IroT/MavN Is a Legionella Transmembrane Fe(II) Transporter: Metal Selectivity and Translocation Kinetics Revealed by in Vitro Real-Time Transport

Abstract

In intravacuolar pathogens, iron is essential for growth and virulence. In Legionella pneumophila, a putative transmembrane protein inserted on the surface of the host pathogen-containing vacuole, IroT/MavN, facilitates intravacuolar iron acquisition from the host by an unknown mechanism, bypassing the problem of Fe(III) insolubility and mobilization. We developed a platform for purification and reconstitution of IroT in artificial lipid bilayer vesicles (proteoliposomes). By encapsulating the fluorescent reporter probe Fluozin-3, we reveal, by real-time metal transport assays, that IroT is a high-affinity iron transporter selective for Fe(II) over other essential transition metals. Mutational analysis reveals important residues in the transmembrane helices, soluble domains, and loops important for substrate recognition and translocation. The work establishes the substrate transport properties in a novel transporter family important for iron acquisition at the host−pathogen intravacuolar interface and provides chemical tools for a comparative investigation of the translocation properties in other iron transporter families.

Iron is an essential transition metal in all living organisms.^1,2 Because of its reactivity, speciation, and insolubility in an oxidative environment (as ferric iron), organisms evolved a sophisticated network of iron-binding proteins and molecules to guarantee iron solubilization, cellular uptake, transfer, and storage.^1,2 These ensure iron bioavailability for enzyme maturation without reaching toxic intracellular levels.^1,2

Legionella pneumophila is the causative agent of Legionnaires’ disease. In nature, L. pneumophila infects amoebae, while in humans, it invades and replicates inside macrophages.³ Survival within macrophages and protozoa requires the formation of a specialized intracellular compartment known as the Legionella-containing vacuole (LCV).³ Nutrient availability in the vacuole is limited, and in intracellular pathogens, including L. pneumophila, iron acquisition is critical for virulence and survival.^4,5 This requires unique molecular processes to allow iron sequestration from the endogenous host cell cytosolic pool hijacking host iron speciation.^4,5

Legionella utilizes the type IV−B Dot/Icm secretion system to inject effector proteins into the host cells and the vacuolar membrane at the host−pathogen interface.^3,6–8 These effectors increase the bacteria’s ability to survive by mediating nutrient transport from the host cytosol through the host-derived vacuolar membrane and by hijacking host vesicle trafficking pathways.^9–11 During intracellular growth, pathogens must either release and reduce ferric iron from host proteins or utilize the soluble and reactive ferrous iron from the regulated cytosolic host labile iron pool (LIP), which maintains bioavailable Fe(II) at low micromolar concentrations.^11–14 The molecular mechanisms utilized to bypass host iron restriction, allowing acquisition and translocation across the host vacuolar membrane, remain poorly understood.

IroT/MavN was identified as a Legionella-derived protein that is inserted in the host-derived vacuolar membrane, enabling iron acquisition into the vacuole^4,5 for subsequent import in Legionella via endogenous transporters (e.g., FeoB).¹⁵ In agreement, IroT is found exclusively in obligate bacterial intracellular parasites (Legionella or Rickettsiella).¹⁶

IroT is a virulence factor.^4,5 ΔIroT mutant cells exhibit iron-starvation before growth is arrested during the early stages of infection.^4,5 The model of a bacterial protein inserting into host membranes to mediate iron translocation provides a paradigm for how intravacuolar pathogens use virulence-associated secretion systems to manipulate and acquire host iron. IroT/MavN has been identified as a metal transporter,^4,17 but the molecular mechanism by which IroT mediates iron acquisition remains elusive.

To investigate the molecular function and transport properties of L. pneumophila IroT (LpIroT), we have developed a workflow for its expression, purification, and reconstitution in artificial lipid bilayers (proteoliposomes). A codon-optimized LpIroT sequence (Uniprot: Q5ZRR5) was utilized for expression in E. coli. Upon membrane isolation, a detergent screening identified detergents for LpIroT extraction and purification in monodisperse form. Fos-choline-14 was selected for solubilization followed by detergent exchange to 7-cyclohexyl-1-heptyl-β-d-maltoside (Cymal-7). LpIroT was purified by immobilized metal affinity chromatography (IMAC) and size exclusion chromatography (SEC) to >95−98% purity as determined by SDS-PAGE (Figure S1A,B). SEC revealed significantly monodisperse LpIroT-Cymal-7 micelles (with a minor shoulder) and no aggregation. Oligomeric state characterization of the LpIroT-Cymal-7 micelles was performed by SEC coupled to multiangle light scattering (SEC-MALS, Figure 1A). SEC-MALS analysis resulted in a protein-detergent complex of 212.1 ± 4.9 kDa, with calculated protein and modifier (lipid and detergent) masses of 85.3 ± 2.0 kDa (theoretical monomer MW = 83.0 kDa) and 126.7 ± 5.7 kDa, respectively. Thus, LpIroT is monomeric in Cymal-7 micelles.

Figure 1.

LpIroT micelles and SUVs. (A) SEC-MALS analysis for LpIroT purified in Cymal-7 micelles. The dashed gray line shows theoretical molar mass. (B) Dynamic light scattering (DLS) size distribution analysis revealing monodisperse control and LpIroT small unilamellar vesicles (SUVs)

To investigate the metal transport properties, we designed a platform to study real-time substrate transport events across membranes, via LpIroT reconstitution into artificial lipid bilayers (proteoliposomes), followed by encapsulation of metal-dependent fluorescent probes in the vesicle lumen. Purified LpIroT in Cymal-7 was reconstituted in unilamellar liposomes via freeze−thaw and extrusion through 200 nm filters, followed by liposome destabilization by detergent addition and removal by Biobeads. Protein incorporation was quantified by SDS-PAGE upon liposome separation by ultracentrifugation, followed by LpIroT quantification in the soluble and proteoliposome fractions. The results revealed >90% incorporation in proteoliposomes (Figure S1C). The control and LpIroT containing vesicles were characterized by dynamic light scattering revealing monodisperse size distribution and average diameters of 155 ± 55 and 189 ± 78 nm, respectively, corresponding to protein-free and protein-embedded Small Unilamellar Vesicles (SUVs) (Figure 1B). On the basis of the IroT proteoliposome average size and protein-to-lipid ratios (1:25 w/w), we estimated the number of IroT molecules per SUV at ~100 molecules/vesicle.

The fluorescent probe PhenGreen SK has been previously utilized to characterize Fe(II) and metal transporters in proteoliposomes.^18–20 However, initial experiments on LpIroT liposomes revealed limiting properties in terms of photo-stability, drifting fluorescence baselines, and sensitivity to trace metal contamination (also in Chelex-treated buffers). In search of alternative turn-on probes (rather than quenching), we identified Fluozin-3 to study LpIroT-mediated metal transport in SUVs. Despite Fluozin-3 being developed as a highly sensitive Zn²⁺ probe,²¹ we tested its turn-on capabilities with Fe(II) and other first/second-row transition metals. Upon mixing equimolar concentrations of Fe(II), Co(II), Ni(II), Zn(II), and Cd(II) (10 μM) with Fluozin-3, fluorescence analysis revealed a metal-dependent turn-on emission response with identical maxima (515 nm), whose position was not affected by the presence of liposomes (Figure 2A and Figure S2A). In agreement to high-affinity metal binding, metal titration of Fluozin-3 (10 μM) revealed linear responses with increasing M(II) concentrations with breakpoints at approximately 1:1 molar ratios, corresponding, to a 335%, 970%, 2000%, 25700%, and 5700% fluorescence increase at saturation over the metal-free probe for the respective metals in absence of lipids (Figure S2B), and 55%, 190%, 340%, 7800%, and 1070% increase in the presence of liposomes (Figure 2B). In light of the role of Mn²⁺ as part of the oxidative burst utilized in the host immune response, we also investigated the Fluozin-3 turn-on response toward Mn²⁺. Addition of Mn²⁺ to Fluozin-3 (Mn²⁺ = 10 μM) resulted in a 130% and 90% fluorescence turn-on response in absence and presence of lipids, respectively. To note, no turn-on saturation was obtained at 1:1 molar ratio in Mn²⁺ titration experiments, suggestive of a lower Mn²⁺ affinity for Fluozin-3 compared to other metals tested. Thus, Fluozin-3 could be overall used to monitor transporter-mediated real-time metal transport across lipid bilayers when encapsulated in proteoliposomes.

Figure 2.

LpIroT-mediated Fe(II) transport in proteoliposomes. (A) Fluozin-3 fluorescence emission turn-on response with equimolar Mn(II), Fe(II), Co(II), Ni(II), Zn(II), and Cd(II) (10 μM; λ_exc = 480 nm) in the presence of 200 nm liposomes (12.5 mg/mL). (B) Fluozin-3 (10 μM) fluorescence response upon metal titration (λ_exc = 480 nm; λ_em = 515 nm) reported as (F−F₀)/F₀ (F₀ is the fluorescence in the absence of metals), in the presence of 200 nm liposomes. (C) Representative fluorescence traces for Fe(II) real-time transport in LpIroT proteoliposomes with Fluozin-3 encapsulated in the lumen as a function of increasing Fe(II) concentrations ((1–40 μM); λ_exc = 480 nm; λ_em = 515 nm). Differential fluorescence at time t (F_t − F₀) was normalized to the fluorescence before Fe(II) addition and corrected for the signals in control liposomes. (D) Fe(II) transport rates in LpIroT proteoliposomes and corresponding fit with a Michaelis−Menten equation: (δF/δt) = (δF/δt)_Max × [Fe(II)]/(K_M+ [Fe(II)]). Inset: Michaelis−Menten Fe(II) transport analysis obtained by fitting the maximum F change (ΔF/F₀) as a function of Fe(II) concentrations (n = 3 ± standard deviation).

To test the LpIroT capability of catalyzing transmembrane translocation of Fe(II) across lipid bilayers, Fluozin-3 was encapsulated in control and LpIroT SUV lumen by freeze−thaw membrane fracture. Control and LpIroT proteoliposomes were exposed to buffered oxygen-free solutions containing Fe(II) (FeSO₄, prepared in an anaerobic atmosphere) and metal fluxes monitored by fluorescence in response to rapid mixing with the proteoliposomes at the emission maximum of Fluozin-3-Fe(II) (515 nm). Kinetic traces in LpIroT revealed a time-dependent development of fluorescent signal, which reached saturation in 8–10 min indicative of real-time detection of IroT-mediated Fe(II) translocation across the proteoliposome lipid bilayer. Recordings of ΔF/F₀ rose exponentially upon rapid exposures of proteoliposomes to extravesicular Fe(II) with increasing concentrations from 0 to 40 μM (Figure 2C). The net Fe(II) influx in LpIroT proteoliposomes (ΔF/F₀ as a function of time, with F₀ the background fluorescence before Fe(II) addition) was estimated by subtracting the background Fe(II) leakage in control liposomes exposed to identical Fe(II) concentrations. The background ion permeation in controls was negligible. These results indicated the successful functional reconstitution of purified LpIroT in proteoliposomes. We performed exponential fitting of the Fe(II) transport kinetic traces and determined initial transport rates (δF/δt) as a function of Fe(II) concentrations. The hyperbolic dependency is suggestive of saturation-dependent carrier-mediated catalytic Fe(II) transport. The curve was fitted with a Michaelis−Menten equation resulting in K_M,Fe(II) = 1.63 ± 0.24 μM (Figure 2D). For comparative analysis with mutants (see below), the Fe(II) translocation was also analyzed as fluorescent change at equilibrium corrected for the fluorescence at t = 0. The ΔF/F₀ values at equilibrium plotted as a function of increasing Fe(II) concentration revealed a similar hyperbolic profile resulting in similar apparent K_M,Fe(II) = 5.99 ± 1.70 μM and (ΔF/F₀)_MAX = 0.25 ± 0.04, in agreement with carrier-mediated Fe(II) translocation (Figure 2D, inset). The overall apparent Fe(II) binding affinity was estimated by Fe(II) titration competition with MagFura-2 to be K_D,Fe(II) = 2.9 ± 0.3 μM, in the same range as the K_M,Fe(II) (Figure S3). From the Fe(II) transport traces at the highest Fe(II) concentration (40 μM), we estimated the Fe(II) translocation velocity. By utilizing the (ΔF/F₀) Fluozin-3 turn-on calibration response at known Fe(II) concentrations in the presence of lipids, and by determining LpIroT concentrations in the Fluozin-3 encapsulated proteoliposomes by SDS-PAGE, we estimated the maximal initial Fe(II) transport at v ≃ 120 nmol Fe(II)/mg⁻¹ s⁻¹.

The determined maximal rates, corresponding to approximately 10 ion/sec per IroT molecule and the hyperbolic dependency of initial rates as a function of substrate concentration^22,23 suggests that IroT utilizes a transport mechanism involving conformational changes (transporters translocation rates, 10⁰ to 10⁴ ions/sec; channels/channel-like transporters, 10⁶ to 10⁸ ion/sec).²⁴

Metal selectivity was investigated by monitoring real-time metal fluxes by fluorescence upon mixing a series of first/second-row essential and toxic metal solutions with LpIroT containing SUVs and control liposomes. The LpIroT transport was determined by subtracting background ion leakage. In light of the different turn-on response of Fluozin-3 toward the metals tested, real-time ΔF/F₀ transport values at equilibrium were corrected with respective response factors calculated from Fluozin-3 titration experiments in the presence of liposomes, providing the relative response of Fluozin-3 toward the different metals (Figure 2A,B). The LpIroT-mediated ΔF/F₀ rose exponentially upon rapid exposures of proteoliposomes exclusively to extravesicular Fe(II), with Co(II), Ni(II), and Zn(II) only eliciting a marginal response (<5%) (Figure 3A,B). Overall, the results indicate that LpIroT is a high-affinity and selective Fe(II) transporter with very limited substrate promiscuity. Control experiments with Fe(III)-citrate as substrate showed no elicited Fluozin-3 response, confirming that iron is transported as a ferrous ion (Figure S4). The experiments were conducted at M(II) concentrations close to the K_M determined for Fe(II). In eukaryotic cells’ cytosol, the labile Fe(II) pool buffers exchangeable Fe(II) at low μM levels.^11,13 Since free Zn(II) (buffered at pM concentrations) and Co(II)/Ni(II) are present at significantly lower levels,¹ we conclude that Fe(II) is the principal physiological substrate translocated across the LCV membrane.

Figure 3.

LpIroT selectivity. (A) Fluorescence real-time transport traces in LpIroT proteoliposomes with encapsulated Fluozin-3 for Mn(II), Fe(II), Co(II), Ni(II), Zn(II), Cd(II) (5 μM). Signals were normalized for the relative fluorescence turn-on response (Figure 2A,B). (B) Metal selectivity determined by maximum fluorescence intensity change (ΔF/F₀) (n = 3 ± standard deviation).

The Legionella containing vacuole features a lumenal acidic environment (pH ~ 6.1), which could be potentially exploited by IroT to drive Fe(II) transport translocation from the host cytosol to the vacuole lumen.²⁵ We have investigated whether proton counter-transport could be coupled to Fe(II) translocation. By encapsulating the fluorescent pH indicator pyranine in the SUV lumen and by triggering Fe²⁺ translocation in IroT proteoliposomes, we observed Fe²⁺-dependent proton counter-flux (resulting in an increased pH detected by pyranine in the lumen). These results suggest that IroT could potentially act as a Fe(II)/H⁺ antiporter (or secondary active transporter) in the LCV membranes (Figure S5).

LpIroT topology prediction (TOPCONS)²⁶ reveals 8 TM helices and a long C-terminal domain (Gln534-Lys683). The topology with N-and C-termini facing the host cytoplasm is supported by in vivo cysteine accessibility investigation.¹⁷ The topology prediction also indicates possible insertion of the N-terminal 20 amino acids present in some IroT homologues (TM0) (Figure 4A). To identify classes of important residues for substrate recognition and transport and to establish the transmembrane Fe(II) translocation pathway we generated a series of mutants in residues potentially involved in Fe(II) coordination, based on relative sequence conservation in IroT homologues or important for in vivo replication,^4,17 and investigated their transport properties (Figure S6). Analysis of Fe(II) binding sites in the PDB indicates that Cys, Glu, Asp, and His are, by far, the most frequent aa residues involved in Fe(II) coordination followed distantly by Met, Gln, Asn, and Tyr, which can provide supporting coordinating ligands.^1,27 On the basis of the topology model, we identified a series of these residues in the TM helices located either: (i) in the proximity of the host cytosol-membrane interface; (ii) deeply embedded in the lipid bilayer; (iii) located at the membrane vacuolelumen interface (Figure 4A). These residues could, therefore, be important coordinating residues in the transmembrane translocation pathway for Fe(II) uptake, binding in the transmembrane core, and release on the opposite side of the bilayer, respectively. We generated a series of combination Ala mutants on residues belonging to each of those classes (Table S1) and reconstituted them in proteoliposomes to investigate their effect on Fe(II) transport (Figure 4B). SDS-PAGE quantification (Figure S7) revealed identical incorporation, while SEC confirmed protein monodispersity as in wtLpIroT. A number of potential Fe(II) coordinating residues are located at the interface between the TM helices and the host cytosol (M24, Y38, Y39, Y81, E82, E84, H214, Y268, M270, N367, Q369, D531, C532) where Fe(II) uptake should take place. Glu, Asp, His and Cys are the most frequent core residues in protein Fe(II) coordination sites. Thus, we have generated a mutant in which E82, E84, H214 (located on adjacent TM2 and TM3 helices and constituting a potential Fe(II) uptake site), and D531 and C532 (on TM8) have been converted to Ala (mutant M1). In addition, the E82A_E84A_H214A mutant (M2) was also generated and investigated. Analysis of Fe(II) transport properties in M1 and M2 proteoliposomes revealed that combined mutations of all the residues (M1) significantly affected the apparent K_M,Fe(II) determined by ΔF/F₀ at equilibrium analysis (K_M,Fe(II) > 25 μM), despite maintaining a V_MAX similar to wtLpIroT (Figure 4B,C and Table S2). As Fe(II) concentrations above 50 μM resulted in leakage in control proteoliposomes, a more accurate K_M,Fe(II) could not be determined. M2 (in which only E82, E84, and H214 are mutated) revealed similar transport properties, suggesting that these three residues are critical for high-affinity Fe(II) uptake, while C532 and D531 might play only a supporting coordination role. Additional mutation of other potential coordinating Met, Tyr, Asn and Gln residues at the TM-host interface (M3) resulted in a decrease in Fe(II) translocation (approximately 30% of wtLpIroT at 40 μM Fe(II)) indicating that one or more of these low-frequency Fe(II) residues might support Fe(II) uptake into the transmembrane domain together with E82, E84 and H214, thus constituting a Fe(II) transmembrane entry pathway (Figure 4B,C and Table S2). We subsequently investigated residues that could be important for intramembrane Fe(II) translocation. Four potential coordinating residues (M227, M260, N355, Y524) embedded at the center of the lipid bilayer were identified. Mutation of these residues (M4) resulted in a reduced Fe(II) transport affinity with an approximately 4-fold change in the apparent K_M,Fe(II) (K_M,Fe(II) = 23.0 ± 14.8 μM) compared to wtLpIroT (Figure 4B–C and Table S2). The hyperbolic transport saturation profile and the reduced K_M,Fe(II) values for M4 are consistent with a transport mechanism that involves conformational changes.

Figure 4.

LpIroT topology and mutants. (A) LpIroT topology model generated with TOPCONS and plotted with Protter (http://wlab.ethz.ch/protter). Mutated residues in this work are highlighted in color as in the legend. (B) Comparative Michaelis−Menten-type Fe(II) transport analysis for IroT mutants in transmembrane residues (mutants M1-M6), by fitting the maximum fluorescence intensity change (ΔF/F₀) as a function of Fe(II) concentrations. Least-squares fitting resulted in the apparent KM,Fe(II) and (ΔF/F₀)_MAX reported in Table S1 (n = 3 ± standard deviation). (C) Relative (ΔF/F₀) Fe(II) transport for M1-M6 in comparison to wtLpIroT (Fe(II) = 40 μM). (D) Fe(II) transport activity as a function of increasing Fe(II) concentrations for the mutants in the Glu/His-rich motif in Loop 7 (mutants M7-M8) and (E) the mutant lacking the C-terminal soluble domain (Δ_541−683-LpIroT).

We then identified potential coordination residues located at the TM-LCV interface, which could be important for Fe(II) release on the vacuole lumen. We generated a mutant in which only high-frequency Fe(II) coordinating residues were mutated (mutant M6, E237A_C324A_H435A) or combined those with mutation of lower frequency Fe(II) coordinating residues (mutant M5). Fe(II) transport analysis in protoeliposomes revealed that the mutation of E237, C324, and H435 (in TM3, TM5, and TM7) was sufficient to abolish Fe(II) transport resulting in a nonfunctional protein (Figure 4B,C and Table S2), thus constituting the putative Fe(II) exit pathway.

Besides transmembrane residues for Fe(II) translocation, sequence analysis reveals a number of Glu-rich motifs, which are required for iron transport activity in numerous iron transporters.^4,28 These are located in soluble loops 2, 6, and 7 and in the C-term soluble domain, with loop 7 and the C-term domain containing 3 and 4 of these motifs, respectively. When each of those were mutated, only mutation of the motif centered on E462 on loop 7 (E439 in original sequence lacking 23 aa, numbering refers to the LpIroT of this work) failed to complement the intracellular growth defect in ΔIroT cells indicating the importance of this motif, but not the others, in iron uptake in vivo.⁴ Loop 7 is also histidine-rich. In vivo analysis revealed the importance of 3 histidines (His468, His473, and His475) in the loop for in vivo activity. We generated E462A, E462A_H468A_H473A_H475A mutants and investigated the Fe(II) transport properties. Our analysis (Figure 4D; Figure S8 confirms incorporation as wtLpIroT) revealed similar transport kinetics as wtLpIroT for these mutants with similar apparent K_M,Fe(II) and V_MAX. Thus, these residues are not essential for Fe(II) transmembrane translocation but important for in vivo activity, possibly by mediating protein−protein interactions or substrate transfer to unrecognized Fe(II) acceptor(s) in the LCV. In light of the presence of four Glu-rich motifs, we also generated a deletion mutant lacking the C-term domain. Fe(II) transport kinetic analysis revealed residual Fe(II) translocation below wtLpIroT K_M,Fe(II) values but sustaining only minimal Fe(II) transport at higher Fe(II) concentrations (Figure 4E). Thus, it appears that the C-terminal domain is critical to guarantee high-capacity translocation. We speculate that this domain, possibly through its Glu-rich motifs, is involved in Fe(II) sequestration from the host cytosolic LIP and delivery to the transmembrane domain. In agreement with a Fe(II) binding role, the deletion of the C-terminal domain resulted in a reduction of overall Fe(II) affinity determined by competition experiments with MagFura-2 compared to wtLpIroT (K_D,Fe(II) = 5.5 ± 0.5 μM). The multiple putative Fe(II) binding motifs potentially playing redundant Fe(II) accepting functions could explain why single motif mutations did not result in intracellular growth defect in LpIroT mutants. Thus, the C-terminal soluble domain is essential for high-capacity Fe(II) translocation.

Overall, by developing a fluorescent in vitro platform for functional reconstitution in lipid bilayers, we established that IroT is a new family of Fe(II) transporters located in the LCV host vacuolar membrane. Together with the mutational analysis, we propose a molecular mechanism responsible for Fe(II) translocation. A number of vacuolar pathogens physically interact with the host endosomal machinery (e.g., Mycobacterium, Salmonella, and Leishmania spp.) and secrete siderophores and/or reductases to allow acquisition of ferric iron from the host glycoproteins.^29–31 Consequently, they do not require specialized transporters in the vacuolar membrane to acquire iron. Contrarily, in L. pneumophila, the LCV does not interact with host endosomes³² (rich in transferrin), requiring a different transport system that can access ferrous iron form the LIP. We demonstrate that IroT is a direct high-affinity and high-capacity Fe(II) transporter capable of mediating Fe(II) translocation, thus allowing Fe(II) acquisition in the vacuole lumen. The LpIroT metal selectivity, order of affinity, and transport kinetics match the cytosolic values for bioavailable iron in the low micromolar range from LIP. Our platform will allow comparative studies toward the characterization of other iron transporter families possessing different transport mechanisms.

ABBREVIATIONS

Cymal-7

7-cyclohexyl-1-heptyl-β-d-maltoside

Icm/Dot

intracellular multiplication/defect in organelle trafficking

LCV

Legionella-containing vacuole

LIP

labile iron pool

MavN

more regions allowing vacuolar colocalization N