Simultaneous multiple element detection by particle beam/hollow cathode-optical emission spectroscopy as a tool for metallomic studies: Determinations of metal binding with apo-transferrin

Abstract

Particle beam/hollow cathode-optical emission spectroscopy (PB/HC-OES) is presented as a tool for the determination of metal ion loading in transferrin (Tf). The elemental specificity of optical emission spectroscopy provides a means of assessing metal ion concentrations as well as the relative amounts of metal per unit protein concentration (up to 2 moles of Fe per mole of protein). The PB/HC-OES method allows for the simultaneous detection of metal content (Fe (I) 371.99, Ni (I) 341.41 nm, Zn (I) 213.86 nm, and Ag (I) 338.28 nm in this case), as well as elemental carbon and sulfur (C (I) 156.14 nm and S (I) 180.73 nm) that are reflective of the protein composition and concentration. Quantification for the metal species is based on calibration functions derived from aqueous solutions, with limits of detection for the entire suite being less than 1.0 μM. Determinations in this manner eliminate much of the ambiguity inherent in UV-VIS absorbance determinations of Tf metal binding. Validation of this method is obtained by analyzing loading response of Fe³⁺ into Tf using the PB/HC-OES method and comparing the results with those of the standard UV-VIS absorbance method. Maximum Fe³⁺ loading of Tf (based on the number of available binding sites) was determined to be 71.2 ± 4.7% by the PB/HC-OES method and 67.5 ± 2.5% for the UV-VIS absorbance method. Element emission ratios between the dopant metals and the carbon and sulfur protein constituents allow for concentration independent determinations of metal binding into Tf. Loading percentages were determined for Ni²⁺, Zn²⁺, and Ag⁺ into Tf with maximum loading values of 19.5 ± 0.4%, 41.0 ± 4.4%, and 141.2 ± 4.3%, respectively. While of no apparent biological significance, Ag⁺ presents an interesting case as a surrogate for Pt²⁺, whose binding with Tf has shown to be quite different from the other metals. A different mode from the others is indeed observed, and is consistent with conjecture on the Pt²⁺ mechanisms. Competitive binding studies not easily performed using absorbance spectroscopy are easily performed by simultaneous, multielement analysis, reflective of the metals and the protein content. In this work, there is clear competition between and Fe³⁺ and Zn²⁺ for binding in the C-terminus lobe of Tf, while Ni²⁺ binds within the N-terminus lobe. Addition of Ag⁺ to this mixture does not affect the other metals’ distributions, but reflects binding at other protein sites.

---

Introduction

The determination of how metals interact with proteins within the human body will help to form a better understanding of their roles in drugs (i.e. titanium cancer drugs),¹ diseases (i.e. iron deficiency anaemia),^1–4 and essential cellular activities. Human serum transferrin (Tf) is the iron transportprotein in serum, having a concentration of ∼35 μM.¹ It is responsible for delivering iron to the cell, where it is used for oxygen transport, electron transfer, and DNA synthesis.³ Transferrin consists of approximately 700 amino acids with a molecular weight of ∼80 kDa and has two metal-binding lobes (referred to as N- and C-terminal lobes), both having similar pockets that bind one Fe³⁺ each. Fig. 1 represents the binding pocket of apo-transferrin (i.e., the metal free form) which coordinates iron in a distorted octahedral geometry and consists of one histidine, one aspartic acid, and two tyrosine residues, as well as a synergistic anion (usually carbonate) that acts as a bidentate ligand to complete the octahedral coordination sphere.⁵

Fig. 1 Structure of the open and closed metal binding sites found in the N-and C-lobes of transferrin (Tf).

Upon Fe³⁺ binding, each lobe of the Tf protein undergoes a change from an open conformation to a closed conformation (aka the Venus fly trap).¹Transferrin receptorproteins on the cell surface only bind the completely closed (diferric) form of Tf, where it is taken into the cell by the process of receptor-mediated endocytosis.^1,3 It has been determined that Tf is only 39% Fe³⁺-saturated in serum, meaning that the empty lobes may be available for binding other metals.⁶ The binding strengths of metal ions other than Fe³⁺ to Tf follows a trend similar to the stability constants for hydroxide binding to metal ions, meaning that the more acidic the metal, the stronger it binds to Tf.^1,7,8 The size of the metal ion also affects binding to Tf as well. In addition, NMR studies have shown that other metal ions that bind to Tf display similar spectral shifts to those corresponding to the open-to-closed conformation change that occurs upon Fe³⁺ binding.^9,10 Because of the selectivity of the Tf receptor for only the closed form of Tf, metal ions other than iron may be taken up into the cell if they bind in the empty lobes of Tf.

Tf is known to bind other metals such as Al³⁺, Ni²⁺, Cu²⁺, Ga³⁺, Cr³⁺, and Zn²⁺ (other metal ions may bind Tf that are not included here).¹ Of these metal ions, nickel forms the weakest Tf complex. This is counterbalanced by the fact that nickel exposure is prevalent (electroplating, alloy production and welding)¹¹ and can be considered hazardous. Research has confirmed that nickel interferes with cellular uptake of iron (presumably through binding to Tf) and ultimately decreases ferritin production within the cell.^11,12 By the same token, zinc-rich diets elevate Zn²⁺ concentrations in plasma and compete with Fe³⁺ for binding by Tf.^13,14 Copper and zinc ions have been reported to block Tf binding to the Tf receptor at the cell surface and thus prevent receptor-mediated endocytosis.¹⁵ These examples point to the importance of knowing binding characteristics of various metals with Tf and how they will compete with Fe³⁺ binding.

Detection and quantification of Fe³⁺ bound to Tf is typically accomplished using UV-VIS absorbance by monitoring the change in the molecular band at ca. 470 nm, which is due to ligand-to-metal charge transfer (LMCT) of tyrosine to Fe³⁺.¹ For the binding of other metal ions to Tf, the bands at ∼255 and ∼295 nm, characteristic of metal coordination to the tyrosine groups in the binding pocket of Tf, are monitored.⁴ The concentration of the bound metal is related to the measured absorbance, and so is intimately tied to the efficiency of chromatographic protein recovery commonly used to purify proteins following the binding experiments. These changes in absorbance are not metal ion specific in the way that optical (atomic) emission spectra are. In addition, the binding of multiple (different) metal ions contributes predominately to the overall strength of the bands, with very little spectral specificity. Thus, it is very difficult to detect multiple metal ions bound to Tf using UV-VIS absorbance measurements. In addition, many metal ion-amino acid complexes are spectroscopically inactive, and so their binding cannot be detected via absorbance. Overall, quantification of metal ion uptake in Tf is performed based on the respective peak absorbance values, which must be differentiated from the general molecular background absorbance that is always present.

In order to obtain unequivocal metal ion identification and quantification, atomic spectroscopy methods are required. Inductively coupled plasma-optical emission spectrometry (ICP-OES) and mass spectrometry (ICP-MS) are capable of quantifying and identifying specific metal ions bound to proteins such as Tf.^16–19 Using these techniques, however, it is not possible to determine if the metal is specifically bound to Tf and not just present as free ions in solution. In principle, free and bound metal ions can be separated chromatographically, but this measurement only identifies the metals present in specific fractions, not the identity of those species. However, the basic nature of the ICP source is such that all species in solution are broken down to elemental form, as are remnants of solvent species and buffer components. Therefore, there are no means of assessing the presence of protein as there is a continuous background of carbon, hydrogen, and other elements in the solvent/buffer system under normal ICP operation conditions. Electrospray ionization mass spectrometry (ESI-MS) also can be used to provide direct information that a metal is associated with a specific protein, but quantification of the metal ions is generally poor with this technique.^20–22 In order to obtain a complete, quantitative picture of the metal–protein interaction, ICP and ESI approaches must accompany each other in a complementary fashion. To address this issue of needing both instruments, a dual ICP/ESI-MS has recently been developed for elemental speciation analysis.²³

An ideal spectroscopic tool for assessing metal ion binding to proteins would provide very high elemental specificity, precise quantification with micromolar sensitivities, and some method of confirming the association of the metal(s) with the protein. Other favorable qualities would include the ability to work with small sample quantities, perform simultaneous multielement determinations, and be readily interfaced with liquid chromatography separations employing various solvent systems. In this paper we present the application of the particle beam/hollow cathode-optical emission spectrometry (PB/HC-OES) method as a means of performing metal–protein binding studies, specifically the metals Fe, Ni, Zn, and Ag with Tf. This instrument operates using a particle beam that interfaces a liquid delivery system (e.g. a liquid chromatography pump) to a hollow cathode glow discharge source operating in an inert atmosphere.²⁴ This approach effectively delivers dry analyte particles to the plasma, where thermal energy induces vaporization and dissociation of proteins and plasma excitation permits detection of metals and nonmetal atoms simultaneously. Optical emission detection provides high specificity and sensitivity for metals analysis, while the operation in an inert plasma permits protein determinations based on the constituent nonmetal (e.g., C, H, O, S, N) responses. This latter attribute makes the PB/HC-OES unique and different from ICP as association with a protein can be directly deduced. More specifically, the ability to obtain emission response ratios of metal to nonmetals (e.g., M⁺/C or M⁺/S) allows for direct stoichiometric determinations²⁵ and makes this technique protein-concentration independent.

In this study, the binding of Fe³⁺, Ni²⁺, Zn²⁺ and Ag⁺ with the iron transportprotein transferrin (Tf), both individually and competitively are described. Iron binding determinations were comparable to that of the standard UV-VIS absorbance method and thus confirmed the ability of the PB/HC-OES to determine metal binding within Tf. The loading of transition metals Ni²⁺ and Zn²⁺ was performed to assess the technique for metals that are spectroscopically difficult to discern by the absorbance method. Silver was chosen as a surrogate for Pt, a common therapeutic metal species that is thought to enter the cell via Tf, though bound to the transport protein in a different manner than the other transition metals. It is believed that the characteristics demonstrated here can be used effectively across a broad range of metal–protein interaction studies.

Experimental

Sample preparation and introduction

High-purity (18.2 MΩ-cm) Barnstead Nanopure (Dubuque, IA) water and methanol (EMD Chemicals, Cincinnati, OH) were used as the mobile phase solution. Samples were prepared and stored in 15 mL centrifuge tubes (VWR, West Chester, PA) that had been rinsed with hydrochloric acid to remove any metal constituents prior to sample preparation. Stock solutions of human apo-transferrin (60 μM, Sigma-Aldrich, St. Louis, MO) were prepared in Tris Buffer (20 mM, TEKnova, Hollister, CA) at pH 7.4 and sodium carbonate (20 mM, Sigma-Aldrich). Iron solutions were prepared by adding iron nitrate (45 μM, Sigma-Aldrich) and nitriloacetic acid (90 μM, Sigma-Aldrich) in a 1:2 ratio respectively to hydrochloric acid (0.1 M, J. T. Baker, Phillipsburg, NJ) and adjusting the pH to 4.0.² Nickel solutions were prepared by adding nickel nitrate (1000 μM, Sigma-Aldrich) to nanopure water, silver solution prepared by adding silver nitrate (1000 μM, Sigma-Aldrich) to nanopure water, and zinc solutions were prepared by adding zinc acetate (1000 μM, Sigma-Aldrich) to nanopure water.^2,4,26 Quantities of 0–4 molar equivalents of Fe³⁺,² and 0–10 molar equivalents of Ni²⁺, Zn²⁺, and Ag⁺ based on preliminary work were added to the apo-transferrin solution, followed by incubation at 37 °C of the samples for 24 hrs. The samples were then desalted using a Sephadex™ G-25 M PD-10 column (GE Healthcare, Buckinghamshire, UK) to remove any unbound metal ions from the protein prior to introduction to the PB/HC-OES system or analysis by UV-VIS absorbance, ensuring that only the metal ions bound to Tf are analyzed. Species inter-conversion is always a concern in metal speciation studies, particularly when chromatographic separations occur at non-physiological pH values or under differing mobile phase conditions. All samples were kept in the buffer solutions throughout the experiment to minimize the chance of species inter-conversion prior to the chemical analysis. A Waters (Milford, MA) model 510 high-performance liquid chromatography pump with a six-port Rheodyne 7125 (Rohnert Park, CA) injection valve and a Rheodyne 200 μL injection loop was used for sample delivery of the protein solutions to the PB interface.

Particle beam interface

A Thermabeam (Extrel Corporation, Pittsburgh, PA) particle beam (PB) interface was used to introduce the sample into the hollow cathode glow discharge source. The interface consists of a thermoconcentric nebulizer, a desolvation chamber, and a two-stage momentum separator. Liquid flow from the HPLC enters the nebulizer through a fused-silica capillary (100 μm i.d.) positioned inside of a metal tube that has a DC potential (∼1 V) applied across its length. This potential creates thermal energy (tip temperature ≈105 °C) that, when coupled with a coaxial flow of He gas, creates an aerosol spray from the capillary tip. The aerosol spray enters the heated (∼150 °C) desolvation chamber to further dry the aerosol particles. The particles are then introduced into a two-stage momentum separator that removes low mass molecules (solvent and He gas) leaving only the dry analyte particles (<10 μm size) to reach the HC source volume.^27,28

Hollow cathode glow discharge source

The copper hollow cathode (HC) is machined to have a cylindrical inner diameter of 3.5 mm and 26.5 mm in length with a particle entry hole of 3.5 mm in diameter in the side of the cathode. The hollow cathode is mounted in the center of a “thermoblock” made of stainless steel. Cartridge heaters (Model SC 2515, Scientific Instruments, Ringoes, NJ) are located at the bottom of the block, yielding uniform heating of the HC, which is required to provide sufficient energy to vaporize and atomize the sample. The block temperature (∼210 °C) is monitored with a W-Re thermocouple and the gas pressure monitored with a Pirani-type vacuum gauge (VRC Model 912011, Pittsburgh, PA). Power is provided to the HC by a Bertan power supply (Model 915-1N, Hicksville, NY) that is operated in constant current (60 mA, ∼350 V) mode. Helium (ultra high purity) is used as the discharge gas (2 Torr).

Optical emission spectrometer and data acquisition

A newly-developed coupling of the PB/HC-OES source to a commercial polychromator system has recently been described.²⁹ A JY RF-5000 (Jobin-Yvon, Division of Instruments, Edison, NJ) was used as the detector for the PB/HC system. The JY RF-5000 polychromator includes a 0.5-m Paschen-Runge polychromator with an ion-etched holographic grating with 2400 grooves/mm, with a practical resolution of ∼0.01 nm. Photomultiplier tubes for each of the 26 optical channels sample data at a rate of 2 kHz. The optical path is nitrogen-purged to enable optical emission detection across the range of 110–620 nm. Spectrometer control and data acquisition were obtained using JY Quantum 2000 version 1.1 software on a Hewlett-Packard Vectra VE computer. Data files were then exported and processed as Microsoft Excel (Seattle, WA) files. Reported analyte responses (peak area) are the average of triplicate determinations (i.e. injections) taken using Quantum 2000 software.

pH measurements

All solutions were pH measured with a Accumet Research AR 10 pH meter (Fisher Scientific) and an Accumet double junction Ag/AgCl pH probe (Fisher Scientific). Adjustments to pH were done with hydrochloric acid (6.0 M, Sigma-Aldrich) and sodium hydroxide (2.0 M, Sigma-Aldrich).

UV-VIS absorbance

All absorbance measurements were performed with a Genesys 10-S UV-VIS spectrometer (Thermo Electron Corporation). The concentration of Tf was determined by measuring the absorbance at 280 nm and using an extinction coefficient of 87 200 M⁻¹ cm⁻¹. Concentration of Fe³⁺ loaded into the Tf was determined by measuring the absorbance at 470 nm and using an extinction coefficient of 4860 M⁻¹ cm⁻¹.²

Optical emission analytical responses

Calibration curves were prepared from aqueous standards of 0.1–500 μM Fe³⁺, Ni²⁺, Zn²⁺, and Ag⁺. An obvious question exists in whether or not response functions for (essentially) free ions are valid for what should be protein-bound metals. Studies by Brewer and Marcus in fact showed that the slopes (and resultant limits of detection) for functions derived for iron in the form of FeCl₂, myoglobin, and holotransferrin differed by less than 10%.³⁰ Typical response functions (derived from triplicate 200 μL injections at each concentration), linearity, and limits of detection for the transition metals and the Tf carbon and sulfur constituents for the present study are presented in Table 1. The less than ideal R² values seen here reflect the diminished transport efficiency of the PB interface for low solutes concentrations, which is vastly improved if needed by use of carrier salts.^31,32 In practice, the instrument was calibrated prior to each set of analyses for highest accuracy. The computed detection limits for the metals are “neat”, whereas those given for C and S refer to the molar concentration of protein that can be determined.

Table 1 Typical PB/HC-OES quantification data for metals and non-metals in present binding studies

Element	Response Function	Correlation (R^2)	LODs
Fe (I)	y = 0.0016x − 0.0018	0.9912	0.17 μM
Ni (I)	y = 0.0311x − 0.269	0.9789	0.08 μM
Zn (I)	y = 0.0518x − 0.5682	0.9874	0.51 μM
Ag (I)	y = 0.0185x − 0.7485	0.9833	0.03 μM
C (I)	y = 0.0047x − 0.0082	0.9694	0.20 μM
S (I)	y = 0.0006x + 0.0016	0.9858	1.86 μM

---

Results and discussion

Method validation for iron loading into Tf

A comparison of UV-VIS absorbance and the PB/HC-OES method was carried out to demonstrate the ability of the system to detect the level of Fe³⁺ loading of Tf. As in other works, the percentage loading is based on the assumption that two molar equivalents of iron can be added to each one of transferrin (i.e., 100% loading would be a 2:1 Fe:Tf molar ratio). Fig. 2, shows the response curves for iron loaded into Tf determined by the UV-VIS absorbance and PB/HC-OES methods. Given the fact that the data are obtained using totally different methods, the agreement is quite good. To be clear, the UV-VIS data reflects changes in optical density of an optical band, regardless of what species contribute to that absorbance. On the other hand, the Fe (I) optical emission response simply reflects the amount of iron that is contained in the sample, regardless of its chemical identity. Neither approach can be said to be certain in terms of obtaining the desired information. Fig. 2 implies that as the amount of Fe³⁺ increases, so does the amount of iron that is taken up by Tf. The UV-VIS data suggests that beyond the addition of 3 molar equivalents the loading reaches a maximum, consistent with similar studies. The PB/HC-OES response is more linear with the amount of Fe³⁺ added. The Fe loading (4 equivalents added) was determined to be 67.5 ± 2.5% by the UV-VIS absorbance method and 71.2 ± 4.7% for the PB/HC-OES method; essentially the same value. This demonstrates that the PB/HC-OES method is capable of determining iron loading and provides comparable results to that of UV-VIS. Furthermore, the complementary use of UV-VIS with the PB/HC-OES method further suggests that iron is incorporated within Tf and not just free in solution.

Fig. 2 Comparison of the results found by UV-VIS absorbance and PB/HC-OES methods for determining Fe³⁺ loading in Tf. UV-VIS absorbance of Fe³⁺ loading monitored by change in absorbance at 470 nm. PB/HC-OES operating parameters: Mobile phase = 50:50 water:methanol, nebulizer gas flow rate = ∼1400 mL min⁻¹ He, nebulizer tip temperature = 105 °C, HPLC flow rate = 1.0 mL min⁻¹, 200 μL injection loop, desolvation chamber temperature = 150 °C, hollow cathode block temperature = 210 °C, discharge current = 60 mA, discharge pressure = 2 Torr He.

Ultimately, the desired information in metal loading studies is the molar stoichiometry reflecting the number of incorporated metal ions per molecule of protein. The data presented in Fig. 2, simply reflects species’ concentrations. Previous studies in this laboratory have demonstrated the ability to determine empirical formulae based on the optical emission intensities of constituent elements in amino acids, metallo-organic compounds, and proteins.^25,33–37 Having the ability to detect carbon, sulfur, and iron simultaneously allows for the determination of the stoichiometry of the metal–protein complex. A subtle, yet very important, point is the fact that the loading results based on ratios are concentration independent; the process recovery can change, but the ratio will stay the same). Fig. 3 shows the element emission ratios determined for Fe (I)/S (I) and Fe (I)/C (I) for each of the equivalent additions of Fe³⁺ to apo-transferrin. As would be anticipated, the ratios increase proportionately with the amount of Fe incorporated in the protein. The fact that the carbon- and sulfur- referenced responses track each other implies that either can be used as the internal standard to compute the empirical formula. This concept is illustrated in Fig. 4, wherein the measured Fe (I)/C (I) and Fe (I)/S (I) ratios are plotted as a function of the previously determined percentage of Fe³⁺ incorporated into Tf (Fig. 2). The degree of correlation between the emission ratios and the percentage loading is excellent (R² = 0.9958 and 0.9884). There is greater variability in responses for both elemental ratios for the ∼40 and 75% Fe loading data, which seems to be a reflection of the sample processing rather than that of the measurements themselves, as seen as well in Fig. 3. Having generated these sorts of plots, they can be used in many situations, with the added advantage that they are independent of the actual metal/protein concentrations. The results of all of the binding studies performed here are summarized in Table 2.

Fig. 3 Element emission ratios of Fe (I)/C (I) and Fe (I)/S (I) normalized to show that the curve characteristics are the same for both elements. PB/HC-OES operating parameters are the same as Fig. 2.

Fig. 4 Comparison of experimentally obtained a) Fe (I)/C (I) and b) Fe (I)/S (I) emission intensity ratios to the actual loading percents of Fe³⁺ into Tf. PB/HC-OES operating parameters are the same as Fig. 2.

Table 2 Summary of metal loading of human serum apo-transferrin

Metal	Loading percentage
Fe^3+	71.2 ± 4.7%
Ni^2+	19.5 ± 0.4%
Zn^2+	41.0 ± 4.4%
Ag^+	141.2 ± 4.3%
1st Competition Study
Fe^3+	33.4 ± 6.1%
Ni^2+	14.0 ± 1.3%
Zn^2+	24.9 ± 0.8%
Total M^+	72.3 ± 6.3 %
2nd Competition Study
Fe^3+	24.2 ± 4.8%
Ni^2+	15.3 ± 1.0%
Zn^2+	19.6 ± 5.2%
Ag^+	101.0 ± 12.3%
Total M^+	160.1 ± 14.2%

---

Nickel loading

Previous work determining binding constants of Ni²⁺–Tf complexes using UV-VIS absorbance determinations reported that only about 1 molar equivalent may bind to Tf.⁴ Ni²⁺ is a d⁸transition metal that often prefers square planar geometry. In the formation of a Ni²⁺–Tf complex, Ni²⁺ would have to adopt a distorted octahedral geometry, which would result in a loss of crystal field stabilization energy (CFSE) compared to square planar geometry, and therefore result in a less favorable interaction than that of the Fe³⁺–Tf complex.⁴ Ultimately, studies by Harris indicated that less than one equivalent of Ni²⁺ was incorporated per Tf molecule. Fig. 5 represents the percentage of Tf loaded with Ni²⁺, reaching a maximum value of ∼19.5% loading (based on two potential sites) at a 10-fold molar excess of Ni²⁺. Harris suggested that a large excess (7×) was needed to drive Ni–Tf complex formation,⁴ but the seeming bi-modal addition depicted in Fig. 5 was not mentioned in that work. In fact, a larger excess was employed here, and the second increase in uptake is beyond the excess range used by Harris. The shape of this response might suggest the sequential filling of the two different sites (N- and C-lobes). Also shown in Fig. 5, the shapes of the Ni (I)/C (I) and Ni (I)/S (I) responses mirror those of the loading curve, providing supporting evidence that the addition of Ni²⁺ to the Tf-containing solution does lead to the assumed metal–protein complex, but only to the extent that the molar ratio is ∼0.4:1 (Ni:Tf).

Fig. 5 Percentage of Tf loaded with Ni²⁺ and relative Ni (I)/C (I) and Ni (I)/S (I) emission intensity ratios as a function of molar excess addition. PB/HC-OES operating parameters are the same as Fig. 2.

Zinc loading

The Zn²⁺ ion is similar in size to Ni²⁺ and has very similar hard soft acid base (HSAB) properties, suggesting that they would have similar binding abilities to Tf. According to Harris, the binding constant of Zn²⁺ to Tf is greater than that of Ni²⁺, but less than Fe³⁺.^4,38Fig. 6 presents the loading response for Zn²⁺ in Tf, which appears to reach a plateau equivalent to a loading of ∼41.0% at an excess of 10 molar equivalents. On a molar basis, this is equivalent to a ratio of ∼0.8:1 (Zn:Tf). Zn²⁺ is a d¹⁰transition metal that upon complexation with Tf has no change in CFSE, making the Zn²⁺–Tf complex more stable than that of Ni²⁺.⁴ The element emission ratios of Zn (I)/C (I) and Zn (I)/S (I) show the same general responses as seen with the loading curve.

Fig. 6 Percentage of Tf loaded with Zn²⁺ and relative Zn (I)/C (I) and Zn (I)/S (I) emission intensity ratios as a function of molar excess addition. PB/HC-OES operating parameters are the same as Fig. 2.

Silver loading

Platinum-based drugs are well known in cancer therapies, and so there is question on the possible role that Tf may play in their transport. Platinum–Tf complexes have been studied by Sadler et al. who suggested that there are 3 potential metal binding locations on Tf, all involving methionine (Met) residues; two of which are on the protein surface and one that is accessible within the N-lobe.³⁹NMR studies ruled out binding in N- and C-lobe sites depicted in Fig. 1. Unfortunately, the preset spectrometer configuration does not permit Pt (I) emission monitoring, but there is an Ag (I) channel. The interaction of silver and sulfur is a favorable interaction due to soft character of both elements (HSAB). To date, there have been no literature reports to our knowledge of Ag⁺–Tf binding or the use of Ag⁺ as a surrogate for Pt²⁺. It was hoped that the chemical similarities between Pt²⁺ and Ag⁺ would yield loading responses that might be relevant. The loading curve for silver seen in Fig. 7 shows some interesting characteristics; far different from the other transition metals presented previously. A steady increase in loading is seen up to molar excesses of 10:1 (Ag:Tf), with a maximum loading value of ∼141.2% based on the two-lobe model. In practice, the data reflects a stoichiometry of ∼2.8 Ag:Tf. This is far greater loading than seen for the other metal ions. The Ag⁺ data here seems to strongly support the Pt²⁺ work of Sadler et al.,³⁹ implying that there are two easily-accessible sites, and one of perhaps more hindered access. This latter supposition will be addressed in the discussion of competitive binding.

Fig. 7 Percentage of Tf loaded with Ag⁺ and relative Ag (I)/C (I) and Ag (I)/S (I) emission intensity ratios as a function of molar excess addition. PB/HC-OES operating parameters are the same as Fig. 2.

Competitive metal binding with Tf

Other than the case of Fe³⁺ which produces a somewhat unique response ca. 470 nm when complexed with Tf, the responses seen in the 255 and 295 nm regions upon metal binding within the Tf lobe structure are completely non-specific. The uniqueness of the PB/HC-OES method is that the response of multiple elements can be detected simultaneously. This allows for competition studies to be carried out, with the OES signals being reflective of the amounts of incorporated metal ions. A mixture containing 10 molar equivalent excesses of Fe³⁺, Ni²⁺, and Zn²⁺ (each) was added to the apo–Tf solution and the mixture allowed to incubate at 37 °C for 24 hours (pH = 7.4) to ensure an equilibrium situation was attained (as was the case in the previous loading experiments). Fig. 8 represents the loading values for each metal ion as well as the total loading. Harris has reported that Ni²⁺ binding prefers the N lobe of Tf, and that Fe³⁺ and Zn²⁺ prefer to bind the C lobe.⁴ As indicated in the figure, the Fe³⁺ and Zn²⁺ loading values are reduced by ∼50% in comparison to those of the single metal ion loading studies, suggesting that Fe³⁺ and Zn²⁺ are competing over the C lobe binding sites available. Ni²⁺ loading values were only lowered by 4% in the competitive binding case versus the Ni²⁺-only study, suggesting that there may not be any competition with Ni²⁺ for the N-lobe binding site. Therefore, the general picture taken from this study is the Ni²⁺ fills the N-lobe, to the exclusion of Fe³⁺ and Zn²⁺, whose total loading is limited by not having access to that region of Tf. The 50% suppression of Fe³⁺ binding is consistent to the amounts observed previously.^11,12 The cumulative loading of the transferrin is only ∼72% (based on two—lobe binding), meaning that the protein loading is still not stoichiometric.

Fig. 8 Results of competitive metal binding study of Fe³⁺, Ni²⁺, and Zn²⁺ simultaneously added to Tf (incubation for 24 hrs at 37 °C) (competition study 1) and those of Fe³⁺, Ni²⁺, Zn²⁺, and Ag⁺ simultaneously added to Tf (incubation for 24 hrs at 37 °C) (competition study 2). PB/HC-OES operating parameters are the same as Fig. 2.

A second competitive metal loading experiment was performed to provide better understanding of the Ag⁺ binding mechanism to Tf. In this case, a 10-fold excess of Ag⁺ was added to the mixed transition metal solution prior to incubation with apo–Tf as described above. The relative binding depicted in Fig. 8 is quite striking. First, the loading percentages of the transition metals are statistically the same as the first study. Second, the relative loadings of Fe³⁺, Ni²⁺, and Zn²⁺ are also not affected by the presence of Ag⁺. Finally, the extent of Ag⁺ incorporation is equivalent to 2 atoms per molecule of Tf (100% based on the two-site model), instead of the ∼2.8 atoms in the Ag⁺-only binding study. These data fully support the findings of Sadler et al. as the Ag⁺ occupies the two surface-Met sites, and does not fill the site located within the region of the N-lobe. This is not surprising as the closing of the N-lobe upon filling with the other transition metals makes the third site (which was not completely occupied in the Ag-only case) inaccessible. Keeping in mind that only closed-lobe transferrin is recognized and passed in to the cell by the Tf receptor, these data imply that Pt²⁺ is transported passively with closed-lobe Tf, and does not compete to the exclusion of Fe³⁺ entering the cell.

Conclusion

The PB/HC-OES method has been shown to be a useful tool for metallomics studies, specifically metal binding to proteins. The ability to unambiguously determine the identity and concentration of incorporated metals (including as mixed metal systems) provides far greater information than can be obtained viaUV-VIS absorbance spectrophotometry. The basic method was validated by comparing the PB/HC-OES results with those of UV-VIS experiments, with the two methods yielding values of 71.2 and 67.5%, respectively. The ability to monitor the optical emission of protein constituent elements (C and S here) provides a means better assuring that the metals are indeed bound to the protein, as well as provide direct stoichiometric evidence of incorporation. Ultimately, the ratio method makes the empirical formulae determinations independent of the actual protein concentrations. Detection limits in the low μM to nM range allow for detection of small sample quantities and the evaluation of metals that may have small binding constants with Tf. Ni²⁺, Zn²⁺, and Ag⁺ were individually exposed to Tf solutions and determined to bind at 19.5 ± 0.4%, 41.0 ± 4.4%, and 141.2 ± 4.3% relative to a two-site Tf model, respectively. Ag⁺ loading suggest that another mode of binding is taking place, presumably reflective of the three Met associations seen in the past for Pt²⁺ binding. Competitive metal binding studies with Tf have shown that Ni²⁺ inhibits the binding of Fe³⁺ and Zn²⁺ by ∼50%, reflecting its occupancy of the N-lobe to exclude the access of the others. Near saturation of the two primary binding lobes has little effect on the Ag⁺ binding to surface Met groups, but hinders addition to the buried Met group. It is believed that the PB/HC-OES method provides a level of specificity and sensitivity not available in more typical methods, on a relatively simple instrumental platform.

Acknowledgements

Financial support from Gaia Herbs, Inc. (Brevard, NC) is greatly appreciated. Thanks to Dr K. Christensen at Clemson University for allowing use of the Genesys 10-S UV-VIS spectrometer.

References

1. H. Sun, H. Li and P. J. Sadler, Chem. Rev., 1999, 99, 2817–2842.
2. E. E. Battin, A. Lawhon, Ashley, J. L. Brumaghim and D. H. Hamilton, J. Chem. Educ., 2009, 86, 969–972.
3. A. C. G. Chua and R. M. Graham, Crit. Rev. Clin. Lab. Sci., 2007, 44, 413–459.
4. W. R. Harris, J. Inorg. Biochem., 1986, 27, 41–52.
5. S. J. Lippard and J. M. Berg, in Principles of Bioinorganic Chemistry, University Science Books, Mill Valley, CA, 1994, pp. 141–148.
6. J. Williams and K. Moreton, Biochem. J., 1980, 185, 483–488.
7. H. Li, P. J. Sadler and H. Sun, Eur. J. Biochem., 1996, 242, 387–393.
8. H. Sun, M. C. Cox, H. Li and P. J. Sadler, Struct. Bond., 1997, 88, 71–102.
9. G. Kubal, A. B. Mason, P. J. Sadler, A. Tucker and R. C. Woodworth, Biochem. J., 1992, 285, 711–714.
10. H. Sun, M. C. Cox, H. Li, A. B. Mason, R. C. Woodworth and P. J. Sadler, FEBS Lett., 1998, 422, 315–320.
11. H. Chen, T. Davidson, S. Singleton, M. D. Garrick and M. Costa, Toxicol. Appl. Pharmacol., 2005, 206, 275–287.
12. T. Davidson, H. Chen, M. D. Garrick, G. D'Angelo and M. Costa, Mol. Cell. Biochem., 2005, 279, 157–162.
13. J. J. Fickel, J. H. Freeland-Graves and M. J. Roby, Am. J. Clin. Nutri., 1986, 43, 47–58.
14. T. L. Wright, P. Brissot, W.-L. Ma and R. A. Weisiger, J. Biol. Chem., 1986, 261, 10909–10914.
15. D. D. McAbee and X. Jiang, J. Biol. Chem., 1999, 274, 14750–14758.
16. A. Sanz-Medel, M. Montes-Bayon and M. L. F. Saanchez, Anal. Bioanal. Chem., 2003, 377, 236–247.
17. C. N. Ferrarello, M. R. Fernandez de la Campa and A. Sanz-Medel, Anal. Bioanal. Chem., 2002, 373, 412–421.
18. A. Sarmiento-Gonzalez, J. R. Encinar, A. M. Cantarero-Roldan, J. M. Marchante-Gayon and A. Sanz-Medel, Anal. Chem., 2008, 80, 8702–8711.
19. M. E. D. Busto, M. Montes-Bayon and A. Sanz-Medel, Anal. Chim. Acta, 2009, 634, 1–14.
20. J. Szpunar and R. Lobinski, Anal. Bioanal. Chem., 2002, 373, 404–411.
21. J. Szpunar, R. Lobinski and A. Prange, Appl. Spectrosc., 2003, 57, 102A–112A.
22. J. Szpunar, Analyst, 2005, 130, 442–465.
23. D. A. Rogers, S. J. Ray and G. M. Hieftje, Metallomics, 2009, 1, 67–77.
24. T. M. Brewer, J. Castro and R. K. Marcus, Spectrochim. Acta, Part B, 2006, 61B, 134–149.
25. T. M. Brewer and R. K. Marcus, Anal. Chem., 2007, 79, 2402–2411.
26. T. Kiss, T. Jakusch, D. Hollender, E. A. Enyedy and L. Horvath, J. Inorg. Chem., 2009, 103, 527–535.
27. M. A. Dempster, W. C. Davis, R. K. Marcus and P. R. Cable-Dunlap, J. Anal. At. Spectrom., 2001, 16, 115–121.
28. J. You, M. A. Dempster and R. K. Marcus, J. Anal. At. Spectrom., 1997, 12, 807–815.
29. C. D. Quarles, Jr. and R. K. Marcus, Spectrochim. Acta Part B-At. Spectrosc. , in press.
30. T. M. Brewer and R. K. Marcus, J. Anal. At. Spectrom., 2007, 22, 1067–1075.
31. T. A. Bellar, T. D. Behymer and W. L. Budde, J. Am. Soc. Mass Spectrom., 1990, 1, 92–98.
32. F. X. Jin, J. J. Hixkman, K. Lenghaus and R. K. Marcus, Anal. Bioanal. Chem., 2004, 380, 204–211.
33. T. M. Brewer, B. Fernandez and R. K. Marcus, J. Anal. At. Spectrom., 2005, 20, 924–931.
34. M. A. Dempster and R. K. Marcus, J. Anal. At. Spectrom., 2000, 15, 43–48.
35. F. Jin and R. K. Marcus, J. Anal. At. Spectrom., 2003, 18, 589–595.
36. F. Jin, K. Lenghaus, J. Hickman and R. K. Marcus, Anal. Chem., 2003, 75, 4801–4810.
37. J. You, M. A. Dempster and R. K. Marcus, Anal. Chem., 1997, 69, 3419–3426.
38. W. R. Harris, Biochemistry, 1983, 22, 3920–3926.
39. M. C. Cox, K. J. Barnham, T. A. Frenkiel, J. D. Hoeschele, A. B. Mason, Q.-Y. He, R. C. Woodworth and P. J. Sadler, JBIC, J. Biol. Inorg. Chem., 1999, 4, 621–631.

---