Two-component regulatory systems in Pseudomonas aeruginosa: an intricate network mediating fimbrial and efflux pump gene expression

Abstract

Pseudomonas aeruginosa is responsible for chronic and acute infections in humans. Chronic infections are associated with production of fimbriae and the formation of a biofilm. The two-component system Roc1 is named after its role in the regulation of cup genes, which encode components of a machinery allowing assembly of fimbriae. A non-characterized gene cluster, roc2, encodes components homologous to the Roc1 system. We show that cross-regulation occurs between the Roc1 and Roc2 signalling pathways. We demonstrate that the sensors RocS2 and RocS1 converge on the response regulator RocA1 to control cupC gene expression. This control is independent of the response regulator RocA2. Instead, we show that these sensors act via the RocA2 response regulator to repress the mexAB-oprM genes. These genes encode a multidrug efflux pump and are upregulated in the rocA2 mutant, which is less susceptible to antibiotics. It has been reported that in cystic fibrosis lungs, in which P. aeruginosa adopts the biofilm lifestyle, most isolates have an inactive MexAB-OprM pump. The concomitant RocS2-dependent upregulation of cupC genes (biofilm formation) and downregulation of mexAB-oprM genes (antibiotic resistance) is in agreement with this observation. It suggests that the Roc systems may sense the environment in the cystic fibrosis lung.

Fig. 1

β-galactosidase activity (Miller Units) of PAK wild type or isogenic deletion mutants carrying chromosomal transcriptional fusions at the attB site (cupB-lacZ or cupC-lacZ) as well as a pMMB67HE vector, either empty or overexpressing rocS2, rocA2 or rocS1 as indicated. Data are the average of biological duplicates, and error bars indicate one standard deviation of the mean. A. rocS2 overexpression induces cupC expression, but this induction is not seen by overexpressing rocA2. The difference between the strain containing the pMMB67HE vector or containing pMMB67-rocS2 is significant with a P-value < 0.001. Furthermore, rocS2 still induces cupC expression in the PAKΔrocA2 mutant (difference between the strain containing the pMMB67HE vector or containing pMMB67-rocS2 is significant with a P-value < 0.001), but the induction is lost in the PAKΔrocA1 mutant indicating that RocA1, but not RocA2, is required for this signalling. B. rocS2 overexpression also induces cupB expression. This induction is independent of both RocA2 and RocA1 as seen in the single and double deletion mutants. All differences between the strain containing the pMMB67HE vector or containing pMMB67-rocS2 are significant with a P-value < 0.05. C. cupC expression can be induced by both rocS1 and rocS2, but in both cases the promoter activity is higher in a PAKΔrocR mutant showing that RocR is a repressor of cupC gene expression. The difference between the strains PAK::cupC-lacZ and PAKΔrocR::cupC-lacZ containing pMMB67-rocS2 is significant with a P-value < 0.02.

Fig. 2

Protein–protein interactions between components of the Roc1 and Roc2 systems as determined by bacterial two-hybrid analysis. Various combinations of recombinant pKT25 and pUT18c plasmids harbouring protein domains of interest (Hpt domains of sensors and D2 domains of response regulators) were co-transformed into E. coli DHM1, and transformants were spotted onto MacConkey agar (A). Dark red colonies indicate a positive interaction. The E. coli two component sensor TorS, together with its cognate response regulator TorR, was used as a positive control. The specificity of interaction was assessed using the unrelated P. aeruginosa TCS GacS/GacA, as well as the unrelated P. aeruginosa response regulator TrpO. The strength of interaction was investigated by measuring the β-galactosidase activity of cells in the respective colonies, and the average activity in Miller Units is indicated next to each colony. (B) Graphic representation of the β-galactosidase activity in panel A. All experiments were carried out in at least duplicates, and error bars represent one standard deviation of the mean. Plasmid combinations of pKT25/pUT18c, respectively, are indicated in panels A and B. It should be noted that E means empty vector. For specific constructs, see Table 1. In panel B, stars indicate a significant difference between a given combination and the control (E/E). Three stars or one star corresponds to P-value < 0.001 or < 0.05 respectively.

Fig. 3

Quantitative RT-PCR analysis of mexA and mexR gene expression. The analysis was done upon overexpression of the sensor-encoding genes rocS2 (A) and rocS1 (B) in a PAK wild type (white bars) and PAKΔrocA2 deletion mutant (grey bars). Expression levels were normalized to the 16SrRNA gene and wild-type levels have been set to 1. The data represent the average of biological triplicates and error bars indicate one standard deviation of the mean. In (A) and (B) the relative expression of mexA is indicated on the left and mexR on the right, as indicated below the figure. In all cases the difference is significant with a P-value < 0.05.

Fig. 4

Antibiotic Disc Diffusion Assay. Discs containing Aztreonam (30 µg), Cefotaxime (300 µg) Cinoxacin (100 µg) or compound sulphonamides (300 µg) were applied to M63 media plates overlaid with PAK wild-type or PAKΔrocA2 harbouring pMMB67-rocS2. Discs containing Carbenicillin (100 µg) were applied to M63 media plates inoculated with PAK wild-type harbouring pMMB67-rocS1 or PAKΔrocA2 harbouring pMMB67-rocS1. The bar in the top left panel is a 1 cm scale.

Fig. 5

Schematic model showing the cross-regulation between Roc1 and Roc2 signalling pathways. The Roc1 and Roc2 components are represented as in Fig. S1. The RocS1 and RocS2 sensors are shown as integral inner membrane proteins. The positive regulation on the cupC gene expression is shown with green arrows. The negative regulation on the mexAB-oprM genes is shown with brown arrows. The positive regulation on cupB genes is shown with blue arrows and involves a yet unknown regulator indicated as X.