Direct Electron Transfer of Human Hemoglobin Molecules on Glass/Tin-Doped Indium Oxide

Interfacial electron transfer kinetics of the haem (Fe III /Fe II ) group in human hemoglobin molecules were investigated on glass/tin-doped indium oxide electrodes. Factors such as surface roughness, crystallinity, hydrophilicity and partial polarization of the working electrode played an important role to provide a more compatible microenvironment for protein adsorption. Results suggested that direct electron transfer from electrode to haem (Fe III )-H 2 O intermediate is coupled to proton at near physiological pH ( I = 0.035, pH = 7.2).


Introduction
Haem-containing proteins such as hemoglobin (Hb), also spelled haemoglobin, are macromolecules that consist in an assembly of four globular polypeptide chains, tightly associated with a nonprotein haem group by means a complex arrangement folding pattern (α-helix). The haem group consists of an iron atom chelated to a porphyrin ring (cf. Figure 1), which allow to carry the oxygen in the red blood cells to whole body of all vertebrates as well as some invertebrates. Although the iron atom can take any of its oxidation states (Fe II or Fe III ), the ferrihemoglobin (methemoglobin, metHb) (Fe III ) cannot bind oxygen [1]. In adult humans, the most common Hb type is a tetramer well-known as Hb A, consisting of two α and two β subunits noncovalently bound (α 2 β 2 ). These subunits are structurally similar to themselves and about the same molecular size. The total molecular weight of the Hb A is ca. 64 kDa.
The four polypeptide chains are bound to each other by salt bridges, hydrogen bonds, and hydrophobic interaction. While Hb does not function physiologically as an electron transfer carrier, it does undergo oxidation and reduction at the haem group in certain cases in vivo [2]. Therefore, the focused research on its electron transfer process might lead to a more profound understanding of electron flow in biological systems.
Throughout almost half of the century, there has been shown that the direct electrochemistry of haem proteins on bare electrodes is fairly difficult [3]. Arrival to this conclusion may be caused by several factors that were overcame to progress, among them: (a) the extended three-dimensional protein conformation due to strong interaction between the protein and the substrate or the lack of an effective microenvironment for adsorption; (b) the inaccessibility of electron communication between the electroactive center of the protein and the electrode due to misalignment of the redox center of the protein; (c) the adsorption of denatured protein onto electrodes, resulting in a loss of bioactivity; and (d) the unsymmetrical distribution of surface charges on protein molecules. According to point (c), a general problem commonly found with used metal electrodes, such as Au, Ag, Pt, and Hg, is that all of them lead to denaturation and irreversible adsorption of the resulting inactive protein, and they are easily fouled by contaminants, i.e., the water molecules that are normally bound at the electrode/electrolyte interface are easily displaced (cf. Figure 2).
Since the pioneering studies of Rusling and co-workers [4,5] in the 1990s, the most successful electrode materials for haem proteins have been carbon or metal oxides, which bear well-defined natural surface functionalities. Semiconducting metal oxides are often optically transparent across the visible spectrum and thus provide additional possibilities for spectral studies, e.g., fluorescence and Raman spectroscopies. In the last decade, a few studies have been conducted on mammalian's Hb, for example, Topoglidis et al. [6] reported that titanium oxide and tin oxide allow the reduction of bovine metHb without the addition of any promoters and mediators. Later, Ayato et al. [7,8] reported that tin-doped indium oxide can induce the electron transfer of the haem (Fe III /Fe II ) redox center in bovine Hb molecules; they also found that the protein directly adsorbed on the electrode surface was not significantly denatured. More recently, Martinez-Mancera and Hernandez-Lopez [9] reported that thin films of solid solutions like In 2-x Sn x O 3 on flat glass substrates can act as both electron acceptors and electron donors, and can be considered a simple model system for mimicking a charge interface of the physiological-binding domain. Herein, the electron transfer properties of the haem (Fe III /Fe II ) redox center in human Hb molecules were investigated, in vitro, on commercial glass/tin-doped indium oxide (ITO) electrodes. Special emphasis is put in theory of cyclic voltammetry and in the Butler-Volmer model, developed by Laviron, for studying the electron transfer between electrode and protein film, the morphological, structural, and surface properties of the electrode, as well as the influence of the physiological milieu that was conditioned into the three-electrode cell system by means of a phosphate-buffered saline (PBS) solution (0.01 mol L −1 Na 3 PO 4 , 0.015 mol L −1 NaCl, pH 7.2) and T = 25°C. To this chapter, we have added supplementary information. Subsection 2.2.1. A procedure of chromatography in-column, which underlines the importance of preparing and purifying the protein solution. Subsection 2.6. A model of theoretical prediction for determining the point of zero charge of the working electrode.

Chemicals
Human hemoglobin (Product No. H7379, pH iep = 6.87, MW = 64.5 kDa) and phosphate-buffered saline (PBS) packs (0.01 mol L −1 Na 3 PO 4 , 0.015 mol L −1 NaCl, pH 7.2), BupH™ were purchased from Sigma-Aldrich ® and Thermo Scientific ® , respectively, and used without further purification. Sodium dithionite (Na 2 S 2 O 4 ), FW = 174.110 g mol −1 was purchased from J.T. Baker and used X Denaturation Physisorption Di usion Figure 2. Cartoon illustrating the importance of the protein adsorption process whose conformation may become distorted on interaction with a metallic surface leading to denaturation. without further purification. BACKBOND spe TM , Sephadex ® G-25, disposable extraction columns were purchased from J.T. Baker. The concentration of Hb was adjusted to 1 × 10 −4 mol Hb L −1 using the PBS solution.

Characterization of the protein by UV-visible spectroscopy
Absorption spectra of human Hb were measured at λ = 200-1000 nm with an UV-Visible spectrophotometer 101 GBS (Cintra), using the following parameters: step size = 0.16 nm, scan speed = 400 nm/min, slit width (SW) = 2 nm. The concentration of Hb referred above was estimated by this technique using the following absorptivity value: A 540 (1%) = 5.97 cm −1 [10] (cf. Figure 3).

Preparation of reduced hemoglobin from oxidized hemoglobin
Reduced hemoglobin can be prepared from oxidized hemoglobin in accordance to the work reported by Dixon and McIntosh [11] with modifications. Briefly, the procedure is as follows:

Electrochemical measurement system
Glass/tin-doped indium oxide (ITO) substrates were purchased from TIRF Technologies, Inc. The ITO film surface was cleaned according to the following standard procedure [15]: immersion for 15 min each in a series of ultrasonically agitated solvents (acetone, ethanol, water) then for 15 min each in ultrasonically agitated: (a) 2.0% (v/v) phosphate-free detergent solution Hellmanex (Hellma TM ; sonification apparatus Super RK510, Sonorex), (b) deionized water type I and (c) ethanol at room temperature. In between the sonification steps, the samples were rinsed in deionized water type I. Finally, the substrates were dried in a stream of nitrogen gas (Praxair, 99.999%) until further processing. The electrode potential was controlled with a potentiostat-galvanostat EW-4960 (Epsilon TM BASi) using a conventional three-electrode cell system supported onto a module C3 (BASi). The latter is coupled to a PC/Processor Intel ® Celeron, 3.06 GHz. A glass/ITO substrate (A g = 1.15 cm 2 ) was used as working electrode. A straight platinum wire (A g = 0.79 mm 2 ) and an electrode of Ag|AgCl in 3 M NaCl solution (E 0 ′ = 0.209 V vs. SHE at 25°C) were used as counter and reference electrodes, respectively.
The cell system was thermostated at 25 ± 0.1°C. Prior to voltammetry, the Hb solution was purged with nitrogen gas (Praxair, 99.999%) for at least 30 minutes; then, a nitrogen atmosphere was maintained over the solution during experiments.

SEM and surface roughness analysis
Scanning electron microscopy (SEM) micrographs were taken with a scanning electron microscope JSM-6510LV (JEOL) operated at an accelerating voltage of 15 KV. The superficial characterization of the electrode's roughness was carried out by means of a surface roughness tester HANDYSURF E-35A (TSK/Carl Zeiss ® ) and performing the norm ASME B46.1-2009 Standard.

XRD analysis
The structural characterization was determined by X-ray powder diffraction (XRD) using a diffractometer D8 Advanced (Bruker AXS), using the following parameters: U = 40 kV, I = 35 mA, Ni-filter, and Cu-Kα radiation, λ = 1.54 Å. A background diffractogram was subtracted using a glass/tin-doped indium oxide slide as blank. For qualitative analysis, XRD diffractograms were recorded in the interval 10° ≤ 2θ ≥ 80° at a scan speed of 2°/min.

Theoretical prediction of the point of zero charge of a glass/ITO electrode
The point of zero charge (PZC) of simple metal oxides can be predicted using an electrostatic model, which takes into account the surface charges originating from the dissociation of amphoteric surface M-OH groups and adsorption of the hydrolysis products of M z+ (OH) z− [16]. In this model, a theoretical value of the PZC can be obtained for a given metal oxide by the following equation with where z is the ionic charge of species indicated by the subscript, i.e., O = O 2 − and M = cation, r is the ionic radius (r O = 0.141 nm, r M = 0.071 nm and 0.081 nm for Sn 4+ and In 3+ , respectively), CFSE is a correction factor called crystal field stabilization energy and the constants A and B are parameters that depend on the coordination number of the cation. By virtue of that Sn 4+ and In 3+ occupy octahedral interstitial sites in SnO 2 and In 2 O 3 , the coordination number for these ionic species is 6. Assuming that CFSE is zero in these calculations, A = 18.6 and B = 0 [16]. The predicted PZC values obtained for SnO 2 and In 2 O 3 are then pH pzc (SnO 2 ) = 5.93 and pH pzc (In 2 O 3 ) = 9.37 [16].
For solid solutions such as ITO, the pH pzc can be considered as a mixture of the constitutive simple oxides and can be calculated by the following equation [17] p H pzc = Σ k s k p H pzc,k where s k represents the molar fraction of each constituting oxide at the surface. It can be defined by where x k is the usual volumetric molar fraction of each constituting oxide.

Morphological and surface roughness analysis
The surface morphology of a pretreated glass/ITO electrode was investigated using SEM (cf. Figure 4). The micrograph, taken at a 500 μm scale, shows a layer of ITO well defined whose thickness was ca. 90 μm. Besides, it is possible to observe a regular, uniform, and flat electrodic surface.
Additionally, the average roughness value obtained in this case was 0.017 μm, which is comparatively lower than the average roughness value obtained for a common glass slide (0.024 μm).

Structural analysis
The structural characterization of a pretreated glass/ITO electrode was investigated using XRD. Figure 5 shows the XRD pattern of a glass/ITO substrate used like a working electrode.   An estimate of the mean crystallite or grain size for a given orientation was determined by using Scherrer's formula [19]: where D hkl is the crystallite size (nm), K is a constant (shape factor, about 0.90), λ is the X-ray wavelength (1.54 Å as mentioned before), β hkl = Δ(2θ) denotes the full width at half maximum (FWHM) or broadening of the diffraction peak (degree), and θ is the diffraction angle (degree). The average D hkl was estimated to be approximately D 222 = 17.9 nm for 2θ = 30.566°. It is worth mentioning that the calculated lattice constant a for the glass/ITO substrate using Bragg's equation was a = 1.01234 nm, which coincides with the reported value in the standard card.

Electroactive surface area determination
By measuring the peak current in cyclic voltammograms (CVs), the electroactive surface area of a pretreated glass/ITO electrode was determined according to the Randles-Ševčik equation for a reversible electrochemical process under diffusive control: where I pa is the anodic peak current (A), n is the number of electrons transferred in the redox reaction, F is Faraday's constant (96,485 C mol −1 of electrons), A e is the electroactive surface area of the electrode (cm 2 ), C OX is the bulk concentration of an oxidant molecule in the solution (mol cm −3 ), ν is the scan rate (V s −1 ), D is the diffusion coefficient of the oxidant molecule in solution, (6.50 ± 0.02) × 10 −6 cm 2 s −1 , for hexacyanoferrate (II) in 0.1 mol L −1 KCl as supporting electrolyte at 25°C [20], R is the gas universal constant (8.314 J K −1 mol −1 ), and T is the absolute temperature (K).

Electrochemical behavior of human hemoglobin molecules
The most popular methods for studying redox enzyme or protein electrochemistry are those based on controlled potential techniques: linear sweep voltammetry (LSV), square wave voltammetry (SWV), and cyclic voltammetry (CV). In the latter, the scan rate, defined as ν = ΔE⁄=Δt, can be varied from less than 1 × 10 −3 V s −1 to 1 × 10 6 V s −1 or more, offering a practical timescale window from minutes to microseconds, which makes to this technique very suitable to study interfacial electron transfer kinetics.
Consider the following hypothetical reversible electrochemical reaction: Ox + ne− ⇌ Red, the interconversion given between oxidized (Ox) and reduced (Red) forms of the protein are fast on the timescale of the voltammogram, as controlled by scan rate. Ideal, reversible voltammograms from a monolayer of electroactive protein on an electrode for a simple electron transfer reaction as it was written above are similar to those of any ultrathin electroactive film. A protein-film voltammetry approach is described in more detail in the following section of this chapter. Figure 6 shows the CVs of glass/ITO electrodes in absence and presence of Hb molecules. In Figure 6a, a CV recorded in PBS solution alone shows a non-Faradaic current behavior. The electrode had the largest background current in the nonelectrolyte solution which reflected the properties of the electric double layer. The double layer capacitance (C dl ) can be estimated by dividing the sum of the anodic and cathodic current with twice the scan rate, i.e., C dl = (I pa + I pc )/2ν [21]. So, the capacitances of the glass/ITO and glass/ITO/Hb electrodes were calculated from Figure 6a and b as 4.8 and 0.4 μF cm −2 , respectively. In Figure 6b, a pair of redox peaks, at around −0.117 V in the cathodic scan and at around −0.097 V in the anodic scan, were found in that Hb-containing solution. This fact indicated that a Faradic current was generated over the glass/ITO electrode, which can be adscribed to the haem (Fe III /Fe II ) redox center in Hb molecules.

Cyclic voltammetry of thin protein films
Once capacitive effects are counted out, the amount of electrochemically active Hb molecules could be estimated from integration of the charge Q (in C) under each peak in those CVs acquired at slow scan rates (i.e., 0.1, 0.3, or 0.5 V s −1 ), given by Faraday's law: where Γ T is total surface concentration of the protein molecule (mol cm The total surface concentration of electroactive Hb molecules was estimated to be Γ T = (4.69 ± 0.52) × 10 −12 mol cm −2 . On the other hand, the theoretical maximum coverage of a protein monolayer on the electrode surface was estimated as 4.89 × 10 −12 mol cm −2 , considering that one human Hb molecule in PBS solution has a Stokes radius of 31.3 Å [22]. These data indicate that θ = (Γ T /Γ T,theo ) = 0.96 of a protein monolayer was achieved.

Figure 7a
shows CVs recorded with different scan rates, from 0.1 to 3.5 V s −1 . Nearly symmetric anodic and cathodic peaks were observed; in addition, they have roughly equal heights. The anodic to the cathodic peak potential difference (ΔE p ) was much greater than the ideal value of zero. At the lower scan rates, i.e., 0.1, 0.3, or 0.5 V s −1 , the smaller redox peak currents were observed, while CVs with the largest redox peak currents corresponded to those acquired at the fastest scan rate, i.e., 3.5 V s −1 .
In such cases, the formal potential E 0 ′ was taken as the midpoint potential between the oxidation and reduction peaks if there is a small separation between them. Considering this criterium, an E 0 ′ = −0.107 V (vs. Ag|AgCl in 3 M NaCl solution) was determined at 0.5 V s −1 . On the other hand, the anodic and cathodic peak currents, I pa and I pc , increased with increasing scan rates as observed in Figure 7b.
These results are characteristic of quasireversible, surface confined electrochemical behavior, in which all electroactive proteins in their haem (Fe III ) forms are reduced on the forward cathodic scan, and the reduced proteins in their haem (Fe II ) forms are then fully oxidized to the haem (Fe III ) forms on the reversed anodic scan.
When the peak currents were plotted against the scan rate, direct linear relationships were obtained, indicating a surface-controlled electrode process. The origin of this process is indicative that the diffusion of H 3 O + ions toward the electrode surface was very fast. Therefore, the electron process can be expressed as proposed in the redox reaction before [23]. The linear regression equations for anodic and cathodic peak currents are as follows: I pa = 2.0305 × 10 −6 ν -1.3809 × 10 −7 , R 2 = 0.9987; -I pc = 1.7896ν -3.0310 × 10 −7 , R 2 = 0.9985.
Linear plots of I p vs. ν were in good agreement with the following equation [24]: However, their width at half height is nearly 200 mV, much larger than the ideal 90.6/n mV at 25°C.
Broadening or narrowing of CV peaks compared to the ideal 90.6/n mV at 25°C suggests a breakdown of the ideal model assumptions of no interactions between redox sites that all have the same E 0 ′. Representative examples arose from studying cytochrome c and myoglobin on Au/alkanethiolate SAMs and OPG/LC surfactants, respectively. Some authors have modeled protein films by LSV, SWV, and CV techniques considering the concepts of spatial distribution of the redox centers, dispersion models of formal potentials (E 0 ′), and electron transfer rate constants to account for the peak broadening [25][26][27]. Other factors, e.g., lack of refinement of mathematical algorithms for the extraction of rate constants or improvements to the goodness of fit of SWV data to pulse heights >50 mV, including counterion transport efficiency, could also influence peak widths, but have not been investigated in detail for protein films.
At scan rates <0.5 V s −1 , ΔE p was nearly constant in the films. As the scan rate increased, the peak potentials shifted negatively (cf. Figure 8). This is consistent with the onset of limiting kinetic effects as scan rates increase.  An increasing ΔE p as the scan rate is increased for an electroactive thin film suggests kinetic limitations of the electrochemistry [28] and is consistent with predictions of the Butler-Volmer model for electron transfer between an electrode and redox sites in a thin film on an electrode. Possible causes could be attributed to: (a) slow electron transfer between electrode and redox centers, (b) slow transport of charge within the film limited by electron or counterion transport, (c) uncompensated voltage drop within the film, and (d) structural reorganization of the protein accompanying the redox reactions.
When nΔE p < 200 mV, the surface electron transfer rate constant (k s ) of the adsorbed Hb on the glass/ITO electrode can be estimated according to Laviron's equation for quasi-reversible thin-layer electrochemistry [29]: Our experimental results showed that the scan rate in the range 0.1-3.5 V s −1 did not affect the k s value, because nΔE p < 200 mV. Assuming a charge-transfer coefficient α of 0.5, the k s of the adsorbed Hb thin film on the glass/ITO electrode was 8.01 s −1 at the onset of limiting kinetic effects (500 mV s −1 ). This value is significantly higher than other previously reported values in the literature for different hemoglobin species and electrode materials (cf. Table 1). For comparison with data on bare or mediator-coated electrodes, k s was converted to the standard heterogeneous rate constant (k 0 ′) by using k 0 ′= k s d, where d is the film thickness [29]. This fact could be attributed to the morphological and structural properties shown by the electrode, i.e., surface roughness, crystallinity, and hydrophilicity [30], as well as the influence of the physiological milieu that was conditioned into the three-electrode cell system, charging positively/negatively to the working electrode (cf. Section 2.6 and Ref. [31]) and negatively to the protein [32]. All these factors played an important role in providing a more favorable microenvironment for the protein.
As indicated in Section 1, Introduction, to facilitate the electron communication between the prosthetic group of haem proteins and an electrode due to misalignment of the haem (Fe III /Fe II ) redox center is a difficult task but could help to advance understanding on biological electron transfer. Techniques such as FT-IR spectroscopy, NMR, ESR anisotropy, polarized reflectance FT-IR, circular dichroism, and calorimetry could yield detailed information on the secondary structure of the protein in regards to its redox state as well as to give some notion of order and specific orientation.

Conclusions
In this study, we clearly demonstrated that human Hb molecules directly physisorbed on glass/tin-doped indium oxide substrates exhibited direct electron transfer (DET) in PBS (0.01 mol L −1 Na 3 PO 4 , 0.015 mol L −1 NaCl, pH = 7.2) solution and T = 25°C.
The experimental results suggest that acid-base equilibria and the water molecule coordinated to the haem group as the sixth ligand might play an important role in the electron transfer process between human hemoglobin and very crystalline and hydrophilic tin-doped indium oxide electrodes.