Vibrio cholerae FeoB hydrolyzes ATP and GTP in vitro in the absence of stimulatory factors

Abstract

Feo is the most widely conserved system for ferrous iron transport in prokaryotes, and it is important for virulence in some pathogens. However, its mechanism of iron transport is not fully understood. In this study, we used full-length Vibrio cholerae FeoB (VcFeoB) as a model system to study whether its enzymatic activity is affected by regulatory factors commonly associated with FeoB proteins from other species or with G-proteins that have homology to FeoB. VcFeoB showed a higher rate of hydrolysis of both ATP and GTP than its N-terminal domain alone; likewise, ions such as K⁺ and Fe²⁺ did not modulate its nucleotide hydrolysis. We also showed that the three V. cholerae Feo proteins (FeoA, FeoB, and FeoC) work in a 1:1:1 molar ratio in vivo. Although both FeoA and FeoC are required for Feo-mediated iron transport, neither of these proteins affected the VcFeoB NTPase rate. These results are consistent with an active transport mechanism independent of stimulatory factors and highlight the importance of using full-length FeoB proteins as a reliable proxy to study Feo-mediated iron transport in vitro.

Graphical Abstract

V. cholerae FeoA, FeoB, and FeoC work at a 1:1:1 molar ratio to mediate ferrous iron uptake through a mechanism driven by FeoB NTP hydrolysis without requiring stimulatory factors.

Introduction

Iron is essential for most living organisms; this element is a co-factor of enzymes involved in critical metabolic pathways, such as respiration and synthesis of nucleic acids.¹ Iron is available in biological systems as ferric iron (Fe³⁺), which is mainly found in insoluble hydroxide complexes or associated with organic molecules; and ferrous iron (Fe²⁺), which is water-soluble in anaerobic and acidic conditions. For pathogens, iron acquisition presents a challenge as they must acquire this element directly from their host. To obtain iron, bacterial pathogens have evolved specialized mechanisms, such as secretion of iron chelators and expression of specific membrane transporters.^1,2 Because of its physicochemical properties, Fe²⁺ is the prevalent form of iron in the human gut; thus, it comprises an important source of iron for gastrointestinal pathogens and those found in reducing environments. Indeed, ferrous iron acquisition is a determining factor for virulence in several bacterial pathogens, including Salmonella enterica,³ Helicobacter pylori,⁴ and Legionella pneumophila.⁵

Feo is the most widespread prokaryotic system exclusively dedicated to ferrous iron uptake.^6-8 This system is encoded in an operon made up of two genes, feoA and feoB, although a few species (mainly belonging to the phylum Gammaproteobacteria) have an additional gene, feoC. FeoA and FeoC are both small (about 8.5 kDa) cytoplasmic proteins, while FeoB is an 83 kDa membrane protein with cytoplasmic terminal domains. The Feo system works as a multimeric transmembrane complex embedded in the inner membrane, and it is involved in the transport of free ferrous iron from the periplasm to the interior of the cell. Feo likely comprises a trimer of trimers of FeoB associated with FeoA and FeoC in unknown ratios.⁹

FeoB is considered the major structural component of the Feo complex; this protein likely forms a trimeric cysteine-gated pore through which iron is transported.¹⁰ FeoB is the only Feo component with known enzymatic activity: the cytoplasmic N-terminus domain of FeoB (referred to as NFeoB) has G-protein homology and hydrolyzes GTP.⁶ Initially, it was proposed that GTP hydrolysis served as the energy source of the complex, and thus, Feo was an active iron importer.⁸ However, the low GTP hydrolysis rate of NFeoB observed in vitro was not consistent with an active transporter, and it was suggested that nucleotide hydrolysis might have a function other than providing the energy for transport.^11-13 Some studies indicate that FeoB may require stimulatory factors, such as metal ions; or that FeoA/FeoC may serve as regulatory proteins that modulate the activity of FeoB.^6,8 Because most of these studies used NFeoB instead of the full-length protein, it was possible that the low GTP hydrolysis rate was a consequence of missing the large transmembrane domain of the protein.¹⁴

Feo-mediated iron uptake is a promising target for novel antibacterial strategies as it is a conserved system in many bacterial pathogens.⁶ A full understanding of the working mechanism of this transporter is then needed. In addition, a deeper understanding of Feo would contribute to our knowledge of the biology G-proteins in bacteria, which is in its infancy.

In this work, we used the FeoB protein from the human pathogen Vibrio cholerae as an in vitro model system to study its catalytic activity and how it may be modulated by stimulatory factors. V. cholerae, which has an absolute requirement for iron, is the etiological agent of cholera, a deadly diarrheal condition historically associated with devastating pandemics.² Previously, we have shown that V. cholerae FeoB (VcFeoB) interacts with FeoA and FeoC in vivo to form a large, transmembrane complex.⁹ Also, we have demonstrated that VcNFeoB, unlike many NFeoB proteins, hydrolyzes both GTP and ATP, and either of these nucleotides is sufficient to support iron uptake¹⁵ (Fig. 1). Nonetheless, we still do not know how the nucleotide hydrolysis leads to iron uptake and how the three proteins work together to orchestrate ferrous iron uptake.

Fig. 1.

Current mechanistic understanding of the V. cholerae Feo system. The transmembrane protein FeoB forms the channel for ferrous iron transport into the bacterial cell. In association with the small cytoplasmic proteins, FeoA and FeoC, FeoB monomers assemble into a large complex, likely comprised of a trimer of trimers of FeoB, embedded in the inner membrane.⁹ Because the exact arrangements of the monomers and the interactions sites have not been defined, a single copy of each is shown. NFeoB exhibits NTPase activity, which is also essential for iron acquisition. FeoA and FeoB are sufficient to assemble the complex, but FeoC is required for function.^9,16 FeoC and FeoA interact with FeoB, and these interactions are critical for iron uptake.^9,16,17

In this study, we characterized the NTPase activity of full-length VcFeoB in terms of its ATP and GTP hydrolysis rate in vitro and how NTPase activity is affected by different potential regulatory factors. We demonstrate that the intrinsic hydrolytic activity of VcFeoB may be consistent with an active transport mechanism, and that neither iron nor potassium affect the activity. We further show that the stoichiometric ratio of the V. cholerae Feo complex is 1:1:1, and that FeoA and FeoC do not modify the NTPase activity of VcFeoB.

Material and Methods

Reagents, Bacterial Strains and Growth Conditions

All chemicals and growth media were obtained from Sigma-Aldrich Chemical Company. Bacterial strain stocks were preserved in 20 % v/v glycerol in tryptic soy broth (TSB) at −80 °C. Unless otherwise stated, E. coli and V. cholerae strains were routinely grown in Luria-Bertani (LB) broth (10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl in double-distilled water, ddH₂O) or on LB agar (1.5 % w/v bacteriological agar) at 37 °C, and 200 rpm for liquid media.

Bacterial strains used in this study are shown in Table S1 (ESI). Heme supplementation consisted of 10 μM heme. For strains harboring a plasmid, antibiotics were used at the following concentrations: For E. coli 50 μg/mL ampicillin, and for V. cholerae, 25 μg/mL ampicillin, and 12.5 μg/mL kanamycin.

Plasmid Construction

All plasmids and primers used in this study are listed in Table S2 (ESI). pET-VcFeoB was constructed by extracting feoB from pFeo101, using the primers FeoB-NdeI-F and FeoB-HindIII-R and inserting the amplified product into pET21a(+), between the NdeI and HindIII restriction sites. To generate pVcFeo(ABC), the stop codon between feoA and feoB, and between feoB and feoC in pFeo101 were respectively deleted through PCR-mediated deletion 18 using the primers fusFeoAB-F, fusFeoAB-R, and fusFeoBC-Fand fusFeoBC-R maintaining the coding sequences in a single ORF. For protein production, Vcfeo(ABC) was transferred from pVcFeo(ABC) to pET21a(+), between the BamHI and XhoI sites, using FeoA_BamHI-F and FeoC-XhoI-R to generate pETVcFeo(ABC).

Purification of FeoB and Feo(ABC)

Full-length VcFeoB and VcFeo(ABC) were purified from inside-out vesicles (IOV) as described by Seyedmohammad and colleagues for the purification of Pseudomonas aeruginosa Feo,¹⁹ with some modifications. Protein synthesis was optimized for the V. cholerae constructs, and we observed the highest protein yield by using Terrific Broth and incubating for 4 h at room temperature after induction with IPTG (Fig. S1A, ESI). Specifically, overnight cultures of E. coli BL21(DE3), in LB broth, harboring the corresponding pET21a(+) derivative with either His-tagged VcfeoB or Vcfeo(ABC) were used to inoculate, in a 1:100 ratio, 1L of Terrific Broth (TB)²⁰ supplemented with ampicillin. TB is made up of 24 g/L yeast extract, 20 g/L tryptone, 4 mL/L glycerol, and 100 mL/L phosphate buffer (0.17 M KH₂PO₄ and 0.72 M K₂HPO₄). The cells were grown at 37 °C and 200 rpm until they reached an OD₆₅₀ of 0.4 – 0.6, and then chilled on ice for 15 min. Expression of cloned genes was induced by adding IPTG to a final concentration of 0.5 mM and incubating the cells for 4 h at 25 °C and 150 rpm. Then, cells were harvested by centrifugation at 6,000 x g at 4 °C for 10 min and resuspended in 20 mL lysis buffer (100 mM K-HEPES pH 7.0, 10 μg/mL DNAse, and 1 Pierce™ Protease Inhibitor Mini Tablet from Thermo Fisher). Henceforth, all the steps were carried out at 4 °C. Cells were lysed by three passages through an Aminco (Thermo) FRENCH^® Press FA-078 at 10,000 psi. Cell debris was removed by centrifuging the lysate at 13,000 x g for 10 min, and the supernatant was subsequently centrifuged at 150,000 x g for 40 min in a TL-100 Beckman Ultracentrifuge equipped with a Type 70.1 Ti Rotor. The pellets, consisting of IOVs, were resuspended in approximately 10 mL 50 mM K-HEPES pH 7.0 + 10 % v/v glycerol and stored at −80 °C until further processing.

For protein purification, IOVs were solubilized by adding 1 volume of solubilization buffer (20 mM K-HEPES, 20 % v/v glycerol, 500 mM NaCl, 2 % w/v DDM, 10 mM imidazole, and 10 mM MgSO₄, pH 7.4) and shaking during 1 h. The solubilized IOV solution was ultra-centrifugated at 150,000 x g for 40 min, and the supernatant was applied to a chromatography column charged with 5 mL of Ni-NTA resin (Qiagen) and previously stabilized with 20 volumes of ddH₂O and 5 volumes of wash buffer (10 mM K-HEPES, 10 % v/v glycerol, 500 mM NaCl, 0.05 % w/v DDM, 10 mM imidazole, and 10 mM MgSO₄, pH 8.0). Protein was allowed to bind the resin by shaking for 1 h, and then the resin was washed with 50 volumes of wash buffer. The protein was eluted with 3 volumes of elution buffer (10 mM K-HEPES, 10 % v/v glycerol, 200 mM NaCl, 0.05 % w/v DDM, 500 mM imidazole, and 10 mM MgSO₄, pH 8.0) that were collected in approximately 15 fractions, discarding the first 0.5 volumes. The fractions containing the protein, as determined by Bradford assay, were mixed and concentrated in elution buffer with no imidazole through an Amicon^® Ultra-15 Centrifugal Filter Unit (Millipore-Sigma) with a 100 kDa cutoff to retain the DDM micelles. Purified protein was aliquoted and stored with 50 % v/v glycerol at −80 °C for further use, or at −20 °C for immediate use (no later than 2 days). From 1L of culture, we obtained 0.40 mg of VcFeoB at a final concentration of 0.51 mg/mL, with high purity as determined by Coomassie staining in SDS-PAGE gel, and immunoblot analysis showed a single band at the expected molecular weight (Fig. S1, ESI). Likewise, for VcFeo(ABC), we obtained 0.18 mg of protein at a final concentration of 0.31 mg/mL verified through SDS-PAGE with Coomassie staining and immunoblot analysis (Fig. S2, ESI).

Purification of FeoA and FeoC

VcFeoA and VcFeoC were purified following a previously standardized protocol¹⁵ with a few modifications: 1L of TB supplemented with ampicillin was inoculated in a 1:100 ratio with an overnight culture of E. coli BL21(DE3) cells transformed with the respective pET16b(+) derivative harboring VcfeoA or VcfeoC (Table S1, ESI), and incubated at 37 °C and 200 rpm until OD₆₅₀ = 0.4 – 0.6. Then, gene expression was induced by adding IPTG to a final concentration of 0.5 mM and incubated 4 h at 25 °C and 150 rpm. Cells were harvested by centrifugation at 7,500 x g for 5 min at 4 °C and pellets were stored at −80 °C until further processing. For protein purification, cell pellets were resuspended in 2 mL of lysis buffer (25 mM Tris-Cl pH 8.0, 100 mM NaCl, and 1 tablet / 10 mL of Pierce™ Protease Inhibitor Mini Tablet from Thermo Fisher Scientific). For VcFeoA, the lysis buffer contained additionally 1 % w/v Na₂CO₃ and 1.0 M urea. The resuspended cells were lysed by sonication and, since VcFeoA and VcFeoC are found primarily in the insoluble fraction, pellets were washed with 1 M urea followed by denaturation with 8 M urea, then refolded by dialysis against 4 L of 25 mM Tris-Cl (pH 8.0), 100 mM NaCl, 10 % v/v glycerol, 0.1 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride (PMSF) pH 8.0 at 4 °C overnight. Cell debris was removed from the solubilized His-tagged protein by centrifugation at 30,000 x g for 30 min at 4 °C, and the supernatant was loaded on 5 mL of Ni-NTA resin (Qiagen) previously charged with Ni²⁺ and equilibrated with 20 volumes of wash buffer (lysis buffer with 20 mM imidazole). Elution was carried out with 3 volumes of elution buffer (25 mM Tris-Cl pH 8.0, 100 mM NaCl, 10 % v/v glycerol, and 500 mM imidazole) distributed in approximately 15 fractions. The fractions containing the protein, as determined by Bradford assay, were mixed and concentrated in elution buffer with no imidazole through an Amicon^® Ultra-15 Centrifugal Filter Unit (Millipore-Sigma) with a 10 kDa cutoff. Purified protein was aliquoted and stored with 50 % v/v glycerol at −80 °C for further use, or at −20 °C for immediate use (no later than 2 days).

Removal of the His-Tag Epitope

The His-tag epitope was cleaved from purified FeoA, FeoB, and FeoC proteins using TEV protease (NEB) and removed through a Ni²⁺ HiTrap column (Sigma-Aldrich). See Fig. S3 in ESI. Both procedures were performed as instructed by the manufacturers.

SDS-PAGE and Immunoblotting

Protein samples from cell cultures or protein purifications were analyzed through 4 % – 7.5 % gradient SDS-PAGE gels.

To evaluate the yield and purity of the isolated proteins, resolved proteins were tank-transferred from the polyacrylamide gel to an Immobilon^®-P PVDF membrane (Merck Millipore). Then, His-tagged proteins were detected using mouse monoclonal anti-6x-His (Thermo Fisher Scientific) and visualized using horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (Bio-Rad). To ensure even levels of loading, total protein content in the immunoblotted samples was assessed by Coomassie staining using R-250 Brilliant Blue (Bio-Rad).

Assessment of NTPase Activity

The malachite green-based colorimetric assay was used to quantify the concentration of inorganic phosphate released from the hydrolysis of ATP/GTP by FeoB or Feo(ABC) as previously described.¹⁵ All the assays were carried out in clear, flat bottom 96-well plates in a final reaction volume of 40 μL with 3 replicates for each condition. The corresponding enzyme at a fixed concentration (1 μM unless otherwise indicated) was mixed with various concentrations of nucleotides in reaction buffer (20 mM Tris, pH 8.0, 5.0 mM MgCl₂, and 200 mM KCl) and incubated at 37 °C for 1 h with constant agitation. Similarly, for determining the effect of potential stimulatory/inhibitory factors, the factors were added to the reaction mixture together with the enzyme and 650 μM of the nucleotide at the following concentrations: K⁺ or Na⁺ (200 mM), 2,2’-bipyridyl (100 μM), Fe²⁺ (1,000 μM), sodium ascorbate (5 mM), and equimolar ratios for the FeoA and FeoC proteins, unless otherwise stated. The reaction was quenched by adding 200 μL of the malachite green solution (1.0 mg/mL malachite green and 50 mg/mL sodium borate in 1.0 M HCl) and incubating at room temperature for 30 min. Then, A_{λ = 620 nm} was measured using a SpectraMax M3 microplate reader (Molecular Devices). For the determination of kinetic parameters, the initial rate data were fitted to a Michaelis-Menten model by a nonlinear regression using GraphPad Prism v. 5.0.0.

Growth Assay of V. cholerae EPV6

To determine the phenotypes associated to specific Feo genotypes, V. cholerae EPV6 harboring a pWKS30 derivative with the Feo construct was grown on LB agar with and without heme supplementation at 37 °C overnight.²¹ The strains under evaluation were streaked in different quadrants of the same plates together with the appropriate positive and negative controls (i.e., a functional Feo system and the empty vector, respectively). These assays were performed in triplicates, this is, three separate plates for each condition inoculated with individual colonies. Observable growth after incubation was considered a positive result.

Growth rate in liquid medium was measured as OD₆₅₀ in triplicate cultures in EZ RDM broth²² supplemented with 2 % w/v sucrose, 5 μM FeSO₄, and the appropriate antibiotics.

Inductively coupled plasma mass spectrometry (ICP-MS)

We adapted the protocol described by Peng et al.²¹ to determine the iron content in 100 μg of the purified VcFeoB protein before and after treatment with 2,2’-bipyridyl. Namely, the protein solutions were digested with 1 mL of 35 % v/v trace-metal-grade nitric acid (Fisher Scientific) in 5 mL Savillex vials at 80 °C for 3 h. Vials were then cooled and allowed to evaporate to dryness at 60 °C overnight. The residue was resuspended in ~ 3 mL of 2% v/v nitric acid to reach a total dissolved solids concentration of approximately 200 ppm, and then incubated at 60 °C overnight. Metal content was measured using an Agilent 7500ce inductively coupled plasma mass spectrometer and analyzed against defined standards at the ICP-MS laboratory in the Jackson School of Geosciences at the University of Texas at Austin. The sample of treated protein was prepared by incubating 100 μg VcFeoB with 100 μM 2,2’-bipyridyl during 1 h at RT, then the sample was filtered through a 50 kDa Amicon^® Ultra-15 Centrifugal Filter Unit (Millipore-Sigma) to remove the chelator agent, and resuspended in the original volume of milliQ water. A blank sample containing no protein was used as control.

Results

Production and purification of full-length VcFeoB

Most studies on the hydrolytic activity of FeoB have used only the catalytic N-terminus domain (NFeoB), but it is not known how reliable a proxy NFeoB is for the full-length protein. Therefore, we purified full-length VcFeoB (Fig. S1, ESI) to determine its activity in vitro.

VcFeoB hydrolyzes both ATP and GTP

VcNFeoB possesses the archetypical G-proteins motifs and, unlike NFeoB from E. coli, which only hydrolyzes GTP, VcNFeoB hydrolyzes both ATP and GTP.¹⁵ To determine whether full-length VcFeoB also has this dual NTPase activity, we assessed its hydrolytic rate towards ATP and GTP through the malachite green assay, which measures the phosphate released from nucleotide hydrolysis. We found that the full-length protein, like its N-terminus counterpart, hydrolyzes both purine nucleotides in vitro with a faster turnover rate for GTP (k_cat[GTP] > k_cat[ATP]), though with a higher specificity for ATP ([k_cat /Km] [GTP] < [k_cat/Km] [ATP]). Specifically, the k_cat values we measured for ATP and GTP hydrolysis by VcFeoB were (2.60 ± 0.39) x 10⁻² s⁻¹ and (4.48 ± 1.01) x 10⁻² s⁻¹, respectively (Fig. 2). This k_cat[GTP] value is approximately 2-fold higher than that of VcNFeoB (2.0 × 10⁻² s⁻¹).¹⁵ This rate is within the same order of magnitude of that of the full-length FeoB from Klebsiella pneumoniae (KpFeoB, 9 × 10⁻² s⁻¹), and is consistent with those k_cat[GTP] values reported for in vitro measurements of some bacterial active transporters.¹⁴ In addition, our confidence interval for VcFeoB k_cat[GTP] overlaps with those estimated for potassium-stimulated NFeoB of both Streptococcus thermophilus (StNFeoB) and E. coli (EcNFeoB) (4.3 × 10⁻² and 1.1 × 10⁻², respectively).^12,15 Altogether, these data using full-length VcFeoB confirm the dual NTPase activity of the V. cholerae Feo system and shows that the presence of the transmembrane region of VcFeoB increases its GTPase rate to values similar to those from other bacterial NFeoB proteins activated by stimulatory factors.^13,23

Fig. 2.

NTPase activity of full-length VcFeoB. VcFeoB hydrolyzes both ATP and GTP, with a higher hydrolytic rate, though lower affinity, towards GTP as determined by the malachite green assay. k_cat and Km values with 95 % CI for both substrates were calculated by fitting initial hydrolysis rates to a Michaelis-Menten model through a non-linear regression (R² = 0.81 and 0.74 for ATPase and GTPase regressions respectively).

Potassium, iron, FeoA, and FeoC do not affect VcFeoB NTPase activity in vitro

G-proteins may require stimulatory factors, such as metal ions or accessory proteins, to reach an activated state with higher hydrolytic activity. Several bacterial NFeoB proteins increase their GTPase rate in the presence of potassium; for instance, EcNFeoB and StNFeoB increase their GTP hydrolysis rate up to 20-fold in the presence of K⁺.^12,13 Similarly, Pseudomonas aeruginosa full-length FeoB (PaFeoB) k_cat[GTP] increases one order of magnitude in the presence of Fe²⁺.¹⁰ To evaluate whether the NTPase activity of VcFeoB is also bolstered by these factors, we assessed the effect of Fe²⁺, K⁺, FeoA, and FeoC on the ATP and GTP hydrolysis by VcFeoB. None of these potential stimulatory factors significantly affected the VcFeoB NTPase rate in vitro. Namely, NTPase rate did not change upon the addition of 200 mM K⁺ or Na⁺ (Fig. 3A), which is the K⁺ concentration at which EcNFeoB and StNFeoB reach their maximum activity.^12,13,15 Addition of ferrous iron in concentrations ranging from 100 to 1,000 μM did not have a significant effect on either ATPase or GTPase activity of VcFeoB (Fig. 3B). We ruled out that oxidation of Fe²⁺ to Fe³⁺ may has prevented activation of VcFeoB by measuring the effect of iron on NTPase activity in the presence of the reducing agent, sodium ascorbate. There was no increase in activity; the GTPase activity was lower under these conditions (Fig. S5, ESI). Since the stimulatory effect of Fe²⁺ on PaFeoB is mediated by a cysteine residue (C675) in the transmembrane portion¹⁰, and since this cysteine is conserved in VcFeoB (C678, Fig. S6, ESI), a possible explanation for our negative results is that our isolated VcFeoB protein is already bound to Fe²⁺ that remained from the purification process, and the k_cat values we measured corresponded to the iron-activated form of VcFeoB. To assess this possibility, we compared the basal NTPase activity of VcFeoB to that in the presence of 2,2’-bipyridyl, a ferrous iron chelator that would compete for the bound iron (Table S3, ESI). The addition of this chelating agent had no effect on the VcFeoB NTPase activity (Fig. 3C), indicating that pre-bound Fe²⁺ did not lead to higher baseline activity.

Fig. 3.

VcFeoB NTPase rate is not affected by stimulatory factors. (A) VcFeoB hydrolysis of ATP and GTP are not stimulated by either K⁺ or Na⁺. (B) Concentration of Fe²⁺ ranging from 0 to 1,000 μM do not affect the NTPase activity of VcFeoB in vitro. (C) 2,2’-bipyridyl does not affect the enzymatic activity of VcFeoB, indicating that it is not already activated by Fe²⁺ remaining from the purification process. Experiments were carried out in triplicate with 650 μM NTP at 37 °C. (D) Neither 1,000 μM Fe²⁺ (Fe) nor the cytoplasmic proteins, FeoA and FeoC in equimolar concentration, lead to a significant change in the NTPase activity of VcFeoB in vitro. 100 % (dashed line) indicates basal hydrolysis towards 650 μM NTP at 37 °C with no added factors. Error bars represent SD, and statistical significance was estimated by one-way ANOVA (A, B, and D) or two-tail t-test (C) at the levels of 0.05, 0.01, or 0.001. None of the differences were significant with these analyses.

We also tested whether a possible stimulatory effect by Fe²⁺ requires the mediation of the accessory proteins FeoA or FeoC. In this regard, the VcFeoB NTPase activity in the presence of Fe²⁺ and FeoC showed no significant changes (Fig. 3D). Notably, estimated confidence intervals for relative NTPase activity under different conditions for FeoB in the presence of FeoA ranged from 90 % basal activity (no effect) to 150 % basal activity (a modest increase), as shown in Fig. 3D. This increase is well below the reported effects of cations on other bacterial FeoBs (i.e., about one order of magnitude), but it is similar to those slight effects we observed from FeoA and FeoC towards NFeoB GTPase activity.¹⁵ To further test whether FeoA may have a stimulatory role on VcFeoB, we repeated the NTPase assessments with and without FeoA using twice the replicates (i.e., N = 6). These data indicate no significant difference in NTPase activity in the presence or absence of FeoA (Fig. S4, ESI).

Thus, our results suggest that nucleotide hydrolysis by VcFeoB is not affected by stimulatory factors typically associated with G-proteins and other FeoBs, and that neither FeoA nor FeoC play a significant role in regulating the NTPase activity of FeoB.

VcFeoB is inhibited by phosphatase inhibitors

In Ras-homologue G-proteins, such as FeoB, the G1-motif (GxxxxGKS/T), stabilizes the β- and γ-phosphates of the guanine base through hydrogen bonds donated by amidic groups in the main chain.¹³ To further establish whether full-length VcFeoB follows this mechanism as suggested by its sequence, we evaluated the effect of phosphatase inhibitors on the NTPase activity of VcFeoB. We hypothesized that if VcFeoB works in a way similar to eukaryotic G-proteins, phosphatase inhibitors interfering with GTP hydrolysis and GDP release should affect VcFeoB likewise. We assessed the VcFeoB NTPase rate in the presence of 100 μM sodium azide (NaN₃) and 100 μM sodium orthovanadate, extensively used phosphatase inhibitors. Our results show that both compounds significantly inhibited the NTP hydrolysis by VcFeoB (Fig. 4). Thus, these data suggest that FeoB might hydrolyze purine nucleotides by binding its phosphates groups through a similar mechanism of that of eukaryotic G-proteins, such as Ras.

Fig. 4.

Phosphatase inhibitors significantly affect the NTPase activity of VcFeoB. Both phosphatase inhibitors, sodium orthovanadate (OV) and sodium azide (NaN₃), inhibit NTP hydrolysis by VcFeoB. 100 % indicates basal hydrolytic activity with no added factors. Experiments were carried out in triplicate with 650 μM NTP, using distilled water instead of inhibitor solution in control reactions. Error bars represent SD, statistical significance was estimated by two-tail t-test one-way ANOVA with Dunnett’s multiple comparison test at the levels of 0.05, 0.01 (**), or 0.001(***).

FeoA, FeoB, and FeoC function in a 1:1:1 stoichiometric ratio in V. cholerae

Although there was no effect of FeoA and FeoC on VcFeoB NTPase activity when those proteins were added in a 1:1:1 molar ratio, we do not know if this the stoichiometry of the proteins in vivo. In previous studies, we found a modest stimulatory effect on VcNFeoB when FeoA and FeoC when added simultaneously at a 2-fold molar excess.¹⁵ Thus, we wanted to determine whether the 1:1:1 ratio recapitulate the intracellular conditions and support iron transport in vivo, a sensitive measure of function of FeoB.

To determine whether the 1:1:1 ratio is functional in vivo, we constructed a strain in which FeoA, FeoB and FeoC are synthesized as a single polypeptide, thus forcing the proteins to be in the 1:1:1 ratio. This was accomplished by deleting both the stop codon between feoA and feoB, and that between feoB and feoC. Then, V. cholerae EPV6 was transformed with the fusion construct. EPV6 is a Feo-deletion strain that also lacks other iron transport systems and is able to grow in LB medium only when it carries a functional ferric or ferrous iron transporter or when the medium is supplemented with heme²¹ (Table S1, ESI). The feoABC fusion construct, Feo(ABC), was able to support growth of V. cholerae EPV6 to the same extent as the plasmid carrying the native feoA, feoB, and feoC genes (pFeo101) in medium without heme supplementation (Fig. 5). The ability of the fusion protein to support iron transport and full growth of the bacteria in vivo indicates that FeoA, FeoB, and FeoC can work in a 1:1:1 molar ratio in V. cholerae; thus, our in vitro experiments on the VcFeoB NTPase are done under conditions that resemble the active complex in vivo.

Fig. 5.

VcFeo(ABC) is a functional Feo system and recapitulates the NTPase activity of VcFeoB. (A) VcFeo(ABC) supports growth of the feo-deletion strain V. cholerae EPV6 in non-heme-supplemented medium. # 1, 2, 3 are biological replicates for the construct pVcFeo(ABC). The empty vector (pWKS30) and the same vector harboring the wild-type V. cholerae Feo operon (pFeo101) were used as negative and positive controls, respectively. (B) The growth rates of V. cholerae EPV6 carrying pFeo(ABC) and pFeo101 are not significantly different in liquid medium supplemented with 5 μM Fe²⁺; while V. cholerae EPV6 carrying the empty vector has reduced growth. The y-axis is shown in log scale.

The fusion protein VcFeo(ABC) hydrolyzes NTP at a rate similar to VcFeoB

To confirm that the fusion construct Feo(ABC) retained full NTP hydrolysis activity in vitro, we purified VcFeo(ABC) produced from the expression vector pET21a(+) (Fig. S2, ESI) and measured its catalytic activity through the same protocols used for VcFeoB. As shown in Fig. 6, k_cat values estimated for VcFeo(ABC) were (3.099 ± 0.445) x 10⁻² s⁻¹ and (5.438 ± 0.476) x 10⁻² s⁻¹ for ATP and GTP, respectively; which are not significantly different from those calculated for VcFeoB (Fig. 2) using a t-test with α = 0.05. Although we cannot rule out that these in vitro assays are still missing intracellular factors important to recapitulate the full activity of the Feo system, this result is consistent with our previous findings insofar as VcFeo(ABC) is a functional Feo system comparable to the wild type, and FeoA and FeoC do not affect the enzymatic activity of FeoB.

Fig. 6.

VcFeo(ABC) hydrolyzes both ATP and GTP, with a higher hydrolytic rate, though lower affinity, towards GTP as determined by the malachite green assay. k_cat and Km values with 95 % CI for both substrates were calculated by fitting initial hydrolysis rates to a Michaelis-Menten model through a non-linear regression (R² = 0.76 and 0.99 for ATPase and GTPase regressions, respectively).

Discussion

A primary limitation of studies on FeoB has been the difficulty of isolating the full-length, catalytically active protein. The use of full-length FeoB proteins may reveal features not recapitulated by their N-termini alone. Syedmohammad and colleagues¹⁹ and Smith and Sestok¹⁴ developed protocols for purifying the proteins from P. aeurginosa and K. pneumoniae, respectively, and we were able to successfully use their approaches to isolate the full-length FeoB protein of V. cholerae.

FeoB contains a homolog of eukaryotic G-proteins⁷ in the N-terminal terminal cytoplasmic domain. G-proteins are characterized by five G-motifs (G1 – G5), involved in nucleotide-binding, magnesium-binding, and nucleotide releasing.^12,13 For these proteins, magnesium is critical; this ion coordinates the nucleotide β- and γ-phosphates inside the active site of the enzyme.²³ Moreover, several NTPases, including some G-proteins and FeoBs, require potassium to be active.¹³ In this regard, bacterial FeoBs can be classified into two groups according to their substrate specificity and their potassium-dependency.²³ EcNFeo and StNFeoB belong to the first group, which comprises potassium-dependent enzymes that hydrolyze GTP exclusively. By contrast, members of the second group, such as VcNFeoB and HpNFeoB, do not require potassium or other monovalent cations, and hydrolyze both ATP and GTP.¹⁵ Two factors seem to be determinant for nucleotide specificity and potassium dependency: The second residue of the G5 motif, which likely governs substrate specificity (serine in GTP-specific FeoBs; and alanine in promiscuous FeoBs), and an asparagine residue in G1 (N11, as firstly described in StNFeoB), which is key in potassium binding, though it is conserved in both groups (Fig. S6, ESI).^13,15,23,24 Therefore, other residues besides N11 must be involved in potassium-mediated activation, and dual NTPase activity of the enzymes belonging to the second group may be an ancestral trait that evolved towards GTP-specificity in certain clades.²³ Our experimental data on full-length VcFeoB confirms it has dual NTPase activity but no potassium-mediated activation. Interestingly, Smith and Sestok indicate that full-length KpFeoB¹⁴ is not stimulated by potassium despite possessing the serine residue in the G5-motif typical of K⁺-dependent GTP-specific FeoBs (Fig. S6, ESI). They hypothesize that K⁺ may not be critical for enzymatic activity, but it rather recapitulates contacts made between the N-terminus of FeoB and the transmembrane domain, which are missing in vitro by using NFeoB alone.¹⁴ Alternatively, our current knowledge about the functional relationship between FeoB G-motifs and nucleotide specificity and potassium binding might be incomplete, so that other factors must be considered. To discern between these scenarios and gain insight into the functional relevance of the G-motifs in FeoB, it would be necessary to compare the nucleotide specificity and K⁺-mediated activation of more full-length FeoB proteins to those from their N-termini.

FeoA and FeoB are structural elements of the Feo complex in V. cholerae, and both FeoA and FeoC interact with VcFeoB.⁹ However, the roles of FeoA and FeoC are not known. Since NFeoB has G-protein homology, numerous authors claim that FeoA may regulate FeoB GTPase activity as a GTPase activating protein (GAP) or Guanine Dissociation Factor (GEF).^6-8 In eukaryotes, G-proteins usually work together this type of accessory proteins, GAPs and GEFs, which accelerate GTP hydrolysis and GDP release, respectively.¹³ Moreover, it has been suggested that FeoC may work as a cytoplasmic iron sensor²⁵, transcription factor, inhibitor of complex assembly¹⁷, or modulator of the FtsH-mediated FeoB degradation.²⁶ Nevertheless, we did not observe any regulatory effects of FeoA and FeoC on full-length VcFeoB NTPase activity, even in our fusion construct VcFeo(ABC). Therefore, our results do not support the hypotheses of FeoA or FeoC as modulators of FeoB NTPase rate, at least in V. cholerae.

Similarly, Smith and Sestok reported that KpFeoA did not cause any change on the KpFeoB GTPase rate in vitro.¹⁴ Taken together, our data suggest that FeoA and FeoC may have regulatory roles in features other than nucleotide hydrolysis, such as complex formation or tuning of protein-protein interactions. Indeed, a hypothesis that still needs to be tested is whether FeoA and FeoC may couple NTP hydrolysis to iron uptake. Other studies have suggested alternative roles for FeoC, including as a modulator of gene expression factor and as an iron sensor, but evidence in this regard is inconclusive.6 Future research must focus on how FeoA and FeoC affect the function of the Feo system beyond the catalytic activity of FeoB.

Fe²⁺ triggers GTP hydrolysis in PaFeoB, and C675, a highly conserved residue, is critical in this response¹⁰; therefore, it is plausible that C675 regulates iron uptake by working as an iron sensor to initiate nucleotide hydrolysis. Since C675 is located in the transmembrane portion of the protein, Fe²⁺-mediated activation could only be tested in vitro by working with full-length FeoB and, indeed, Fe²⁺ does not affect the NTP hydrolysis in vitro by NFeoB.²³ Although VcFeoB possesses the putative cysteine sensor (C678, Fig. S6, ESI), Fe²⁺ did not lead to any change in its NTPase rate in vitro (Fig. 3B and D), however, we cannot rule out a sensing function in vivo. Since the mutation C678S renders the V. cholerae’s Feo system non-functional²⁷, this residue must play some role in iron uptake. Likewise, the role of iron availability in NTP hydrolysis regulation must be assessed in other full-length FeoBs to establish whether the response to iron is conserved in other species and whether it always depends on transmembrane cysteine residues.

FeoB proteins have been considered “living fossils” of the eukaryotic G-proteins.⁷ FeoB has the archetypic G-protein motifs and shares a similar mechanism for nucleotide hydrolysis. This mechanism involves GTP binding, phosphate stabilization and hydrolysis, and GDP release. Here we show that NaN₃ and sodium vanadate, which are known phosphatase inhibitors,^28-30 also inhibit VcFeoB NTPase activity. We have noted inhibitory activity of these compounds towards VcNFeoB as well (unpublished work). Considering that these compounds function by stabilizing the transition state of phosphate hydrolysis, our results support the hypothesis of a G-protein-like mechanism for FeoB, which is also supported by the structural characterizations available for NFeoB. Further functional assessments using full-length FeoB may be insightful regarding the mechanistic traits that remain obscure, namely, the role of transmembrane residues and potassium in FeoB-mediated NTP hydrolysis.

Whether the Feo system is an active or a passive transporter is another question that remains unanswered. Given that FeoB has a slow intrinsic GTPase turnover rate and a weak nucleotide binding affinity, it has been extensively proposed that nucleotide hydrolysis might not be the energy source of the complex.^6,23 Equally possible is that stimulatory factors, such as K⁺ or GAPs, are needed to activate FeoB. Smith and Sestok found that the basal activity of full-length KpFeoB may be consistent with an active transport mechanism.¹⁴ The k_cat values we estimated for VcFeoB and VcFeo(ABC) are in the same order of magnitude that KpFeoB and those of activated NFeoBs (i.e., 10⁻² s⁻¹), which could indicate that nucleotide hydrolysis drives active transport. For example, the hemolysin B transporter, HlyB; the polypeptide transporter, SecA; and the antibacterial peptide exporter, McjD, all active transporters of E. coli, have k_cat[ATP] values of 3.3 × 10⁻² s⁻¹, 1.7 × 10⁻² s⁻¹, and 5.8 × 10⁻² s⁻¹, respectively.^31-33 In this regard, Smith and Sestok14 as well as Seyemohammad et al.,¹⁹ who studied full-length PaFeoB, point out that the hydrolytic activity of full-length FeoB proteins highly depends on the experimental conditions, such as pH, the detergent or polymer used to stabilize the protein, and whether the protein is reconstituted in liposomes or micelles. Membrane proteins are more stable and active when reconstituted in bilayer membranes³⁴; indeed, PaFeoB achieves its maximum GTPase rate in liposomes.¹⁹ Therefore, it is plausible that if FeoB is an active transporter; its actual NTPase activity may be significantly higher in vivo. The Km[GTP] value we estimated for VcFeoB (655 μM) is similar to that previously reported for VcNFeoB (650 μM)¹⁵. Likewise, both Km[ATP] (280 μM) and Km[GTP] are low relative to the physiological concentrations of these nucleotides in bacteria (3 mM ATP and 1 mM GTP in E. coli).³⁵ Thus, FeoB activity may not be limited by the intracellular availability of substrates. Also, the enzymatic efficiency values (k_cat/Km) of VcFeoB are in the same order of magnitude that those reported for other NFeoBs (~ 10–5 μM⁻¹ s⁻¹)²³, and show that this enzyme has a higher affinity towards ATP, similar to its N-terminal counterpart.¹⁵ Future studies must determine the optimal conditions to resemble the cellular environment of FeoB; this is, establish a reliable proxy to study the enzymatic activity of this protein.

We found that fusing the ORFs of feoA, feoB, and feoC in V. cholerae results in synthesis of a single protein, VcFeo(ABC) in which the molar stoichiometry of the three components is 1:1:1. The finding that the fusion protein is fully functional in vivo indicates that all the protein-protein interactions and structural organization required for iron transport can be carried out by the fusion polypeptide to the same extent as the wild type operon in which the components are encoded by three separate genes. Using the fusion construct not only showed that a 1:1:1 stoichiometry of the Feo system components was functional, but also verified that FeoA and FeoC do not enhance FeoB NTPase rate. We observed a similar Km[ATP], but higher Km[GTP], for the fusion protein relative to VcFeoB (Fig. 2 and 6), indicating that this artificial arrangement of the Feo proteins may affect nucleotide binding differently. Interestingly, several Feo operons contain fusion genes, such as two copies of feoA together or a FeoB protein containing an N-terminal FeoA-like domain.⁶ Since gene fusion is a mechanism involved in operon evolution, we can gain insight into the evolutional history of this ancient system by understanding how artificial fusion Feo proteins, such as VcFeo(ABC), work.

Conclusions

In this study, we characterized full-length VcFeoB in terms of its hydrolytic activity towards ATP and GTP, and how it is affected by K⁺, Fe²⁺, and the accessory proteins, FeoA and FeoC. Altogether, our results suggest that in V. cholerae FeoA, FeoB, and FeoC work at a 1:1:1 molar ratio to mediate ferrous iron uptake through a mechanism driven by FeoB NTP hydrolysis without requiring stimulatory factors.