Iron acquisition in Pseudomonas aeruginosa by the siderophore pyoverdine: an intricate interacting network including periplasmic and membrane proteins

Pyoverdine (PVDI) has been reported to act both as a siderophore for scavenging iron (a key nutrient) and a signaling molecule for the expression of virulence factors. This compound is itself part of a core set of virulence factors produced by Pseudomonas aeruginosa during infections. Once secreted into the bacterial environment and having scavenged ferric iron, PVDI-Fe3+ is taken back into the P. aeruginosa periplasm via the outer membrane transporters FpvAI and FpvB. Iron release from PVDI in the bacterial periplasm involves numerous proteins encoded by the fpvGHJKCDEF genes and a mechanism of iron reduction. Here, we investigated the global interacting network between these various proteins using systematic bacterial two-hybrid screening. We deciphered a network of five interacting proteins composed of two inner-membrane proteins, FpvG (iron reductase) and FpvH (unknown function), and three periplasmic proteins, FpvJ (unknown function), FpvF (periplasmic PVDI-binding protein), and FpvC (iron periplasmic-binding protein). This interacting network strongly suggests the existence of a large protein machinery composed of these five proteins, all playing a role in iron acquisition by PVDI. Furthermore, we discovered an interaction between the periplasmic siderophore binding protein FpvF and the PvdRT-OpmQ efflux pump, also suggesting a role for FpvF in apo-PVDI recycling and secretion after iron delivery. These results highlight a multi-protein complex that drives iron release from PVDI in the periplasm of P. aeruginosa.


Interaction between FpvG and FpvH and formation of an inner-membrane complex.
Previous studies have demonstrated that expression of the FpvH, FpvK, and FpvJ proteins is required for optimal reductase activity of FpvG 28 . Moreover, genes encoding the FpvG, FpvH, FpvJ, and FpvK proteins are organized in an operon 31 and it is well recognized that adjacent genes tend to encode interacting proteins 32 . FpvG, FpvH, and FpvK have been predicted to be inner-membrane proteins and FpvJ periplasmic, because of a signal peptide 28 . We deciphered the interacting network between the membrane proteins FpvG, FpvH, and FpvK by performing systematic BACTH screening in E. coli, which is based on the reconstitution of adenylate-cyclase activity. The full-length fpvG, fpvH, and fpvK genes were fused to the T25/T18 domains of adenylate cyclase in the two-hybrid vectors. Screening of the possible protein-protein interactions between FpvG, FpvH and FpvK on indicator plates containing X-gal highlighted an interaction between FpvG and FpvH (Fig. 2), whereas no interaction could be observed for FpvK. FpvG interacted with itself, suggesting at least dimerization of this protein (Fig. 2). As FpvG was already been demonstrated to be an inner-membrane protein 28 , we investigated the subcellular localization of FpvH. Cell fractionation experiments showed that FpvH is also an inner-membrane protein (Fig. S1A in Supplemental Material).
We expressed FpvG and FpvH proteins with C-terminal His 6 and Strep-tag sequences, respectively, in E. coli to validate their interaction. Bacterial membranes were solubilized in detergent and the complex purified by Strep-trap affinity followed by size-exclusion chromatography (Fig. 3A). The presence of both FpvG and FpvH proteins in the elution peak was confirmed by Coomassie-blue staining and immunoblot analysis using specific anti-His 6 and anti-Strep antibodies (Fig. 3B,C). Isolation of the FpvG-FpvH complex confirmed the BACTH results and revealed the ability of FpvG reductase and FpvH to interact and form an inner-membrane complex.

Figure 2.
Interacting network with the membrane proteins FpvG, FpvH and FpvK. DHM1 cells producing the protein of interest fused to the T18 or T25 domain of adenylate cyclase were spotted on indicator plates containing X-gal for BATCH screening. Each protein was tested with an empty vector and the RetS protein, which is not related to the PVDI pathway. RetS is able to form dimers and serves as a positive control 37 . The blue color indicates an interaction between the two proteins of interest. The experiment was repeated three times with each time 10 colonies as described in Materials and Methods. A representative image is shown.  (Table S3 in Supplemental Material). Entire SDS-PAGE and blots are shown in Fig. S2 in Supplementary Information.
Interaction between the inner-membrane proteins FpvG and FpvH with the periplasmic proteins FpvJ, FpvC, and FpvF. Finally, we also tested the interactions between the inner-membrane and periplasmic proteins. BACTH analysis showed that both FpvG and FpvH membrane proteins interact with the three periplasmic proteins FpvJ, FpvF, and FpvC (Fig. 4). As with the membrane proteins, none of the periplasmic proteins interacted with FpvK in this two-hybrid approach (Fig. 4). Overall, BACTH screening revealed the existence of an interaction network between FpvG, FpvH, FpvJ, FpvF, and FpvC.
We next attempted to isolate all five proteins by pulldown experiments using an anti-Flag resin. The periplasmic fraction of E. coli overproducing FpvJ His6 , FpvC HA , and FpvF Flag was incubated with solubilized membranes containing FpvG His6 and FpvH Strep , and the mixture incubated with an anti-Flag resin. FpvG His6 , FpvJ His6 , and FpvC HA co-precipitated with FpvF Flag (Fig. 6). Equivalent results were obtained when fractions were incubated with PVDI-Fe ( Fig. S5 in Supplemental Material). None of the non-Flag-tagged proteins were retained on the anti-Flag resin when incubated alone ( Fig. S6 in Supplemental Material). We were unable to detect FpvH Strep either due to immunodetection problems or because this protein is not present in the complex. Overall, these results confirm the existence of at least a four-protein complex, linking the inner-membrane FpvG protein and three periplasmic components of the PVDI pathway. In addition, this complex could be isolated in the presence or absence of PVDI-Fe.

FpvF interacts with the membrane protein PvdT of the PvdRTOpmQ efflux pump.
Previous studies have shown that FpvF can form dimers that bind apo-PVDI 25 and that PVDI recycling is altered in a ∆fpvF mutant 28 . Based on these observations, it seemed possible that FpvF is involved in PVDI recycling by interacting with proteins of the efflux pump PvdRT-OpmQ. Indeed, our BACTH analysis revealed an interaction between FpvF and the membrane protein PvdT (Fig. 7).

Discussion
One of the major particularities of the PVDI-dependent iron acquisition pathway in P. aeruginosa, and probably conserved among fluorescent Pseudomonads, is that this siderophore delivers iron into the bacterial periplasm, with siderophore-free iron then being transported further by an ABC transporter into the cytoplasm. This mechanism is completely different from that described previously for other siderophore-dependent iron-uptake pathways, such as the enterobactin and ferrichrome pathways in E. coli, two archetypes in the field of bacterial iron homeostasis, which deliver iron directly into the bacterial cytoplasm 34 . After the uptake of PVDI-Fe 3+ across the outer membrane by FpvAI or FpvB 21-23 , iron release from PVDI in the bacterial periplasm requires the FpvGHJKCDEF proteins 28 . Moreover, the molecular mechanism involved implies both iron reduction by FpvG reductase to decrease the affinity of PVDI for the metal and an iron chelator, FpvC 26-28 .
We used a systematic BACTH assay to unravel the interactions between these proteins and highlight specific interactions between FpvG reductase, the inner-membrane protein FpvH, and the three periplasmic proteins FpvJ, FpvC, and FpvF (Fig. 8). Although our BACTH screening was carried out using both N-and C-terminal T18/T25 tags, we observed no interactions with FpvK, suggesting that either (i) FpvK does not interact with the three other proteins, (ii) the interactions are transient or of weak affinity, (iii) the interaction requires a third protein partner, or (iv) the interaction is just not detectable by BACTH. Indeed, fusion to the T18 or T25 domains may affect the folding of the protein or prevent interactions 35 .
We biochemically confirmed the interaction between the two inner-membrane proteins FpvG-FpvH by affinity and size-exclusion chromatography, but we still know nothing about the stoichiometry of the FpvG-FpvH complex, except that FpvG is able to form dimers based on the BACTH data. Previously, the in vivo kinetics of iron dissociation from PVDI showed that FpvG activity is dependent on FpvH expression 28 . FpvJ and FpvK expression also affect FpvG activity, but clearly to a lower extent 28 . The ability of FpvK to affect FpvG reductase activity, like FpvH and FpvJ, highly suggests that it also interacts with the other inner-membrane proteins.
BACTH screening also showed an interacting network between the three periplasmic proteins FpvC, FpvF, and FpvJ. This complex was validated by pulldown experiments. FpvC and FpvF are two periplasmic-binding proteins associated with the ABC transporter FpvDE. Purified FpvC was shown to chelate ferrous iron in an in vitro PVDI-Fe dissociation assay using DTT as the iron reducer 28 . Mass spectrometry approaches under native conditions have shown that FpvF can bind PVDI, and that FpvC and FpvF are both able to form the tetrameric  Interacting network with periplasmic (FpvC, FpvF and FpvJ) and membrane proteins (FpvG, FpvH and FpvK). Bacterial two-hybrid assays for proteins were quantified by measuring the β-galactosidase activity, as described in Materials and Methods. Zip, which is not related to the PVDI pathway, served as a positive control 38 . ND: not determined because for FpvF we were unable to obtain the pKT25-FpvF vector. The experiment was repeated three times independently. Error bars represent the standard errors of the means.
The function of FpvJ is currently unknown, but this protein may allow interaction of the FpvC-FpvF-PVDI-Fe complex with the FpvG-FpvH inner-membrane complex to achieve iron reduction and the transfer of ferrous iron from PVDI to FpvC. BACTH screening showed that all three periplasmic proteins FpvJ, FpvC, and FpvF can interact with the two-protein FpvG-FpvH complex in the absence or presence of PVD-Fe, forming an inner-membrane machinery. We were able to isolate four of the five proteins by pulldown assay, confirming the existence of a complex between the inner-membrane reductase, FpvG, and the three periplasmic proteins, FpvF, FpvC, and FpvJ. Immunodetection of FpvH Strep with anti-Strep antibodies revealed non-specific bands of various molecular weights, preventing us from assessing the presence of FpvH in the pulldown assay (data not shown). The exact stoichiometry of this complex is still unknown.
Moreover, FpvDE is the putative ABC transporter that allows the translocation of ferrous iron across the inner membrane into the cytoplasm, and its deletion affects iron acquisition by PVDI 25,28 . We also evaluated the possible interactions of FpvDE with the periplasmic FpvC and FpvF and the membrane proteins FpvG, FpvH, and FpvK, but were unable to detect any interaction (data not shown). However, FpvC probably plays a role in bringing ferrous iron to the permease, FpvE, but further biochemical studies will be necessary to demonstrate this.
Finally, previous studies have demonstrated that FpvF dimers can bind apo-PVDI and apo-PVDI recycling is partially abolished in a ∆fpvF mutant 25,28 . We thus investigated whether FpvF can interact with the PvdRT-OpmQ efflux pump involved in PVDI recycling. We found that FpvF interacts with PvdT, the inner-membrane protein of the efflux pump. This result strongly supports the hypothesis that FpvF or FpvF 2 binds apo-PVDI in a FpvF-PVDI or FpvF 2 -PVDI 2 complex and brings the apo-siderophore to PvdT for its recycling into the extracellular medium.  www.nature.com/scientificreports www.nature.com/scientificreports/ In conclusion, this study provides new insights about the possible interacting network of the various proteins involved in iron release from PVDI in the periplasm of P. aeruginosa (Fig. 8). These interactions have been highlighted using a two-hybrid approach and confirmed in vitro using purified or pulldown experiments; but  www.nature.com/scientificreports www.nature.com/scientificreports/ they existence still need to be confirmed in P. aeruginosa cells. This complex interacting network strongly suggests a multi-protein complex at the inner membrane, allowing iron to be removed from PVDI. Consequently, the molecular mechanisms for iron acquisition via PVDI involves the following steps (Fig. 8D). As detailed in the introduction, PVDI-Fe 3+ is imported across the outer membrane into the bacterial periplasm by FpvAI and FpvB. In the periplasm, PVDI forms a FpvC-FpvF-PVDI-Fe complex with the two periplasmic-binding proteins. As FpvJ was found to interact with both periplasmic and membrane proteins, it may help in the interaction of this periplasmic complex with the inner-membrane FpvG-FpvH complex. Iron reduction by FpvG decreases the affinity of PVDI for the metal and a transfer of iron to the periplasmic-binding protein FpvC, which likely brings iron to the FpvDE ABC transporter. Apo-PVDI is most likely bound to FpvF, which is able to interact with PvdT, allowing the recycling of apo-PVDI to the extracellular medium by the efflux pump PvdRTOpmQ. Moreover, this FpvGHJCF complex is the first example to be described of a complex between an inner-membrane reductase and two periplasmic-binding proteins associated with an ABC transporter.
Deciphering and understanding the protein-protein interacting network is an important piece in the understanding of the PVDI-Fe 3+ uptake pathway puzzle. Our work will undoubtedly initiate a number of future directions like chemical crosslinking experiments in P. aeruginosa cells with tagged proteins to assess the existence of this interacting network in the pathogen. Electron microscopy approaches are planned to obtain structural information on these different protein complexes. At last, further studies are also needed to understand the exact role of FpvH, FpvJ, and FpvK within the complexes and PVDI-Fe dissociation.

Materials and Methods
Chemicals, bacterial strains and growth medium. Medium culture Lysogeny Broth (LB) and LB-agar were purchased from Difco. Detergent n-dodecyl-ß-D-maltoside (DDM) was purchased from Anatrace, N-Lauroylsarcosine sodium (SLS) and Tween 20 from Sigma. The strains used in this study are listed in the Supplementary Table S1. Briefly, TOP10 and DH5α strains were used for cloning procedures, TOP10 and BL21 strains for protein production and DHM1 strains for bacterial two-hybrid assays. E. coli strains were routinely grown in LB medium at 37 °C and on LB-agar for solid culture. Plasmids were maintained by the addition of antibiotics such as ampicillin (100 µg/ml), kanamycin (50 µg/ml), chloramphenicol (50 µg/ml) and streptomycin (100 µg/ml). plasmid construction. Plasmids used in this study are listed in Supplementary Table S1. All the PCRs were performed with DNA Phusion high-fidelity polymerase (Thermofischer Scientific). DNA sequences from Pseudomonas aeruginosa PAO1 were taken from Pseudomonas Genome DataBank (www.pseudomonas.com). Oligonucleotides were purchased from Sigma and are listed in Supplementary Table S2. All primers used introduced restriction sites.
All constructs were screened with colony PCR and plasmids were purified with the Macherey Nagel Nucleospin Plasmid kit in accordance with the manufacturer's instructions. All constructions were verified by DNA sequencing (Eurofins).
Bacterial two-hybrid assay. For plate-BACTH assay, two compatible vectors producing proteins fused to T18 or T25 domain were co-transformed into DHM1 cells that were incubated at 30 °C for 16 h. Ten independent colonies of each transformation were inoculated together into 2 ml of LB medium supplemented with ampicillin, kanamycin and 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG, Sigma) and incubated at 30 °C for 16 h. 5 µl of each culture was spotted onto LB-agar plate supplemented with appropriate antibiotics, 0.5 mM IPTG and 40 µg/ml 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal, Sigma). The plate was incubated for 16 h at 30 °C.
For liquid medium assay, two compatible vectors producing proteins fused to T18 or T25 domain were co-transformed into DHM1 cells that were incubated at 37 °C for 16 h. Ten independent colonies of each transformation were inoculated into 2 ml of LB-medium supplemented with appropriate antibiotics and were Scientific RepoRtS | (2020) 10:120 | https://doi.org/10.1038/s41598-019-56913-x www.nature.com/scientificreports www.nature.com/scientificreports/ incubated at 37 °C during 24 h. The next day, 20 µl of each culture were inoculated in 2 ml of LB supplemented with appropriate antibiotics and 0.5 mM IPTG and incubated at 37 °C for 16 h. 100 µl of each culture was used for the ß-galactosidase assay using Miller Protocol 36 .
ß-galactosidase dosage. 100 µl of bacterial culture were added to 900 µl of Z Buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 ,10 mM KCl, 1 mM MgSO 4 , pH 7.0, 0.2% β-mercaptoethanol). 1 µl of 0.1% sodium dodecylsulfate and 50 µl of chloroform were added to the suspension that was mixed vigorously for 10 seconds. The suspension was then incubated for 5 min at 28 °C. 200 µl of 4 mg/ml 2-nitrophenyl ß-D-galactopyranoside (ONPG, Sigma) were added to the cells. Reaction was stopped by adding 500 µl of 1 M Na 2 CO 3 . The suspension was centrifuged at 14,000 g for 3 min and the optical density of the supernatant was read at 420 and 550 nm. The ß-galactosidase activity was then calculated in Miller Unit (MU) according to the following equation:  . 200 µg/ml of lysozyme were added to the suspension, incubated for 1 h at 4 °C and the cells were centrifuged at 6,700 g for 10 min at 4 °C to remove unbroken cells and insoluble fraction (like insoluble proteins). The supernatant corresponding to the periplasm was ultracentrifuged at 100,000 g for 40 min at 4 °C. Spheroplasts were washed three times with Tris-Sucrose buffer and re-suspended into chill water and treated with benzonase (Sigma, 250U/µl). After incubation for 1 h at 37 °C, suspension was centrifuged for 40 min at 100,000 g at 4 °C to collect the cytoplasm (supernatant). The pellet, corresponding to the total membranes, was re-suspended in 50 mM Tris-HCl pH 8.0.
Membrane isolation. E. coli strains overproducing proteins of interest were pelleted, re-suspended in 50 mM Tris-HCl pH 8.0 and lysed by sonication. Unbroken cells were removed by centrifugation at 12,000 g for 15 min. Supernatant was centrifuged at 100,000 g for 40 min. The membranes (pellet) were solubilized in 50 mM Tris pH 8.0, 100 mM NaCl, 0.1% SLS for 16 h at 4 °C and ultracentrifuged at 100,000 g for 40 min at 4 °C. The pellet corresponds to the outer membranes and the supernatant to the inner membranes.
pulldown experiments. Periplasmic proteins. BL21 (DE3) cells were transformed with pCDF-FpvJ His6 -FpvF Flag and pRSF-FpvC HA or with pCDF-FpvF Flag or pRSF-FpvC HA only. Overnight culture was inoculated into LB medium with appropriate antibiotics and grown at 37 °C until OD 600 reached 0.6. Then protein production was induced by adding 0.4 mM IPTG at 22 °C for 16 h. Cells were re-suspended in buffer A (50 mM Tris pH 8.0, 250 µM EDTA, 20% sucrose), subjected to cellular fractionation and the periplasmic fraction was recovered. 100 µl of the periplasmic fraction were mixed with 50 µl agarose beads charged with nickel (Sigma) and incubated on a rotating wheel for 1 h at room temperature. The mixture was then centrifuged 2 min at 2,000 rpm