Exploitation of an iron transporter for bacterial protein antibiotic import.

Unlike their descendants, mitochondria and plastids, bacteria do not have dedicated protein import systems. However, paradoxically, import of protein bacteriocins, the mechanisms of which are poorly understood, underpins competition among pathogenic and commensal bacteria alike. Here, using X-ray crystallography, isothermal titration calorimetry, confocal fluorescence microscopy, and in vivo photoactivatable cross-linking of stalled translocation intermediates, we demonstrate how the iron transporter FpvAI in the opportunistic pathogen Pseudomonas aeruginosa is hijacked to translocate the bacteriocin pyocin S2 (pyoS2) across the outer membrane (OM). FpvAI is a TonB-dependent transporter (TBDT) that actively imports the small siderophore ferripyoverdine (Fe-Pvd) by coupling to the proton motive force (PMF) via the inner membrane (IM) protein TonB1. The crystal structure of the N-terminal domain of pyoS2 (pyoS2NTD) bound to FpvAI (Kd = 240 pM) reveals that the pyocin mimics Fe-Pvd, inducing the same conformational changes in the receptor. Mimicry leads to fluorescently labeled pyoS2NTD being imported into FpvAI-expressing P. aeruginosa cells by a process analogous to that used by bona fide TBDT ligands. PyoS2NTD induces unfolding by TonB1 of a force-labile portion of the plug domain that normally occludes the central channel of FpvAI. The pyocin is then dragged through this narrow channel following delivery of its own TonB1-binding epitope to the periplasm. Hence, energized nutrient transporters in bacteria also serve as rudimentary protein import systems, which, in the case of FpvAI, results in a protein antibiotic 60-fold bigger than the transporter's natural substrate being translocated across the OM.

Bacteriocins are peptide or protein antibiotics produced by bacteria to kill their neighbors, usually in response to environmental stress, that play a fundamental role in shaping bacterial communities (1–3) and are implicated in the invasion mechanisms of pathogens (4, 5). Bacteriocins are currently the focus of intense efforts to develop them as much needed antibiotics against multidrug-resistant bacteria (6, 7) and as antiinfectives for use in agriculture (8). The present work centers on the mode of action of protein bacteriocins, which are species-specific protein antibiotics that are widespread agents of competition in gram-negative bacteria (9). Protein bacteriocins are known to parasitize a variety of cell envelope proteins (10), but how they exploit these systems to promote their import has remained unresolved since their discovery (11). We reveal the mechanism by which the nuclease bacteriocin pyocin S2 (pyoS2) crosses the outer membrane (OM) of Pseudomonas aeruginosa, a path that is likely to be used by other TonB-dependent protein bacteriocins.

Protein bacteriocins are 40- to 80-kDa toxins that carry a single cytotoxic domain at their C terminus into the cell. Cell death ensues through depolarization of the cell by an ionophore or enzymatic cleavage of peptidoglycan precursors in the periplasm or nucleic acids (DNA, tRNA, or rRNA) in the cytoplasm (12). The best studied of the protein bacteriocins are the colicins that target and kill Escherichia coli. Colicins exploit a variety of OM proteins as their primary receptor, including the vitamin B12 transporter BtuB and the siderophore transporter FepA (13, 14). Colicin entry into cells requires contact with proton motive force (PMF)-linked systems in the inner membrane (IM) that span the periplasm: Tol-Pal for group A colicins and Ton for group B colicins (15). Tol-Pal is a multiprotein complex involving three IM proteins (TolA, TolQ, and TolR), a periplasmic protein (TolB), and an OM lipoprotein (Pal). The Ton system comprises three IM proteins (TonB, ExbB, and ExbD). Both systems, which are virulence factors in pathogenic bacteria, energize processes at the OM: Tol-Pal stabilizes the membrane, while Ton catalyzes import of scarce nutrients such as iron and vitamins across the membrane. Group A colicins generally require additional OM proteins, usually porins, to contact the Tol-Pal system (16). Few translocator proteins have been identified for group B colicins (10). How the Ton or Tol-Pal system catalyzes import of a protein bacteriocin across the OM is unknown. Once in the periplasm, nuclease colicins are translocated across the IM by the AAA+ ATPase FtsH, which also proteolytically releases the nuclease domain to the cytoplasm (17, 18). In the present work, we reveal how pyoS2, a Ton-dependent protein bacteriocin, translocates across the OM through its receptor and show that this mechanism has strong parallels with that used by the endogenous ligand for the receptor.

Results

The N-Terminal Domain of PyoS2 Mimics Ferripyoverdine Binding to FpvAI.

PyoS2 is a 74-kDa endonuclease bacteriocin that is effective in the treatment of P. aeruginosa-induced pneumonia in mice and in eradicating P. aeruginosa biofilms (6, 19). The primary receptor for pyoS2 is FpvAI (20–22), a TonB-dependent transporter (TBDT) (23) that actively imports ferripyoverdine (Fe-Pvd) (24). FpvAI is a classical 22-stranded β-barrel with a central channel that is completely occluded by a globular “plug” domain, which must be reconfigured for substrate transport by coupling to the PMF via TonB1 (25).

Our starting point for elucidating the pyoS2 mechanism of entry was to delineate the FpvAI-binding domain by limited trypsin digestion (residues 1–209; ). The isolated 23-kDa N-terminal domain of pyoS2 (pyoS2NTD) was folded (Fig. S1) and outcompeted Fe-Pvd for FpvAI (Fig. 1), which it bound with high affinity in competition isothermal titration calorimetry (ITC) experiments (Kd = 240 pM) (Fig. 1). We solved the crystal structure of the FpvAI–pyoS2NTD complex by molecular replacement to a resolution of 2.8 Å (details are provided in ). Continuous electron density was observed for residues 11–208 of the pyoS2NTD, and near-complete density was observed for FpvAI (Fig. 1 and Table S1). PyoS2NTD is composed of five α-helices, including a C-terminal helix extending ∼80 Å from the interface and a short antiparallel β-hairpin at the N terminus (residues 11–26) that docks onto the helical bundle. Spectroscopic measurements indicated that this N-terminal region is folded in solution and contributes to the stability of the domain, but high B-factors in the crystal structure suggest this region is likely to be dynamic (Fig. S1). An 11-amino acid proline-rich region (PRR; residues 35–45) contacts primarily the FpvAI plug domain. Deletion of the first 45 residues, including the PRR (Δ1–45 pyoS2NTD), decreased pyoS2NTD binding ∼1,000-fold (Fig. S2). Importantly, the PRR not only occupies the same binding site as Fe-Pvd but also resembles its shape (Fig. 1 and Fig. S3). Consequently, pyoS2NTD induces conformational changes within the plug that are very similar to those triggered by Fe-Pvd, as evidenced by comparable backbone rmsds for the two complexes relative to unliganded receptor and by the equivalent reconfiguring of the same secondary structure elements (Fig. S3 ). In summary, the structure of the complex points to pyoS2 exploiting FpvAI both as a receptor and translocator, in contrast to bacteriocins that recruit additional OM proteins for import (3).

Fig. 1.

High-affinity binding of pyoS2NTD to FpvAI mimics that of Fe-Pvd. (A) PyoS2NTD outcompetes Fe-Pvd for FpvAI binding in analytical gel-filtration chromatography. Protein elution was monitored at A280 (solid line), and Fe-Pvd elution was monitored at A400 (dashed line). mAu, milliabsorbance units. (B) ITC data for pyoS2NTD binding FpvAI in the presence of a lower affinity competitor, Δ1–45 pyoS2NTD (details are provided in ). Data from two independent experiments were fitted to a competitive binding model from which the following thermodynamic parameters were extracted: ΔH = −15.8 ± 1.4 kcal/mol, ΔS = −9.1 ± 4.6 cal⋅mol−1⋅K−1, n = 0.78 ± 0.02 sites, Kd = 0.24 ± 0.004 × 10−9 M. (C) The 2.8-Å crystal structure of pyoS2NTD (blue) in complex with FpvAI (gray) highlighting the pyoS2NTD β-hairpin (yellow), the FpvAI plug (brown) and signaling domains (black), and the TonB1 box (red). Several of the β-strands that make up the β-barrel of FpvAI have been removed to highlight the plug domain. The signaling domain of FpvAI is involved in activating its own expression and that of Pvd (41). (D and E) Comparison of FpvAI plug domain surface when bound by the PRR of pyoS2NTD and Fe-Pvd, respectively, highlighting the movement of FpvAI Tyr231 from its position in the unliganded state (light blue).

PyoS2NTD Translocates into P. aeruginosa Cells.

We developed a fluorescence-based assay using Alexa Fluor 488-conjugated pyoS2NTD (pyoS2NTD-AF488) to probe pyoS2NTD import in vivo. Addition of pyoS2NTD-AF488 to live P. aeruginosa PAO1 cells yielded fluorescent bacteria, whereas those lacking FpvAI were not labeled (Fig. 2). Fluorescence recovery after photobleaching (FRAP) was used to monitor pyoS2NTD-AF488 translocation. Since OM proteins have restricted lateral diffusion (26), the expectation was that pyoS2NTD-AF488 bound to FpvAI would not show fluorescence recovery in FRAP experiments. Instead, we observed rapid fluorescence recovery (Fig. 2), suggesting pyoS2NTD-AF488 was translocating to the periplasm, where it could diffuse freely. We investigated the energetics of import as a test of this hypothesis. Dissipation of the PMF by the addition of the protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP) to P. aeruginosa before labeling with pyoS2NTD-AF488 resulted in no recovery of fluorescence in FRAP experiments (Fig. 2), consistent with pyoS2NTD-AF488 remaining bound to FpvAI in the OM of a deenergized cell. TBDT ligands are transported through their receptors in conjunction with the PMF, which is transduced into mechanical force by engagement of the IM protein TonB with a TonB box motif of the transporter (23) (Fig. 1). The TonB box of FpvAI is known to be required for both Fe-Pvd import and pyoS2 toxicity (27). Of the three tonB genes in P. aeruginosa, only tonB1 was required for pyoS2 toxicity (Fig. S4), as has been reported for Fe-Pvd import (28). We also identified a putative TonB box at the N terminus of pyoS2NTD (M11VITH15) that comprises the first β-strand of the β-hairpin. TonB-dependent bacteriocins such as colicin B, which targets E. coli, generally have their own TonB box (29). Direct binding of pyoS2NTD to TonB1 was shown by ITC (Kd ∼ 1 μM), as well as by cross-linking, and was abolished when the β-hairpin was deleted (Δ1–30 pyoS2NTD) (Fig. S4 ). Deletion of the TonB1 box (Δ1–21 pyoS2NTD-AF488) also abolished fluorescence recovery in FRAP experiments, as well as pyoS2 cytotoxicity (Fig. S5 ). Finally, Δ1–21 pyoS2NTD-AF488 fluorescence, when bound to cells, was completely removed by trypsin treatment, in contrast to pyoS2NTD-AF488 (Fig. S5 ). We conclude that as well as being the receptor-binding domain for the bacteriocin, pyoS2NTD carries all of the information necessary for translocation across the OM and that the process requires at least FpvAI, the PMF, and TonB1 in the IM. Thereafter, import to the cytoplasm is mediated by the bacteriocin nuclease domain FtsH and a conserved DPY motif present in all nuclease bacteriocins (9, 17, 30).

Fig. 2.

PyoS2NTD translocates across the P. aeruginosa OM. (A) Fluorescence labeling of live FpvAI-expressing P. aeruginosa PAO1 with pyoS2NTD-AF488. The PAO1 fpvA− transposon mutant PW5036 (fpvA-H02::ISlacZ/hah) exhibits no labeling. (Scale bars, 5 μm.) FRAP experiments on P. aeruginosa PAO1 labeled with pyoS2NTD-AF488 (B) and pyoS2NTD-AF488 (C) are shown, where cells were first treated with 100 μM CCCP. The bleached region is highlighted (dashed circle). The absence of FRAP in these experiments suggests pyoS2NTD remains bound to FpvAI in the OM, whereas FRAP suggests pyoS2NTD-AF488 has translocated to the periplasm, where it can diffuse laterally. (Scale bars, 1 μm.)

PyoS2NTD Translocation Through FpvAI Mirrors That of a TBDT Ligand.

We developed a cross-linking strategy to map the pyoS2 import route that capitalized on its PMF dependence and involved trapping pyoS2NTD during translocation. The PMF delivers ∼20 pN mechanical force (31), and so we fused pyoS2NTD to GFP (pyoS2NTD-GFP), which can resist up to 100 pN (32). In contrast to pyoS2NTD-AF488, FRAP experiments with pyoS2NTD-GFP bound to P. aeruginosa cells did not show fluorescence recovery, consistent with a block in translocation (Fig. 3). We substituted 13 pyoS2NTD residues, distributed throughout the pyoS2NTD-GFP construct, for the UV-inducible cross-linker para-Benzoylphenylalanine (pBpa) ( and Fig. 3). The interfacial site Tyr46pBpa was included as a positive control. All pBpa variants were purified and first complexed with FpvAI in vitro. Of the 13 variants, cross-linking was only observed for Tyr46pBpa, its cross-link to FpvAI Met431 at the protein–protein interface confirmed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) (Fig. S6). After ensuring pBpa labeling did not interfere with pyoS2 cytotoxicity (Fig. S4), we developed a method for cross-linking and purifying pyoS2NTD-GFP translocation intermediates from live P. aeruginosa cells and identified cross-linking sites by LC-MS/MS (). In vivo cross-links were observed for all 13 pBpa variants, suggesting that translocation intermediates were indeed trapped by this strategy (Fig. 3). We were able to map cross-links for seven pBpa sites (Gln17, Ile23, Ala29, Lys70, Ala87, Gln135, and Gln184), all of which were to transmembrane regions of FpvAI that are inaccessible to pyoS2NTD in its ground state complex with FpvAI (Fig. 1). PyoS2NTD must translocate significant distances to satisfy the cross-linking data, as in the case of pyoS2NTD Gln184, which moves 76 Å to meet FpvAI Val197 (Fig. 3 ). We identified six principal cross-link sites in FpvAI, although LC-MS/MS data suggested residues to either side of these sites are also hit, but at lower frequency (Figs. S7 and S8). Five of six residues are within the plug domain that occludes the channel, and one was to the barrel wall (Met766) (Fig. 3). Four pBpa sites (Gln17, Ile23, Ala87, and Gln135) cross-link to the same FpvAI plug domain residue, Met177, suggesting much of the pyoS2NTD polypeptide chain translocates past this residue.

Fig. 3.

Mapping the pyoS2NTD translocation pathway by photo cross-linking. (A) FRAP experiment on a P. aeruginosa PAO1 cell labeled with pyoS2NTD-GFP demonstrates stalled import. The bleached region is highlighted (dashed circle). (Scale bar, 1 μm.) (B) Structure of the pyoS2NTD–FpvAI complex showing the sites of pBpa incorporation in pyoS2NTD-GFP (yellow diamonds) and the major cross-link sites in FpvAI (yellow circles), most of which are located in the plug domain of the TBDT. The red dotted line indicates the cross-link detected between pBpa184 in pyoS2NTD and Val197 in FpvAI, which requires this region of the pyocin to translocate at least 76 Å. (C) Ten percent SDS/PAGE gel showing Coomassie-stained, in vivo cross-linked adducts of translocated pyoS2NTD-GFPpBpa variants following purification by nickel affinity chromatography from OM extracts of P. aeruginosa PAO1 cells. Only cross-links to FpvAI were observed by LC-MS/MS, confirming that GFP traps the translocating pyoS2NTD within FpvAI (details are provided in ). (D) Fragmentation spectrum of pBpa184-Val197 cross-linked peptides (X = pBpa) with b- and y-ions indicated, along with the E-value, precursor ion mass (MH+), and precursor ion charge.

We have interpreted our cross-linking data in the context of recent atomic force microscopy (AFM) studies on other TBDTs, the vitamin B12 transporter BtuB, and the ferric hydroxamate receptor FhuA from E. coli (Fig. 4). Hickman et al. (33) have shown that the plug domains of BtuB and FhuA are composed of force-resistant and force-labile subdomains. Displacement of the force-labile subdomain by engagement of TonB with the ligand-bound transporter opens an ∼13-Å-wide channel. Structural alignment of the plug domains of BtuB (34) and FpvAI identifies an equivalent force-labile subdomain (Fig. S9). PyoS2NTD translocation therefore involves at least three steps (Fig. 4). Step 1 is binding of pyoS2NTD to FpvAI, where Fe-Pvd mimicry induces recruitment of TonB1 in the periplasm. Step 2 is PMF-driven unfolding of the FpvAI force-labile subdomain, release of the N-terminal β-hairpin, and delivery of the pyocin’s TonB1 box through the channel, where it engages either the same or another copy of TonB1. The pBpa substitutions at pyoS2NTD residues Gln17, Ile23, and Ala29 all cross-link to sites within the channel, either to the nonlabile subdomain (Asp239) or to the barrel wall (Met766). Displacement of the β-hairpin from the helical bundle of pyoS2NTD destabilizes the domain (Fig. S1), expediting import. Step 3 is TonB1-dependent unfolding of pyoS2NTD. The majority of observed cross-links are for this final state in which five pyoS2NTD residues spanning >100 amino acids become cross-linked to three hyperreactive FpvAI labile subdomain residues (Pro166, Met177, and Leu182). PyoS2NTD residues Gln17 and Ile23 are among the sites that cross-link to FpvAI Met177, which, along with their cross-links to Met776, explains why these residues give two cross-linked adducts on SDS/PAGE (Fig. 3). Finally, a single cross-link was observed between pyoS2NTD Gln184pBpa and FpvAI Val197, likely denoting the entrapment by GFP of the translocating pyocin within the FpvAI channel.

Fig. 4.

Model for pyoS2NTD translocation through FpvAI. The three-step model of pyoS2NTD translocation through FpvAI supported by pBpa cross-linking data (Fig. 3 and Figs. S6–S9) is shown. In step 1, binding of pyoS2NTD to FpvAI mimics Fe-Pvd, activating the receptor for substrate transport and recruiting the C-terminal domain of TonB1 in the periplasm. In step 2, a PMF-dependent mechanical force, applied via the ExbB–ExbD–TonB1 complex in the IM (not shown), drives unfolding of the labile half of the plug domain. The N terminus of pyoS2NTD enters the ∼13-Å-wide cavity that is created, allowing it to present its own TonB1 box in the periplasm. In step 3, the pyoS2NTD TonB1 box is bound by another copy of TonB1; at this time, the mechanical force is used to drive translocation of pyoS2NTD through the FpvAI lumen. Further translocation is blocked by the force-resistant GFP. The multiple cross-links observed between the labile portion of the FpvAI plug domain and translocated pyoS2NTD residues are presumed to involve unfolded polypeptide chains in the periplasm.

Discussion

Our data show that PMF-coupled TonB1 imports pyoS2NTD by dragging it through a narrow channel in FpvAI created by TonB1-induced displacement of the labile plug subdomain, a pathway ordinarily used by its ligand Fe-Pvd. TBDT plug displacement is thought to be facilitated by waters at the plug–barrel interface, which could lubricate pyoS2 translocation (35). Two mechanisms seem reasonable as to how the remaining ∼500 amino acids of pyoS2 translocate through FpvAI; either the PMF continues to be involved or refolding of pyoS2NTD in the periplasm provides the driving force. Force-dependent unfolding of the bacteriocin during import disrupts the high-affinity complexes of the pyocin with FpvAI (Kd ≃ pM) and the C-terminal nuclease with its immunity protein (Kd ≃ fM) (36). Immunity proteins protect nuclease bacteriocin-producing cells but are displaced at the cell surface during import (37). AFM studies have shown that force-dependent remodeling of a colicin nuclease accelerates the immunity protein dissociation rate (31). PyoS2 unfolding during import could provide this remodeling force. Ton-dependent bacteriocins are widely distributed in bacteria, delivering different cytotoxic activities (nucleases, peptidoglycan hydrolases, and ionophores) (12). The subversion of FpvAI by pyoS2 demonstrates a mechanism that is likely to be the basis of import for many Ton-dependent bacteriocins and reveals how TBDTs moonlight as protein transporters.

The protein import mechanism we have uncovered in bacteria has similarities to the classical protein import pathway of mitochondria catalyzed by the translocase at outer membrane complex (38). Both systems rely on an N-terminal sequence motif, the presequence in mitochondria and the TonB box in bacteria, for translocation across the OM. These sequences direct protein import through β-barrel transporters, driven by protein complexes in the IM coupled to the PMF. Unlike mitochondria, however, where the presequence is recognized by a receptor within the translocase, in bacteria, the imported proteins themselves are recognized by specific receptors that also act as translocases.

Materials and Methods

Protein Expression and Purification.

His-tagged pyoS2NTD and derivatives were expressed in BL21 (DE3) cells and were purified by nickel-affinity chromatography and size-exclusion chromatography (SEC). The TonB1 periplasmic domain (residues 109–342) was purified using the same method with the addition of tobacco etch virus-protease removal of the His-tag. FpvAI was expressed in E. coli TNE012 cells (ompA, ompB, and tsx−) (39) transformed with pNGH183 carrying the fpvAI gene from P. aeruginosa PAO1 with an E. coli ompF signal sequence. After isolation of the OM fraction, FpvAI was purified by anion exchange chromatography and SEC. Details are described in .

Limited Trypsin Proteolysis and Peptide Mass Fingerprinting.

A total of 2 mg of full-length pyoS2 (purified using the same method as described for the pyoS2NTD) in a 1:1 molar complex with FpvAI was digested with 2 μg of Sequencing Grade Modified Trypsin (Promega) at room temperature overnight in 25 mM Tris⋅HCl (pH 8.0), 150 mM NaCl, and 1% (wt/vol) n-octyl-β-d-glucopyranoside (β-OG). Digestion was stopped with 1 mM PMSF, and fragments were copurified with FpvAI on a Superdex 200 10/300 GL column (GE Healthcare). Fragments were resolved by SDS/PAGE, and bands were excised for in-gel trypsin digestion and peptide mass fingerprinting as described in .

Analytical SEC.

PyoS2NTD at 5 μM was added to a 5 μM stoichiometric complex of FpvAI and Fe-Pvd (Sigma) in 25 mM Tris⋅HCl (pH 8.0), 150 mM NaCl, and 1% (wt/vol) β-OG, and was incubated at room temperature for 6 h. The mixture was separated on a Superdex 200 10/300 GL column equilibrated in the same reaction buffer monitoring for protein elution at A280 and for Fe-Pvd elution at A400.

Crystallization and Structure Determination.

The FpvAI–pyoS2NTD complex at 8 mg/mL, purified as described in , was crystallized by the sitting-drop vapor diffusion method in 96-well MRC two-drop plates (SWISSCI) at 18 °C. Drops consisted of 100 nL of protein and 100 nL of crystallization solution dispensed using a Mosquito robot (TTP Labtech). Crystals of the complex were grown in 0.35 M ammonium sulfate, 13.5% (wt/vol) PEG3350, and 0.05 M sodium acetate (pH 4.0), and were harvested after 30 d. Crystals were cryoprotected in the crystallization solution supplemented with 1% (vol/vol) n-octyl-polyoxyethylene and 25% (vol/vol) ethylene glycol before flash-cooling into liquid nitrogen. X-ray data were collected at the European Synchrotron Radiation Facility on beamline ID23-2 from a single cryocooled crystal (100 K) using a 225-mm MarMOSAIC CCD detector. Data collection, processing, refinement, and model building are described in . The atomic coordinates and structure factors for the pyoS2NTD–FpvAI complex [Protein Data Bank (PDB) ID code 5ODW] have been deposited in the PDB.

Fluorescence Microscopy.

Alexa Fluor 488 was conjugated onto the C terminus of pyoS2NTD as described in . P. aeruginosa PAO1 cells were grown in M9-glucose media (6.78 g/L Na2HPO4, 3 g/L KH2PO4, 0.5 g/L NaCl, 10 mM d-glucose, 1 mg/mL NH4Cl, 2 mM MgSO4) at 37 °C, and were labeled with 1 μM fluorophore-conjugated pyoS2 construct for 15 min (details are provided in ). FRAP experiments were performed using a PerkinElmer spinning disk confocal microscope with a 100× oil-immersion objective (1.4 N.A.). Images were acquired using the 488-nm laser at 10% power, and bleaching was performed at 50% laser power at maximum speed. Recovery images were acquired over a time course up to 2 min. Bright-field images were recorded for each FRAP experiment.

ITC.

For FpvAI-pyoS2NTD–binding experiments, proteins were prepared in 50 mM potassium phosphate (pH 7.0) and 1% (wt/vol) β-OG. Experiments were performed using a MicroCal iTC200 thermostat at 25 °C. The cell contained FpvAI at 3.5 μM, and the syringe contained pyoS2NTD at 70 μM. Competition ITC experiments were performed with 10 μM FpvAI and 15 µM Δ1–45 pyoS2NTD in the cell and with 70 μM pyoS2NTD in the syringe. Because of the small ΔH, TonB1-pyoS2NTD–binding experiments were performed using a MicroCal PEAQ-ITC isothermal titration calorimeter. Proteins were prepared in 50 mM potassium phosphate (pH 7.0) and 150 mM NaCl. The cell contained TonB1 at 15 μM, and the syringe contained either pyoS2NTD or Δ1–30 pyoS2NTD at 150 μM. The first injection of 0.5 μL was followed by 19 injections of 2 μL, with each injection spaced by 180 s. Binding isotherms were fitted using the manufacturer’s software.

Growth Inhibition Assays.

PyoS2 cytotoxic activity was assayed by plate-based growth inhibition assays. Typically, a 10-mL culture of P. aeruginosa was grown at 37 °C to an OD600 of 0.6. Lawns were prepared by addition of 200 μL of culture to 5 mL of molten soft LB-agar [0.75% (wt/vol) agar in LB] at 42 °C and were poured over LB-agar plates. Once set and dry, 2 μL of serially diluted wild-type or variant pyoS2 was applied to the lawn. Lawns were allowed to grow overnight at 37 °C, and cytotoxicity was determined by observation of clearance zones.

LC-MS/MS Identification of in Vivo Photoactivated Cross-Links.

A 500-mL culture of P. aeruginosa PAO1 was grown for 24 h at 30 °C in iron-free succinate medium (6.1 g/L K2HPO4, 3 g/L KH2PO4, 4.1 g/L succinic acid) supplemented with 7.5 mM (NH4)2SO4, 1 mM MgSO4, and 1× trace metal solution [prepared as PTM4(Fe−) as described elsewhere (40)]. Cells were washed and resuspended in 35 mL of fresh medium, incubated for 60 min in the presence of 100 nM pyoS2NTD-GFPpBpa variants, and UV-irradiated at 365 nm for 30 min in a CL-1000 UV cross-linker. Cells were then resuspended in 10 mM Tris⋅HCl (pH 8.0), 0.25% (wt/vol) lithium diiodosalicyclic acid, and 2% (vol/vol) Triton X-100, and were lysed through sonication. The cleared lysate was ultracentrifuged at 200,000 × g for 45 min to isolate the OM fraction, which was solubilized in 10 mM Tris⋅HCl (pH 8.0), 5 mM EDTA, and 2% (wt/vol) β-OG. Cross-linked complexes were purified using EDTA-resistant cOmplete His-Tag Purification Resin (Roche) and were resolved by SDS/PAGE. Cross-linking sites were mapped by LC-MS/MS and pLink software as described in .