Effects of in Vitro Glycation on Fe Binding and Fe ... - Semantic Scholar

Papers in Press. First published July 1, 2004 as doi:10.1373/clinchem.2004.033811

Clinical Chemistry 50:9 000 – 000 (2004)

Endocrinology and Metabolism

Effects of in Vitro Glycation on Fe3⫹ Binding and Fe3⫹ Isoforms of Transferrin Ann Van Campenhout,1 Christel Van Campenhout,2 Albert Rene´ Lagrou,3 and Begon˜a Manuel-y-Keenoy1* at 64% theoretical Fe3ⴙ saturation [27 (0.7)% vs 23 (1.1)% of the Fe1C-Tf isoform; P ⴝ 0.009]. Conclusions: Glycation impairs Fe3ⴙ binding and affects Fe3ⴙ-Tf isoform distribution depending on concentration. The diagnostic implications of these results need further elucidation in clinical studies.

Background: In diabetes, protein function is altered by glycation, but the impact on the Fe3ⴙ binding and antioxidant functions of transferrin (Tf) is largely unknown. The aim of the present study was to investigate the effects of glycation on the distribution of Fe3ⴙ on the two Fe3ⴙ-binding sites of Tf. Methods: In vitro glycation of Tf was accomplished by preincubation with glucose for 14 days. Tf was loaded with Fe3ⴙ compounds to achieve theoretical Tf Fe3ⴙ saturations of 32%, 64%, and 96% (monitored by spectrophotometry). Fe3ⴙ-Tf isoforms were separated by isoelectric focusing. Results: Fe3ⴙ binding was highest when Tf was incubated with Fe:nitrilotriacetic acid and reached a steady state overnight. Increasing the Fe3ⴙ load led to a shift of isoform profile toward the diferric form (Fe2-Tf): in freshly prepared Tf, Fe2-Tf represented 6.3%, 30%, and 66% of all isoforms at 32%, 64%, and 96% theoretical Fe3ⴙ saturation, respectively. Fe3ⴙ was equally distributed to the monoferric Tf forms with Fe3ⴙ bound to the amino (Fe1N-Tf) and carboxy termini (Fe1C-Tf). Glycation decreased binding of Fe3ⴙ to Tf (monitored at 450 nm). At low theoretical Fe3ⴙ saturation (32%), glycation increased the mean (SD) proportion of Fe2-Tf by 18 (3)% in the presence of 33.3 mmol/L glucose vs 12 (4)% with 0 mmol/L glucose (P ⴝ 0.01). In contrast, at 96% theoretical Fe3ⴙ saturation, Fe2-Tf decreased linearly with increasing glycation (r ⴝ 0.97; P ⴝ 0.008). Preincubation, independent of glycation, favored the Fe1N-Tf isoform

© 2004 American Association for Clinical Chemistry

Transferrin (Tf)4 is the most important extracellular iron transport protein in humans. This glycoprotein consists of a single polypeptide chain with two homologous lobes: the N-terminal (residues 1–336) and the C-terminal lobe (residues 337– 679). The latter contains two N-linked oligosaccharide chains of complex structure that differ in the number of sialic acid residues (0 – 8). Because each domain contains a Fe3⫹-binding site, four isoforms of Tf can be distinguished depending on their iron content: no Fe3⫹ bound (Fe0-Tf or apoTf), one Fe3⫹ ion bound to the N-terminal lobe (Fe1N-Tf), one Fe3⫹ ion bound to the C-terminal lobe (Fe1C-Tf), and both binding sites occupied (Fe2-Tf) (1 ). Differences in the iron and sialic acid content of the protein affect the pI of the molecule: binding of each Fe3⫹ ion and sialic acid residue leads to pI decreases of 0.2 and 0.1 pH units, respectively (2, 3 ). These differences in pI allow separation of the different Tf isoforms by isoelectric focusing (IEF) (4 ). In addition to its role in regulating the iron fluxes between the sites of absorption, storage, and utilization, Tf has also been attributed a very important antioxidant function in plasma. Free Fe2⫹ is very reactive and capable of producing free radicals that cause oxidative damage to biomolecules (5 ). Typically, no free Fe2⫹ is present because all iron is bound to Tf in a redox-inactive Fe3⫹ form. The exact ways in which modifications or abnormalities of

1 Laboratory of Endocrinology, Antwerp Metabolic Research Unit, University of Antwerp, Wilrijk, Belgium. 2 Laboratory of Immunology and Protein Chemistry, University Hospital of Antwerp, Edegem, Belgium. 3 Laboratory of Human Biochemistry, Department of Pharmaceutical Sciences, University of Antwerp, Antwerp, Belgium. *Address correspondence to this author at: Laboratory of Endocrinology, Antwerp Metabolic Research Unit, University of Antwerp T 4.37, Universiteitsplein 1, B-2610 Wilrijk, Belgium. Fax 32-3-820-2574; e-mail begona. [email protected]. Received March 9, 2004; accepted June 3, 2004. Previously published online at DOI: 10.1373/clinchem.2004.033811

4 Nonstandard abbreviations: Tf, transferrin; Fe0-Tf/apoTf, transferrin with no Fe3⫹ ions bound; Fe1N-Tf, monoferric transferrin with Fe3⫹ bound to the amino-terminal lobe; Fe1C-Tf, monoferric transferrin with Fe3⫹ bound to the carboxy-terminal lobe; Fe2-Tf, diferric transferrin; IEF, isoelectric focusing; NTA, nitrilotriacetic acid; TIBC, total iron-binding capacity; and NA, neuraminidase.

1 Copyright © 2004 by The American Association for Clinical Chemistry

2

Van Campenhout et al.: Glycation and Tf Isoforms

Tf affect its Fe3⫹-binding and antioxidant capacity are not yet fully elucidated. In diabetes mellitus, protein structure and function are significantly affected by glycation. The aldehyde group of glucose reacts nonenzymatically with amino groups of proteins to form Schiff’s bases, which undergo Amadori rearrangements, forming fructosamines. These modified proteins degrade slowly and irreversibly to advanced glycated end products (6 ). Because diabetes is also associated with both an increase in oxidative stress (7, 8 ) and disturbances in iron metabolism (9 –11 ), it is of great interest to investigate the hypothesis that glycation of Tf can contribute to oxidative stress by impairment of its antioxidant function. In a recent study we showed that in vitro glycation of Tf decreases its Fe3⫹-binding capacity and makes it less effective in protecting against in vitro lipid peroxidation (12 ). Therefore, in the present study we aimed to gain more insight into the mechanisms underlying the effects of glycation on the Fe3⫹-binding capacity of Tf. For this purpose, we investigated the distribution of Fe3⫹ between the two binding sites by studying the IEF patterns of in-vitro-glycated Tf.

Materials and Methods in vitro glycation of Tf

Human apoTf (purity ⱖ97%, iron free; cat. no. T2036; Sigma-Aldrich) at a concentration of 5 g/L was dissolved in sodium phosphate buffer (0.1 mol/L, pH 7.4) containing different concentrations of d-glucose (0, 5.6, 13.9, 22.2, 33.3, or 1000 mmol/L) and incubated for 14 days at 37 °C under sterile conditions. Thereafter, the remaining free glucose was removed by passing the reaction mixtures through a Sephadex G-25 column (PD-10; Amersham Bioscience) equilibrated with sodium phosphate buffer. The protein concentration in each of the collected fractions was determined by the Bradford method (13 ) with weighed solutions of apoTf as calibrators. Aliquots of 120 ␮L (3 g/L) were stored at ⫺70 °C until further use. Tf integrity after storage was verified nephelometrically (BN II Nephelometer; Dade Behring) with specific antibodies against Tf (OSAX 15) and calibrated against a commercial standard (CRM 470, IFCC-validated). We determined the extent of glycation by measuring the fructosamine concentration with the nitroblue tetrazolium colorimetric assay (assay A11A00350, Cobas Mira; ABX Diagnostics). This method is based on the reducing ability of fructosamines in alkaline solution. At 37 °C the sample is added to carbonate buffer (pH 10.35) containing nitroblue tetrazolium, which is subsequently reduced to formazan. This causes an increase in the absorbance at 550 nm that is measured spectrophotometrically between 10 and 15 min.

bicarbonate (final concentration, 30 mmol/L) and 10 ␮L of different concentrations of freshly prepared FeCl3, iron citrate, or Fe:nitrilotriacetic acid [Fe:NTA, molar ratio 1:7, prepared by adding equal volumes of 10 mmol/L ferrous ammonium sulfate and 70 mmol/L NTA (adjusted to pH 7.0) according to the specifications of Breuer and Cabantchik (14 )]. The pH of the reaction mixture was stable at 7.4. Final Fe3⫹ concentrations of 0, 20, 40, 60, 150, 300, 600, and 1000 ␮mol/L theoretically achieved Tf saturations of 0%, 32%, 64%, 96%, 240%, 480%, 960%, and 1600% as calculated by the formula: Tf saturation ⫽

iron concentration TIBC

Where the total iron-binding capacity (TIBC) ⫽ 25 ⫻ the Tf concentration, in g/L (15 ). The time course of the formation of the Tf-Fe complexes was monitored spectrophotometrically either in cuvettes at 470 nm (UV2 Series; Unicam) or in 96-well plates at 450 nm (ELX808 Ultra Microplate Reader; BioTek Instruments Inc.). To avoid interference in the IEF band pattern of Fe3⫹-Tf isoforms, we removed sialic acid residues enzymatically by adding neuraminidase (NA; Clostridium perfringens; Beckman Coulter Inc.). Varying the incubation time revealed that the most efficient removal occurred after overnight incubation of apoTf with a concentration of 0.1 U NA/mg of Tf. This was illustrated by shifts in the bands toward the cathode (higher pI) with increasing duration of incubation (Fig. 1). IEF of the different Tf solutions was performed on the PhastSystemTM with PhastGel IEF pH 5– 8 (Pharmacia LKB). The PhastSystem Separation Technique File No 100 (from the User’s Manual) was used with slight modification of the standard program according to van Noort et al. (16 ): Sample applicator down at 1.2 0 V 䡠 h Sample applicator up at 1.3 0 V 䡠 h Extra alarm to sound at 1.1 450 V 䡠 h Sep 1.1: 2000 V, 2.0 mA, 3.5 W at 15 °C for 585 V 䡠 h Sep 1.2: 200 V, 2.0 mA, 3.5 W at 15 °C for 15 V 䡠 h Sep 1.3: 2000 V, 5.0 mA, 3.5 W at 15 °C for 450 V 䡠 h Eight samples were applied per gel with an applicator that was placed at the cathodic end of the gel between the extra alarm and Sep 1.2. Immediately after focusing, the gels were fixed for 10 min in 100 g/L trichloroacetic acid

Fe3⫹ binding and ief of Fe3⫹-Tf isoforms To identify the conditions for efficient Fe3⫹ binding, we incubated 100 ␮L of apoTf in sodium phosphate buffer, pH 7.4 (final concentration, 2.5 g/L) with 10 ␮L of sodium

Fig. 1. Effect of duration of incubation with NA. ApoTf was incubated with 0.1U of NA/mg of apoTf at 37 °C overnight (lane 1) or for 4 h (lane 2), 2 h (lane 3), or 1 h (lane 4).

3

Clinical Chemistry 50, No. 9, 2004

and stained overnight by incubation with EZBlueTM Gel Staining Reagent (cat. no. G1041; Sigma-Aldrich) and washed in distilled water. The isoforms were identified by measuring the distance of each band from the cathode and calculating the pI from a calibration line that was generated with pI markers (calibration set pH 5.20 to 7.35; Amersham Pharmacia Biotech UK Limited). On the basis of our the knowledge that binding of each Fe3⫹ causes a decrease in pH units (⬃0.2) double that caused by each sialic acid residue (⬃0.1) and according to the diagrammatic representation of de Jong and van Eijk (17 ), we assigned each pI value a isoform identification as shown in Fig. 2. Digital images of the Phastgels were obtained by scanning with HP Scanjet 4P, and the relative distribution of the different isoform bands was assessed with use of UN-SCAN-IT gelTM digitizing software for Windows (Ver. 5.1; Silk Scientific Inc). The area enclosed by each peak was given as total pixel intensity and expressed as percentage of the sum of all peaks. This method demonstrated the acceptable reproducibility and stability of the frozen Tf samples. Quantification of the isoform bands was linear at Tf concentrations of 0.187–1 g/L (16 ).

statistical methods Results are expressed as the mean (SD). The statistical significance of the differences was evaluated by paired and unpaired Student t-tests for two-group comparisons, one-way ANOVA, and Pearson correlation coefficients to compare the different glycation conditions (Excel software). Two-tailed P values ⬍0.05 were considered significant.

Results Preliminary experiments were conducted to find the optimum conditions of Fe3⫹ binding before we separated the Fe3⫹-Tf isoforms by IEF.

effect of the type and concentration of Fe3⫹ compound on Fe3⫹ binding In a first series of experiments, we investigated the optimum conditions for Fe3⫹ binding. Various Fe3⫹ compounds at different concentrations were incubated with fresh apoTf up to 18 h, and formation of the iron-Tf complex was monitored at 470 nm. Binding was highest in the case of Fe:NTA. For example, after 18 h, the absorbance was 0.041, 0.055 and 0.063 arbitrary absorbance units for FeCl3, iron citrate, and Fe:NTA, respectively, at 32% theoretical Fe3⫹ saturation. At 96% theoretical Fe3⫹ saturation, these values were 0.117, 0.103, and 0.173 absorbance units, respectively. After overnight incubation of apoTf with these different Fe3⫹ compounds, the samples were subjected to IEF to separate the different Fe3⫹-Tf isoforms. As illustrated in Fig. 3, there was a shift of the bands toward the anode with increasing Fe3⫹ concentrations. This shift from apoTf (cathodal position) toward Tf with one or two Fe3⫹ ions bound (more anodal position) was observed for the three Fe3⫹ compounds but was more pronounced for Fe:NTA. Fe2-Tf represented 6 (6)%, 30 (13)%, and 66 (27)% at 32%, 64%, and 96% theoretical Fe3⫹ saturation. The isoform proportion in the N-band was not different from that in the C-band: respectively, 21.9 (4.0)% and 22.2 (5.8)% at 32% theoretical Fe3⫹ saturation, 20.8 (2.8)% and 26.3 (7.6)% at 64% theoretical Fe3⫹ saturation, and 12.1 (9.3)% and 17.2 (13.3)% at 96% theoretical Fe3⫹ saturation (not significantly different in the paired comparison; n ⫽ 5). In subsequent experiments, the effect of glycation on Fe3⫹ isoforms of Tf was investigated after overnight incubation of nonglycated or glycated Tf with both NA and Fe:NTA.

effects of glycation

Fig. 2. Identification of the Tf isoform pattern based on pI. Calculation of the pI was based on the calibration line generated by measuring the distance from the cathode of known pI markers.

Incubation of human apoTf (half-life ⫽7 days (18 )] for 14 days with different concentrations of glucose led to a concentration-dependent increase in the fructosamine content (13.0, 15.5, 22.5, 47.0, and 102.5 ␮mol/L after incubation with 0, 5.6, 13.9, 22.2, and 33.3 mmol/L glucose, respectively). These glycated Tf solutions were incubated with three different concentrations of Fe:NTA to obtain theoretical Fe3⫹ saturations of 32%, 64%, and 96%.

4


Fig. 3. Fe3⫹-Tf isoforms obtained after incubation of apoTf with different Fe3⫹ compounds. ApoTf was incubated overnight with FeCl3 (A), iron citrate (B), and Fe:NTA (C) to obtain theoretical Fe3⫹ saturations of 0%, 32%, 64%, 96%, 240%, 480%, 960%, and 1600% (lanes 1– 8, respectively).

Time course of Fe3⫹ binding. We monitored Fe3⫹ binding spectrophotometrically by measuring the increases in absorbance at 450 nm after subtraction of the blank reaction with no iron added (Fig. 4). The mean (SE) increase in absorbance after 90 min was 0.038 (0.005), 0.051 (0.004), and 0.066 (0.005) absorbance units for the various Tf solutions (n ⫽ 42) at, respectively, 32%, 64%, and 96% theoretical Fe3⫹ saturation. There was no further significant increase in absorbance during the overnight incubation. We observed no difference between fresh Tf and Tf preincubated for 14 days in the absence of glucose. In contrast, glycation of Tf tended to decrease Fe3⫹ binding, but this difference reached statistical significance only for the Tf glycated with 1 mol/L glucose (P ⬍0.05 compared with Tf preincubated in the absence of glucose). Fe3⫹-Tf isoforms. After glycation and additional incubation in the presence of Fe:NTA, the different Fe3⫹-Tf isoforms were separated by IEF. A representative example is shown in Fig. 5. The percentage distribution of the various Fe3⫹-Tf isoforms was not significantly different between the fresh apoTf and apoTf incubated for 14 days without glucose. Preincubation with an extremely high glucose concentration (1 mol/L) produced a diffuse pattern with no detectable individual isoforms. The results of the quantitative analysis of the IEF patterns by densitometry are summarized in Fig. 6. Increasing concentrations of Fe:NTA also led to more Fe3⫹ binding in glycated Tf as indicated by the increasing percentages of Fe2-Tf at the expense of Fe0-Tf. For example, at 32%, 64%, and 96% theoretical Fe3⫹ saturation, the mean Fe2-Tf in the preincubated Tf solutions (n ⫽ 15) represented, respectively, 15 (4)%, 36 (5)%, and 76 (14)%, and Fe0-Tf represented 38 (7)%, 14 (6)%, and 0% (not detectable) of all isoforms. When we compared the Tf glycated at different glucose concentrations with the Tf preincubated without glucose, at a theoretical Fe3⫹ saturation of 32%, Fe2-Tf was significantly higher after the glycations performed at 13.9, 22.2,

and 33.3 mmol/L glucose [13 (3)%, 17 (3)%, and 18 (3)%, respectively, vs 12 (4)% for the nonglycated Tf; P ⫽ 0.06, 0.02, and 0.01 respectively; n ⫽ 3]. At the higher theoretical Fe3⫹ saturation of 64%, the proportion of Fe2-Tf was not significantly different among the various glycation conditions. At 96% theoretical Fe3⫹ saturation, in contrast, Fe2-Tf decreased linearly with increasing glycation [83 (14)%, 75 (16)%, 72 (21)%, and 68 (11)% after preincubation with 5.6, 13.9, 22.2, and 33.3 mmol/L glucose, respectively; r ⫽ 0.97; P ⫽ 0.008). In contrast to the results obtained with freshly prepared Tf, the proportions of the Fe1N-Tf and Fe1C-Tf isoforms were influenced by the 14-day preincubation depending on the Fe3⫹ saturation. At 32% theoretical Fe3⫹ saturation, the proportions of Fe1N-Tf and Fe1C-Tf were 23.9 (1.7)% and 23.0 (1.0)% respectively (not significant). At 64% theoretical Fe3⫹ saturation, the proportion of Fe1N-Tf was higher [27.3 (0.7)%] than the proportion of Fe1C-Tf isoform [23.1 (1.1)%; P ⫽ 0.009; n ⫽ 15]. At 96% theoretical Fe3⫹ saturation, these proportions were 11.8 (3.0)% and 12.2 (4.2)% (not significant). This shift was not affected by the degree of glycation.

Discussion The analysis of Tf isoforms has led to abundant, often controversial literature. The separation of these isoforms is based on pI differences caused by genetic variation, iron load, and sialic acid content (4 ). Analysis of the sialo-Tf isoforms has already demonstrated its use as a diagnostic marker of congenital disorders of glycosylation (19 ) and heavy alcohol consumption (20, 21 ), both conditions that decrease the amount of sialylation and even lead to a loss of one or two N-glycans. In this study, we focused on the iron content of Tf as source of its microheterogeneity. Urea gel electrophoresis has been successfully applied to differentiate between various Fe3⫹-Tf isoforms and is precise enough to detect differences in the ratio of occupancy of the N- and C-terminal sites (22 ). In the search for


5

Fig. 4. Comparison of Fe3⫹ binding measured at 450 nm during 90 min for fresh Tf (f), Tf preincubated for 14 days with 0 (solid line) 5.6 (〫), 13.9 (dashed line), 22.2 (⫻), 33.3 (F), or 1000 (⫹) mmol/L glucose. ApoTf (2.5 g/L) was incubated with Fe:NTA to achieve theoretical Fe3⫹ saturations of 32% (A), 64% (B), and 96% (C). Results show the increase in absorbance after addition of Fe:NTA and are expressed as mean (SE; error bars) of three experiments with measurements in duplicate. ⴱ, P ⬍0.05 in the paired comparison with preincubation with 0 mmol/L glucose.

6


Fig. 5. Effect of glycation on Fe3⫹-Tf isoforms after incubation with Fe:NTA to obtain theoretical Fe3⫹ saturations of 32% (A), 64% (B), and 96% (C). Tf was preincubated for 14 days with 0, 5.6, 13.9, 22.2, 33.3, and 1000 mmol/L glucose (lanes 1– 6, respectively).

semiautomated and less time-consuming alternatives, one of the currently preferred methods is IEF because of its high resolving power. van Noort and van Eijk (16 ) described an IEF method using the PhastSystem that allowed fast separation and easy estimation of the relative proportions of Tf isoforms. However, and partly because of its high resolving power, interpretation of results obtained by IEF presents several hazards. For example, great care must be taken to avoid overlapping of the different types of isoforms. To overcome interference on the IEF profiles as a result of the presence of sialic acid residues, these were removed by preincubation with NA. Incubation under conditions described in the literature [0.06 U/mg of Tf at 37 °C for 1 h (16 )] did not succeed in removing all sialic acid residues from our Tf samples. Prolonging the incubation time and increasing the NA concentration led to more efficient removal, but even after overnight incubation with 0.1 U NA/mg of Tf, there was still a detectable monosialo-Tf isoform band. This incomplete hydrolysis has also been observed by other authors (16, 23 ) and may be attributable to several causes, all related to the properties of this enzyme. One cause may be that the pH optimum for the enzyme is 5.8 – 6.0 in phosphate buffer. However, because of the importance of pH in Fe3⫹-loading experiments and the goal to mimic physiologic conditions, experiments with Tf and Fe3⫹ loading were performed at pH 7.4. Another cause could be that Fe3⫹, as well as other heavy metals and oxygen, inhibit the enzymatic activity of NA. Finally, the enzyme may lose activity on incubation (by as much as 50% after incubation at 37 °C for 24 h) and thus be insufficient to achieve total hydrolysis at the minimal concentrations of

sialylated Tf that remain toward the end of the overnight incubation. The second methodologic concern when using IEF to separate Fe3⫹ isoforms is the care that should be taken to perform optimum and accurate Fe3⫹ binding, as investigated extensively by Hackler et al. (24 ). Although Tf is also capable of binding other ions (e.g., Zn2⫹, Ga3⫹, Al3⫹, and Cu2⫹) (2, 25–28 ), Fe3⫹ is by far the most important metal that binds to Tf. However, as data in the literature show, Fe3⫹ binding is very sensitive to changes in pH (29 –31 ), and the presence of a synergistic anion, usually, but not necessarily, bicarbonate, is an absolute requirement (32, 33 ). Moreover, it is difficult to estimate the exact extent of Fe3⫹ binding to apoTf, which differs among the various types of Fe3⫹ salts (22, 34 ). We addressed this issue by investigating the binding of Fe3⫹ to Tf by spectrophotometry and the formation of Fe3⫹-Tf isoforms by IEF. To mimic the in vivo situation as closely as possible, we incubated a mixture containing physiologic concentrations of apoTf (2.5 g/L) and sodium bicarbonate (30 mmol/L) with different Fe3⫹ compounds at various concentrations and at a pH of 7.4. In these conditions, formation of the Fe3⫹-Tf complex was faster and more pronounced for the incubation with Fe:NTA than with FeCl3 or iron citrate. In a solution at physiologic pH, the greater part of Fe:NTA exists in the monomeric form. In this way Fe3⫹ is immediately available for uptake by Tf in a reaction lasting 10 s (35 ). Iron citrate, in contrast, forms polymeric iron complexes and thus releases low-molecular-weight iron species more slowly. Because the availability of low-molecular-weight iron is rate limiting, Fe3⫹ binding to Tf requires several hours (36 ). When FeCl3 was


7

Fig. 6. Distribution of Fe3⫹ isoforms of Tf glycated in vitro. Glycated apoTf, obtained by preincubation with glucose (0 –33.3 mmol/L) for 14 days, was incubated overnight with Fe:NTA at 32% (A), 64% (B), and 96% (C). Percentage distributions of the different isoforms were calculated by densitometry. Results are the mean (SE; error bars) of three separate experiments. ⴱ, P ⬍0.02 in the paired comparison with preincubation with 0 mmol/L glucose.

8


used, Fe3⫹ binding was also less effective. Published observations show that only 5–25% of Fe3⫹ binds when 1 equivalent of FeCl3 is added to apoTf (34 ). Each Tf molecule can bind a maximum of two Fe3⫹ ions. Depending on the iron supply of the organism, Tf molecules are either iron-free or loaded with one or two Fe3⫹ ions. In healthy controls, Tf saturation is ⬃30% and represents a mixture of four-ninths apoTf, four-ninths Tf with one Fe3⫹ ion, and one-ninth Tf with two Fe3⫹ ions bound. In our experiments, similar proportions were reached when Tf was incubated with Fe3⫹ to achieve a theoretical saturation of 32%. Two monoferric isoforms of Tf are distinguished: Tf with Fe3⫹ bound to the Nterminal binding site (Fe1N-Tf) or bound to the C-terminal binding site (Fe1C-Tf). In our experiments, the proportions of the Fe1N-Tf and Fe1C-Tf isoforms were similar in freshly prepared Tf. In Tf that was preincubated for 14 days in the presence or absence of glucose, a slightly higher proportion of the Fe1N-Tf isoform was observed at 64% theoretical Fe3⫹ saturation. This shift was not affected by the degree of glycation. It is not known whether the two binding sites have different functional roles. This issue has led to abundant but often contradictory literature. The N- and C-terminal binding sites have similar structures, but they are not equivalent in terms of Fe3⫹ uptake and release because they differ in their accessibility to iron chelates, binding strength, spectroscopic properties, kinetic lability, and response to changes in pH (29, 30, 37, 38 ). For example, it is known that the C-terminal site can hold Fe3⫹ at a lower pH (39, 40 ). Fe1N-Tf binds the Tf receptor on cells with a much lower affinity than does Fe1C-Tf (41 ). Some reports have concluded that in plasma Fe3⫹ is randomly distributed to the two sites (42– 44 ), whereas others have suggested that one site is preferentially occupied (22, 45, 46 ). As expected, increasing concentrations of iron led to more Fe3⫹ binding, as illustrated by the progressively higher percentages of the Fe2-Tf isoform and lower percentages of Fe0-Tf. Despite optimal Fe3⫹ binding conditions, complete Fe3⫹ saturation of Tf was not achieved. For example, at 96% Fe3⫹ saturation, only the Fe2-Tf isoform should be present, but in our experiments we also observed monoferric isoforms, albeit in small amounts. When Fe3⫹ saturation was calculated from the proportions of the isoform observed in the gels, values ranged from 91% in the apoTf incubated with no glucose to 83.9% in that preincubated with 33.3 mmol/L glucose. These values contrast with the theoretical 96% and suggest that during the experimental procedure (IEF) there was a loss of 5% of the nonglycated Tf and an additional 7% attributable to glycation. This observation suggests that release of Fe3⫹ from the Fe2-Tf isoform might have occurred during IEF. Release might occur at the anodal, more acidic position of the Phastgel where Fe2-Tf is located because it is well known that release of iron from Tf occurs at pH ⬍6.3 (47 ). In addition, sequestering of iron by the ampholytes during IEF might also occur. Zak and Aisen (22 )

observed a 25% loss in absorbance at 470 nm of saturated transferrin in 2% Ampholine. Because of the observed difference between the expected saturation and the actual saturation, we have consistently used the term “theoretical saturation” when describing Fe3⫹ concentrations in relation to Tf. In view of this observation, extrapolation of in vitro results of iron-binding experiments to the in vivo situation requires great care. Indeed, iron metabolism in vivo is subject to many factors, some as yet unidentified, that could affect both Fe3⫹ binding and release of free Fe2⫹. These issues have important implications with respect to both the amount of iron delivered to tissues and the release of redox-active free Fe2⫹, which is an important source of oxidative stress. It is not known whether the increased lipid peroxidation observed in diabetes mellitus is associated with an impairment of Fe3⫹ binding to the two sites on Tf. This hypothesis was suggested by our previous observation that glycation of apoTf led to a decrease in both its TIBC and in its capacity to prevent iron-induced lipid peroxidation (12 ). In the present study we aimed to investigate how glycation of Tf affects Fe3⫹ binding on the two sites by analyzing the Fe3⫹-Tf isoforms of in-vitro-glycated Tf. Our results indicate that the effect of glycation on Fe3⫹ binding is dependent on the degree of Fe3⫹ saturation of the Tf molecule. At 32% theoretical Fe3⫹ saturation, the Fe2-Tf isoform was more abundant in the Tf with higher degrees of glycation. Other authors have observed that glycated proteins bind more Fe3⫹ than nonglycated proteins. In the resulting glycochelate complex, however, the metal is more loosely bound and may thus retain redox activity (48 ). Moreover, glycated holo-Tf (fully Fe3⫹-saturated Tf) is also known to facilitate the production of free oxygen radicals (49 ), which can amplify the oxidative effects of iron further. In contrast, when Tf was almost totally (96%) saturated with Fe3⫹, the proportion of the Fe2-Tf isoform decreased linearly with the degree of glycation. This effect is in accordance with the decrease in TIBC we observed with increasing degrees of glycation (12 ). The measurement of TIBC is based on the addition of an excess of iron (iron/Tf molar ratio ⫽ 2.4). It is known that the antioxidant capacity of Tf (by Fe3⫹ binding) decreases with increasing Fe3⫹ saturation (50 ). Furthermore, the stability constant for Fe3⫹ is slightly lower for the second binding site (28 ), and binding of Fe3⫹ leads to rotation of the domains and conformational changes in the Tf molecule (51 ). This information, together with our observations, implies that the impairment of Fe3⫹ binding by glycation is more apparent at higher Fe3⫹ saturation. The relationship between glycation and Fe3⫹ binding is also seen in the in vivo situation in both rat models and humans. In patients with diabetes, the proportion of glycated Tf is three times higher than in controls (5.2% vs 1.6%) (52 ) and the TIBC is lower (49, 53 ).


In conclusion, and notwithstanding the analytical shortcomings of the method used to measure Fe3⫹-Tf isoforms, our results suggest that glycation facilitates the initial binding of Fe3⫹ to Tf but hinders further binding when high amounts of iron are present. In either case, the iron is bound more loosely and is thus more redox-active. This interpretation could explain the decrease in Fe3⫹-binding antioxidant capacity we observed in glycated apoTf (12 ). The impairment of Tf function may have important consequences with regard to the appearance of oxidative stress in vivo in diabetes and its involvement in the pathogenesis of diabetic complications. The relationship between the Fe3⫹-Tf isoforms and glycemia, glycation, oxidative stress, and complication profiles of patients with diabetes requires a more robust method to measure Fe3⫹-Tf isoforms as well as further investigation in clinical studies. These results could aid in the development of more specific and sensitive techniques capable of identifying patients with a disturbed iron-oxidant-antioxidant balance who are at increased risk of severe diabetic complications.

We thank P. Aerts, M. Vinckx, and J. Vertommen of the Laboratory of Endocrinology for their technical support. This work was supported financially by the Flemish Institute for Scientific-Technological Research.

References 1. de Jong G, van Dijk JP, van Eijk HG. The biology of transferrin. Clin Chim Acta 1990;190:1– 46. 2. van Eijk HG, van Noort WL, Kroos MJ, van der Heul C. Analysis of the iron-binding sites of transferrin by isoelectric focussing. J Clin Chem Clin Biochem 1978;16:557– 60. 3. van Noort WL, de Jong G, van Eijk HG. Purification of isotransferrins by concanavalin A Sepharose chromatography and preparative isoelectric focusing. Eur J Clin Chem Clin Biochem 1994;32: 885–92. 4. van Eijk HG, van Noort WL. The analysis of human serum transferrins with the Phastsystem: quantitation of microheterogeneity. Electrophoresis 1992;13:354 – 8. 5. Gutteridge JM, Rowley DA, Halliwell B. Superoxide-dependent formation of hydroxyl radical and lipid peroxidation in the presence of iron salts. Detection of ‘catalytic’ iron and anti-oxidant activity in extracellular fluids. Biochem J 1982;206:605–9. 6. Singh R, Barden A, Mori T, Beilin L. Advanced glycation endproducts: a review. Diabetologia 2001;44:129 – 46. 7. Baynes JW. Role of oxidative stress in the development of complications in diabetes. Diabetes 1991;40:405–12. 8. Wolff SP. Diabetes mellitus and free radicals. Free radicals, transition metals and oxidative stress in the aetiology of diabetes mellitus and complications. Br Med Bull 1993;49:642–52. 9. Cutler P. Deferoxamine therapy in high-ferritin diabetes. Diabetes 1989;38:1207–10. 10. Tuomainen TP, Nyysso¨ nen K, Salonen R, Tervahauta A, Korpela H, Lakka T, et al. Body iron stores are associated with serum insulin and blood glucose concentrations. Population study in 1,013 eastern Finnish men. Diabetes Care 1997;20:426 – 8. 11. Fernandez-Real JM, Lopez-Bermejo A, Ricart W. Cross-talk between iron metabolism and diabetes. Diabetes 2002;51:2348 – 54.

9

12. Van Campenhout A, Van Campenhout CM, Lagrou AR, Manuel-yKeenoy B. Transferrin modifications and lipid peroxidation: implications in diabetes mellitus. Free Radic Res 2003;37:1069 –77. 13. Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of proteindye binding. Anal Biochem 1976;72:248 –54. 14. Breuer W, Cabantchik ZI. A fluorescence-based one-step assay for serum non-transferrin-bound iron. Anal Biochem 2001;299:194 – 202. 15. Gambino R, Desvarieux E, Orth M, Matan H, Ackattupathil T, Lijoi E, et al. The relation between chemically measured total ironbinding capacity concentrations and immunologically measured transferrin concentrations in human serum. Clin Chem 1997;43: 2408 –12. 16. van Noort WL, van Eijk HG. Separation of human serum and cerebrospinal fluid transferrins in subfractions by isoelectric focusing with PhastsystemTM and quantitations with the UltroscanTM XL densitometer. Sci Tools 1992;36:1– 6. 17. de Jong G, van Eijk HG. Microheterogeneity of human serum transferrin: a biological phenomenon studied by isoelectric focusing in immobilized pH gradients. Electrophoresis 1988;9:589 – 98. 18. Potter BJ, Chapman RWG, Nunes RM, Sorrentino D, Scherlock S. Transferrin metabolism in alcoholic liver disease. Hepatology 1985;5:714 –21. 19. Jaeken J, Carchon H. The carbohydrate-deficient glycoprotein syndromes: an overview. J Inherit Metab Dis 1993;16:813–20. 20. Stibler H. Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clin Chem 1991;37:2029 –37. 21. Whitfield JB, Fletcher LM, Murphy TL, Powell LW, Halliday J, Heath AC, et al. Smoking, obesity, and hypertension alter the doseresponse curve and test sensitivity of carbohydrate-deficient transferrin as a marker of alcohol intake. Clin Chem 1998;44: 2480 –9. 22. Zak O, Aisen P. Nonrandom distribution of iron in circulating human transferrin. Blood 1986;68:157– 61. 23. Rauch SD. Transferrin microheterogeneity in human perilymph. The Laryngoscope 2000;110:545–52. 24. Hackler R, Arndt T, Tilmann KO, Gressner AM. Effect of separation conditions on automated isoelectric focusing of carbohydratedeficient transferrin and other human isotransferrins using the PhastSystem. Anal Biochem 1995;230:281–9. 25. Aasa R, Malmstroem BG, Saltman P. The specific binding of iron(III) and copper(II) to transferrin and conalbumin. Biochim Biophys Acta 1963;75:203–22. 26. Harris WR, Pecoraro VL. Thermodynamic binding constants for gallium transferrin. Biochemistry 1983;22:292–9. 27. Harris WR. Thermodynamic binding constants of the zinc-human serum transferrin complex. Biochemistry 1983;22:3920 – 6. 28. Martin RB, Savory J, Brown S, Bertholf RL, Wills MR. Transferrin binding of Al3⫹ and Fe3⫹. Clin Chem 1987;33:405–7. 29. Baldwin DA, De Sousa DM, Von Wandruszka RM. The effect of pH on the kinetics of iron release from human transferrin. Biochim Biophys Acta 1982;719:140 – 6. 30. Congiu-Castellano A, Barteri M, Castagnola M, Bianconi A, Borghi E, Della Longa S. Structure-function relationship in the serotransferrin: the role of the pH on the conformational change and the metal ions release. Biochem Biophys Res Commun 1994;198: 646 –52. 31. Messori L, Poggetto GD, Monnanni R, Hirose J. The pH dependent properties of metallotransferrins: a comparative study. Biometals 1997;10:303–13. 32. Schlabach MR, Bates GW. The synergistic binding of anions and

10

33.

34. 35.

36.

37.

38.

39.

40.

41.

42.

43.


Fe3⫹ by transferrin. Implications for the interlocking sites hypothesis. J Biol Chem 1975;250:2182– 8. Zak O, Ikuta K, Aisen P. The synergistic anion-binding sites of human transferrin: chemical and physiological effects of sitedirected mutagenesis. Biochemistry 2002;41:7416 –23. Bates GW, Schlabach MR. The reaction of ferric salts with transferrin. J Biol Chem 1973;248:3228 –32. Bates GW, Wernicke J. The kinetics and mechanism of iron(III) exchange between chelates and transferrin. IV. The reaction of transferrin with iron(III) nitrilotriacetate. J Biol Chem 1971;246: 3679 – 85. Bates GW, Billups C, Saltman P. The kinetics and mechanism of iron(III) exchange between chelates and transferrin. I. The complexes of citrate and nitrilotriacetic acid. J Biol Chem 1967;242: 2810 –5. Bali PK, Aisen P. Receptor-modulated iron release from transferrin: differential effects on N- and C-terminal sites. Biochemistry 1991;30:9947–52. Mason AB, He QY, Halbrooks PJ, Evers SJ, Gumerov DR, Kaltashov IA, et al. Differential effect of a His tag at the N- and C-termini: functional studies with recombinant human serum transferrin. Biochemistry 2002;41:9448 –54. Aisen P, Leibman A, Zweier J. Stoichiometric and site characteristics of the binding of iron to human transferrin. J Biol Chem 1978;253:1930 –7. Thompson CP, Grady JK, Chasteen ND. The influence of uncoordinated histidines on iron release from transferrin. A chemical modification study. J Biol Chem 1986;261:13128 –34. Zak O, Trinder D, Aisen P. Primary receptor-recognition site of human transferrin is in the C-terminal lobe. J Biol Chem 1994; 269:7110 – 4. Huebers HA, Josephson B, Huebers E, Csiba E, Finch CA. Occupancy of the iron binding sites of human transferrin. Proc Natl Acad Sci U S A 1984;81:4326 –30. DiRusso SC, Check IJ, Hunter RL. Quantification of apo-, mono-,

44.

45.

46.

47.

48.

49.

50.

51. 52.

53.

and diferric transferrin by polyacrylamide gradient gel electrophoresis in patients with disorders of iron metabolism. Blood 1985;66:1445–51. Beguin Y, Huebers HA, Finch CA. Random distribution of iron among the two binding sites of transferrin in patients with various hematologic disorders. Clin Chim Acta 1988;173:299 –304. Leibman A, Aisen P. Distribution of iron between the binding sites of transferrin in serum: methods and results in normal human subjects. Blood 1979;53:1058 – 65. Williams J, Moreton K. The distribution of iron between the metal-binding sites of transferrin human serum. Biochem J 1980; 185:483– 8. Surgenor DM, Koechlin BA, Strong LE. Chemical, clinical and immunological studies on the products of human plasma fractionation. XXXVII. The metal-combining globulin of human plasma. J Clin Invest 1949;28:73. Qian M, Liu M, Eaton JW. Transition metals bind to glycated proteins forming redox active “glycochelates”: implications for the pathogenesis of certain diabetic complications. Biochem Biophys Res Commun 1998;250:385–9. Fujimoto S, Kawakami N, Ohara A. Nonenzymatic glycation of transferrin: decrease of iron-binding capacity and increase of oxygen radical production. Biol Pharm Bull 1995;18:396 – 400. Gutteridge JM, Quinlan GJ. Antioxidant protection against organic and inorganic oxygen radicals by normal human plasma: the important primary role for iron-binding and iron-oxidising proteins. Biochim Biophys Acta 1992;1159:248 –54. Hirose M. The structural mechanism for iron uptake and release by transferrins. Biosci Biotechnol Biochem 2000;64:1328 –36. Austin GE, Mullins RH, Morin LG. Non-enzymic glycation of individual plasma proteins in normoglycemic and hyperglycemic patients. Clin Chem 1987;33;2220 – 4. Mauricio D, Pe´ rez A, Riera A, Arroyo JA, de Leiva A, Gimferrer E. Serum parameters of iron metabolism in type I (insulin dependent) diabetes mellitus. Diabetes Nutr Metab 1995;8:315– 6.