Differential proteolysis of sigma regulators controls cell-surface signalling in Pseudomonas aeruginosa


  • Richard C. Draper,

    1. Department of Biochemistry, University of Otago, PO Box 56, Dunedin 9054, New Zealand.
    Search for more papers by this author
  • Lois W. Martin,

    1. Department of Biochemistry, University of Otago, PO Box 56, Dunedin 9054, New Zealand.
    Search for more papers by this author
  • Paul A. Beare,

    1. Department of Biochemistry, University of Otago, PO Box 56, Dunedin 9054, New Zealand.
    Search for more papers by this author
    • Present address: National Institute of Allergy and Infectious Diseases, Hamilton, MT 59840, USA.

  • Iain L. Lamont

    Corresponding author
    1. Department of Biochemistry, University of Otago, PO Box 56, Dunedin 9054, New Zealand.
      E-mail iain.lamont@otago.ac.nz; Tel. (+64) 3 479 7969; Fax (+64) 3 479 7866.
    Search for more papers by this author

E-mail iain.lamont@otago.ac.nz; Tel. (+64) 3 479 7969; Fax (+64) 3 479 7866.


Cell-surface signalling systems are widespread in Gram-negative bacteria. In these systems gene expression occurs following binding of a ligand, commonly a siderophore, to a receptor protein in the outer membrane. The receptor interacts with a sigma regulator protein that extends from the periplasm into the cytoplasm to control the activity of a cognate sigma factor. The mechanisms of signal transduction in cell-surface signalling systems have not been determined. Here we investigate signal transduction in the pyoverdine, ferrichrome and desferrioxamine siderophore systems of Pseudomonas aeruginosa. When pyoverdine is present the sigma regulator FpvR undergoes complete proteolysis resulting in activation of two sigma factors PvdS and FpvI and expression of genes for pyoverdine synthesis and uptake. When pyoverdine is absent subfragments of FpvR inhibit PvdS and FpvI. Similarly, subfragments of the sigma regulators FoxR and FiuR are formed in the absence of desferrioxamine and ferrichrome. These are much less abundant when the siderophores are present and downstream gene expression takes place. In all three systems RseP (MucP/YaeL) is required for complete proteolysis of the sigma regulator and sigma factor activity. These findings indicate that regulated proteolysis is a general mechanism for signal transduction in cell-surface signalling.


Pseudomonas aeruginosa are widespread, diverse and versatile bacteria and are a leading cause of morbidity and mortality in patients with cystic fibrosis or severe burns (Driscoll et al., 2007). To grow and to cause disease bacteria must overcome the hurdle of low iron bio-availability in host tissues (Ratledge and Dover, 2000). Many bacteria, including P. aeruginosa, address this challenge by secreting low-molecular-weight compounds (siderophores) that chelate iron ions in the extracellular environment and over 500 chemically distinct siderophores are known (Drechsel and Winkelmann, 1997). In Gram-negative bacteria ferrisiderophores are imported by a class of outer membrane receptors, each specific for a single ferrisiderophore or group of chemically similar siderophores, in a process dependent on the energy-transducing protein TonB (Noinaj et al., 2010). Iron ions are subsequently released from the ferrisiderophores for incorporation into proteins. P. aeruginosa secretes two siderophores, pyoverdine and pyochelin, and can also utilize a wide range of exogenous siderophores made by other bacteria and fungi (Poole and McKay, 2003). Emphasizing this ability, and the importance of ferrisiderophore uptake to the bacteria, the genome of P. aeruginosa PAO1 contains 34 genes encoding TonB-dependent receptors (Cornelis and Matthijs, 2002).

Expression of many ferrisiderophore receptors, in P. aeruginosa and in many other species, is strongly induced by the presence of the cognate ferrisiderophore. A widespread mechanism for achieving this is cell-surface signalling (Visca et al., 2002; Braun and Mahren, 2005). In this mechanism gene expression is induced by the binding of ferrisiderophore to its cognate receptor in the outer membrane. Binding, accompanied by importation of the ferrisiderophore into the periplasm, results in an altered interaction between a periplasmic domain of the receptor protein and a signal-transducing protein (sigma regulator) that spans the cytoplasmic membrane. This results in activation of a sigma factor protein, located in the cytoplasm, which directs expression of specific genes (often only the gene encoding the receptor protein). In the absence of the ferrisiderophore signal there is only low-level expression of the gene encoding the receptor protein as the sigma factor protein is inhibited, or at least not activated, by the sigma regulator. Genome analyses have shown that surface-signalling systems of this sort are widespread in Gram-negative bacteria – for example, there are predicted to be 13 such systems in P. aeruginosa (Visca et al., 2002) and the human gut symbiont Bacteroides thetaiotamicron contains approximately 30 (Braun and Mahren, 2005).

In one of the best-characterized systems, pyoverdine is secreted by P. aeruginosa and chelates extracellular ferric ions. The resulting ferripyoverdine is transported into P. aeruginosa by a cell-surface receptor protein, FpvA. Pyoverdine also induces a cell-surface signalling pathway that controls the activity of two sigma proteins FpvI and PvdS (Lamont et al., 2002; Beare et al., 2003). FpvI directs the expression of the fpvA gene that encodes the ferripyoverdine receptor, while the PvdS regulon includes genes for pyoverdine synthesis (pvd) and additionally genes important for virulence, exotoxin A and PrpL protease. When pyoverdine is not available, FpvI and PvdS activity is inhibited by the sigma regulator protein FpvR. FpvR is thought to sequester FpvI and PvdS at the inner membrane and also renders PvdS susceptible to proteolysis (Lamont et al., 2002; Beare et al., 2003; Spencer et al., 2008; Tiburzi et al., 2008). The binding of pyoverdine to FpvA results in a signal being transferred through a periplasmic domain of FpvA to FpvR (Shen et al., 2002; James et al., 2005). This relieves the inhibitory activity of FpvR, resulting in FpvI and PvdS becoming active and consequent expression of their target genes. Other surface signalling systems that have been characterized in P. aeruginosa enable uptake of the heterologous siderophores desferrioxamine and ferrichrome, via the FoxA and FiuA receptor proteins respectively (Llamas et al., 2006; Mettrick and Lamont, 2009). In each case binding of the ferrisiderophore to its receptor results in signal transduction via a sigma regulator protein (FoxR or FiuR) to a sigma factor (FoxI or FiuI) that directs increased expression of the relevant ferrisiderophore receptor. Although a number of surface signalling systems have been studied, the mechanism underlying signal transduction from the periplasm to the cytoplasm is not known.

Gram-negative bacteria contain other regulatory systems in which a signal is transduced from the periplasm to the cytoplasm resulting in altered activity of a sigma factor protein. A well-studied example is the sigma E (σE) stress response pathway in Escherichia coli (reviewed in Ades, 2008; Heinrich and Wiegert, 2009; Chen and Zhang, 2010). σE activity is inhibited by the membrane-spanning anti-sigma factor RseA. Stress signals, such as the presence of misfolded outer membrane proteins, elicit cleavage of RseA by a periplasmic protease DegS at a specific location (‘site-1’) within the periplasmic domain of RseA. ‘Site-1’ proteolysis renders the remaining truncated RseA susceptible to ‘site-2’ proteolysis by RseP, which is located in the cytoplasmic membrane, cleaving RseA within the transmembrane domain. The resulting cytoplasmic domain of RseA is degraded by cytoplasmic proteases, resulting in activity of σE and expression of the σE stress-response regulon. In P. aeruginosa, homologues to DegS/RseP/RseA/σE[AlgW, RseP (also known as MucP), MucA and AlgU (also known as AlgT)] control production of the exopolysaccharide alginate (Reiling et al., 2005; Wood et al., 2006; Qiu et al., 2007; Cezairliyan and Sauer, 2009). Addition of cycloserine or overproduction of MucE, a periplasmic peptide, induce MucA degradation by AlgW and RseP in a similar manner to the E. coli heat-shock response (Qiu et al., 2007; Cezairliyan and Sauer, 2009; Wood and Ohman, 2009). In Bordetella bronchiseptica expression of an outer membrane haem receptor BhuR requires an RseP homologue (HurP) in conjunction with a cell-surface signalling pathway (King-Lyons et al., 2007). There is also evidence that RseP is required for gene expression in the ferric citrate transmembrane signalling system in E. coli (Braun et al., 2005). Systems where regulated intramembrane proteolysis controls gene expression have also been described in many other bacteria, in viruses, fungi and protozoa and in human cells (for recent reviews see Heinrich and Wiegert, 2009; Urban, 2009; Chen and Zhang, 2010) although in many cases the mechanisms controlling protease activity are not fully elucidated.

Here, we demonstrate the involvement of a proteolytic cascade in the pyoverdine signalling system and identify RseP as a key component. Furthermore we provide evidence that regulated proteolysis represents a general mechanism for the control of cell-surface signalling systems in P. aeruginosa.


A cytoplasmic subfragment of FpvR is associated with inhibition of sigma factor activity

The effect of pyoverdine and the associated signalling pathway on the FpvR protein was examined using a mAb directed against the N-terminal (cytoplasmic) domain. FpvR has a predicted mass of 37 kDa but the predominant form of FpvR that was detected, as exemplified in the pyoverdine-deficient mutant (pvdF), is a 20 kDa protein (Fig. 1A). In wild-type bacteria in which fpvR was overexpressed from a plasmid (FpvR++) FpvR was also mostly in the 20 kDa form although a protein corresponding in size to full-length FpvR was detected. As the mAb recognizes the N-terminal (cytoplasmic) domain of FpvR, the 20 kDa fragment must contain at least part of the N-terminal domain but lack much of the C-terminal (periplasmic) domain. Hence we refer to this fragment as FpvR(N-term). The FpvR(N-term) fragment was not observed in the pyoverdine-producing WT strain and addition of pyoverdine to cultures of the pvdF mutant resulted in rapid disappearance of this fragment, with a significant reduction after 5 min and no detectable protein after 30 min (Fig. 1A and B). The FpvR(N-term) fragment was detected in the absence of the ferripyoverdine receptor protein (FpvA), even in the presence of pyoverdine, consistent with a requirement for FpvA for signal transduction (Shen et al., 2002; Beare et al., 2003; James et al., 2005). In all strains and conditions where FpvR(N-term) was detected, expression from the PvdS-dependent pvdE promoter and from the FpvI-dependent fpvA promoter was inhibited relative to wild type. In cases where no FpvR protein or subfragment was detected, gene expression occurred (Fig. 1A). Expression of parts of the fpvR gene corresponding to both the N-terminal and C-terminal regions was unaffected by addition of pyoverdine (Fig. 2G and H) so that inter-strain differences in FpvR protein subfragments are not a consequence of changes in transcription of fpvR.

Figure 1.

The effect of pyoverdine on proteolytic processing of FpvR. A. Cell extracts were prepared from wild-type bacteria, strains carrying the mutations shown or wild-type bacteria overexpressing FpvR (FpvR++) and analysed by Western blot with anti-FpvRN antibody that is specific for the N-terminal (cytoplasmic) part of FpvR. The positions of full-size FpvR and an N-terminal fragment FpvR(N-term) are indicated. Reporter gene data for each strain for the PvdS-dependent promoter PpvdE (Lamont et al., 2002) and the FpvI-dependent promoter PfpvA (Beare et al., 2003) are shown. B. Cells of P. aeruginosa pvdF mutant were grown to late exponential phase and pyoverdine added. Samples were taken at the times indicated and analysed by Western blot as above, along with samples from cells with no added pyoverdine. C. Cell extracts were prepared from bacteria expressing FpvR–FLAG fusion and analysed by Western blot with anti-FpvR(N-term) (left panel) or anti-FLAG (right panel) antibody. D. Schematic representing FpvR from the N- to C-termini, with the predicted cytoplasmic domain (dark grey), trans-membrane (TM) region (black), periplasmic domain (pale grey) and location of the fused FLAG epitope shown. The location of the epitope for the anti-FpvRN monoclonal antibody and approximate locations of FpvR(N-term) and FpvR(C-term) are shown. ND, not determined.

Figure 2.

The effects of protease mutations on processing of FpvR and gene expression. A. Cell extracts were prepared from wild-type P. aeruginosa (WT) and from strains carrying mutations in protease-encoding genes, and analysed by Western blot with anti-FpvRN antibody. The position of FpvR(N-term) is shown and novel FpvR subfragments are indicated by arrowheads. B–H. RT-qPCR was carried out on the mutant strains shown following growth with or without pyoverdine (pvd) (20 µM). The amounts of pvdL, pvdS, fpvA, fpvI, fptA and fpvR transcripts are shown relative to the reference genes clpX and oprL. For fpvR, primers were specific to regions of the gene encoding the N-terminal (cytoplasmic) (G) or C-terminal (periplasmic) (H) parts of FpvR. Data are representative of three biological replicates ± standard deviation.

We hypothesized that FpvR(N-term) arises due to proteolysis of FpvR. To investigate this further and to detect the C-terminus of the FpvR protein we introduced a C-terminal FLAG tag into the chromosomal fpvR gene by allele exchange. The addition of the tag did not affect the ability of the FpvR protein to inhibit sigma activity or respond to pyoverdine, as measured by PvdS- or FpvI-dependent reporter assay for PpvdE and PfpvA respectively (Fig. S1). A protein of approximately 17 kDa [FpvR(C-term)] corresponding to the C-terminus of FpvR was detected in the pvdF background, and in small amounts in wild-type bacteria (Fig. 1C). The size of this protein added to that of FpvR(N-term) (20 kDa) corresponds to that of full-size FpvR (37 kDa) suggesting that the N-term and C-term fragments are formed as the result of a single proteolytic cleavage event (Fig. 1D).

Taken together these data suggest that the activity of FpvR is governed by proteolytic processing and that different processing pathways occur, dependent on the presence or absence of pyoverdine.

RseP protease is involved in degradation of FpvR

The periplasmic proteases AlgW, MucD and Prc and the intramembrane protease RseP have been implicated in the regulation of gene expression in P. aeruginosa (Reiling et al., 2005; Wood et al., 2006; Qiu et al., 2007; Cezairliyan and Sauer, 2009; Damron and Yu, 2011). To investigate if any of these proteases were involved in processing of FpvR we examined FpvR in strains lacking the algW, rseP, mucD or prc genes. Mutations in algW, mucD or prc did not alter proteolysis of FpvR (Fig. 2A). However, in the rseP mutants, additional FpvR subfragments (∼13–14 kDa) were detected whether or not pyoverdine was present (Fig. 2A), suggesting a role for RseP in the processing of FpvR. The sizes of the FpvR degradation products observed in the rseP mutants suggests that they are likely to include the N-terminal (cytoplasmic) region (10.7 kDa), the membrane spanning region (2.4 kDa), and possibly a small part of the periplasmic domain (see Fig. 1D). RseP cleaves substrates within the membrane-spanning region, which would be present in these subfragments.

RseP determines the activity of FpvR

Proteases involved in degradation of FpvR would be predicted to affect the activity of PvdS and FpvI. The effect of protease mutations on PvdS- and FpvI-dependent gene expression was therefore assessed. As expected, transcription of the PvdS-dependent gene pvdL and FpvI-dependent gene fpvA was induced by the addition of pyoverdine to the pvdF strain with no protease mutations (Fig. 2B and D). The pvdF rseP strain showed significantly reduced pvdL and fpvA transcription compared with the parental (RseP+) strain (P < 0.01). This shows that RseP is involved in the induction of gene expression. Equivalent findings were obtained using luciferase reporter gene assays, with an rseP mutation inhibiting pyoverdine from causing increased expression from the PvdS-dependent pvdE promoter and from the FpvI-dependent fpvA promoter (Fig. S2). Mutations in algW, mucD or prc did not prevent pyoverdine-mediated expression (Figs 2 and S3).

Transcription of pvdS or fpvI was not reduced in the rseP mutant (Fig. 2C and E), showing that reduction in sigma factor activity was not due to reduced pvdS or fpvI expression in this strain. To determine whether the rseP mutation has a general effect on expression of ferrisiderophore receptor genes, we investigated expression of fptA that encodes FptA, the receptor for a second siderophore (pyochelin) that is made by P. aeruginosa. Expression of fptA is induced by the presence of pyochelin in a process dependent on PchR, an AraC-type regulator that responds to the presence of intracellular pyochelin (Michel et al., 2005) and does not involve a cell-surface signalling mechanism. Transcription of fptA was increased in the rseP mutant (Fig. 2F), in direct contrast to the effect of the rseP mutation on expression of PvdS- and FpvI-dependent genes, showing that RseP is not required for its expression.

Together these results show that RseP is required for induction of gene expression in the pyoverdine system and are consistent with a model in which degradation of FpvR by RseP leads to release of PvdS and FpvI and consequent expression of downstream genes (Fig. 4).

Figure 4.

Model for the role of proteolysis in the ferripyoverdine cell-surface signalling systems. A. Inhibition of gene expression. Proteolysis of FpvR by a periplasmic protease results in formation of a membrane-spanning subfragment (∼20 kDa) that binds to and inhibits the cognate σ factors (PvdS and FpvI), as well as a periplasmic subfragment (∼17 kDa). B. Activation of gene expression. Binding of ferripyoverdine to the FpvA receptor protein causes it to interact with FpvR, resulting in proteolysis by a site-1-like protease to give membrane-spanning subfragments (∼13–14 kDa). These are susceptible to proteolysis by RseP giving cytoplasmic subfragments (∼10 kDa) which are degraded by cytoplasmic protease(s), releasing active σ factors. Subfragments indicated by [] are predictions of this model that have not been observed experimentally.

Proteolysis regulates other cell-surface signalling systems

The exogenous siderophores desferrioxamine and ferrichrome induce expression of cognate receptor genes (foxA and fiuA) through cell-surface signalling pathways analogous to the pyoverdine system (Llamas et al., 2006; Mettrick and Lamont, 2009). The membrane-spanning sigma regulators in these pathways, FoxR and FiuR, have 42% and 44% sequence similarity to FpvR, with the same predicted topology and similar molecular weights. In addition, the N-terminus of each protein has been shown to contain the region required for anti-sigma activity (Mettrick and Lamont, 2009).

These parallels suggested that RseP may also be involved in controlling the activities of the desferrioxamine and ferrichrome signalling pathways. To test this hypothesis, desferrioxamine- and ferrichrome-mediated gene expression was measured in the pvdF rseP mutant. Expression of the receptor genes fiuA and foxA was strongly induced by desferrioxamine and ferrichrome respectively in the RseP+ (pvdF) strain (Figs 3A and S2), as expected (Llamas et al., 2006; Mettrick and Lamont, 2009). This response was completely abolished in the rseP mutant, implying a strict requirement for RseP in the induction of these systems. These data suggest the molecular mechanism underlying the regulation of the Fpv, Fox and Fiu systems is conserved, and that in all cases induction of sigma activity is dependent on RseP.

Figure 3.

Conservation of molecular mechanism in cell-surface signalling systems. A. RT-qPCR was carried out on the strains indicated grown in the presence or absence of siderophore (20 µM) as indicated. The amounts of fiuA and foxA transcript are shown relative to the reference genes clpX and oprL. Data are representative of three biological replicates ± standard deviation. B. Bacteria expressing plasmid-borne FiuR–FLAG (left panel) or FoxR–FLAG (right panel) were grown in the presence or absence of the cognate siderophores ferrichrome (fch) or desferrioxamine (dfx) (20 µM) as indicated. Cell extracts were analysed by Western blot with the anti-FLAG antibody and the positions of C-terminal protein fragments are indicated.

To investigate proteolysis of FoxR and FiuR we complemented foxR and fiuR mutations with alleles encoding FoxR and FiuR fused to a C-terminal FLAG tag. FoxR–FLAG and FiuR–FLAG fusion proteins were active and displayed comparable responses to the untagged alleles as measured by reporter assay with foxA:lux and fiuA::lux fusions (Fig. S1). Similar to FpvR, the full-length FoxR or FiuR proteins (predicted MW ∼39 kDa) were not detected (Fig. 3A). C-terminal FoxR and FiuR fragments (∼18 kDa) were detected in the absence of siderophore with lower amounts present in bacteria grown in the presence of the cognate siderophore, mirroring findings for FpvR (Fig. 3B).


Cell-surface signalling systems that span both membranes and control the activity of sigma factors in response to an extracellular signal are common in Gram-negative bacteria. However, the mechanisms whereby a signal is transduced from a cell-surface receptor via a sigma regulator to a sigma factor are not well understood. Here we show that signal transduction involves proteolysis of sigma regulator proteins, with the intramembrane protease RseP being involved in the degradative pathway. This is the case for all three cell-surface signalling systems investigated here and a partial model for the role of proteolysis in signal transduction is shown in Fig. 4. Given the clear parallels in different cell-surface signalling systems (numbers, sizes, sequence similarities and predicted subcellular locations of proteins) it seems very likely that regulated proteolysis also plays a role in other systems.

Our data show that in the presence of pyoverdine, FpvR is rapidly degraded and neither the N-terminal (cytoplasmic) or C-terminal (periplasmic) domains can be detected in wild-type bacteria. This is consistent with PvdS and FpvI sigma factors being active under these conditions. In cells lacking pyoverdine, a truncated form of FpvR is present that contains the cytoplasmic (sigma-binding) component. This is consistent with FpvR-mediated inhibition of PvdS and FpvI in these cells, both by sequestration from RNA polymerase and because PvdS undergoes FpvR-dependent proteolysis in the absence of pyoverdine (Spencer et al., 2008). However, the form of FpvR that is present in the absence of pyoverdine is shorter than full-length FpvR suggesting that FpvR can enter two alternative processing pathways, one leading to complete degradation of the protein and one leading to the formation of a relatively stable fragment that interacts with the target sigma factors (Fig. 4). The size of this fragment (∼20 kDa) indicates that it would be very unlikely to interact with FpvA as it would lack the C-terminal region thought to be required for this interaction (Schalk et al., 2009) but it would still be associated with the cytoplasmic membrane. This is consistent with the observation that PvdS is preferentially associated with the membrane only in cells lacking pyoverdine (Tiburzi et al., 2008). In the absence of RseP, subfragments of FpvR were detected whether or not pyoverdine was present showing that RseP is involved in degradation of FpvR in both situations.

How might pyoverdine influence which pathway occurs? Binding and internalization of ferripyoverdine by FpvA results in movement of the periplasmic signalling domain that is required for signalling and is thought to interact with FpvR (Schalk et al., 2009). One possibility is that the interaction of this domain with FpvR results in a conformational change in FpvR, initiating a proteolytic cascade that includes a protease analogous to the site-1 protease of the σE stress response pathway (Ades, 2008; Heinrich and Wiegert, 2009; Chen and Zhang, 2010) and RseP, which leads to the complete degradation of FpvR (Fig. 4B). In the absence of interaction with the periplasmic domain of FpvA, FpvR is cleaved by a different periplasmic protease leading to a relatively stable 20 kDa membrane-anchored cytoplasmic subfragment that binds to and inhibits the target sigma factors (Fig. 4A). If this model is correct at least two periplasmic proteases are involved in degradation of FpvR. Following initial cleavage, degradation may also involve non-specific proteases that degrade FpvR subfragments from the C-terminus towards the membrane to make it a substrate for RseP, as occurs in an anti-sigma-dependent stress response pathway in Bacillus subtilis (Heinrich et al., 2009). Mutations in genes encoding candidate proteases (AlgW, MucD and Prc) did not significantly affect gene expression or the FpvR fragments detected by Western blotting so that other proteases must be involved and it will be of interest to identify these. Intriguingly, a C-terminal (periplasmic) fragment of FpvR was present in much higher amounts in the absence than in the presence of siderophore (Figs 1 and 3). It remains to be determined whether this fragment has any function.

An equivalent mechanism can be postulated for the desferrioxamine (Fox) and ferrichrome (Fiu) systems in which RseP is required for induction of gene expression, and C-terminal fragments of the sigma regulators are detected, in the same way as for the pyoverdine system. The effect of an rseP mutation on gene expression in these systems was even more striking than in the pyoverdine system (Figs 3 and S2). RseP has also been reported as being required for signalling in the ferric citrate uptake pathway in E. coli (Braun et al., 2005). An RseP-like protease, HurP, has been implicated in regulation of a haem-uptake pathway in B. bronchiseptica (King-Lyons et al., 2007). The mechanism was not investigated but this research provides experimental support for the proposal that HurP catalyses degradation of HurR, the sigma regulator in this system, in a haem-dependent manner. Collectively, these findings indicate that proteolytic degradation of sigma regulators occurs in many and perhaps all cell-surface signalling systems.

The role of regulated proteolysis in controlling cell-surface signalling is clearly analogous to the regulatory mechanisms that are involved in the general stress response (and production of alginate) in Gram-negative bacteria. In each case a sigma factor is inhibited by a sigma regulator that spans the cytoplasmic membrane and in response to an environmental signal, a periplasmic protease acts on the sigma regulator making it susceptible to proteolysis by RseP. This leads to the sigma factor being released and becoming active. However there are also some clear differences. In the cell-surface signalling systems proteolysis also occurs in the absence of environmental signal (ferrisiderophore) and the periplasmic protease DegS that carries out site-1 proteolysis of RseA in the general stress response apparently does not contribute to proteolysis of the cell-surface signalling regulators. The involvement of RseP in cell-surface signalling pathways greatly extends the range of substrates of this intramembrane protease to the large class of sigma regulators that interact with cell-surface receptors in Gram-negative bacteria. P. aeruginosa contains 13 sigma regulators that are thought to be involved in cell-surface signalling pathways and other Gram-negative bacteria contain many more (Visca et al., 2002; Campbell et al., 2007). All of these may require RseP activity to control their activities. RseP-like intramembrane proteases are also involved in regulation of multiple activities involving transmission of signals from the periplasm of Gram-negative bacteria, or the external surface of Gram-positive bacteria, to the cytoplasm (Ades, 2008; Heinrich and Wiegert, 2009; Chen and Zhang, 2010). In each case studied so far, substrate proteins are thought to have a single-membrane-spanning domain with an extracytoplasmic ‘input’ domain. Proteolysis of the input domain in response to an external signal is required for efficient activity of the intramembrane protease, providing a mechanism for responding to environmental conditions.

In conclusion, this study represents the first demonstrated role for regulated proteolysis in the control of cell-surface signalling systems. Our results also suggest that regulated intramembrane proteolysis of sigma regulators represents a general mechanism for controlling the activities of sigma factors.

Experimental procedures

Growth of bacteria

Bacterial strains used in this study are listed in Table S1. Bacteria were routinely grown in LB medium or on LB agar at 37°C. P. aeruginosa was grown using King's B medium (King et al., 1954) for Western blotting, reverse transcription quantitative PCR (RT-qPCR) and luciferase reporter gene assays. Media were supplemented with ampicillin, carbenicillin, tetracycline or gentamicin as required, at the same concentrations as previously (Mettrick and Lamont, 2009). Pyoverdine, purified from P. aeruginosa PAO1, desferrioxamine and ferrichrome were added as required to a final concentration of 20 µM as described previously (Mettrick and Lamont, 2009).

Genetic manipulation

Plasmids used in this study are listed in Table S2. Restriction of DNA molecules and DNA cloning were carried out using standard methods (Sambrook et al., 2000) with enzymes purchased from Roche Molecular Biologicals. DNA fragments required for strain construction were amplified from genomic DNA of P. aeruginosa PAO1 by PCR with Phusion (NEB) or Supermix High Fidelity (Invitrogen) using appropriate primers (details available on request) and cloned into the required vectors. All plasmid constructs were verified by DNA sequencing.

Methods for transferring promoter-luciferase reporter constructs into P. aeruginosa and for construction of unmarked deletions in P. aeruginosa genes have been described previously (Hoang et al., 1998; Mettrick and Lamont, 2009). To engineer deletions of algW, mucD, prc or rseP, fragments of DNA flanking the deletion site were amplified by PCR, ligated together and cloned into the allele replacement vectors pEX18Gm or pEX18Tc to give plasmids pEX18Gm::ΔalgW, pEX18Gm::ΔmucD, pEX18Tc::Δprc and pEX18Gm::ΔrseP respectively (Table S2). Chromosomal allele replacement was then carried out (Hoang et al., 1998). In each case, over 90% of the target gene was deleted. A chromosomal allele of fpvR expressing the FpvR–FLAG fusion was engineered by amplifying and joining DNA fragments flanking the stop codon of fpvR, resulting in a DNA construct with a KpnI site between the second last (CTC) and last (TGA) codons of fpvR. This construct was cloned into pEX18Tc and ligated with oligonucleotides FLAGfwd (5′-CGACTACAAAGACCATGACGGTGATTATAAAGATCATGATATCGATTACAAGGATGATGATGACGGGTGAGGTAC-3′) and FLAGrev (5′-CTCACCCGTCATCATCATCCTTGTAATCGATATCATGATCTTTATAATCACCGTCATGGTCTTTGTAGTCGGTAC-3′) that encode the Triple-FLAG epitope tag (Hernan et al., 2000) and that had been annealed together and were compatible for ligating with KpnI-digested DNA. Allele replacement in P. aeruginosa was then carried out as described above. All deletions and allele replacements were confirmed by PCR and/or Southern blotting.

Reporter gene assays

Methods for luciferase reporter gene assays have been described previously (Mettrick and Lamont, 2009). Briefly, reporter gene constructs containing the promoters of the pvdE, fpvA, foxA or fiuA genes (Table S2) were transferred into P. aeruginosa by conjugation. Bacteria were then grown in Kings B medium in microtitre plates for 30 h, with luciferase and absorbance readings being taken every 30 min. Luciferase activity at 20 h, normalized to absorbance, is representative of the complete assay and is shown here.

Western blotting

Bacteria were grown in Kings B medium, containing siderophores as required, to late exponential phase (OD600 between 1.8 and 2). Samples (2 ml) were centrifuged in a bench-top microcentrifuge (13 000 r.p.m., 2 min) and resuspended in 0.5 ml of phosphate-buffered saline (pH 7.5) containing Complete Protease Inhibitor Cocktail (Roche) prepared as described by the manufacturer. Cell extracts were prepared by sonication in an ice bath and cellular debris was removed by centrifuging at 10 000 r.p.m. for 2 min. Alternatively, samples of culture (100 µl) were centrifuged, resuspended in SDS-PAGE loading buffer (20 µl) and immediately heated at 100°C for 5 min. Proteins (20 µl) were separated by electrophoresis on a 15% SDS-PAGE gel and transferred to nitrocellulose membrane using standard methods (Harlow and Lane, 1999). Equal protein loadings were confirmed by Coomassie staining of identically loaded gels run in parallel. Blots were probed with anti-FLAG M2 (Sigma F-3165) or monoclonal antibody anti-FpvRN (purchased from GenScript, New Jersey) raised against a peptide corresponding to residues 62–75 within the N-terminal (cytoplasmic) part of FpvR. Secondary detection was carried out with anti-Mouse HRP conjugates (Sigma), Super Signal ECL (Pierce) and a Fuji LAS-1000 Imager.

Reverse transcription quantitative PCR (RT-qPCR)

Bacteria were grown at 22°C using Kings B medium. RNA was extracted using an RNAeasy kit (Qiagen) following the manufacturer's protocol, and yield and quality assessed by Nanodrop spectrophotometry. DNase I treatment (Invitrogen) and cDNA synthesis (Roche ‘Transcriptor’) were performed as per the manufacturers' instructions. qPCR was carried out on the Roche 480 platform using LightCycler SYBR Green I master mix using the protocol described by the manufacturer. Primers and PCR cycling conditions are listed in Table S3. Primers were validated by calibration curve and gave linear amplification efficiencies of between 1.9 and 2.1. No template control and no RT reactions were included for each primer pair to control for the presence of contaminating DNA and Melt analysis was carried out to ensure the intended product was amplified. Relative quantification was performed using the second derivative maximum method corrected for primer efficiencies, with clpX and oprL being reference genes (Palmer et al., 2005; A.F. Konings, L.W. Martin, D.W. Reid and I.L. Lamont, in preparation).


This research was supported in part by a grant from the New Zealand Marsden Fund, administered by the Royal Society of New Zealand. We are grateful to Catherine Day for providing suggestions during this research and for commenting on an earlier version of the manuscript.