Investigating the roles of the conserved Cu²⁺-binding residues on Brucella FtrA in producing conformational stability and functionality

Abstract

Brucella is a zoonotic pathogen requiring iron for its survival and acquires this metal through the expression of several high-affinity uptake systems. Of these, the newly discovered ferrous iron transporter, FtrABCD, is proposed to take part in ferrous iron uptake. Sequence homology show, FtrA, the proposed periplasmic ferrous-binding component, is a P19-type protein (a periplasmic protein from C. jejuni which shows Cu²⁺ dependent iron affinity). Previous structural and biochemical studies on other P19 systems have established a Cu²⁺ dependent Mn ²⁺ affinity as well as formation of homodimers for these systems. The Cu²⁺ coordinating amino acids from these proteins are conserved in Brucella FtrA, hinting towards similar properties. However, there has been no experimental evidence, till date, establishing metal affinities and the possibility of dimer formation by Brucella FtrA. Using wild-type FtrA and Cu²⁺-binding mutants (H65A, E67A, H118A, and H151A) we investigated the metal affinities, folding stabilities, dimer forming abilities, and the molecular basis of the Cu²⁺ dependence for this P19-type protein employing homology modeling, analytical gel filtration, calorimetric, and spectroscopic methods. The data reported here confirm a Cu²⁺-dependent, low-μM Mn ²⁺ (Fe ²⁺ mimic) affinity for the wild-type FtrA. In addition, our data clearly show the loss of Mn²⁺ affinity, and the formation of less stable protein conformations as a result of mutating these conserved Cu²⁺-binding residues, indicating the important roles these residues play in producing a native and functional fold of Brucella FtrA.

Graphical abstract

Using wild-type and mutant Brucella FtrA, a P19 (a C. jejuni protein showing Cu²⁺ dependent iron affinity) homolog, this investigation reports the roles of the conserved Cu²⁺-binding residues in producing the native fold for FtrA, metal binding (Cu²⁺ and Mn²⁺) affinities, and dimer forming ability.

1. Introduction

Brucella is a Gram-negative pathogen, capable of directly infecting a wide range of ruminant mammalian hosts (eg., B. melitensis, B. abortus, and B. suis, infecting goat and sheep, cattle and swine; B. pinnipedialis and B. ceti, infecting marine animals), causing infertility, abortion, and an illness known as brucellosis (1, 2). On the other hand, Brucella infection in humans is caused by consumption of, or contact with these infected natural hosts (with B. melitensis, B. abortus, and B. suis), and is characterized by a chronic, debilitating febrile-like illness (1). Although there are successful animal vaccines available to prevent brucellosis in its natural hosts, there is no vaccine available for humans (2, 3). The chronic debilitating nature of the human brucellosis, the difficulty treating it with antibiotics, and the lack of available vaccination to prevent the human infection, in addition to the highly infectious nature of B. melitensis, B. abortus, and B. suis strains by the aerosol route, are the main reasons that make these strains as drug targets and potential bioweapons, respectively (4–6). To complicate things further, this Gram-negative pathogen survives and grows intracellularly in host macrophages, where the intracellular pathogen acquire its nutritional requirements by expressing dedicated uptake pathways and circumventing several immune response strategies provided by the host (7–16). One such essential micronutrient for nearly all living organisms is iron which plays signaling, catalytic, and various other important roles in living systems (17, 18). Under aerobic and biological conditions (pH ~ 7.3), the predominant oxidation state of iron is Fe³⁺, which is extremely insoluble and takes part in reactive oxygen species (ROS) generating Fenton type reactions, producing toxic oxygen-based radical species. To reduce this toxic radical generating property as well as to increase its aqueous solubility, this essential micronutrient is tightly regulated during its uptake, transport, and utilization in all biological systems, making the concentration of “available iron” extremely low for invading pathogens (19–23). As a result, pathogens growing intra- or intercellularly, their ability to get hold of the host’s iron pool often determines their ability to stay viable as well as cause infection (24). Not surprisingly, invaded host systems further withhold this precious micronutrient by upregulating expression of transferrin, which lowers available iron concentration, downregulating the expression of transferrin receptors on the cell surface etc., thus starving the pathogen of iron and providing a protective measure that has been termed as the host’s “nutritional immunity” in literature (25). To counteract these host immune strategies that withhold the desired bacterial iron, pathogens employ the following iron hijacking uptake pathways to circumvent the said nutritional immunity: a) the production and extracellular secretion of siderophores (siderophores are small organic molecules rich in O and N donor atoms that preferentially bind to Fe³⁺ in the host systems) and expression of Fe³⁺-siderophore transporters; b) the expression of high-affinity heme uptake systems on bacterial membrane; and c) the direct transportation of inorganic Fe³⁺ through high-affinity TonB-dependent membrane uptake systems (19–32). Like most other organisms, Brucella also require iron for survival and growth. However, being an intracellular pathogen, encounters a very low concentration of this nutrient and combats this nutritional deficiency by expressing two well described iron uptake pathways, consisting of siderophore and heme transport systems (33–38). In addition, our recent work has identified a pH (these genes are better expressed under acidic conditions) and iron-regulated gene cluster, BAB2_0837–0840, from the Brucella abortus 2308 genome, which is predicted to express a dedicated four-component Fe²⁺ transport system (39). A similar Fe²⁺ uptake system has also been identified in the human pathogen Bordetella spp. and based on the sequence homology with this Fe²⁺-uptake system from Bordetella, the gene products of BAB2_0837–0840 from Brucella have also been named FtrABCD (39, 40). In our previous study, we have shown that a Brucella ΔftrA strain loses viability in Fe²⁺/Fe³⁺-restricted media compared to when grown in heme, confirming the role of this protein in inorganic-iron utilization (39). Further, the ΔftrA mutant strain could only transport the radiolabeled ⁵⁵Fe³⁺ compared to ⁵⁵Fe²⁺, suggesting that FtrA plays a role in ferrous iron transport (39). Finally, the same study has also shown significant attenuation of the ΔftrA strain in cultured murine macrophages and experimentally infected mice, supporting the importance of this proposed ferrous-iron transporter in Brucella virulence (39).

These bacterial FtrABCD-type four-component ferrous iron transport systems from Brucella and Bordetella are homologs to eukaryotic ferrous iron transporters, Ftr1p-Fet3P (41, 42). The best studied of all the eukaryotic Ftr1p-Fet3p transporter is from the yeast Saccharomyces cerevisiae. This ferrous iron transporter is characterized by an oxidation dependent translocation of Fe³⁺ through the permease protein, Ftr1p, where Fet3p, in addition to sequestering Fe²⁺, also acts as a ferroxidase (43, 44). The bacterial homologs of these two-component eukaryotic Ftr1p-Fet3p systems have been identified as two-, three-, and four-component systems, where an inner-membrane permease protein can only translocate Fe³⁺(acting as a Ftr1P homolog), whereas the Fe²⁺ sequestration and oxidation are performed by the same or different proteins (45). For example, in the four-component FtrABCD systems found in Brucella and Bordetella, the periplasmic FtrA is proposed to sequester Fe²⁺, whereas FtrB, a new class of CupII-type ferroxidase, has been hypothesized to catalyze the iron oxidation (45). The oxidized Fe³⁺ is subsequently transported through the membrane permease, FtrC, and FtrD is thought to serve as an electron sink, resetting the reaction (Figure 1) (45).

Figure 1.

A schematic representation of the ferrous ion transporter, FtrABCD, from Brucella spp. which is a bacterial homolog of the ferroxidase-dependent eukaryotic Ftr1p-Fet3P system (39, 40, 45). The operons responsible for expressing the individual components of the FtrABCD system are presented at the top as blue arrows. It is hypothesized that Fe²⁺enters the periplasm through an unassigned outer membrane receptor where it is sequestered by the periplasmic FtrA in a Cu²⁺-dependent manner, like P19-type proteins. FtrB, the CupII-type multicopper ferroxidase catalyzes the conversion of this bound Fe²⁺to Fe³⁺ (shown by the arrows with electron, e^-, flow), whereas FtrC, the Ftr1p homolog is proposed to transport this oxidized metal, and FtrD acts as the electron sink and resets the ferrous uptake system (39, 40, 45).

Based on sequence homology, these FtrA proteins from Brucella and Bordetella are designated as a P19-type system (45). P19 is an iron regulated periplasmic protein from C. jejuni which has been reported to show Cu²⁺ dependent iron affinity and is crucial for the viability of this bacteria under iron restricted condition (46). P19-type proteins are well documented in literature and are characterized by their Cu²⁺-dependent iron affinity (46–48). X-ray crystal structures for several P19-type proteins have identified stable homodimer formation, showing extensive H-bonding interactions, as well as defining the Cu²⁺environment in these systems as His₃Met/ His₃Glu (46–48). Interestingly, these studies also show each Cu²⁺ ion being shared by a dimeric unit, where three coordinating amino acids (H, H, and E/M) come from one monomeric unit and the last coordination site is provided by a conserved histidine coming from the symmetry opposed monomeric chain of the same dimer (46–48). Our recent report has shown that these Cu²⁺-binding residues from other P19-type proteins are completely conserved in Brucella and Bordetella FtrA (conserved Cu²⁺-binding residues from Brucella FtrA: H65, M111/E67, H118, and H151) hinting that similar roles are played by these residues (39, 40).

Although, the Cu²⁺-dependent iron affinity for these P19-type proteins, as well as their abilities to form stable homodimers are well documented in literature, to our knowledge, till date there has been no study on any P19-type proteins that has evaluated these conserved properties from a molecular level. Further, despite identifying the Cu²⁺-binding residues from several P19-type proteins (46–48), no previous work has examined the effects that mutating these residues has on these proteins’ metal affinities or dimer forming abilities. Thus, in this work using wild-type and four Cu²⁺-binding mutants (H65A, E67A, H118A, and H151A) of recombinant FtrA from Brucella we have investigated the relative sizes of these proteins, their Cu²⁺ and Mn²⁺ sequestration abilities, as well as overall folding stabilities of these proteins utilizing analytical gel filtration, isothermal titration calorimetry (ITC), and differential scanning calorimetry (DSC), respectively. Analytical gel filtration data reported here indicate wild-type FtrA purifies as a dimer whereas H118A elution profiles indicate protein degradation, which we attribute to a result of this mutation. ITC experiments on wild-type FtrA conducted at two different pH values establish Brucella FtrA as a functional P19-type protein showing a Cu²⁺-dependent low μM Mn²⁺ (an Fe²⁺ mimic) affinity. No Mn²⁺ affinity was observed for any of the Cu²⁺-binding mutants tested in this investigation, irrespective of their Cu²⁺ binding ability. Taken together with the analytical gel filtration data, we interpret our DSC data as indicative of the fact that only wild-type FtrA, with all four conserved copper binding residues, can form stable homodimers within the concentration range tested. Our DSC data also show that the wild-type FtrA is further stabilized successively by sequestering Cu²⁺ and Mn²⁺, whereas the Cu²⁺-binding mutants did not show this property.

2. Materials and methods

All chemicals and materials used in this study are of the highest purity grade. CuCl₂, GdCl₃, and MnCl₂ were purchased from Fisher Scientific and Sigma. N-(2-acetamido)-2-aminoethanesulfonic acid (ACES), isopropyl-β-D-thiogalactoside (IPTG), Tris base, Tris hydrochloride, and sodium chloride were purchased from Fisher Scientific. Imidazole was purchased from Acros.

2.1. Bacterial cloning and site-directed mutagenesis

A list of the bacterial plasmids used in this study is provided in Supplementary Table 1. B. abortus 2308 ftrA (BAB2_0840) was cloned by PCR from genomic DNA without the signal sequence into pET32a (Novagen) using NcoI and XhoI restriction sites making pET32a-WTFtrA. Point mutations were performed according to manufacturer’s instructions using the Q5 Site-Directed Mutagenesis Kit (NEB) and using the constructed pET32a-WTFtrA as the template. Plasmids were DNA sequenced for confirmation. Primers for site directed mutagenesis were developed using NEB Basechanger (https://nebasechanger.neb.com/) and are listed in Supplementary Table 2. Plasmids were transformed into Escherichia coli strains DH5α for cloning and BL21 (DE3) pLysS for recombinant protein expression. These strains were cultivated on tryptic soy agar at 37°C, or in LB (Lysogeny broth or Luria Broth) at 37°C with shaking, supplemented with 100 μg/ml ampicillin. Stock cultures were maintained in LB with 25% glycerol at −80° C and streaked for isolation before use.

2.2. Purification of recombinant Brucella FtrA

Bacterial cells were shaken at 37° C in LB broth supplemented with 100 μg/ml ampicillin. Induction of the plasmids was performed at A₆₀₀=0.6–0.8 by adding 2mM isopropyl-β-D-thiogalactoside (IPTG) for 4 hours maintained at 37° C. Cells were subsequently harvested by centrifugation at 10,000 × g at 4°C for 10 min followed by resuspending in 1/10 volume Histidine-Wash buffer (50 mM sodium phosphate, 300 mM sodium chloride, 10 mM imidazole, pH 7.4). Cells were lysed by sonication and the crude lysate was centrifuged at 25,000 × g for 20 min at 4°C. The supernatant was applied to HisPur Resin in a gravity flow Cobalt-column (Thermo Scientific). The column was prepared by washing it first with 3 column volumes of Histidine-Wash buffer before applying the crude lysate. The desired protein was eluted with 2 column volumes of Histidine-Elution Buffer (50 mM sodium phosphate, 300 mM sodium chloride, 100 mM imidazole, pH 7.4). This purified tagged-FtrA was then extensively dialyzed for 72 hours at 4°C in recombinant enterokinase buffer (500 mM sodium chloride, 200 mM Tris Cl, 20 mM calcium chloride, pH 7.4). The N-terminal tag was cleaved with 200 IU Histidine-tagged recombinant bovine enterokinase (GenScript). The subsequent tag-free protein solutions were reapplied to HisPur Resin in a gravity flow Cobalt-column (Thermo Scientific) where the enterokinase tag bound to the Co-column, and tag-free FtrA was collected in the flow-through. The tag free FtrA was then dialyzed 3X against 25 mM ACES (N-(2-Acetamido)-2-aminoethanesulfonic acid) buffer (pH 6.3 or pH 7.3) to get stock protein solutions. Protein samples were assessed for purity by SDS-PAGE (Supplementary Figure 1) and concentrations were determined using A₂₈₀ on a Biospec NanoDrop instrument (Thermo Scientific).

2.3. Analytical Gel Filtration

An ÄKTA pure 25L FPLC (GE Healthcare) and a Superdex™ 75 Increase 10/300 GL (GE Healthcare) column with a flow rate of 0.5 mL/min in HEPS Buffer Saline or HBS (10mM HEPES pH7.3, 140mM NaCl) were used for analytical gel filtration experiments. Approximately 0.5 mg of as-isolated FtrA proteins (wild-type and mutant) were run on the column with an isocratic elution volume (V_e) of 36.5 mL. Relative Molecular Weights (rMW) were determined using a semi-log linear fit using Gel Filtration Standard (Bio-Rad) using the last 4 globular proteins γ-globulin, ovalbumin, myoglobin, and vitamin B12 with rMW of 132kDa, 56kDa, 17kDa, and 1.3kDa respectively with an R²= 0.9924. Stoke’s radii were determined on the linear range among ovalbumin, myoglobin, and vitamin B12 with radii of 30.5, 20.5, and 1.6 Å, respectively (Bio-Rad), with the assumption that the shape of FtrA proteins are approximately spherical globular proteins.

2.4. Circular Dichroism

Circular dichroism (CD) was used to investigate secondary structural changes in wild-type FtrA concomitant to metal titration. Experimental data were collected at 25°C on a JASCO J-815 CD spectrometer in 3 mM Phosphate Buffer Saline (PBS) at pH 7.3 using a 1 mm quartz cuvette and recording absorbance in the 200–280 nm range. The raw data, after correcting for buffer and metal absorptions, were converted to mean residue ellipticity ([θ]) using equation 1 (49–51):

$$[\mathrm{\theta }]=100\frac{\text{signal}}{(\text{C})(\text{n})(\text{l})}$$ (1)

where C stands for the concentration of the protein (mM), n is the number of amino acid residues in FtrA (without the signal peptide), and l is the path length (cm).

2.5. Preparing protein and ligand solutions for ITC and DSC experiments

Prior to conducting metal titration experiments using ITC and DSC, purified wild-type and mutant FtrA samples were dialyzed (3X) at 4 °C against 25 mM ACES buffer at pH 6.3 or 7.3, and the 3^rd dialysate was preserved to prepare the metal stock solutions for buffer matching. Working solutions of Cu²⁺ and Mn²⁺ were freshly prepared before each titration by diluting stock metal solutions in this dialysate. To prepare metal free (apo) wild-type FtrA, the as-isolated protein (mentioned simply as wild-type protein for the rest of this report) was dialyzed 3X in 25 mM ACES buffer at pH 7.3 containing excess EDTA (Ethylenediaminetetraacetic acid) followed by 3X cleaning dialysis in the same buffer to get rid of the excess EDTA.

2.6. Differential Scanning Calorimetry (DSC)

DSC experiments on wild-type and mutant-FtrA (H65A, E67A, H118A, and H151A) dissolved in 25 mM ACES buffer at pH 6.3 or 7.3 were performed on the TA-instruments Nano-DSC to determine protein conformational stabilities (indicated by melting temperatures, T_m) as well as mechanism (indicated by the numbers and shapes of melting peaks) (51–55). In a typical DSC experiment with the wild-type FtrA, the protein solution (20–70 μM) under investigation was degassed for 5 mins prior to injecting in the sample compartment. The temperature was scanned in the 40–98 °C interval under 3 atmosphere pressures at 1 °C/min scan rate. To determine thermodynamic reversibility of the process, all experiments were performed in a heating-cooling followed by a reheating cycle, where the two heating steps were carried out at the same scan rate and the cooling was conducted at a faster rate (2 °C/min) to ensure the least degradation to the protein structure upon unfolding. A wide range for concentration of the wild-type protein was selected for recording DSC data to determine the effect of concentration on melting temperature (T_m). The raw heat capacity data obtained as a function of temperature scan was analyzed by using fourth-order polynomial baseline and the best fit was obtained by using Gaussian fit model available with the Nano Analyze program. All experiments were performed in triplicate and ΔH_cal as well as ΔH_cal /ΔH_vH values reported here are the average of three reproducible experimental data.

2.7. Isothermal Titration Calorimetry

ITC experiments described in this work were collected on either Affinity-ITC or LV Nano-ITC (TA) instruments. For Cu²⁺-titration, the metal solution was incrementally injected into a solution of purified wild-type or mutant FtrA (Supplementary Figure 1), whereas Mn²⁺ was injected into protein solution previously saturated with Cu²⁺ (5X). All solutions used in ITC experiments were degassed for 7 mins prior to starting the titration. A typical ITC titration experiment was performed at 25 °C with a 200–320 Revolution Per Minute (RPM) stirring rate, by filling the sample cell with 350 μL of protein and incrementally injecting 50 μL of pure metal solution in it. The injection interval was set between 180–440 seconds and was controlled by the Nano-run software. The heats of dilution (for ligand into buffer titrations) were determined in separate experiments and were subtracted from the integrated heats for metal-protein titrations during data analysis. The best fit to the ITC raw data were obtained by using one-site binding model (Nano Analyze Software) yielding stoichiometry (n), dissociation constant K_d (where, K_d = 1/K_a) and ΔH for the protein-ligand reactions, and the integrated heat data were graphically represented in exo-down mode (56, 57). Each metal-protein titration experiment was performed in multiple replicates yielding average thermodynamic binding parameters reported in this work.

2.8. Mn²⁺ as an Fe²⁺ mimic for experiments described in this work

Extremely low solubility of Fe³⁺ (K_sp Fe³⁺ = 10⁻³⁸), the predominant oxidation state of this metal under neutral pH and oxic conditions, is one of the major reasons for biological systems to express specialized iron sequestration and uptake pathways (19, 26, 27). This aqueous insolubility and the spontaneous oxidation of Fe²⁺ to Fe³⁺ under physiological and oxic conditions make biochemical experimentations that require Fe²⁺ especially challenging for protein-Fe²⁺ reactions with weak/slow binding thermodynamics and/or kinetics, as iron oxidation/precipitation to Fe³⁺-oxo-hydroxo species takes place faster than protein binding. To avoid interference from such competing precipitation and oxidation/reduction reactions, we have used Mn²⁺ as an Fe²⁺ mimic in this work. Other investigators working with the homologs of FtrA, P19, and FetP have also used Mn²⁺ as an iron mimic (46, 47). In addition, Fe²⁺ and Mn²⁺, having similar charge to size ratio, compete for metal-binding proteins in biological systems where the metal-protein affinity is dictated by the Irving-William series, predicting slightly lower affinity for Mn²⁺-protein compared to Fe²⁺-protein complexes (58, 59). Further, previous experimental and theoretical work has identified the Mullikan atomic charges of aquated Fe²⁺ and Mn²⁺ species as well as the pKa values for their hexa-aquo complexes to be very close to each other, indicating very similar aqueous coordination properties (60). Lastly, many bacterial species are known to have substituted iron with manganese in typically Fe-dependent metalloproteins to avoid such iron toxicity, indicating the functional similarity between these two ions in biological systems (61). Thus, the use of Mn²⁺ in our experiments is consistent with previous studies and is biologically relevant.

2.9. Homology Modeling and Structural alignment

The model structure of FtrA and its mutants were constructed by using the homology-modelling server SWISS-MODEL (https://swissmodel.expasy.org). The modelled structure of FtrA was done based on the template structure of P19 protein (PDB 3lzl), identified from the BLAST (basic local alignment search tool) analysis using default parameters. Further analysis of the molecular structures was done with CHIMERA. The structural alignment of the models of FtrA and its mutants is carried out by MATRAS (http://strcomp.protein.osaka-u.ac.jp/matras/matras_multi.html) with default parameters (62–64).

3. Results and discussion

3.1. Elution profile for H118A is different compared to wild-type FtrA

Analytical gel filtration experiments were performed on wild-type and all four Cu²⁺-binding mutants of FtrA. As mentioned in the Introduction, previously reported P19-type protein crystal structures show stable homodimer formation (46–48), and being that FtrA is a P19 homolog, we wanted to investigate if FtrA forms a similar structure. Based on the amino acid sequence of the polypeptide chain of mature wild-type FtrA the molecular weight of the monomeric unit is 17.36 kDa, which was confirmed on the SDS-PAGE (Supplementary Figure 1). As is indicated in Table 1 (and Supplementary Figure 2), wild-type FtrA, as well as H65A, E67A, and H151A mutants all eluted at similar V_e ~10.65 ml as a single sharp monodisperse-peak, corresponding to relative molecular weight ~45.6 kDa, and shared a very similar Stoke’s radius (Table 1), indicating that the wild-type protein and these three mutants are relatively the same size. In contrast, the H118A mutant eluted at multiple elution volumes (Table 1) corresponding to a range of molecular weights between 11.94–167.07 kDa. The fact that H118A had an elution peak corresponding to a relative molecular weight of 11. 94 kDa, which is less than the theoretical molecular weight of a single polypeptide chain for wild-type FtrA (Supplementary Figure 2), in addition to showing laddering in gel electrophoresis experiments (Supplementary Figure 1), can be interpreted as a result of protein degradation. Formation of similar stable homodimers have also been observed previously, using analytical gel filtration data, for P19 from Campylobacter jejuni, however, this report did not investigate the effect of Cu²⁺-binding residue mutation on dimer formation ability of this protein (46).

Table 1.

Analytical gel filtration data on wild-type and Cu²⁺-binding mutants of FtrA. Relative Molecular Weights (rMW) were determined using a semi-log linear fit using Gel Filtration Standard (Bio-Rad) with the last 4 globular proteins γ-globulin, ovalbumin, myoglobin, and vitamin B12 with rMW of 132kDa, 56kDa, 17kDa, and 1.3kDa respectively with an R²= 0.9924. Approximately 0.5 mg of wild-type and recombinant FtrA samples (1 mg of H118A was used) were eluted separately at a rate of 0.5 mL/min in HBS (10mM HEPES pH 7.3, 140mM NaCl) with an isocratic elution volume of 36.5 mL.

Protein	Elution Volume (mL)	Relative Molecular Weight (kDa)	Stoke’s Radius (Å)
WT-FtrA	10.65	45.64	28.65
H65A	10.63	46.04	28.72
E67A	10.65	45.64	28.65
H151A	10.65	45.64	28.65
H118A Peak 1	7.66	167.07	38.58
H118A Peak 2	10.51	48.50	29.12
H118A Peak 3	12.83	17.72	21.42
H118A Peak 4	13.7	11.94	18.53

3.2. Secondary structure of wild-type FtrA is not altered by metal titration

CD spectroscopy was performed on wild-type FtrA both with and without added Cu²⁺ and Mn²⁺ to evaluate changes in average secondary structure as a result of metal binding (as confirmed by ITC data presented above). Wild-type FtrA (without any added metal) showed two strong peaks centered at 220 nm (positive) and 200 nm (negative). Compared to this, when Cu²⁺ and Mn²⁺ were titrated into the wild-type protein solution in separate experiments (final metal concentration 3X with respect to the protein), there was little to no changes in the peak position or intensity, indicating insignificant secondary structural differences between metal-free and metal-bound wild-type FtrA (Supplementary Figure 4). Although, CD data does not show any average global structural changes upon metal binding to wild-type FtrA, this does not preclude the possibility of local structural changes in the metal binding site, as well as changes in protein folding stability, as would be discussed in the DSC result sections. Additionally, this data is in accordance with the findings from the X-ray crystal structure for P19, showing little secondary structural changes in the solid-state for the metal-free and metal-coordinated protein (46–48).

3.3. FtrA homology modeling predicts structure alteration for H118A

The homology modeled structure (with GMQE 0.8, QSQE 0.98, and sequence similarity of 0.48) of wild-type Brucella FtrA is shown in Figure 2a). The two monomeric chains (chain-A and chain-B) of FtrA model as a dimer and indicate two coordinated Cu²⁺ ions coordinated in H₃E environment, very similar to other P19-type proteins. Moreover, the homology structure also indicates that each Cu²⁺ is covalently connected to three amino acid residues from one chain - His65, Glu67, and His118, while the fourth coordination site being filled by a histidine residue (H151) coming from the other chain (Figure 2b), similar to other P19-type proteins (46–48).

Figure 2.

a) Homology modeled structure of FtrA (Template 3lzl with GMQE 0.8, QSQE 0.98, and sequence similarity of 0.48). The Cu²⁺-coordinating residues from chain A and chain B are indicated from this model (46–48). b) Cartoon representation of the coordination of Cu²⁺in the FtrA model structure between chain A (His65, Glu67, and His118) and chain B (His151). The Cu²⁺ interacting amino acids are indicated in ball and stick model. c) Cartoon diagram of Cu coordination in FtrA showing the N-H…O interaction between H118 of chain A and Gly116 of chain A.

Interestingly, closer inspection of the local coordination environment of each of these Cu²⁺ ions show residues His65 and Glu67 residing within the same beta-strand (β4) while the His118 residue resides within a different beta-strand (β7) (Figure 2b). Based on this model, the local structural scaffold of the copper-binding site is strictly dependent on the intermolecular H-bonds among the residues of three antiparallel beta-strands β4, β6, and β7. However, as shown in Figure 2b), a sudden ‘twist’ in the local structural scaffold of the strand β7 perhaps results in an ‘energetically unfavorable state’ for the copper-binding site. The modeled structure also shows an important N-H-O-interaction (2.58Å) between the conserved Cu²⁺-binding H118 residue from the β7 strand and G116 from the main chain (Figure 2c). Although a modeled structure, this possible interaction can be taken as preliminary indication of a local and global structural stabilizing effect provided by the H118 residue. Finally, a structural alignment of the wild-type FtrA model with the four mutants reported here showed a structural alteration/disruption for the H151A mutant (Supplementary Figure 2), implying possible loss of native fold of the protein.

3.4. Wild-type FtrA shows Cu²⁺-dependent Mn²⁺ binding affinity

ITC experiments were performed on wild-type FtrA by incrementally injecting aliquots of Cu²⁺ and Mn²⁺ into protein solutions. Wild-type ITC metal-titration experiments were performed at two pH values (6.3 and 7.3) based on the previous observation that the expression of the ftrA gene is better induced by an acidic pH (39). As can be seen from Table 2 and Figure 3, the wild-type protein shows exothermic heats upon addition of Cu²⁺ at both pH 6.3 and 7.3. The buffer subtracted integrated heat data for Cu²⁺ titration is presented in the lower panel of the ITC thermogram (the black dots). The best fit (solid lines) to experimental data were obtained by using one-site independent binding model, yielding K_d values of 4.9 and 5.3 μM; n (per monomer of wild-type FtrA) = 0.53 and 0.50; and ΔH= −3.3 and −3.9 kcal/mol at pH 6.3 and 7.3, respectively (Table 2), which shows a pH invariant Cu²⁺ affinity. Similar μM Cu²⁺-binding affinities and exothermic heat changes have also been reported for the FtrA homolog FetP, another P19-type protein, using ITC experiments in 25 mM Bis-Tris buffer at pH 7.2 (47). Interestingly, in the same study, Cu²⁺ did not show any binding affinity towards FetP in ACES buffer at pH 7.2 in contrast to our observation (47). Moreover, many of the experimental n values reported in this previous work were not obtained by iterative data fitting method (47). Based on the X-ray structures of the P19-type proteins, each Cu²⁺ ion is shared between two monomeric protein chains via a conserved histidine residue coming from the symmetry opposed unit (designated by H151 in Brucella FtrA), and would explain the ~0.5 stoichiometry for Cu²⁺ binding to wild-type FtrA (46–48).

Table 2.

Average K_d, n, ΔH, and ΔS obtained from various ITC titration experiments involving wild-type FtrA (WT-FtrA) and metal-binding mutants (H65A, E67A, H118A, and H151A) with Cu²⁺ and Mn²⁺ in 25 mM ACES buffer at pH 6.3 and 7.3 (25 °C). The standard deviations in the measurements are calculated from the average of three trials of the same titration.

Protein + metal	pH	n (monomer of FtrA)	Kd (μM)	ΔH (kcal/mol)	ΔS (cal/mol K)
WT-FtrA + Cu2+	6.3	0.53 ± 0.03	4.9 ± 2.0	−3.3 ± 1.2	13.3 ± 3.0
WT-FtrA + Cu2+	7.3	0.50 ± 0.05	5.3 ± 1.4	−3.9 ± 1.5	11.05 ± 5
(WT-FtrA + Cu2+) + Mn2+	6.3	0.51 ± 0.05	20.00 ± 0.002	−1.2 ± 0.2	17.5 ± 0.8
(WT-FtrA + Cu2+) + Mn2+	7.3	1.1 ± 0.2	8.4 ± 2.3	+0.4 ± 0.2	24.6 ± 0.9
WT-FtrA + Mn2+	7.3	No binding	----	----	----
H65A + Cu2+	7.3	0.70 ± 0.03	7.6 ± 1.6	−4.1 ± 0.5	9.6 ± 1.1
(H65A + Cu2+) + Mn2+	7.3	No binding	----	----	----
E67A + Cu2+	7.3	1.40 (0.03)	2.05 ± 0.04	−5.3 ± 0.6	8.4 ± 2.3
(E67A + Cu2+) + Mn2+	7.3	No binding	----	----	----
H118A + Cu2+	7.3	No binding	-----	-----	-----
H151A + Cu2+	7.3	No binding	-----	-----	-----

Figure 3.

Representative ITC isotherms for Cu²⁺ titrations into solutions of wild-type FtrA at a) pH 6.3, [FtrA] = 20 μM, [Cu²⁺] in the syringe = 150 μM and b) pH 7.3, [FtrA] = 47 μM, [Cu²⁺] in the syringe = 350 μM. The upper panels show exothermic heat peaks per injection as a function of time, whereas the dots on the lower panel are the integrated heat data for each Cu²⁺ addition. The solid red line represents the best fit to the integrated heat data using independent one site binding model. All experiments were performed in 25 mM ACES buffer at 25 °C. See Table 2 for thermodynamic binding parameters.

When the Mn²⁺ solution was titrated into buffered solutions of the wild-type FtrA incubated with 5X Cu²⁺, the heat per injection was exothermic at pH 6.3. In contrast, similar experiment at pH 7.3 produced small endothermic heats (Figure 4). We have been able to fit these raw Mn²⁺ heat data using a one-site independent binding model, yielding K_d = 20.00 and 8.4 μM, n (per monomer of wild-type FtrA) = 0.51 and 1.1, ΔH = −1.2 and +0.4 kcal/mol; ΔS = 17.5 and 24.6 cal/molK, at pH 6.3 and 7.3 respectively. Similar exothermic heat upon injecting Mn²⁺ to FetP (saturated with either Cu²⁺ or Zn²⁺ at pH 7.3 in ACES buffer) has been reported earlier, however with n ~ 0.5 per monomer of FetP (47). Although the calculated free energy change (ΔG) for Mn²⁺ binding to Cu²⁺-saturated wild-type FtrA under these experimental conditions are nearly the same (~−6.5 kcal/mol), looking at the ITC thermograms, it is evident that two different binding mechanisms are responsible for Mn²⁺ affinity at these two pH values. As is evident from these data, the small positive ΔH is off set by a larger entropic gain at pH 7.3 as a result of metal titration. Interestingly, the speciation of Mn²⁺ at these experimental pH’s is [Mn(H₂O)₆]²⁺ with experimental pKa ~ 11 (60). Binding of Mn²⁺ (aq) to the wild-type FtrA should release the same number of water molecules per metal ion, giving similar entropic gain. We speculate that this greater entropic contribution of Mn²⁺ binding to wild-type FtrA at a more basic pH could be attributed to a difference in solvation/desolvation of the metal binding pocket as a result of metal chelation. Multiple sequence alignment of Brucella FtrA with other structurally characterized P19-type proteins have identified two conserved residues (E and D) as Mn²⁺ chelators (39). Both E and D residues are deprotonated at these experimental pH values and should contribute equally to metal binding interaction. However, a stronger affinity for Mn²⁺ and a higher n, as observed at the basic pH, can be taken as an indication of some yet unidentified amino acids (with pKa’s closer to this pH) from wild-type FtrA as taking part in metal sequestration. It is however important to emphasize here that these are indirect evidences for structural data and more direct experimental evidence are required to establish these speculations.

Figure 4.

Representative ITC isotherms for the Fe²⁺ mimic (Mn²⁺) titrations into solutions of wild-type as-isolated FtrA saturated with Cu²⁺, at a) pH 6.3, [FtrA] = 45 μM, [Mn²⁺] in the syringe = 550 μM and b) pH 7.3, [FtrA] = 58 μM, [Mn²⁺] in the syringe = 550 μM. The upper panels show exothermic heat peaks per injection as a function of time, whereas the dots on the lower panel are the integrated heat data for each Cu²⁺ addition. The solid red line represents the best fit to the integrated heat data using independent one site binding model. All experiments were performed in 25 mM ACES buffer at 25 °C. The inset in figure b) shows Mn²⁺ titration into wild-type FtrA without adding any Cu²⁺ ions, indicating no binding between the Fe²⁺ mimic and the wild-type protein under these experimental conditions.

Finally, when Mn²⁺ solution was titrated into wild-type FtrA without any Cu²⁺ present, no binding heats were observed (inset of Figure 4), indicating, like other reported P19-type proteins, wild-type FtrA cannot bind to Mn²⁺ without Cu²⁺ (47).

Our previous in-cell study indicated a more detrimental effect on viability of Brucella as a result of ftrA deletion at a lower pH, suggesting a greater importance of this gene product at an acidic pH (39). In contrast, the ITC data presented here with Mn²⁺ plus the wild-type protein show a higher affinity (K_d) and protein-to-metal stoichiometry (n) at a basic pH (Table 2). We attribute this apparent contradiction under cellular vs in vitro conditions based on the higher availability of Fe²⁺ at an acidic pH, as well as the observation that the ftrA gene is better expressed under an acidic condition (39). Mn²⁺, the Fe²⁺ mimic used in our ITC experiments, is the predominant species under both experimental pH conditions (based on the established Frost diagrams) and could not resolve a possible metal speciation dependent affinity of FtrA as a function of pH fluctuation.

3.5. FtrA Cu²⁺-binding mutants cannot bind the Fe²⁺ mimic

To investigate the contribution of the individual Cu²⁺-binding residues on FtrA towards its metal affinity, ITC titration assays were performed on all four mutants (H65A, E67A, H118A, and H151A) at pH 7.3. Figure 5 and Table 2 shows that the ITC thermograms for H65A and E67A mutants when aliquots of Cu²⁺ was incrementally titrated in protein solutions in separate experiments, indicating exothermic peaks. Best fit to these Cu²⁺ titration data for H65A and E67A mutants were obtained by using independent single site binding model, yielding K_d and n values 7.6 μM and 0.70 (per monomer); and 2.05 μM and 1.40 (per monomer), respectively. These Cu²⁺-protein stoichiometries are significantly different compared to wild-type FtrA n data (n (per monomer) = 0.50) at pH 7.3. These higher n values for the mutants can be a result of the contribution of some other amino acids from the mutant proteins that can coordinate with Cu²⁺ compared to the wild-type Cu²⁺ coordination sphere. Our DSC data discussed below indicates altered conformational folds for these mutants (Table 3), which can be taken as further support of different coordination environments around Cu²⁺ in these mutants. In contrast to the H65A and E67A Cu²⁺ ITC data, the other two mutants, H118A and H151A, did not show any binding heat with Cu²⁺ solution, indicating complete loss of Cu²⁺ binding affinity as a result of these two mutations and also shedding light to the fact that these two conserved residues have important roles in Cu²⁺ sequestration.

Figure 5.

Representative ITC isotherms for Cu²⁺ titrations into solutions of a) [H65A] =70 μM, [Cu²⁺] in the syringe = 600 μM and b) [E67A] = 75 μM, [Cu²⁺] in the syringe = 650 μM at pH 7.3. The upper panels show exothermic heat peaks per injection as a function of time, whereas the dots on the lower panel are the integrated heat data for each Cu²⁺ addition. The solid red line represents the best fit to the integrated heat data using independent one site binding model. All experiments were performed in 25 mM ACES buffer at 25 °C. See Table 2 for thermodynamic binding parameters.

Table 3.

Protein melting data, T_m, obtained from DSC experiments on wild-type (WT) and Cu²⁺-binding mutants of FtrA (H65A, E67A, H118A, and H151A) with and without (5X) Cu²⁺, and Cu²⁺ + Mn²⁺ in 25 mM ACES buffer at pH 6.3 and 7.3. Each DSC experiment was performed at least three times to get average T_m values. ΔH_cal were obtained by fitting the raw data with a fourth-order polynomial baseline and ΔH_cal/ΔH_vH were calculated by fitting these baseline corrected data using Gaussian fits.

Protein	pH	Tm (°C)	ΔHcal kJmol−1	ΔHcal/ΔHvH
WT-FtrA	6.3	75, 84	321.10 ± 3.87	0.99
WT-FtrA	7.3	74, 83	260.22 ± 6.47	1.00
WT-FtrA + Cu2+	6.3	79, 85	348.43 ± 3.41	1.00
WT-FtrA + Cu2+	7.3	76, 83	323.10 ± 7.50	0.98
WT-FtrA + Cu2+ + Mn2+	6.3	83, 86	382.61 ± 3.50	0.98
WT-FtrA + Cu2+ + Mn2+	7.3	85, 87	345.20 ± 2.80	1.00
H118A	7.3	56, 60, 64	127.26 ± 9.70	0.98
H151A	7.3	60, 62	113.59 ± 6.71	0.99
H65A	7.3	73, 80	126.52 ± 1.60	0.99
E67A	7.3	78, 80	59.40 ± 3.30	1.00

Additionally, none of the Cu²⁺-binding mutants, irrespective of their ability to sequester Cu²⁺, showed any heat upon injection of Mn²⁺ solution in ITC experiments, indicating complete loss of Mn²⁺ affinity as a result of mutating conserved Cu²⁺-residues. The failure of H118A to perform as a functional FtrA is not surprising given the analytical gel filtration, SDS-PAGE, as well as the homology modeling studies all indicated protein degradation, suggesting that H118 is one of the most important conserved Cu²⁺-binding amino acids for integrity of this protein. On the other hand, the analytical gel filtration indicated H65A, E67A, H151A as having similar size to the wild-type protein (Table 1). However, the fact that none of these mutants could bind to Mn²⁺, despite being able to form dimeric structures, can be interpreted as a result of the formation of protein dimers with different conformations than the wild-type protein. This interpretation gains further support from the DSC melting data for these mutants as described in the next sections.

3.6. Wild-type FtrA show two symmetric melting peaks in DSC experiments

We performed DSC experiments on wild-type FtrA, at two experimental pH values to evaluate the effect of protein folding stability as result of metal binding as well as changes in pH. Further, considering protein concentration can influence its ability to form stable dimers, we conducted these wild-type DSC experiments in a range of concentrations (20–70 μM).

Wild-type FtrA, without any added Cu²⁺ or Mn²⁺, displayed two symmetrical melting peaks, implying the presence of two independent folding-unfolding events represented by each peak. Data fitting with a fourth-order polynomial baseline (after buffer heat subtraction) produced ΔH_cal = 321.10 ± 3.87 and 260.22 ± 6.44 kJmol⁻¹ at pH 6.3 and 7.3, respectively. This DSC information is the first evidence showing a greater thermal stability of wild-type FtrA at an acidic pH and might explain the more detrimental effect of ΔftrA mutation reported earlier (39). These baseline corrected peaks were fitted using two Gaussian models yielding melting peaks centered at 74 and 83 °C (pH 7.3); and 75 and 84 °C (pH 6.3). (Table 3). Further, ΔH_cal/ ΔH_vH calculated from these models were equal to 0.99 and 1.10 at pH 6.3 and 7.3, respectively, confirming the two two-state folding unfolding event in the metal-free wild-type FtrA. Additionally, T_m values obtained from these fits of the DSC runs involving different protein concentrations showed negligible variation in the concentration range of our experiments, indicating little to no effect on thermal denaturation as a result of dilution (Table 3). Finally, DSC reheating scans of the wild-type protein exhibited irreversibility, indicating once wild-type FtrA is thermally denatured, it could not fold back to produce the same native structure.

Two symmetrical melting peaks persisted when 5X Cu²⁺ was added into wild-type FtrA at both pH values. Similar data fitting with fourth-order polynomial baselines yielded ΔH_cal = 348.10 ± 3.41 and 323.09 ± 7.50 kJmol⁻¹ at pH 6.3 and 7.3, respectively. When these baseline corrected wild-type FtrA + 5X Cu²⁺ DSC data were fit using two Gaussian models, the protein melting peaks appeared at 79 and 85 °C (pH 6.3) and 76 and at 83 °C (pH 7.3) (Figure 7, Table 3). Comparing these ΔH_cal and T_m values with wild-type FtrA data with no added metal clearly shows the gain in overall folding stability of wild-type FtrA as a result of binding to Cu²⁺. The calculated ΔH_cal/ ΔH_vH obtained from these models were equal to 1.00 and 0.98 at pH 6.3 and 7.3, respectively, once again indicating the presence of two two-state unfolding events in the presence of Cu²⁺. Similar to the data of the wild-type FtrA with no Cu²⁺ added, these DSC melting events also showed irreversibility, indicating, once the Cu²⁺-bound FtrA unfolds, lowering the temperature fails to fold the protein back to its original conformation.

Figure 7.

Representative DSC thermograms for wild-type FtrA + 5X Cu²⁺ at a) pH 6.3 and b) pH 7.3 in 25 mM ACES buffer and indicating similar folding for the wild-type protein in the presence of excess Cu²⁺ at two pH tested. Temperature scans were done between 40–98 °C under 3 atmosphere pressure at a scan rate of 1°C/min. The solid thick lines represent the raw melting data, the dashed lines represent the model sum obtained by fitting the raw data using two Gaussian models, and the thin solid lines represent individual two-state model fits (Table 3).

Finally, single asymmetric peaks (Figure 8, Table 3) under both pH conditions were obtained upon addition of 5X Mn²⁺ to solutions of wild-type FtrA, saturated with Cu²⁺. Similar fourth-order polynomial baseline fit for these data yielded ΔH_cal = 382.10 ± 3.50 and 345.20 ± 2.80 kJmol⁻¹ at pH 6.3 and 7.3, respectively, indicating an increase in energy required to unfold the protein in contact with both Cu²⁺ and Mn²⁺, compared to the metal-free DSC data (Table 3). Although, asymmetry in melting peaks could indicate the existence of non-two-state protein folding mechanisms, the best fit for these baseline corrected raw data were obtained by using two closely spaced independently-folding two-state models appearing at 83, 86 °C (pH 6.3) and 85, 87 °C (pH 7.3) (Figure 8, Table 3). Further reliability of these two closely spaced two-state unfolding events is obtained by the observation that ΔH_cal/ ΔH_vH from these fits are also very close to 1 (Table 3).

Figure 8.

Representative DSC thermograms for wild-type FtrA + 5X Cu²⁺+ Mn²⁺ at a) pH 6.3 and b) pH 7.3 in 25 mM ACES buffer indicating similar folding stability of the wild-type protein in the presence of excess Mn²⁺ and Cu²⁺ at both tested pH’s. Temperature scans were done between 40–98 °C under 3 atmosphere pressure at a scan rate of 1°C/min. The solid thick lines represent the raw melting data, the dashed lines represent the model sum obtained by fitting the raw data using two Gaussian models, and the thin solid lines represent individual two-state model fits (Table 3).

In contrast, DSC thermograms of wild-type FtrA with 5X Mn²⁺ (without any added Cu²⁺), showed no changes in melting temperatures when compared to the data obtained for wild-type FtrA by itself (Figure 6). This observation is a crucial one, as it clearly indicates Cu²⁺ binding to wild-type FtrA produces a stabilized conformational fold that is required for the recognition of Mn²⁺ and can explain the molecular basis of the Cu²⁺-dependent Mn²⁺ affinity for these proteins.

Figure 6.

Representative DSC thermograms for wild-type FtrA at a) pH 6.3 and b) pH 7.3 in 25 mM ACES buffer and indicating pH invariant folding for the wild-type protein. Temperature scans were done between 40–98 °C under 3 atmosphere pressure at a scan rate of 1°C/min. The solid thick lines represent the raw melting data, the dashed lines represent the model sum obtained by fitting the raw data using two Gaussian models, and the thin solid lines represent individual two-state model fits (Table 3).

A previous DSC report on a different protein homodimer interpreted the appearance of two peaks as a result of the dimer first dissociating into monomers (yielding the lower T_m) and then subsequently undergoing denaturation (yielding the higher T_m) (65). Based on the experimental evidence presented above we can conclude that Brucella FtrA, like its homologs, forms homodimers, where we interpret our data as indicative of a similar dimer ⬄ monomer (yielding the lower T_m) and the monomer ⬄ denaturation (yielding the higher T_m) unfolding model. This model gets further support based on the fact that these melting peaks shift to higher temperatures (Table 2 and Figure 7) in the presence of 5X Cu²⁺, as the homology modeling for wild-type FtrA as well as the reported X-ray structures for other P19-type proteins, show the H151 residue to cross link the two monomeric units by coordinating with Cu²⁺, resulting in a more stable protein dimer (46–48). Further, a qualitative comparison of all DSC data for wild-type FtrA, wild-type FtrA with 5X Cu²⁺, and wild-type FtrA with 5X Cu²⁺ + 5X Mn²⁺ indicate a progressive broadening of the lower melting peak (based on the width of the peak at the half-peak-height) representing the dimer ⬄ monomer conversion, and indicating loss of some cooperativity of this process as a result of ligand addition (54). However, more structural experiments are required to be performed to validate and explain this observation.

3.7. FtrA Cu²⁺-binding mutants have altered folding stabilities

Similar DSC experiments were performed on all four Cu²⁺-binding mutants to evaluate the effect of mutation on the folding stability as well as the proposed dimer to monomer to denatured model proposed above for wild-type FtrA.

Raw DSC data for the H151A mutant (at pH 7.3) showed a broad melting peak, which after similar baseline correction yielded ΔH_cal = 113.59 ± 6.70 kJmol⁻¹ indicating a significant loss of folding stability as a result of this mutation, compared to the wild-type data (Table 3). The best fit to this baseline corrected data was obtained by using two Gaussian models, yielding two melting temperatures taking place at 60 and 62 °C, which are much lower temperatures compared to the wild-type data (Figure 9a, Table 3), indicating loss of protein folding stability. Interestingly, a structural alignment of the wild-type FtrA model with the four mutants reported in here showed a structural alteration/disruption for the H151A mutant (Supplementary Figure 3), which can be taken as indirect evidence supporting this DSC loss of folding stability data.

Figure 9.

Representative DSC thermograms for a) H151A and b) H118A mutants at pH 7.3 in 25 mM ACES buffer, showing the effect of mutation of the metal-binding mutant on the folding stability of the protein. All mutants tested in this work show reduced thermal stability compared to the wild-type data. Temperature scans were done between 40–98 °C under 3 atmosphere pressure at a scan rate of 1°C/min. The solid thick lines represent the raw melting data, the dashed lines represent the model sum obtained by fitting the raw data using Gaussian models, and the thin solid lines represent individual two-state model fits (Table 3).

DSC thermograms for H65A and E67A displayed one asymmetrical peak each, with the baseline corrected data showing ΔH_cal = 126.52 ± 3.30 and 59.40 ± 1.65 kJmol⁻¹, respectively, at pH 7.3, once again indicating loss of protein folding stabilities as a result of these mutations. Interestingly, the loss of this overall folding stability for the E67A mutant is observed to have the highest value. Based on multiple sequence alignment of Brucella FtrA with other P19-type proteins, the E67 residue is predicted to take part in both Cu²⁺ and Fe²⁺ sequestration, making it one of the most important metal-binding residues, and as a result, this observation of the loss of largest conformational stability for E67A, from DSC experiments, can be rationalized (39). The best fit to these asymmetric DSC peaks for H65A and E67A were obtained by using two closely spaced Gaussian models yielding melting temperatures of 73 and 80 °C; 78 and 80 °C, and ΔH_cal/ΔH_vH = 0.99 and 1.00, respectively (Table 3). Although these two mutants retained the two unfolding events, similar to the wild-type protein, closer inspection of the positions of the melting peaks confirm that the higher melting peaks for the wild-type protein appeared at significantly lower temperatures for these two mutants, confirming loss of native folding stability. In contrast to the wild-type data, DSC data for these two mutants in the presence of 5X Cu²⁺ were identical to the Cu²⁺-free thermograms. Interestingly, these two mutants showed wild-type affinity for this metal in ITC (Table 2) and analytical gel filtration indicated formation of protein structures, which are similar in size to wild-type FtrA. However, these DSC data indicate that H65A and E67A must coordinate Cu²⁺ in a different environment than is described for the wild-type protein, and as a result does not gain further conformational stability from the addition of Cu²⁺.

Finally, H118A showed three symmetrical melting peaks (Figure 9b, Table 3) arising at 56, 60 and 64 °C, representing three independent unfolding events. Using the observations from analytical gel filtration (Table 1) and gel electrophoresis (Supplementary Figure 1), we attribute these peaks to protein degradation products.

4. Conclusion

Previous biochemical and crystallographic data on other P19-type systems identified three conserved properties for these proteins: their ability to form stable homodimers, the Cu²⁺-dependent iron affinity, and the Cu²⁺ coordination sphere, although no previous work investigated a molecular basis of these properties (46–48). In addition to these P19-type proteins, several other iron uptake systems show similar Cu²⁺ connection, making a detailed molecular level investigation towards such dependence important (66, 67). Thus, in this work using established biochemical protocols we conducted experiments on wild-type and Cu²⁺-binding mutants (H65A, E67A, H118A, and H151A) of Brucella FtrA to investigate this Cu²⁺ dependency for its functionality. Our ITC metal binding assays confirmed that wild-type FtrA is a functional P19-type protein, showing Cu²⁺-dependent μM Mn²⁺ affinity which was absent in all Cu²⁺-binding mutants, irrespective of their ability to sequester Cu²⁺. This is an important conclusion as previous studies on other P19-type proteins suggested the only requirement for these systems to bind iron is to have previously sequestered Cu²⁺ (46–48). Using the results from our DSC and analytical gel filtration data we further conclude that this inability of the Cu²⁺-binding mutants to sequester Mn²⁺ is due to their inability to form a stable and native conformational dimer. This is a crucial finding considering all P19-type proteins are known to form similar dimeric structure and coordinate Cu²⁺ ions in the same His₃Glu environment. The structural homology models for the wild-type protein also sheds light onto the reason for loss of conformational stabilities for at least two of the mutants, H181A and H151A, although direct experimental evidence is required to establish the validity of these predictions. Taken together, our data suggest a dual role for the conserved Cu²⁺-binding residues, H65, E67, H118, and H151: a) producing a stable and native functional (capable of showing Cu²⁺ dependent Mn²⁺ affinity) dimers for Brucella FtrA, b) taking part in directly coordinating to Cu²⁺.

Highlight.

Brucella FtrA protein, protein dimer formation ability, role of Cu2+ in iron uptake, FtrA Cu²⁺ mutants, bacterial Ft1p-Fet3p analog

List of abbreviations

ITC

Isothermal titration calorimetry

DSC

Differential scanning calorimetry

CD

Circular dichroism

ACES

N-(2-acetamido)-2-aminoethanesulfonic acid

SDS-PAGE

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

BLAST

asic local alignment search tool