A novel anti-anti-activator mechanism regulates expression of the Pseudomonas aeruginosa type III secretion system

Authors

  • Nandini Dasgupta,

    1. Department of Microbiology, W. M. Keck Microbial Communities and Cell Signaling Program, University of Iowa, 540B Eckstein Medical Research Building, Iowa City, IA 52242, USA.
    Search for more papers by this author
  • Guinevere L. Lykken,

    1. Department of Microbiology, W. M. Keck Microbial Communities and Cell Signaling Program, University of Iowa, 540B Eckstein Medical Research Building, Iowa City, IA 52242, USA.
    Search for more papers by this author
  • Matthew C. Wolfgang,

    1. Department of Microbiology and Immunology and Cystic Fibrosis/Pulmonary Research and Treatment Center, University of North Carolina, Chapel Hill, NC 27599, USA.
    Search for more papers by this author
  • Timothy L. Yahr

    Corresponding author
    1. Department of Microbiology, W. M. Keck Microbial Communities and Cell Signaling Program, University of Iowa, 540B Eckstein Medical Research Building, Iowa City, IA 52242, USA.
    Search for more papers by this author

Summary

Expression of the Pseudomonas aeruginosa type III secretion system (TTSS) is coupled to the secretion status of the cells. Environmental signals such as calcium depletion activate the type III secretion channel and, as a consequence, type III gene transcription is derepressed. Two proteins, ExsA and ExsD, were shown previously to play a role in coupling transcription to secretion. ExsA is an activator of TTSS gene transcription, and ExsD is an anti-activator of ExsA. In the absence of environmental secretion cues, ExsD binds ExsA and inhibits transcription. Here, we describe the characterization of ExsC as an anti-anti-activator of TTSS expression. Transcription of the TTSS is repressed in an exsC mutant and is derepressed upon ExsC overexpression. The dependence on exsC for transcription is relieved in the absence of exsD, suggesting that ExsC and ExsD function together to regulate transcription. Consistent with this idea, ExsC interacts with ExsD in bacterial two-hybrid and co-purification assays. We propose a model in which the anti-anti-activator (ExsC) binds to and sequesters the anti-activator (ExsD) under low Ca2+ conditions, freeing ExsA and allowing for transcription of the TTSS. The P. aeruginosa system represents the first example of an anti-activator/anti-anti-activator pair controlling transcription of a TTSS.

Introduction

Pseudomonas aeruginosa is an opportunistic pathogen of plants, invertebrates and vertebrates (Rahme et al., 2000). Most human P. aeruginosa infections result from predisposing host conditions such as immunodeficiency, cystic fibrosis, severe burns or corneal ulcerations (Richards et al., 1999). P. aeruginosa is also a leading cause of nosocomial pneumonias and urinary tract infections (Torres et al., 1990; Richards et al., 1999). Treatment of these infections is often complicated by the tendency of P. aeruginosa to form antibiotic-resistant biofilms (Stover et al., 2000).

One important virulence determinant of P. aeruginosa is a type III secretion system (TTSS) (Frank, 1997). These systems are used to translocate bacterial effector proteins (toxins) directly to the interior of eukaryotic host cells (Hueck, 1998; Galan, 2001; Cornelis, 2002). The P. aeruginosa TTSS is used to deliver at least four effectors (ExoS, ExoT, ExoU and ExoY) to host cells (Frank, 1997). These effectors promote pathogenesis by disabling the local immune response, destroying epithelia and inhibiting wound healing processes (Finck-Barbancon et al., 1997; Ganesan et al., 1999; Garrity-Ryan et al., 2000; Kaufman et al., 2001).

The P. aeruginosa TTSS consists of ≈ 35 co-ordinately regulated gene products required for secretion and translocation of the effector proteins (Frank, 1997). At least three environmental signals induce transcription of the P. aeruginosa TTSS: (i) contact of P. aeruginosa with eukaryotic cells; (ii) growth in the presence of serum; or (iii) exposure to low Ca2+ concentrations (Iglewski et al., 1978; Vallis et al., 1999). Induction by low Ca2+ is the best understood mechanism and is mediated by at least two distinct signalling pathways. One of these pathways involves a recently described adenylate cyclase signalling cascade (Wolfgang et al., 2003). Under low-Ca2+ conditions, a membrane-associated adenylate cyclase (CyaB) catalyses the formation of cyclic AMP (cAMP). The rise in intracellular cAMP activates the cAMP-dependent transcriptional factor Vfr, resulting in transcription of genes encoding the TTSS, type IV pilus-associated twitching motility functions and a type II secretion system (Wolfgang et al., 2003). Although the mechanism by which Vfr controls expression of these genes remains to be determined, the Vfr-dependent pathway appears to be a mechanism for co-ordinating the expression of related virulence functions on a global scale (Suh et al., 2002).

The second mechanism of coupling low-Ca2+ conditions to TTSS gene transcription is mediated through the type III secretion channel itself. Under high-Ca2+ conditions, the secretion channel is inactive, and transcription of the TTSS is repressed (McCaw et al., 2002). This is consistent with the observation that mutants lacking a functional secretion channel are repressed for type III gene transcription irrespective of Ca2+ and suggests that a functional secretion channel is required for activation of transcription (Yahr et al., 1996; McCaw et al., 2002). Under low-Ca2+ conditions, the secretion channel becomes activated and, as a consequence, type III gene transcription is derepressed (McCaw et al., 2002). The coupling of transcription with secretion is common to many TTSSs (Yersinia sp., Salmonella typhimurium, Shigella flexneri) and the related flagellar biosynthesis systems; however, the regulatory mechanisms appear to be largely unrelated (Miller, 2002).

At least two proteins, ExsA and ExsD, are directly involved in coupling transcription of the P. aeruginosa TTSS with secretion. ExsA is a positive activator of transcription required for the expression of all TTSS genes, and ExsD is an anti-activator (negative regulator) of ExsA (Frank et al., 1994; Yahr and Frank, 1994; Hovey and Frank, 1995; McCaw et al., 2002). ExsD inhibits transcription of the TTSS by virtue of its interaction with ExsA (McCaw et al., 2002). Under low-Ca2+ conditions, the negative regulatory activity of ExsD is suppressed through an unknown mechanism (McCaw et al., 2002). As ExsD is not itself secreted, we postulated that a competition exists between ExsD and a secreted protein for binding to a hypothetical type III-specific chaperone (McCaw et al., 2002). The chaperone was proposed also to function as an inhibitor of ExsD. Under conditions in which the secretion channel is inactive (high Ca2+ or by mutation), the chaperone is bound to the secreted protein and is unavailable to inhibit ExsD. Activation of the secretion channel by low Ca2+ results in export of the secreted protein and makes the chaperone available to inhibit the negative regulatory activity of ExsD. Lacking from this model was the identity of the chaperone and its cognate secreted protein.

One candidate for the chaperone activity is ExsC. ExsC is encoded as the first gene of the exsCBA operon and has the biochemical characteristics of a TTSS-specific chaperone including low molecular mass (16.3 kDa), an acidic isoelectric point (4.6) and a predicted carboxy-terminal amphipathic alpha-helix (Frank and Iglewski, 1991; Yahr and Frank, 1994; Page and Parsot, 2002; Feldman and Cornelis, 2003). Although exsC was one of the first identified genes of the P. aeruginosa TTSS, its function has been unclear based in part on its lack of homology with other proteins (Frank and Iglewski, 1991). With the recent release of the Vibrio parahaemolyticus and Photorhabdus luminescens genomes, however, homologues of both exsC and exsD have been identified (Duchaud et al., 2003; Makino et al., 2003). In both organisms, the ExsC and ExsD homologues are located near one another on the chromosome and are associated with a cluster of TTSS genes. This suggested to us that ExsC and ExsD may function together to regulate type III gene transcription. In the present study, we characterize the role of ExsC and report that it is required for transcription of the P. aeruginosa TTSS and functions as an inhibitor (anti-anti-activator) of ExsD.

Results

exsC is required for expression of the TTSS

To determine whether exsC is involved in the regulation of TTSS transcription, a markerless in frame deletion of exsC (codons 8–141) was constructed, returned to the chromosome of the wild-type strain PA103 by allelic exchange and confirmed by Southern blot analysis (data not shown). To assess TTSS function, the ΔexsC mutant was grown under non-inducing (high Ca2+, in the absence of the chelator EGTA) and inducing (low Ca2+, in the presence of EGTA) conditions for TTSS expression and secretion. Culture supernatant and cell-associated protein samples were analysed by SDS–PAGE and immunoblotting (Fig. 1A–E). When grown under low-Ca2+ conditions, wild-type strain PA103 secretes type III-related exoproducts including ExoU, ExoT, PopB and PcrV (Fig. 1A and B, lanes 1 and 2). In contrast, both an exsA mutant (Frank et al., 1994) and the ΔexsC mutant are defective for expression and secretion of TTSS-related exoproducts (Fig. 1A–E, lanes 3, 4, 7 and 8).

Figure 1.

Characterization of the ΔexsC mutant. Wild-type PA103 (lanes 1 and 2) and the exsA::Ω (lanes 3 and 4), ΔexsD (lanes 5 and 6) and ΔexsC mutants (lanes 7 and 8) were grown under high (–EGTA, lanes 1, 3, 5 and 7) and low (+EGTA, lanes 2, 4, 6 and 8) Ca2+ conditions. At A540 of 1.0, supernatant and cell-associated fractions were prepared as described in Experimental procedures and subjected to SDS–PAGE, immunoblot analysis and β-galactosidase assay.
A. Silver-stained gel of concentrated culture supernatant material. Representative type III-related exoproducts are indicated on the right side of the gel.
B–E. Immunoblots of supernatant (B) and cell-associated (C–E) fractions probed for ExoU (B and C), ExsC (D) or ExsA (E). The asterisk in (D) indicates a cross-reactive band that serves as a loading control. (E) was overexposed to demonstrate the low levels of ExsA expression in the ΔexsC mutant.
F. Strains carrying the mini-pC-lacZ and mini-pD-lacZ transcriptional reporters were grown in the absence or presence of EGTA and assayed for β-galactosidase. The values (Miller units) are the average of three separate experiments, and the standard deviation is indicated by error bars.

To measure the effect of the ΔexsC mutation on transcription of TTSS genes, three previously characterized lacZ transcriptional reporters (mini-pC-lacZ, mini-pD-lacZ and mini-pS-lacZ) (McCaw et al., 2002) were integrated into the unoccupied chromosomal CTX phage attachment site of the ΔexsC mutant. The pC reporter contains the promoter located upstream of the exsCBA operon, the pD promoter controls expression of the 12-gene exsD–pscL operon (pscB–L encode structural components of the TTSS), and the pS promoter controls the transcription of exoenzyme S (exoS) (Yahr and Frank, 1994; Yahr et al., 1995; 1996). Transcription from the mini-pD-lacZ (Fig. 1F), mini-pC-lacZ (Fig. 1F) and mini-pS-lacZ (data not shown) reporters is induced by low Ca2+ and is dependent upon exsA (Yahr and Frank, 1994; Yahr et al., 1995; McCaw et al., 2002). As reported previously, transcription from each of the reporters is derepressed in an exsD background irrespective of the Ca2+ concentration (McCaw et al., 2002). In the ΔexsC mutant, expression of the reporters is reduced to levels similar to those seen in the mutant lacking exsA. These data indicate that the ΔexsC mutant is defective for transcription of the TTSS. At least four explanations might account for this defect. One possibility is that deletion of exsC from the chromosome destabilizes the polycistronic exsCBA mRNA, thereby impairing exsA expression. A second is that ExsC may function as a cofactor for ExsA-dependent transcription. A third possibility is that ExsC may be required for ExsA translation and/or stability. Finally, we demonstrated previously that an ExsD-dependent inhibitory feedback mechanism prevents transcription of the TTSS in type III secretion mutants (Yahr et al., 1996; McCaw et al., 2002). The transcriptional defect in the ΔexsC mutant could result if ExsC was required for secretion. The diminished levels of ExsA observed in the ΔexsC mutant are consistent with each of the above models (Fig. 1E, lanes 7 and 8).

exsC overexpression results in derepression of the TTSS

To determine whether the in frame deletion of exsC had a polar effect on exsA expression, we performed a complementation analysis. An expression plasmid (pexsC) consisting of exsC under the transcriptional control of its native promoter (pC) was introduced into the ΔexsC mini-pD-lacZ reporter strain. While a vector control (pUCP18) failed to restore expression of the TTSS in the ΔexsC mutant, the pexsC plasmid restored the transcriptional activity of the pD promoter (Fig. 2G), ExsA expression (Fig. 2E, lane 2 versus 8) and expression/secretion of type III-related exoproducts (Fig. 2A–C, lane 6 versus 8). Complementation by a plasmid-encoded copy of exsC (pexsC) suggests that a polar effect on exsA expression is an unlikely explanation for the transcriptional defect seen in the ΔexsC mutant.

Figure 2.

Complementation of the ΔexsC mutant. Wild-type PA103 mini-pD-lacZ (lanes 1–4, 9 and 10) and ΔexsC mini-pD-lacZ (lanes 5–8) carrying a vector control (lanes 1, 2, 5 and 6), an exsC expression plasmid (lanes 3, 4, 7 and 8) or an exsD expression plasmid (lanes 9 and 10) were grown under non-inducing (–EGTA, lanes 1, 3, 5, 7 and 9) and inducing (+EGTA, lanes 2, 4, 6, 8 and 10) conditions for type III secretion. At A540 of 1.0, supernatant and cell-associated fractions were prepared and subjected to SDS–PAGE, immunoblot analysis and β-galactosidase assay.
A. Silver-stained gel of concentrated culture supernatant material. Representative type III-related exoproducts are indicated on the right side of the gel.
B–F. Immunoblots against supernatant (B) and cell-associated (C–F) fractions probed for ExoU (B and C), ExsC (D), ExsA (E) or ExsD (F).
G. β-Galactosidase activity of the mini-pD-lacZ reporter after growth in the absence or presence of EGTA. The values (Miller units) are the average of three separate experiments, and the standard deviation is indicated by error bars.

In addition to complementing the ΔexsC mutant for transcription, the pexsC expression plasmid resulted in ExsC overexpression (Fig. 2D, lanes 1 and 2 versus 3, 4, 7 and 8) and a derepressed phenotype in both wild-type PA103 and the ΔexsC mutant. Characteristics of this phenotype include constitutive transcription from the pD promoter (Fig. 2G), cytoplasmic accumulation of ExoU when cells are grown under high-Ca2+ conditions (Fig. 2C, lanes 1 versus 3 and 7) and elevated levels of type III-related exoproducts when cells are grown under low-Ca2+ conditions (Fig. 2A and B, lanes 2 versus 4 and 8). This is similar to the previously reported phenotype of an ΔexsD mutant (Fig. 1) in which transcription of the TTSS is derepressed irrespective of Ca2+, but secretion of type III-related exoproducts remains largely dependent upon Ca2+ chelation (McCaw et al., 2002). Interestingly, the opposite phenotype is seen after deletion of exsC or expression of ExsD from a plasmid (Fig. 2A–G, lanes 5–6 and 9–10) (the diminished expression of plasmid-encoded ExsD stems from its dependence on ExsA for transcription and has been described previously; McCaw et al., 2002). Under those conditions, expression of the TTSS is repressed even when cells are exposed to low-Ca2+ conditions (McCaw et al., 2002). The reciprocity of the ΔexsC and ΔexsD mutant phenotypes suggests that these two proteins function together to regulate expression of the TTSS. Although more complicated models can be envisaged to account for the ΔexsC phenotype, the simplest explanation is that ExsC antagonizes the negative regulatory activity of ExsD.

ExsC antagonizes the negative regulatory activity of ExsD

If the function of ExsC is to antagonize the negative regulatory activity of ExsD, we hypothesized that: (i) ExsA overexpression should restore expression of the TTSS in the ΔexsC mutant; (ii) ExsD-dependent repression should be reversed by concomitant overexpression of either ExsC or ExsA; and (iii) the dependence on exsC for transcription should be relieved in the absence of exsD. To test the first hypothesis, arabinose-inducible exsC and exsA expression plasmids (pJN-exsC and pJN-exsA) were constructed and introduced into the ΔexsC mini-pD-lacZ reporter strain. Similar to the results seen with pexsC, pJN-exsC complemented the exsC mutation and derepressed transcription of the TTSS when cells were grown in the presence of arabinose (Fig. 3A). The exsA expression construct (pJN-exsA) also complemented the exsC mutation and derepressed transcription of the TTSS (Fig. 3A). Two important conclusions can be drawn from this result. First, ExsC is not required as a cofactor for ExsA-dependent transcription. Secondly, the transcriptional defect in the ΔexsC mutant may result from the unmitigated activity of ExsD. In other words, in the absence of ExsC, the negative regulatory activity of ExsD cannot be suppressed by low Ca2+.

Figure 3.

ExsC and ExsA suppress ExsD-dependent repression. ΔexsC mini-pD-lacZ(A) and wild-type mini-pD-lacZ(B) carrying the indicated plasmids were grown under non-inducing (–EGTA) or inducing (+EGTA) conditions for expression of the TTSS. As indicated, 0.5% or 0.1% arabinose was added to induce expression of ExsC or ExsA respectively. At A600 of 1.0, cells were harvested and assayed for β-galactosidase activity. The values are reported in Miller units and represent the average of three separate experiments. The standard deviation is indicated by error bars.

To determine whether ExsD-dependent repression is reversed by either ExsC or ExsA overexpression, pexsD or a vector control (pUCP18) and pJN-exsC, pJN-exsA or a vector control (pJN105) were introduced into the wild-type mini-pD-lacZ reporter strain. With the vector control (pJN105), pD transcription is induced under low-Ca2+ conditions; however, the addition of arabinose consistently reduced transcription by about twofold (Fig. 3B). The basis for this is unclear. Nevertheless, the repression of pD transcription imparted by the pexsD expression plasmid is relieved after co-expression of either ExsC or ExsA and is enhanced by the addition of arabinose (Fig. 3B). In the case of ExsA overexpression, restoration of pD transcription is likely to be the result of ExsA being in excess of ExsD. Restoration of pD transcription by ExsC overexpression is consistent with a model in which ExsC inhibits the negative regulatory activity of ExsD.

To determine whether ExsC activity is dependent on ExsD, a double mutant lacking both exsC and exsD was constructed by introducing the ΔexsC deletion allele into the ΔexsD mini-pD-lacZ background. Consistent with our model, the phenotype of the resulting double mutant (ΔexsCD) resembles that of the ΔexsD mutant. As described previously, the characteristics of this phenotype include derepressed transcription of the TTSS (Fig. 4C), cytoplasmic accumulation of ExoU after growth in the presence of Ca2+ (Fig. 4B, lanes 1 versus 5 and 7) and elevated levels of ExoU in the culture supernatant after growth under low-Ca2+ conditions (Fig. 4A, lanes 2 versus 6 and 8) (McCaw et al., 2002). These results demonstrate that exsC is not required for transcription of the TTSS or as a structural component of the secretion apparatus. Furthermore, these data support the conclusion that ExsC is not required as a cofactor for ExsA-dependent transcription, for ExsA translation or for ExsA stability as ExsA expression levels are similar in both ΔexsD and ΔexsCD mutants (data not shown).

Figure 4.

exsC is not required in the absence of exsD.
A–C. Wild-type PA103 (lanes 1 and 2), ΔexsC (lanes 3 and 4), ΔexsD (lanes 5 and 6) and ΔexsCD (lanes 7 and 8) carrying the mini-pD-lacZ transcriptional reporter were grown under non-inducing (–EGTA, lanes 1, 3, 5 and 7) and inducing (+EGTA, lanes 2, 4, 6 and 8) conditions for type III secretion. At A540 of 1.0, supernatant and cell-associated fractions were prepared and subjected to immunoblot analysis and β-galactosidase assay.
A and B. Immunoblots against supernatant (A) and cell-associated (B) fractions probed for ExoU.
C. β-Galactosidase activity of the mini-pD-lacZ transcriptional reporter.
D. β-Galactosidase activity of ΔexsCD mini-pD-lacZ carrying a vector control (pUCP18), pexsC or pexsD. The values are the average of three separate experiments, and the standard deviation is indicated by error bars.

Both ExsC overexpression and loss of exsD result in derepression of the TTSS. If derepression results from distinct mechanisms, then overexpression of ExsC in the ΔexsD mutant might be additive, resulting in further derepression of the TTSS. Conversely, if ExsC and ExsD regulate the TTSS through a common mechanism, overexpression of ExsC in the ΔexsD mutant should have no significant effect on expression of the TTSS. To test this idea, the ΔexsCD mutant was transformed with the pexsC and pexsD expression plasmids and a vector control. Transformants were grown under high- and low-Ca2+ conditions and assayed for the activity of the mini-pD-lacZ transcriptional reporter and secretion of type III-related exoproducts. Both transcription from the pD promoter (Fig. 4D) and the levels of ExoU expression (data not shown) were nearly identical with both pexsC and the vector control. These data suggest that derepression resulting from ExsC overexpression is dependent upon ExsD; however, the negative regulatory activity of ExsD is not dependent upon ExsC.

ExsC interacts with ExsD

One mechanism by which ExsC could antagonize ExsD would be through a direct protein–protein interaction. To determine whether ExsC interacts with ExsD, the bacterial LexA two-hybrid assay was used (Dmitrova et al., 1998; Daines and Silver, 2000). This system makes use of the LexA repressor, a mutant form of the repressor (LexA408) with altered DNA binding specificity, and a hybrid operator with half-sites specific for LexA and LexA408. The hybrid operator controls the expression of a chromosomally encoded lacZ reporter. Only heterodimers of LexA and LexA408 can bind to the hybrid operator and repress transcription of the reporter. To test for an interaction between ExsC and ExsD, the LexA/LexA408 DNA-binding domains were fused to ExsC or ExsD. The expression plasmids were introduced into the reporter strain and assayed for β-galactosidase activity. In this assay, a positive interaction is indicated by repression of the lacZ reporter. Repression of the lacZ reporter was not observed with either the vector controls or the individual fusions of ExsC/ExsD to LexA/LexA408 (Fig. 5A). The combinations of LexA–ExsC with LexA408–ExsD and LexA408–ExsC with LexA–ExsD, however, resulted in 92% and 83% inhibition of lacZ expression respectively. A positive control, containing fusions of LexA and LexA408 to the dimerization domains of Fos and Jun, respectively, gave 98% repression (Fig. 5A). The observation of the ExsC–ExsD interaction in Escherichia coli suggests a direct interaction between the two proteins.

Figure 5.

ExsC interacts with ExsD.
A. The E. coli two-hybrid reporter strain SU202 carrying expression plasmids expressing the indicated fusions of LexA to Fos, ExsC or ExsD (Fusion partner 1) and LexA408 to Jun, ExsC or ExsD (Fusion partner 2) were grown in LB medium and assayed for β-galactosidase activity. β-Galactosidase activity (Miller units) was normalized to the value obtained with the vector controls. The values are the average of three separate experiments.
B–F. The ΔexsC mini-pD-lacZ reporter strain carrying either pJN-exsC or pJN-exsCHis6 was grown in the absence or presence of EGTA with the indicated concentrations of arabinose. Cell extracts were assayed for β-galactosidase activity (in Miller units) (B) and subjected to Ni-NTA affinity chromatography. Aliquots of the starting extracts (C and D) and the column eluates (E and F) were immunoblotted for ExsC and ExsD. Bands corresponding to ExsC, ExsCHis6 and ExsD are indicated by arrows. The transcription data are the average of at least three experiments.

To determine whether the ExsC–ExsD interaction occurs in P. aeruginosa, a hexahistidine-tagged ExsC derivative was placed under the transcriptional control of an arabinose-inducible promoter, and the resulting plasmid (pJN-exsCHis6) was introduced into the ΔexsC mini-pD-lacZ reporter strain. To test for an ExsC–ExsD interaction, cells were grown under increasing concentrations of arabinose, and cell extracts were subjected to β-galactosidase assay. Both tagged (ExsCHis6) and untagged ExsC complement the ΔexsC mutant for transcription of the mini-pD-lacZ reporter after growth in the presence of arabinose, indicating that ExsCHis6 is fully functional (Fig. 5B, lanes 1 and 2). Aliquots of the cell extracts were immunoblotted for ExsD (Fig. 5C), ExsC (Fig. 5D) and ExsA (data not shown), and the remainder was applied to a Ni2+-NTA affinity column. The columns were washed, and the eluates were immunoblotted for ExsD (Fig. 5E), ExsC (Fig. 5F) and ExsA (data not shown). With extracts prepared from cells expressing ExsCHis6, co-purification of ExsD is observed at levels that parallel ExsD expression levels (Fig. 5C and E, lanes 5–10). Co-purification of ExsD was not observed with cells expressing untagged ExsC despite the presence of ExsD in the cell extracts (Fig. 5C and E, lanes 1 and 2). Combined, the bacterial two-hybrid and co-purification data provide strong evidence that ExsC interacts with ExsD. ExsA was not detected in the ExsC–ExsD complex (data not shown), suggesting that ExsC functions by interfering with the interaction of ExsD and ExsA.

ExsC is a soluble, cell-associated protein

Our data demonstrate that ExsC is required for transcription of the TTSS and are consistent with ExsC relieving ExsD-dependent repression. The next question we wished to address was the relationship between low Ca2+, the activity of ExsC and transcription of the TTSS. One explanation for induction of transcription by low Ca2+ might be changes in the relative expression levels of ExsC, ExsA and ExsD. We viewed this as an unlikely explanation as all three genes are co-ordinately regulated by ExsA (Yahr and Frank, 1994; McCaw et al., 2002); however, it was important to eliminate this possibility. Cells were grown under high- and low-Ca2+ conditions and normalized to cell numbers. Extracts were prepared, subjected to immunoblot analysis, and the induction ratio of each protein (ExsC, ExsA, ExsD) was determined by densitometry from three independently prepared sets of extracts. Although the expression levels of all three proteins increased after growth in the presence of low Ca2+, there was no significant change in the relative induction ratios of ExsC (3.7-fold ± 0.8), ExsA (3.8-fold ± 1.0) or ExsD (2.8-fold ± 0.7) by  low  Ca2+.  These  data  suggest  that  the  activity  of  at least one of these proteins might be subject to a post-translational regulatory event that is coupled to the activation state of the type III secretion channel.

Another mechanism for regulation by Ca2+ might result from the physical separation of ExsC and ExsD under high-Ca2+ conditions. To investigate this, we determined the subcellular localization of ExsC. Wild-type PA103 was grown in the presence or absence of Ca2+ and fractionated into secreted (s), cell-associated (wce), soluble (sol) and insoluble (insol) fractions. As reported previously, ExsD is found in the soluble, cell-associated fraction (Fig. 6) (McCaw et al., 2002). Likewise, both ExsC and ExsA also fractionate to the soluble cell-associated fraction (Fig. 6), indicating that changes in the subcellular localization of ExsC cannot account for the induction of transcription after growth in the absence of Ca2+.

Figure 6.

A. Co-localization of ExsA, ExsC and ExsD. Supernatant (sup) and whole-cell extracts (wce) were prepared from wild-type PA103 grown under non-inducing (–EGTA) and inducing (+EGTA) conditions for expression of the TTSS. The whole-cell extract was further fractionated into soluble (sol) and insoluble (insol) fractions as described in Experimental procedures and subjected to anti-ExsA, -ExsC or -ExsD immunoblot analysis.
B. The ExsC–ExsD interaction is insensitive to Ca2+ concentrations. P. aeruginosa extracts were mock treated (lanes 1, 2, 5 and 6) or treated with 1 mM EGTA (lanes 3 and 7) or 1 mM CaCl2 (lanes 4 and 8), incubated with 1 µg of purified ExsCHis6 (lanes 2–4 and 6–8) and subjected to Ni-NTA chromatography. Aliquots of the starting extracts (lanes 1–4) and the column eluates (lanes 5–9) were immunoblotted for ExsC and ExsD. Bands corresponding to ExsD and ExsCHis6 are indicated by arrows.

A final possibility is that the ExsC–ExsD interaction may be sensitive to Ca2+ concentrations. To test this idea, a cell-free P. aeruginosa extract prepared from cells expressing ExsD was mock treated or treated with 1 mM EGTA or 1 mM CaCl2. The treated extracts were incubated with 1 µg of purified ExsCHis6 for 30 min and then applied to Ni2+-NTA affinity columns. The columns were washed, and the eluates were immunoblotted for ExsD and ExsC. As seen earlier, the presence of ExsD in the eluates is dependent upon ExsCHis6 (Fig. 6, lane 5 versus 6). The co-purification of ExsD with ExsC, however, is not sensitive to EGTA or CaCl2 treatment (Fig. 6, lanes 6 versus 7–8), suggesting that the interaction of these two proteins is not directly regulated by Ca2+ concentration.

Discussion

There are at least two physiological responses to low Ca2+: activation of the Vfr-dependent adenylate cyclase signalling cascade (Wolfgang et al., 2003) and activation of the type III secretion channel (McCaw et al., 2002). We demonstrated previously that the activity of the anti-activator, ExsD, is co-ordinated with the activity of the secretion channel (McCaw et al., 2002). Under high-Ca2+ conditions, ExsD prevents transcription by binding ExsA. Activation of secretion by low Ca2+, however, suppresses the negative regulatory activity of ExsD. We hypothesized that a specific inhibitor of ExsD is made available under low-Ca2+ conditions (McCaw et al., 2002). We now report that ExsC is the ExsD inhibitor based on the following data. First, in the absence of exsC or upon ExsD overexpression, transcription of the TTSS is repressed. In both cases, repression results from the unchecked activity of ExsD. Conversely, in the absence of exsD or upon overexpression of ExsC, transcription of the TTSS is derepressed. Under this scenario, ExsC overexpression neutralizes ExsD such that the phenotype resembles that of an exsD mutant. Secondly, the dependence on exsC for expression of the TTSS is relieved in a mutant lacking both exsC and exsD. This demonstrates that, in the absence of exsD, the activity of ExsC becomes dispensable. Thirdly, either ExsC or ExsA overexpression relieves ExsD-dependent repression and also leads to constitutive expression of the TTSS. In the case of ExsA overexpression, derepression most probably results from ExsA being in excess of ExsD. Finally, ExsC interacts directly with ExsD in both bacterial two-hybrid and co-purification assays. Combined, these data demonstrate that ExsC functions as an inhibitor of ExsD.

A second potential role for ExsC is that of a type III-specific chaperone. Although such activity has yet to be demonstrated, ExsC has biochemical characteristics typical of TTSS chaperones including a low molecular mass, an acidic isoelectric point and a carboxy-terminal amphipathic alpha-helix (Page and Parsot, 2002; Feldman and Cornelis, 2003). The chaperone-like properties of ExsC suggest a possible mechanism of coupling activation of secretion with transcription. In such a model, ExsC would serve as both an inhibitor of ExsD and a chaperone for a type III secreted protein (protein X). Before the activation of secretion, protein X remains cytosolic and sequesters ExsC from ExsD (Fig. 7A). This might result if protein X has a higher affinity for ExsC or if protein X is in large excess over ExsD. Activation of secretion initiates a sequence of events. First, protein X is secreted leaving behind free ExsC (Fig. 7B). Liberated ExsC then binds to and inhibits ExsD. Finally, ExsA-dependent transcription of the TTSS ensues. This model fits well with our findings that ExsC and ExsD are not secreted or differentially expressed, and that the ExsC–ExsD interaction is not directly sensitive to Ca2+ concentrations.

Figure 7.

Model for regulation of the P. aeruginosa TTSS. See text for details.

Assuming the model to be correct, the major piece of the puzzle remaining is the identity of the type III secreted protein. If ExsC functions as a chaperone for a type III-related exoproduct, then secretion of that protein might be impaired in the absence of ExsC. To address this possibility, the extracellular protein profile of the ΔexsCD and ΔexsD mutants was compared by SDS–PAGE and silver staining. No significant change in the amounts of known type III-related exoproducts was detected (data not shown). Likewise, co-purification studies using the histidine-tagged form of ExsC have thus far failed to identify additional ExsC interaction partners. Future work will be directed towards the identification of other proteins that interact with ExsC.

Although the structural components of the type III secretion and translocation apparatus are well conserved across species, the regulatory mechanisms controlling expression of TTSSs are diverse (Hueck, 1998; Francis et al., 2002). This diversity is likely to be the result of adaptation to suit the individual lifestyles of specific pathogens. One feature common to many systems is a mechanism of coupling gene expression to the activation state of the secretion channel (Miller, 2002). Despite the conservation of this feature, the mechanisms of accomplishing this task are quite varied. The Salmonella typhimurium and Shigella flexneri systems make use of a co-activator and an activator (Darwin and Miller, 2001; Mavris et al., 2002). The co-activator also functions as a chaperone for a secreted protein. Upon activation of secretion, the co-activator is released and is free to interact with the activator, resulting in transcription of the TTSS. The Yersinia sp. systems use a negative regulator that is secreted upon activation of the secretion channel (Pettersson et al., 1996).

As presented here, the P. aeruginosa system requires the activity of at least three proteins, a transcriptional activator (ExsA), a non-secreted anti-activator (ExsD) and a non-secreted anti-anti-activator (ExsC). This mechanism of gene regulation appears to be novel given that a literature search failed to identify other examples of anti-activator/anti-anti-activator partners. The closest corollary is seen in the regulatory system controlling Bacillus subtilis sporulation, where a sigma factor, an anti-sigma factor and an anti-anti-sigma factor control gene transcription in the forespore (Margolis et al., 1991; Min et al., 1993). The forespore compartment sigma factor (σF) is subject to regulation by the anti-sigma factor SpoIIAB (Duncan and Losick, 1993). SpoIIAB prevents transcription of sporulation genes in the mother cell. In the forespore, the anti-anti-sigma factor (SpoIIAA) is dephosphorylated and subsequently binds to SpoIIAB (Duncan et al., 1996). This liberates σF to bind RNA polymerase and activate the transcription of forespore compartment genes.

Although there are no other reports of TTSSs that make use of an anti-activator and an anti-anti-activator, blast searches with ExsC and ExsD reveal homologues of each in the recently completed genome sequences of V. parahaemolyticus and P. luminescens (Duchaud et al., 2003; Makino et al., 2003). This finding suggests that other organisms may use a mechanism similar to that reported here to regulate the transcription of TTSS genes.

Experimental procedures

Bacterial strains and culture conditions

Pseudomonas aeruginosa strains were maintained on Vogel–Bonner minimal medium (Vogel and Bonner, 1956) with antibiotics as required (300 µg ml−1 carbenicillin, 100 µg ml−1 tetracycline, 100 µg ml−1 gentamicin). The exsA::Ω and ΔexsD mutants have been described previously (Frank et al., 1994; McCaw et al., 2002). For expression of the TTSS, strains were grown at 30°C with vigorous aeration in trypticase soy broth (TSB) supplemented with 1% glycerol, 100 mM monosodium glutamate, 2 mM EGTA and antibiotics at the concentrations indicated above.

Construction of the PA103 ΔexsC and ΔexsCD mutants

Polymerase chain reaction (PCR) primers were used to amplify upstream (5′-SstI-TGTCGAGCTCTGTCGCACTC GCTGAAACTCGG and 5′-BamHI-ACTAGGATCCGACCT TGCTCGTTAAATCCATAGGG) and downstream (5′-BamHI-TAGTGGATCCGTCGGCATGAGGGTTTGAGCG and 5′-HindIII-GTGAAAGCTTCATGCACAAGCAAT TCCT TCAACC) flanking regions of exsC. PCR products were sequentially ligated into pEX18Tc (Hoang et al., 1998), resulting in an in frame deletion of codons 8–141. The resulting plasmid (pEX18TcΔexsC) was mobilized to wild type and ΔexsD mini-pD-lacZ by conjugation, and tetracycline-resistant merodiploids were isolated. Merodiploids were resolved as described previously and confirmed by PCR and Southern analyses (Sawa et al., 1999). The resulting mutants, PA103 ΔexsC and PA103 ΔexsCD, had no discernable growth defects when compared with wild type.

Expression and purification of ExsC and ExsA

For polyclonal antiserum production, exsC and exsA were PCR amplified as NdeI–XhoI fragments and cloned into the respective sites of the expression vectors pET23b and pET16b (Novagen). The pET23b-exsC construct incorporates a carboxy-terminal hexahistidine epitope tag, and the pET16b-exsA construct incorporates an amino-terminal 10-residue histidine tag. Histidine-tagged proteins were purified by Ni2+-NTA affinity chromatography, and antisera were generated in New Zealand White rabbits as described previously (Yahr et al., 1998). Antibody against ExsC and ExsA was affinity purified by immobilizing recombinant ExsCHis6 and ExsA His10 to Sulfo-link (Pierce) resin as described previously (Duong and Wickner, 1997).

ExsC and ExsA expression plasmids

The pexsC expression plasmid was constructed by PCR amplification of exsC under the transcriptional control of its native promoter pC with the following primers (5′-HindIII-ACGCAAGCTTATGAAGGACGTCCTGCAGCTCATCC and 5′-SacI-AGT TGAGCTCCGCTGCA AGGCCTGCTCACTACG). The PCR product was cloned into the corresponding sites of pUCP18 (West et al., 1994). pJN-exsC was constructed by PCR amplification of exsC (5′-XbaI-ACAGTCTAGAGGCGC CCCCATGGATTTAACG and the SacI primer described above). The PCR product was cloned into the corresponding sites of pJN105 (Newman and Fuqua, 1999). pJN-exsCHis6 was constructed by PCR amplification of exsCHis6 (5′-XbaI-ACAGTCTAGAGGCGCCCCCATGGATTTAACGAG and 5′-SacI-ACAGGAGCTCAAAAAACCCCTCAAGACCCG) from pET23-exsC and cloning into pJN105. Construction of the pexsD expression plasmid was reported previously (McCaw et al., 2002).

SDS–PAGE, immunoblots and β-galactosidase assays

For analysis of secreted and cell-associated fractions, cells were grown in TSB to an absorbance (A540) of 1.0. Supernatant fractions were prepared by pelleting 1.5 ml of culture (12 500 g) for 4 min. Supernatant fractions (1 ml) were transferred to a fresh microfuge tube, and protein was precipitated by the addition of 350 µl of 50% trichloroacetic acid. After incubation on ice (30 min), precipitated protein was collected by centrifugation (10 min, 12 500 g), washed with 1 ml of acetone and suspended in 15 µl of SDS–PAGE sample buffer. Cell-associated fractions were prepared by pelleting 1.25 ml of cell culture (A540 = 1.0), suspending in 300 µl of SDS–PAGE sample buffer and sonicating for 10 s. Supernatant (derived from 3.6 × 108 cfu) and cell-associated (equivalent to 3.3 × 107 cfu) samples were heated for 5 min at 95°C and electrophoresed on Hi-Tris (to visualize ExsC) or 15% SDS-polyacrylamide gels. Immunoblots were developed using ECL and ECF reagents (Amersham Pharmacia Biotech) and anti-rabbit horseradish peroxidase or alkaline phosphatase conjugants (Chemicon) respectively. For determination of ExsC, ExsA and ExsD induction ratios, twofold serial dilutions of cell extract were subjected to SDS–PAGE and immunoblot analyses. Immunoblots were scanned for chemifluorescence and analysed using a Molecular Dynamics Typhoon Phosphorimager and software. The mini-pC-, mini-pD- and mini-pS-lacZ transcriptional reporters were constructed as described previously (Becher and Schweizer, 2000; McCaw et al., 2002). Strains harbouring the reporters were grown under the indicated conditions until the absorbance (A540) reached 1.0, and β-galactosidase activity was measured using the substrate ONPG (Miller, 1972). The reported values (Miller units) represent the average of at least three independent experiments.

Fractionation

Strains were grown in TSB to an absorbance (A540) of 1.0. Supernatant fractions were prepared as described above. For preparation of the soluble and insoluble fractions, 1.7 × 109 cells were sedimented by centrifugation (4 min, 12 500 g), washed with 1 ml of 10 mM Tris (pH 8.0) and resedimented. The cell pellet was suspended in 300 µl of 10 mM Tris (pH 8.0), 20 mM EDTA, sonicated for 20 s and centrifuged (12 500 g) for 4 min to remove unbroken cells. The cell extract was transferred to a fresh tube and centrifuged at 100 000 g for 30 min at 4°C to separate the soluble fraction from the insoluble membrane fraction (pellet).

Bacterial two-hybrid screen

The exsC two-hybrid expression plasmids were constructed by PCR amplification of exsC (5′-SacI-TGTAGAGCTCATG GATTTAACGAGCAAGGT and 5′-KpnI-GTGAGGTACCT CAAACCCTCATGCCGACC) and cloning into the respective sites of the pSR658 and pSR659 two-hybrid expression vectors (Dmitrova et al., 1998; Daines and Silver, 2000). Clones were confirmed by nucleotide sequence analysis and transformed into the E. coli SU202 reporter strain (Dmitrova et al., 1998). To assay for β-galactosidase activity, overnight cultures were back-diluted to an A600 of 0.05 in LB containing 1 mM IPTG and the appropriate antibiotics. Cultures were incubated at 30°C until the A600 reached 1.0. Cells were harvested and assayed for β-galactosidase activity as described earlier.

Ni 2+-NTA affinity chromatography

The ΔexsC mini-pD-lacZ reporter strain carrying either pJN-exsC or pJN-exsCHis6 was grown under non-inducing (–EGTA) or inducing (+EGTA) conditions for expression of the TTSS in the presence of the indicated concentration of l-arabinose to absorbance (A540) of 1.0. Cells (1.7 × 109 cfu) were sedimented by centrifugation (5 min, 12 500 g) in microfuge tubes and frozen at −70°C. For Ni2+-NTA chromatography, cell pellets were suspended in 250 µl of IP buffer [50 mM Tris, pH 7.5, 150 mM NaCl, 20% glycerol (w/v), 0.5% Tween 20] and 10 µl of a protease inhibitor cocktail [0.5 µg ml−1 leupeptin (Sigma), 25 mM 1,10-phenanthroline (Sigma), 25 µg ml−1 pepstatin A (Roche Applied Science) and 5 mM Pefabloc SC (Roche)]. Cells were lysed by sonication (two 8 s pulses, 50% duty cycle), and unbroken cells were removed by centrifugation (10 min, 12 500 g, 4°C). The volume of the cell extract was adjusted to 500 µl with the addition of 250 µl of IP–BSA buffer (IP buffer containing 0.5% bovine serum albumin). Ni2+-NTA spin columns (Qiagen) were equilibrated with IP–BSA buffer, and the lysate was applied according to the manufacturer's instructions. The columns were washed twice with 600 µl of IP buffer containing 20 mM imidazole and eluted with 200 µl of elution buffer (20 mM Tris, pH 8.0, 150 mM NaCl, 300 mM imidazole). Protein in the eluate was precipitated by the addition of 70 µl of 50% trichloroacetic acid and incubated on ice for 30 min. Precipitates were collected by centrifugation (12 min, 13 000 g, 4°C) and washed with 1 ml of ice-cold acetone. The protein pellet was dried, suspended in 20 µl of SDS–PAGE sample buffer, heated to 95°C for 5 min and subjected to immunoblot analysis.

Acknowledgements

We thank Phoebe Lostroh and Mark Urbanowski for their many suggestions throughout the evolution of this work, and Elisabeth Naylor for excellent technical assistance. Support for these studies was provided by the Howard Hughes Medical Institute Biomedical Research Support Faculty Start-up Program, the University of Iowa W. M. Keck Microbial Communities and Cell Signaling Program, the Cystic Fibrosis Foundation Research Development Program and the National Institutes of Health (RO1-AI055042).

Ancillary