Density Functional Theory Study on Metal-Binding Energies for Human Serum Transferrin-Metal Complexes

Abstract

The absolute values of the metal-binding energies of human serum transferrin (Tf) N-lobe, |ΔE|, were calculated using the density functional theory and were found to increase in magnitude in the following order: Fe(III)>Ga(III)>Al(III)>Cu(II)>Zn(II)>Ni(II). The calculated energies were well correlated with the logarithmic values of the reported metal-binding constants of Tf, which had been experimentally determined, with a correlation coefficient of 0.96. In the estimation of the binding energies, the solvation energies (solvent effect) of free metal ions were a very important factor. The results provide a theoretical explanation for the binding of Fe(III) to Tf, which produces sufficient energy to induce a conformational transition of the Tf molecule, making it possible to interact with Tf receptor 1.

1. Introduction

Serum transferrin (Tf), the iron-transport protein in vertebrates, is a glycosylated single polypeptide with a molecular mass of approximately 80 kDa [1]. It consists of two major lobes, N- and C-lobe, which have about 40% homology between their amino acid sequences [2]. Each lobe is further subdivided into two similarly sized domains (N1 and N2; C1 and C2) and possesses a single iron-binding site in the interdomain cleft (N- and C-site) [2]. Iron bound at each site is octahedrally coordinated by six ligands: two tyrosine residues, one aspartic acid residue, one histidine residue, and a synergistic anion (a bidentate carbonate ion in a physiological condition), of which one oxygen atom is in interaction with η NH of arginine residue [3, 4].

In addition to ferric ions, the iron-binding sites are capable of binding various other metal ions [5, 6]. Among these, ferric ions most tightly bind to Tf, giving a binding constant of 10^20~23 M⁻¹ [5, 6]. Diferric Tf (Fe(III)₂Tf) binds to Tf receptor 1 (TfR1) at the cell surface (pH 7.4) [7–9]. The binding of the Fe(III)₂Tf complex to TfR1 is followed by endocytosis and iron release in the acidic environment (pH 5.6) of the endosome. The iron-free Tf (apoTf) remains bound to the TfR1 in the endosome. This adduct returns to the cell surface, and the apoTf is dissociated from the TfR1 at pH 7.4, which allows the recycling of both the proteins [7].

Many studies have been carried out on the physicochemical properties of Tf-metal complexes. In particular, regarding the stability of a Tf-metal complex, the metal ion radius has been thought to be one of the most important factors [10]. For instance, in the lanthanum series, wherein the elements have very similar chemical properties, the binding stabilities of the metal ions are correlated with their radii, that is, a Tf complex with a lanthanoid ion with a large radius is more unstable than one with a small radius [10]. However, for Fe(III), which is a key metal ion, although the radius of Fe(III) (0.64 Å) is larger than that of Al(III) (0.54 Å) [6], the Tf-Fe(III) complex is much more stable than the Tf-Al(III) complex (binding constant: 10^13~14 M⁻¹ [6]). Thus, the metal ion radius alone is not necessarily the dominant factor in the stability of a Tf-metal complex.

The binding of Fe(III) to Tf induces a great conformation change that interacts strongly with TfR1 [11–13], that is, upon Fe(III)-loading, the N1 and N2, and C1 and C2 domains undergo conformational changes, which bring about a transition of the cleft from the open form of apoTf to the closed form of Fe(III)₂Tf.

In human circulation, Tf is saturated with iron to only approximately 30% [14]. This suggests that there is a possibility of binding various metal ions to Tf in the circulation. Cobalt (III), gallium (III), and bismuth (III) tightly bind to Tf with metal-binding constants of 10^~22 M⁻¹ [15], 10^19~20 M⁻¹ [5, 6], and 10^~19 M⁻¹ [6], respectively. The Tf complexes of these metal ions interact with TfR1 at pH 7.4, but the binding affinities of Co(III)₂Tf, Ga(III)₂Tf, and Bi(III)₂Tf to TfR1 are approximately 11-, 500-, and 2000-fold lower than that of Fe(III)₂Tf, respectively [11]. Al(III)₂Tf, which has a binding constant of 10^13~14 M⁻¹ [6], does not interact with TfR1 [11, 16]. Recent studies [12, 13] by our group showed that the structure of Al(III)₂Tf has an intermediate form between the closed and open forms of the Tf domains, resulting in the absence of the binding of Al(III)₂Tf to TfR1. Thus, the Tf/TfR1-mediated uptake of a biologically toxic element such as Al by cells is prevented. On the other hand, radioisotopes such as ⁶⁷Ga(III) [17] and ⁶⁰Co(III) [18] are able to be used as medical diagnostic and/or therapeutic regents by forming stable complexes with Tf for use against tumor cells, which contain an abundance of TfR1.

Tf complexes of metal ions having high affinities to Tf can bind to TfR1. The binding energy of a metal ion to Tf is thought to contribute to the conformational change in the Tf molecule. Therefore, it is physiologically important to understand the factors responsible for the formation of the Tf-metal complex.

In order to analyze these factors, we calculated the intrinsic binding energies of several metal ions, including Fe(III), using the density functional theory (DFT) by (1) considering in vacuo conditions for the binding site and free metal ions and then (2) taking the solvent effect for the free metal ions in water into consideration. The calculated results were evaluated by investigating their correlation with the reported experimental binding constants [6]. The solvation energy of the metal ion was shown to be an important factor in its affinity to Tf.

2. Methods

The 3D coordinates of Fe(III)-binding Tf N-lobe were obtained from the X-ray data of the crystal structure of iron-binding human N-lobe Tf (PDB ID: 1A8E) [3] and used for the DFT calculation. The metal-binding ligands and their surrounding atomic groups used for the calculation were defined as follows: functional groups of the amino acid residues (Asp-63, Tyr-95, Tyr-188, His-249) and CO₃₂₋, Arg-124 interacting with CO₃₂₋, and two water molecules [3]. We denote the atomic groups in the metal-binding environment in total as the binding site (BS) for this calculation.

The binding energy between the BS and each metal in vacuo or with the solvation effect is defined by the following formula: Δ𝐸(binding energy) = 𝐸(metal-binding BS in vacuo) − [𝐸(BS in vacuo) + 𝐸(free metal ion in vacuo or water)]. We employed Gaussian03W with the B3LYP/6-31G** basis set [19] for the ab initio DFT calculations of the complexes with Fe(III), Ga(III), Al(III), Cu(II), Zn(II), and Ni(II). The solvation energy of each free metal ion was calculated using the DFT method with a metal ion coordinated by six water molecules in the octahedral geometry as a solvation model. Furthermore, IPCM [20] with the B3LYP/6-31G** basis set was used as a reaction field model of the solvation effect for the surrounding water molecules of the solvent model at a solvent dielectric constant of 78.39.

3. Results and Discussion

First, we estimated the Fe(III)-ligand distances by geometry optimization of the Fe(III)-BS complex with an octahedral structure. Figure 1 shows the calculated bond lengths between the Fe(III) and oxygen atoms of the ligand residues, O of Asp-63, O of Tyr-95, and O of Tyr-188, as well as between a nitrogen atom of His-245 and oxygen atoms of CO₃₂₋, that is, a synergistic anion. All these lengths concurred with those obtained from the X-ray crystal data (1.6 Å resolution) [3] within a deviation of only 0.1 Å (Table 1). The results of the present geometrical optimization using only the ligands directly participating in the BS as a model showed that the complex structure was not disturbed by other parts of the protein. Consequently, this metal-binding site model was evaluated to be adequate for the theoretical calculation in this study. The bond lengths between each metal ion and its ligands for the Ga(III)BS, Al(III)BS, Cu(II)BS, Zn(II)BS, and Ni(II)BS complexes are listed in Table 1. For the Zn(II)BS complex, the bond distance between Zn(II) and Tyr-188 (O) was estimated to be 7.04 Å. This may be a five-coordinate complex.

Table 1 

Bond lengths between each metal ion and its ligands (calculation data and experimental data of the crystal structure).

Figure 1 

Optimized structure of Fe(III)-BS complex. The unit of the bond lengths of ligand-Fe(III) is Å.

We then calculated the metal-binding energy employing the BS model defined above. This calculation was carried out for the intrinsic metal-binding energies (Δ𝐸_𝑖) under the conditions of no hydration for the free metal ions and a dielectric constant of 𝜀=1 (in vacuo). The calculated metal-binding energies in vacuo showed a linear relationship with the logarithmic values (log𝐾₁) of the reported metal-binding constants of Tf (Table 2), which had been experimentally determined [6], with a correlation coefficient of 𝛾=0.863. As shown in Figure 2, the order of the absolute values of the calculated binding energies in vacuo (|Δ𝐸_𝑖|) with respect to the metal ions is as follows: Ga(III)>Fe(III)>Al(III)>Cu(II)>Zn(II)>Ni(II). On the other hand, the order of the experimentally obtained log𝐾₁ values is as follows: Fe(III)>Ga(III)>Al(III)>Cu(II)>Zn(II)>Ni(II) [6]. The contradiction between these orders requires the calculation of metal-binding energies under conditions that are similar to the physiological ones. Therefore, we recalculated the binding energies by considering the solvation effect for free metal ions and the dielectric constant of water (𝜀=78.39). These results are also shown in Figure 2. The absolute values of calculated metal-binding energies (real metal-binding energies, Δ𝐸_𝑟) showed the following order: Fe(III)>Ga(III)>Al(III)>Cu(II)>Zn(II)>Ni(II). This order was consistent with that of the experimentally obtained log𝐾₁ values [6]. The correlation between them was improved, with 𝛾=0.958 (Figure 2).

Table 2 

Binding energies and binding constants (experimental data).

Figure 2 

Plot of the absolute value of the calculated binding energy (|Δ𝐸|, kcal/mol) between each metal ion and BS versus the Experimentally obtained log𝐾1 value [6]. The open squares denote the absolute values of the binding energies in vacuo (intrinsic binding energies, |Δ𝐸𝑖|), and the correlation coefficient is 𝛾=0.863. The filled circles denote the absolute values of the binding energies with the solvation effect (real binding energies, |Δ𝐸𝑟|) and 𝛾=0.958.

The solvation energy of Ga(III) in solution was calculated to be −1196 kcal/mol (Table 2). The absolute value of this energy was much larger than that calculated for Fe(III), that is, |−783| kcal/mol. During the binding of a metal ion to the BS, the metal ion has to undergo dehydration. The dehydration energy (minus solvation energy) is supplied from the metal-binding energy in vacuo (Δ𝐸_𝑖). The energy difference, that is, the intrinsic metal-binding energy − solvation energy, is the real binding energy (Δ𝐸_𝑟). The real binding energy of Fe(III) to the BS had the largest absolute value. This binding energy was consumed to induce a conformational transition to the completely closed form of Fe(III)₂Tf from the open form of apoTf. Our group [13] observed the hydrodynamic radii to be 42.6±0.1 Å for apoTf, 38.8±0.2 Å for Al(III)₂Tf, and 37.2±0.3 Å for Fe(III)₂Tf, by using dynamic light scattering. These observations support the results of the present study. Table 2 lists the values of the calculated intrinsic binding energies, solvation energies, real binding energies, and the reported experimental binding constants [6] in the log𝐾 form for the metal ions used in the present study.

In conclusion, we calculated the binding energies of several metals to Tf by DFT using the model shown in Figure 1 and by considering the solvent effects for free metal ions. This provided fairly good results to theoretically explain the binding of Fe(III) to Tf, which produces sufficient energy to induce a conformational change in the Tf molecule, making it possible to interact with TfR1.

Acknowledgment

This work was supported in part by Morioka University Academic grant.