directed by the LightShift Chemiluminescent EMSA kit protocol (Pierce).
3D modelling of Rns.
3D models of the N terminus and C terminus of Rns were generated using phyre and Swiss-model, respectively (Bordoli et al., 2009; Kelley & Sternberg, 2009).
RESULTS
Generation of a panel of Rns mutants
To characterize the structure–function relationship of Rns, the GPS-LS linker scanning system (NEB) was used to mutagenize the rns-encoding plasmid pSS2192. Mutant derivatives of this plasmid that displayed reduced activation of the coo : : gfp fusion in plasmid pCooGFP (as defined by lower fluorescence) were screened for and sequenced (Table 2⇓). The jpred and PROF programmes (Cuff et al., 1998; Rost et al., 2004) were used to obtain a consensus predicted secondary structure of Rns to better understand the regions where the insertions had occurred and to relate this to the effects of the insertions on Rns activity (Fig. 1⇓).
Predicted secondary structure of the Rns protein. The numbered amino acid sequence of Rns is shown on the top line. Secondary structure predictions from jpred and PROF are shown on the second and third lines, respectively. H, α-helix; E, β-strand. The consensus prediction is depicted on the bottom line. α, α-helix; β, β-strand. Helices predicted to be involved in the HTH motifs are shown in grey. There was no consensus prediction for the extreme N terminus of Rns but subsequent analysis revealed that this area is predicted to be unstructured or disordered.
Non-functional insertion mutants resulting from mutagenesis of pSS2192
A number of the mutated rns ORFs encoded truncated forms of Rns (indicated by Δ) as the 15 bp insertion generated by the GPS-LS mutagenesis system introduces a premature TAA stop codon in two of the six possible reading frames. Over 90 % of the protein was missing in truncates RnsI17Δ and RnsH20Δ, while most of the predicted DNA-binding domain had been lost in truncates RnsI192Δ and RnsI195Δ. These truncated variants were not studied further.
Transactivation of the coo promoter by Rns is reduced by mutations in, or in the vicinity of, the predicted HTH motifs
The fluorescence levels in cultures encoding Rns derivatives with pentapeptide insertions in, or in the vicinity of, the protein's predicted HTH motifs were similar to those of cultures harbouring the empty vector pCL1920, i.e. less than 5 % of that detected in the presence of wild-type Rns (Fig. 2⇓). This was also true for the two truncated variants tested, RnsS239Δ and RnsV253Δ. The mutant protein RnsC102 retained an intermediate level of activity at the coo promoter. Bacteria expressing this protein were 12 % as fluorescent as those expressing wild-type Rns.
Transactivation of the coo promoter by Rns and mutant derivatives. Fluorescence levels of E. coli strain XL-1 cultures harbouring pCooGFP and the empty vector negative control pCL1920 or a pSS2192 derivative expressing the indicated mutant Rns were calculated as percentages of the level of fluorescence of cultures expressing wild-type Rns. The data represent averages of duplicate cultures each measured in triplicate. Error bars indicate sd. Measurements were performed independently at least twice; a representative dataset is shown.
The intracellular levels of the Rns mutant proteins were examined by Western immunoblotting with anti-Rns antiserum. While Rns protein was observed in lysates of E. coli strain XL-1 containing wild-type plasmid pSS2192, it was not detected in lysates prepared from cells harbouring the mutant derivatives of plasmid pSS2192. However, RT-PCR analysis confirmed that these mutants were at least transcribed (results not shown). Rns is expressed from its native, positively autoregulated promoter in pSS2192-based plasmids. The insertions may have prevented the mutants from auto-activating the rns promoter to produce the levels of protein necessary for detection by Western immunoblotting.
Therefore, a selection of the mutant rns genes were subcloned downstream of the T5 promoter in the His-tagged Rns expression vector pHisRns. His-tagged forms of mutants C102, L198, F205, Q227, Y242 and V253Δ were detectable (by SDS-PAGE and/or Western immunoblotting), thus ruling out the possibility that the lack of coo promoter activation observed in the presence of these mutants was due to their mutations preventing expression of the encoded Rns protein (see supplementary Fig. S1, available with the online version of this paper). Accordingly, the ability of these Rns mutants to activate a coo : : gfp fusion was assessed (Fig. 3⇓). Again, RnsC102 exhibited a reduced ability to activate the coo promoter. Cultures of E. coli XL-1 containing plasmids pCooGFP-2 and pHisC102 were 27 % as fluorescent as those containing wild-type pHisRns. As before, the levels of expression from the coo promoter in the presence of the truncate RnsV253Δ or the mutants with insertions in or near the predicted HTH motifs of Rns were similar to those found in the presence of the rns-free control.
Transactivation of the coo promoter by His-tagged Rns and mutant derivatives. Fluorescence levels of E. coli strain XL-1 cultures harbouring pCooGFP-2 and the empty vector control pQE-30 or a pHisRns derivative expressing the indicated mutant Rns were calculated as percentages of the level of fluorescence of cultures containing wild-type His-tagged Rns. The data represent averages of duplicate cultures measured in triplicate. Error bars indicate sd. Measurements were performed independently at least twice; a representative dataset is shown.
Pentapeptide insertions also affect Rns autoregulation
Two pentapeptide insertion mutants were chosen for further analysis. RnsQ227 was selected as a representative of the mutants with insertions in the vicinity of the predicted HTH motifs of Rns and RnsC102 was selected due to its unique status of retaining a partial level of activity at the coo promoter. DNA fragments containing the 15 bp insertions encoding the relevant mutations were subcloned into the MBP : : Rns fusion expression vector pMRns5 to create the plasmids pM-Q227 and pM-C102. Fusion of MBP to the amino terminus of Rns does not affect the activity of the protein in vivo or in vitro, but does improve its solubility such that it can be overexpressed and purified (Munson & Scott, 1999).
The pRnsLacZ-2 plasmid contains a transcriptional fusion of the rns promoter to the lacZ reporter gene. This plasmid, along with plasmid pMAL-c2, pMRns5, pM-C102 or pM-Q227, was transformed into the ETEC lac deletion strain LMC10 to assess the ability of the mutants to activate the rns promoter. The levels of β-galactosidase activity in cells containing the positive control plasmid pMRns5 were consistently 1.6-fold higher than those in cells containing the negative empty vector control pMAL-c2 (Fig. 4⇓). This is due to positive regulation of the rns promoter by the MBP : : Rns fusion protein. The RnsC102 insertion mutant also activated transcription from the rns promoter, but to a level lower than wild-type Rns. The β-galactosidase expression levels in cells containing MBP : : RnsC102 were approximately 1.3-fold less than those in cells containing the MBP : : Rns fusion protein (P<0.005). The RnsQ227 mutant did not activate rns transcription. β-Galactosidase activity levels in cells containing plasmid pM-Q227 were similar to those in cells containing plasmid pMAL-c2. Therefore, RnsC102 and RnsQ227 displayed similar levels of activity at the rns promoter as they had at the coo promoter.
Transactivation of the rns promoter by MBP : : Rns fusion protein and mutant derivatives. The β-galactosidase activities of ETEC strain LMC10 harbouring plasmid pRnsLacZ-2 and the MBP : : Rns-expressing plasmid pMRns5, the empty vector control plasmid pMAL-c2 or a plasmid pMRns5 derivative expressing the indicated mutant were measured. The data represent averages of duplicate assays on duplicate cultures. Error bars indicate sd. Measurements were performed independently at least twice; a representative dataset is shown. Statistical significance is indicated by **, where P<0.005.
A pentapeptide insertion in the vicinity of HTH2 of Rns eliminates DNA binding by the protein
EMSAs were performed to study the interaction of Rns and RnsQ227 with the coo and rns promoters and thus determine if the inactivity of the mutant was due to the pentapeptide insertion adversely affecting its ability to bind to DNA. A 195 bp region of the coo promoter (positions −196 to −2 relative to the transcription start site) that includes the two Rns-binding sites previously identified by DNase I footprinting (Munson & Scott, 1999) was used as the coo promoter DNA probe. Three major protein–DNA complexes were detected when EMSAs were performed with this probe and a range of concentrations of the wild-type MBP : : Rns fusion protein (Fig. 5a⇓). The highest mobility complex results from Rns initially occupying binding site I or binding site II alone. The next highest mobility complex is due to simultaneous occupation of both of these sites at the coo promoter (V. Mahon and S. G. J. Smith, unpublished observations). It is likely that the third, low mobility complex is due to non-specific interactions that occur in the presence of high protein concentrations. RnsQ227 did not bind specifically to the coo promoter probe (Fig. 5a⇓). As determined by densitometry, the amount of coo DNA that remained unbound in the presence of 171 nM MBP : : RnsQ227 fusion protein was equal to the amount of probe present in the protein-free control. At the highest concentration of protein tested – 514 nM MBP : : RnsQ227 fusion protein – only the non-specific protein–DNA complex was formed.
Interaction of MBP : : Rns and MBP : : RnsQ227 fusion proteins with DNA. EMSAs were carried out with approximately 20 pg of (a) a biotinylated coo promoter DNA probe or (b) a biotinylated rns promoter DNA probe and a range of concentrations of MBP : : Rns or MBP : : RnsQ227 fusion proteins as indicated at the top of each panel. The unbound DNA fragments and protein–DNA complexes are indicated on the left.
A 434 bp sequence of the rns promoter (positions −305 to +129 relative to the transcription start site), which includes the three sites shown by DNase I footprinting to be bound by Rns (Munson & Scott, 2000), was used as the rns promoter DNA probe. Three protein–DNA complexes were observed when EMSAs were performed with this probe and a range of concentrations of wild-type MBP : : Rns fusion protein (Fig. 5b⇑). It can be postulated that the complexes are a result of Rns binding to one site initially, then at higher protein concentrations to two sites and finally at even higher protein concentrations binding to all three of its sites at the rns promoter. Protein–DNA complexes were not detected when EMSAs were conducted with MBP : : RnsQ227 fusion protein and rns promoter DNA even in the presence of 171 nM MBP : : RnsQ227 fusion protein (Fig. 5b⇑). Thus, the pentapeptide insertion after position Q227 of Rns results in a mutant protein that has lost the ability to bind to DNA and, therefore, cannot activate transcription from the coo and rns promoters.
A predicted area of disorder around residue C102 of Rns is essential for full activity
The uncharacterized area of Rns around residue C102 is not predicted to be of α-helical or β-sheet structure and the pentapeptide insertion within RnsC102 is not predicted to introduce a novel α-helix or β-sheet in the protein. To better understand why this insertion had reduced but not abolished protein activity, the effects of more subtle point mutations in the region were assessed. Site-directed mutagenesis was employed to individually alter the C102 residue itself and the neighbouring residue R103 to alanine (there is already a naturally occurring alanine residue at position 101 of Rns). Neither of these alanine substitutions affected the ability of Rns to transactivate the coo promoter.
As secondary structure predictions could not provide a reason for the effects of the mutations in the area around the C102 residue of Rns, alternative prediction programmes were employed. GlobPlot () is a web-based tool that plots the tendency of protein sequences to be ordered or disordered. GlobPlot analysis of the primary sequence of Rns revealed that the extreme N- and C-termini of Rns and the region encompassing residues 100–104 are predicted to be disordered. DRIP-PRED () and Spritz (), two other computational methods for protein disorder prediction, also predict that an area of disorder is present around residues 100–104 of Rns (Vullo et al., 2006). However, when the GlobPlot analysis was performed on the sequence of Rns including the pentapeptide insertion after position C102, the disorder in this region was no longer predicted to be present. In contrast, the single amino acid changes of C102A and R103A were not predicted to disrupt the disorder in this area. It is possible that to be fully functional as a transcriptional activator, it may be necessary for residues 100–104 of Rns to be disordered.
To determine the importance of predicted protein disorder for Rns activity a mutant protein, RnsΔNACRS, was produced in which the amino acid sequence that comprises the predicted disordered region between positions 100 and 104 was deleted. To test the possibility that the identity of the amino acids in the potentially disordered region is not critical, another Rns derivative was created in which LENSASR, the seven-residue linker sequence of the AraC family member RhaS of E. coli, was substituted for Rns residues 100–104. The RhaS linker was selected because in its native setting, the LENSASR sequence has been suggested to function only to flexibly link the two functional domains of RhaS. Furthermore, a derivative of the related RhaR protein containing the RhaS linker in place of its own linker sequence was functional with only small deficits in its activity (Kolin et al., 2007).
GlobPlot analysis revealed that replacing the NACRS residues of Rns with the LENSASR linker sequence restored the predicted disorder in this region. The fluorescence levels in cultures of E. coli XL-1/pCooGFP-2 harbouring the deletion mutant encoded by pM-ΔNACRS were similar to those of cultures harbouring the empty vector pMAL-c2 (21±1 % and 17±3 % of the level of fluorescence in the presence of wild-type Rns, respectively). Thus, deletion of the potentially disordered area of Rns eliminated the ability of the protein to activate the coo promoter. Cultures of E. coli strain XL-1 harbouring plasmids pCooGFP-2 and pM-RhaSLinker were as fluorescent as those containing pMRns5 (the relative fluorescence levels of the cultures were 108±3 % and 100±3 %, respectively). Therefore, the Rns derivative containing the RhaS linker appeared to be fully functional at the coo promoter.
EMSAs were performed to study the binding of purified MBP fusions of these Rns derivatives to the coo promoter (Fig. 6a⇓). The RnsC102 insertion mutant bound to coo promoter DNA. However, it did so with less affinity than wild-type Rns. Approximately 20 % (as determined by densitometry) of the coo promoter probe remained unbound in the presence of 171 nM MBP : : RnsC102 fusion protein, while at the same concentration of the MBP : : Rns fusion, no free probe remained. MBP : : RnsC102 fusion protein also displayed a greater tendency than wild-type Rns to form the low-mobility non-specific protein–DNA complex with the coo promoter. In contrast, the MBP : : RnsΔNACRS fusion protein did not specifically bind the coo promoter DNA probe. Even at the highest concentration of the deletion mutant tested, the majority of the probe remained unbound and only a small amount of the non-specific complex had been formed. However, MBP : : (Rns with RhaS linker) fusion protein bound the coo promoter DNA probe and formed specific protein–DNA complexes at a similar rate to wild-type MBP : : Rns fusion protein.
Interaction of MBP : : Rns or its mutant derivatives with coo and rns promoter DNA. EMSAs were carried out with approximately 20 pg of (a) a biotinylated coo promoter DNA probe or (b) a biotinylated rns promoter DNA probe and a range of concentrations of MBP : : Rns or its mutant derivatives as indicated at the top of each panel. The unbound DNA fragments and protein–DNA complexes are indicated on the left.
Similar results were observed when the EMSAs were repeated with the rns promoter probe (Fig. 6b⇑). The MBP : : RnsC102 fusion protein bound to rns promoter DNA but with less affinity than wild-type MBP : : Rns fusion protein. Unlike wild-type MBP : : Rns fusion protein, in the presence of 17 nM MBP : : RnsC102 fusion protein only the first protein–DNA complex was formed and at 68 nM MBP : : RnsC102 fusion protein, approximately 40 % (as determined by densitometry) of the rns probe remained unbound. While the MBP : : (Rns with RhaS linker) fusion protein was found to complex with the rns promoter DNA in a manner similar to that observed for MBP : : Rns fusion, the MBP : : RnsΔNACRS fusion protein did not form protein–DNA complexes.
Therefore, the EMSAs revealed that disruption of the NACRS sequence of Rns reduces the affinity of the protein for DNA, while elimination of the sequence results in a mutant protein that is incapable of binding to DNA and consequently cannot activate transcription. However, replacement of the NACRS sequence with an alternative putative disordered region produced a form of Rns that displayed wild-type activity at both the coo and rns promoters. This supports the theory that Rns activity is dependent on the presence of an area of disorder in the vicinity of residues 100–104.
Homology modelling of Rns was attempted to determine the structure of the region in the vicinity of aa 100–104. Using both PHYRE and Swiss-model, we were unable to model the entire structure of Rns. In particular, no structure could be determined for aa 73–154. However, we could generate models of the N terminus and C terminus with these tools, respectively. The N terminus from aa 15–72 was predicted to have a Cupin-like fold (see Supplementary Figs S2 and S3, available with the online version of this paper). The C terminus (from amino acids 155–264) was modelled on the ToxT protein of Vibrio cholerae (Lowden et al., 2010) and broadly agrees with our secondary structure predictions (see Supplementary Figs S4 and S5, available with the online version of this paper).
DISCUSSION
In the absence of a 3D structure, mutagenesis was employed to gain insight into the structure–function relationship of Rns. The C-terminal pentapeptide insertions had a severe negative effect on the activity of Rns at the coo promoter, signifying that the affected areas (two predicted HTH motifs, a long α-helix between these motifs and the C-terminal α-helix of the protein) are vital for Rns activity. It is postulated that these areas are involved in DNA binding. This was confirmed for RnsQ227, which contains an insertion immediately upstream of the predicted HTH2 and did not bind specifically to either the coo or the rns promoter DNA probes. The affected areas may participate directly in DNA binding, i.e. form specific interactions with nucleotides in Rns-binding sites. This is most likely for the two predicted HTHs.
Therefore, Rns may use both HTHs to bind DNA as has been proposed or demonstrated for the AraC family members ToxT (Childers et al., 2007), VirF (Porter & Dorman, 2002) and PerA (Porter et al., 2004). The other C-terminal α-helices of Rns may not be directly involved in DNA binding. It is proposed that, like the α-helix observed between the two HTHs in the crystal structure of the AraC family protein Rob (Kwon et al., 2000), they correctly orientate or stabilize the HTHs to ensure they are in the configuration necessary to interact with DNA.
All truncated variants were inactive. This is unsurprising in mutants lacking over 90 % of the protein (RnsI17Δ and RnsH20Δ) or most of the predicted DNA-binding domain (RnsI192Δ and RnsI195Δ). That the RnsS239Δ variant was inactive is further proof that HTH2 is critical for Rns function. However, the loss of 26 aa may disrupt the folding of the protein's entire C-terminal domain. Only the last 12 residues of Rns are missing in the RnsV253Δ mutant. The inactivity of this truncated variant supports the theory that the final predicted α-helix of Rns plays an essential role.
The sole mutant with an N-terminal pentapeptide insertion, RnsC102, had a reduced ability to activate the coo and rns promoters and a decreased affinity for DNA binding. The insertion after residue C102 had eliminated an area of predicted disorder within Rns. The effects of further mutagenesis and deletion of this potentially disordered area, the NACRS amino acid sequence, indicate that to be fully functional, Rns requires a disordered region between its N- and C-terminal domains. It is postulated that this area of Rns forms, or is part of, a flexible interdomain linker that is necessary to correctly orientate the N- and C-terminal regions of the protein to enable them to participate in their functional roles. The NACRS sequence of Rns has some features in common with the flexible linkers of the AraC, RhaR and RhaS proteins (Eustance & Schleif, 1996; Kolin et al., 2007). The activity of other AraC-like proteins may also depend on the presence of a central disordered/flexible connecting sequence. Residues 100–131 of Rns have been previously suggested to probably function as a flexible linker (Basturea et al., 2008). This study has specifically identified the NACRS sequence that may comprise the minimal linker region of Rns or at least a segment within a larger linker in which disorder or flexibility is vital.
In summary, this analysis has defined regions critical for the activity of Rns and serves as a basis for future research on this protein and other AraC family members.
Acknowledgments
This work was supported by an Irish Research Council for Science, Engineering and Technology (IRCSET) Studentship and an Enterprise Ireland Basic Research Grant (SC-2002-535).