Bacterial TonB-dependent transducers interact with the anti-σ factor in absence of the inducing signal protecting it from proteolysis

Abstract

Competitive bacteria like the human pathogen Pseudomonas aeruginosa can acquire iron from different iron carriers, which are usually internalized via outer membrane TonB-dependent receptors (TBDRs). Production of TBDRs is promoted by the presence of the substrate. This regulation often entails a signal transfer pathway known as cell-surface signaling (CSS) that involves the TBDR itself that also functions as transducer (and is thus referred to as TBDT), a cytoplasmic membrane-bound anti-σ factor, and an extracytoplasmic function σ (σECF) factor. TBDTs contain an extra N-terminal domain known as signaling domain (SD) required for the signal transfer activity of these receptors. In the current CSS model, presence of the signal allows the interaction between the TBDT and the anti-σ factor in the periplasm, promoting the proteolysis of the anti-σ factor and in turn the σECF-dependent transcription of response genes, including the TBDT gene. However, recent evidence shows that σECF activity does not depend on this interaction, suggesting that the contact between these 2 proteins fulfills a different role. Using the P. aeruginosa Fox CSS system as model, we show here that the SD of the FoxA TBDT already interacts with the C-terminal domain of the FoxR anti-σ factor in absence of the signal. This interaction protects FoxR from proteolysis in turn preventing transcription of σFoxI-dependent genes. By structural modeling of the FoxR/FoxASD interaction, we have identified the interaction sites between these 2 proteins and provide the molecular details of this interaction. We furthermore show that to exert this protective role, FoxA undergoes proteolytic cleavage, denoting a change in the paradigm of the current CSS model.

Fig 1. Effect of overproducing the SDs of CSS receptors on CSS activity and anti-σ factor proteolysis.

In both panels, P. putida KT2440 or the P. aeruginosa PAO1 wild-type strains bear the empty pBBR1MCS-5 plasmid (-) or the pBBR1MCS-5-derived plasmid expressing the indicated SD (+ Pp- or Pa-SD) (S1 Table). Strains were grown in iron-restricted conditions (- signal) or iron-restricted medium supplemented with the cognate CSS inducing signal (+ signal), i.e., 1 μm ferrioxamine B for the Pp and Pa Fox systems, 40 μm ferrichrome for the Pp and Pa Fiu systems, aerobactin-containing supernatant for the Pp Iut system, and 20 μm haem for the Pa Hxu system. 1 mM IPTG was added to the cultures used in panel B. (A) β-galactosidase activity of the indicated lacZ fusion genes produced from pMP220-derived plasmids (S1 Table). Data are means ± SD from 3 biological replicates (N = 3). P-values were calculated by two-tailed t test by comparing the value obtained in the strain overexpressing the SD with that of the wild-type strain in the same growth condition and are represented in the graphs by ***, P < 0.001; and ****, P < 0.0001. (B) Western-blot analyses of P. aeruginosa HA-FoxR anti-σ factor and P. putida HA-IutY σ/anti-σ factor hybrid protein produced from pMMB67EH-derived plasmids (S1 Table). Proteins were immunoblotted against the HA-epitope using a monoclonal antibody. Positions of the protein fragments and the molecular size marker (in kDa) are indicated. Presence of the HA-tag adds ∼1 kDa to the molar mass of the protein fragments. Blots are representatives of at least 3 biological replicates (N = 3). The raw data underlying the graphs shown in the figure can be found at Mendeley Data repository (Mendeley Data, V1, 10.17632/nxh4c8ymnn.2). Western blot can be found in S1 Raw Images. CSS, cell-surface signaling; SD, signaling domain.

Fig 2. Analysis of the FoxR/FoxA^SD interaction in vivo by two-hybrid and in vitro by analytical ultracentrifugation.

(A) Schematic representation of the FoxR protein and the protein fragments used in the two-hybrid assay. The amino acids contained in each fragment are indicated in brackets. (B) Bacterial two-hybrid assay to test interactions between FoxA^SD and the different FoxR fragments shown in A. E. coli BTH101 bearing the pUTC18C and pKNT25 empty plasmids (- /-) or its derivatives encoding the indicated FoxR and FoxA^SD proteins were grown in LB with 0.5 mM IPTG. A strain bearing the pUTC18C-zip and pKNT25-zip plasmids (zip / zip) was used as a positive control. β-galactosidase assay was performed, and data are means ± SD from at least 3 biological replicates (N = 3). P-values were calculated by two-tailed t test by comparing the value obtained for one strain with that of the negative control (- / -) and are represented in the graphs by ***, P < 0.001; and ****, P < 0.0001. (C) Biophysical characterization of the FoxR/FoxA^SD interaction. The upper panel shows the sedimentation velocity AUC analysis of FoxR^peri, FoxA^SD, and the FoxR^peri/FoxA^SD complex. Values given correspond to experiments conducted at 8°C in Tris 20 mM (pH 8.3), NaCl 500 mM, DTT 1 mM, Glycerol 3% buffer and have not been standardized to s_20,w. The lower panel shows the sw isotherm analysis of the sedimentation coefficients from a FoxR^peri/FoxA^SD titration. The solid curve indicates the best fit of a global analysis of 2 sw-isotherms (absorbance and interference data). The bottom panel shows the residual plot. The raw data underlying the graphs shown in the figure can be found at Mendeley Data repository (Mendeley Data, V1, 10.17632/nxh4c8ymnn.2). AUC, analytical ultracentrifugation;

Fig 3. Structural prediction of the FoxR/FoxA interaction and effect of key residues on FoxR/FoxA interaction.

(A) The signaling domain of FoxA (in orange) interacts with FoxR (in blue) via their STN-like domains. Close-up view involving the residues predicted to participate in the FoxR/FoxA interaction and the distances between the interacting residues are shown. The interaction model was generated with Alphafold2 using the FoxR^peri domain (amino acids 107–328, Fig 2A) and FoxA^1-516. (B, C) Bacterial two-hybrid assay to test interactions of the 2 subdomains of FoxR^C with FoxA^SD (B), and between the FoxR^C and FoxA^SD protein variants bearing the indicated single residue change (C). E. coli BTH101 bearing the pUTC18C and pKNT25 empty plasmids (- /-) or its derivatives encoding the indicated FoxR^C and FoxA^SD proteins were grown in LB with 0.5 mM IPTG. A strain bearing the pUTC18C-zip and pKNT25-zip plasmids (zip / zip) was used as a positive control. β-galactosidase assay was performed, and data are means ± SD from 3 (N = 3) biological replicates. P-values were calculated by two-tailed t test by comparing the value obtained in the strain bearing the changed protein variant(s) with that of the strain bearing the wild-type FoxR^C / FoxA^SD constructs and are represented in the graphs by ns, no significant; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. The raw data underlying the graphs shown in the figure can be found at Mendeley Data repository (Mendeley Data, V1, 10.17632/nxh4c8ymnn.2).

Fig 4. Role of FoxA^SD in CSS activity.

(A) β-galactosidase activity of the P. aeruginosa foxA::lacZ fusion gene in the PAO1 wild-type bearing the empty pBBR1MCS-5 plasmid (-) or the pBBR1MCS-5-derived plasmid expressing the wild-type FoxA^SD protein or the indicated protein variant (S1 Table). Strains were grown in iron-restricted conditions without or with 1 μm ferrioxamine B. Data are means ± SD from 5 (N = 5) biological replicates. P-values were calculated by two-tailed t test by comparing the value obtained in the strain overproducing the single change residue protein variant with that of the wild-type FoxA^SD protein in the same growth condition and are represented in the graphs by ****, P < 0.0001. (B) Western blot analyses of P. aeruginosa PAO1 producing an N-terminally HA-tagged FoxR anti-σ factor from a pMMB67EH-derived plasmid (S1 Table). Strains also bear the pBBR1MCS-5 empty plasmid (-) or the pBBR1MCS-5-derived plasmid expressing the wild-type FoxA^SD protein or the indicated protein variants (S1 Table) and were grown in iron-restricted conditions without (-) or with (+) 1 μm ferrioxamine B and 1 mM IPTG. Proteins were immunoblotted against the HA-epitope using a monoclonal antibody. Positions of the protein fragments and the molecular size marker (in kDa) are indicated. Presence of the HA-tag adds ∼1 kDa to the molar mass of the protein fragments. Blots are representatives of 3 biological replicates (N = 3). (C) P. aeruginosa PAO1 wild-type strain and the indicated isogenic mutants were grown to late log-phase under iron-restricted conditions and in the absence (-) or presence (+) of 1 μm ferrioxamine. Total proteins were prepared and immunoblotted against the FoxA TBDT using a polyclonal antibody directed against a peptide of the β-barrel of the protein. Positions of the protein fragments and the molecular size marker (in kDa) are indicated. Blot is a representative of 4 biological replicates (N = 4). The raw data underlying the graphs shown in the figure can be found at Mendeley Data repository (Mendeley Data, V1, 10.17632/nxh4c8ymnn.2). Western blot can be found in S1 Raw Images. CSS, cell-surface signaling; TBDT, TonB-dependent transducer.