Reactive Oxygen Species Damaged Hemoglobin Presents Unique Epitopes for Type 1 Diabetes Autoantibodies

Oxidative stress has been implicated in a wide range of diseases including atherosclerosis, hypertension, diabetes, ischemia reperfusion injury and a range of neurodegenerative disorders (Flora, 2007). Metal interactions with hydrogen peroxide have been frequently implicated in catalyzing free radical-mediated damage. Redox-active transition metals, containing unpaired electrons in their d-orbital, are able to generate, or remove, reactive species with unpaired electrons. In particular free ferrous iron (Gutteridge, 1986) has been suggested to catalyze tissue damage by reacting with hydrogen peroxide to generate the highly reactive hydroxyl radical (OH^•). However, the relevance of this Fenton chemistry to systems where strong redox-inactive chelaters exist to bind iron is not clear.

There has been accumulating evidence that an additional mechanism to generate reactive oxygen species occurs in vivo (Moore et al., 1998). This involves the interaction of the ferric state of heme proteins with hydrogen peroxide:

Both the ferryl iron (Fe⁴⁺ = O^2–) and the protein-bound free radical (R^•+) are able to initiate a range of oxidative chemistry with similar reactivity to the hydroxyl radical (Sutton and Winterbourn, 1989). Heme proteins like hemoglobin (Hb) and myoglobin (Mb) contain redox-active transition metal iron that makes them susceptible to causing oxidative damage. Although the structure of the globin chain allows heme to bind oxygen with minimal oxidation of ferrous to ferric iron, autoxidation is not entirely prevented; low concentrations of ferric heme are normally present in vivo (methemoglobin, metmyoglobin). These can then react with the peroxides formed during the autoxidation process itself or elsewhere in the protein`s vicinity. It is well documented that oxidative damage of protein and DNA presents unique neoepitopes, which is one of the factors in antigen-driven autoimmune response in various diseases (Scofield et al., 2005; Habib et al., 2006).

Recent studies from our laboratory have demonstrated that after modification with ROS, proteins became highly immunogenic in experimental animals and oxidative by-products with proteins could lead to neoantigens which would initiate autoimmunity in various diseases (Rasheed and Ali, 2006; Rasheed et al., 2006, 2007a,b). In view of these, the present study was hypothesized that oxidative by-products like hydrogen peroxide, would react with Hb and help to initiate autoimmunity in patients with type 1 diabetes mellitus. To test this hypothesis, Hb was modified by reactive oxygen species, generated by in vitro reaction of hemoglobin with hydrogen peroxide. Immunogenicity of ROS modified and unmodified Hb was investigated in rabbits and found that autoimmunity was established faster and more vigorously in the animals that were immunized against ROS-modified Hb. We have studied the presence of circulating autoantibodies in type 1 diabetes sera directed against native and damaged Hb.

Materials
Anti-rabbit-IgG-alkaline phosphatase conjugate, p-nitrophenyl phosphate, Tween-20, hemoglobin were purchased from Sigma Chemical Company, USA Protein A-Sepharose CL-4B was purchased from Genei, India. Polystyrene microtitre flat bottom ELISA plates and modules were purchased from NUNC, Denmark. All other chemicals were of analytical grade.

Protein was estimated by the methods of Lowry et al. (1951).

Human Subjects
Blood was collected from voluntary donors with history of type 1 diabetes mellitus (n = 39, male = 23, female = 16, aged 49±16) under treatment in J.N. Medical College Hospital A.M.U., Aligarh. The control samples (n = 22, male = 17, female = 5, aged 43±17) were collected from healthy subjects. The average (±SD) of post prandial blood sugar level for 31 diabetic patients was 354±84 mg dL^-1, while it was 110±8 mg dL^-1 for 22 healthy subjects (without family history of diabetes). The glycated hemoglobin (HbA_1C) was 8.0±1.5 for patients, whereas 4.7±0.4 for healthy subjects. Glutamic Acid Decarboxylase Antibodies (GADA), insulin antibodies (IAA), anti-tyrosine phosphatase antibodies (IA-2A) were found in 43, 74 and 39% of the cases, respectively. Immunopositivity for at least one of these autoantibodies was found in all patients. Fifty one percent male patients were having smoking history. All other information on demographics and cumulative clinical and laboratory manifestations over the course of disease was obtained by both chart review and discussion with the patient and his/her family members. The informed consent was obtained from each patient or from patients family members.

All sera were decomplemented by heating at 56°C for 30 min and stored in aliquots at -20°C with sodium azide as preservative.

Modification of Hemoglobin
Hemoglobin from Sigma was used exactly in a similar way as cited by various investigators (Shimada et al., 2005). Hb (7.8 μM) and hydrogen peroxide (2.1 mM) were mixed and incubated at 37°C for 12 h. After incubation, the excess hydrogen peroxide was removed by extensive dialysis against PBS (10 mM sodium phosphate buffer containing 150 mM NaCl, pH 7.4). Hb incubated without hydrogen peroxide served as control. The UV absorption characteristics of native and modified Hb were recorded on a Shimadzu spectrophotometer.

Fluorescence Measurements
Fluorescence measurements were performed on Hitachi F-200 spectrofluorimeter (Japan). The fluorescence spectra were measured at 25±0.1°C with a cell of 1 cm path length. The excitation and emission slits were set at 5 and 10 nm, respectively. Intrinsic fluorescence was measured by exciting the protein solution at 280 or 295 nm and emission spectra were recorded in the range of 300-400 nm. Increase of fluorescence intensity (F.I.) was calculated using the following equation.

Absorption-Temperature Scan
Thermal denaturation profile of protein was performed on a Shimadzu UV-240 spectrophotometer equipped with a temperature programmer and controller assembly (Hasan and Ali, 1990). Protein samples were in PBS, pH 7.4 and absorbance was recorded at a fixed wavelength of 280 nm. Percent denaturation was calculated using the equation:

Assay of Carbonyl Formation
Carbonyl contents of native and ROS-modified Hb were analyzed according to Levine et al. (1990) with slight modifications. The reaction mixture containing 15.6 μM native Hb or ROS-Hb, 0.5 mL of 10 mM 2, 4-dinitrophenylhydrazine (DNPH)/2.5 M HCl was added and thoroughly mixed. After addition of 250 μM TCA (20%) and centrifugation, the pellet was collected and washed three times with 1 mL ethanol: Ethylacetate (1:1) mixture. The pellet was then dissolved in 1 mL of 6 M guanidine solution and incubated at 30°C for 15 min. After centrifugation, the supernatant was collected and carbonyl contents were estimated from the absorbance at 370 nm using a molar absorption coefficient of 22,000 M^-1 cm¹. Samples were spectrophotometrically analyzed against a blank of 1 mL of guanidine solution (6 M). Protein concentration was determined in the samples by the method of Lowry et al. (1951). Carbonyl contents were expressed as nmol mg^-1 protein.

Immunization Schedule
The immunization of random bred, New Zealand white rabbits was performed as described previously (Rasheed et al., 2006). Briefly, rabbits (n = 4; two each for native and ROS-Hb antigens) were immunized intramuscularly at multiple sites with 100 μg of antigen, emulsified with an equal volume of Freund`s complete adjuvant. The animals were boosted in Freund`s incomplete adjuvant at weekly intervals for 6 weeks with the same amount of antigen. Test bleeds were performed 7 days post boost, which gave appropriate titre of the antibody. The animals were bled and the serum separated from the blood was heated at 56°C for 30 min to inactivate complement proteins and stored at -20°C.

Purification of IgG
Immunoglobulin G (IgG) was isolated from immune and preimmune rabbit sera on a protein A-sepharose CL-4B column (Rasheed and Ali, 2006). The homogeneity of isolated IgG was checked by polyacrylamide gel electrophoresis.

Enzyme Linked Immunosorbent Assay
Enzyme linked immunosorbent assay (ELISA) was carried out on polystryrene plates with slight modification (Rasheed and Ali, 2006; Rasheed et al., 2006, 2007a, b). Polystryrene polysorp immunoplates were coated with 100 μL of native or modified Hb (10 μg mL^-1) in carbonate-bicarbonate buffer (0.05 M, pH 9.6). The plates were coated for 2 h at 37°C and overnight at 4°C. Each sample was coated in duplicate and half of the plates served as control devoid of only antigen coating. Unbound antigen was washed thrice with TBS-T (20 mM Tris, 150 mM NaCl, pH 7.4 containing 0.05% Tween-20) and unoccupied sites were blocked with 2% fat free milk in TBS (10 mM Tris, 150 mM NaCl, pH 7.4) for 4-6 h at 37°C. After incubation the plates were washed four times with TBS-T. The test serum serially diluted in TBS-T or affinity purified IgG in TBS (100 μL well^-1) was adsorbed for 2 h at 37°C and overnight at 4°C. Bound antibodies were assayed with anti-rabbit alkaline phosphatase conjugate using p-nitrophenyl phosphate as substrate. The absorbance (A) of each well was monitored at 410 nm on an automatic microplate reader. Each sample was run in duplicate. The control wells were treated similarly but were devoid of antigen. Results were expressed as a mean of A_test-A_control.

Competition ELISA
The antigenic specificity of the antibodies was determined by competition ELISA (Rasheed et al., 2006; Rasheed and Ali, 2006). Varying amounts of inhibitors (0-20 μg mL^-1) were mixed with a constant amount of antiserum or purified IgG. The mixture was incubated at room temperature for 2 h and overnight at 4°C. The immune complex thus formed was coated in the wells instead of the serum/IgG. The remaining steps were the same as in direct binding ELISA. Percent inhibition was calculated using the formula:

Statistical Analysis
Data are presented as mean±SD. Significance of differences from control values were determined with the Student`s t-test (Statgraphics, Origin 6.1). A value of p<0.05 was considered to indicate statistical significance.

Hb was modified by ROS and the induced changes in Hb were analyzed by UV-visible absorption spectroscopy. The absorption spectra of ROS damaged Hb showed 64% hypochromicity at 280 nm. A sharp peak at 410 nm was observed, which is characteristic of native Hb. Upon ROS modification, peak at 410 nm was completely disappeared (Fig. 1). The oxidation of tryptophan residues on damaged Hb was evident by the increase of 51% fluorescence intensity at 330 nm using an excitation wavelength of 295 nm (Fig. 2A). The damage to tryptophan residues were confirmed by the increase of 28% fluorescence intensity at 330 nm using an excitation wavelength of 280 nm (Fig. 2B). Stability of native and modified Hb was checked by temperature-induced denaturation experiments, the protein absorbance changes with temperature until the process of unfolding is complete. Thermal melting profile of native and ROS-damaged Hb showed melting temperature (Tm) of native and modified HAS as 59 and 65.5°C, respectively (Fig. 3). The complete unfolding of native and ROS-damaged Hb was occurred at 75 and 80°C, respectively. After these temperatures both protein samples showed precipitation or aggregation as abnormal increase in their absorbance was noticed. Oxidation of protein typically results in an increase in carbonyl contents, a known biomarker of oxidative stress. The oxidation of ROS-Hb was further evident by the estimation of carbonyl contents. The average carbonyl contents (±SD) of five independent assays of ROS-Hb and native Hb were 28.7±1.2 and 10.4±0.7 nmol mg^-1 protein, respectively, a p-value of<0.001 indicates significant difference in the carbonyl contents of ROS-Hb and native Hb.

Fig. 1:	UV-visible spectra of native (—) and ROS-modified (---) Hb. The samples were in PBS, pH 7.4 at a concentration of 7.8 μM. The spectra are the average of three determinations

Fig. 2:	Fluorescence emission spectra of native (—) and ROS-modified (---) Hb. Protein was in PBS, pH 7.4 at a concentration of 1.5 μM. The excitation wavelength was 295 nm (A) and 280 nm (B). The spectra are the average of three determinations

Fig. 3:	Thermal denaturation profile of native (■) and ROS-modified (●) Hb. The samples were in PBS, pH 7.4 at a concentration of 3 μM

Direct binding ELISA showed ROS-modified Hb as a potent immunogen in rabbits inducing high titre (> 1: 12800) antibodies, whereas with native Hb the titre was low (~ 1: 6400). Preimmune serum serve as negative control, as it did not showed appreciable binding with the respective immunogens under identical experimental conditions (Fig. 4).

To substantiate the above results, immunoglobulin G (IgG) against ROS-Hb and Hb were isolated on Protein-A Sepharose CL-4B affinity column. IgG from preimmune sera was also purified under identical experimental conditions. The purified IgG was found to elute in a single symmetrical peak. SDS-PAGE under non-reducing conditions of affinity purified IgG showed a single homogenous band (data not shown). Direct binding ELISA of purified anti-ROS-Hb IgG and anti-Hb IgG showed strong binding with their respective immunogens. Preimmune IgG showed negligible binding under identical conditions (data not given). The specificity of the purified IgG was further evaluated by competitive inhibition assays. Affinity purified anti-ROS-Hb IgG showed the average percent inhibition (±SD) of 98±2.2 at 20 μg mL^-1 of inhibitor (ROS-Hb) concentration. The concentration of competitor (ROS-Hb) required for 50% inhibition was only 0.5 μg mL^-1. However, with native Hb a maximum of 73±2.2 inhibition of the anti-ROS-Hb IgG was observed. The microtitre plates were coated with ROS-Hb (Fig. 5A). Whereas, anti-Hb IgG showed the average percent inhibition (±SD) of 85±2.2 at 20 μg mL^-1 of inhibitor (native Hb) concentration. The concentration of competitor (native Hb) required for 50% inhibition was 2 μg mL^-1, ROS-Hb showed average percent inhibition 71±2.2 of anti-Hb IgG. The microtitre plates were coated with native Hb (Fig. 5 B).

Preimmune IgG with ROS-Hb and Hb showed negligible inhibition under identical experimental conditions (data not given). The results were similar to those obtained with serum and point to that autoimmunity was established faster and more vigorously in the animals that were immunized with ROS-Hb.

Fig. 4:	Level of induced antibodies against ROS-modified and native Hb. Direct binding ELISA of (A) ROS-Hb and (B) Hb with immune (▲) and preimmune (Δ) sera. Microtitre plates were coated with the respective antigen (10 μg mL-1). Each point represents a mean±SD of three independent assays

Present study comprised serum samples from patients suffering from type 1 diabetes mellitus (DM). Each sample was obtained after careful clinical examination of patients with proven diagnosis at J.N. Medical College Hospital, A.M.U., Aligarh. Sera from thirty nine type 1 DM and twenty two normal human subjects were tested for binding to native and ROS modified Hb by direct binding ELISA (Table 1). A majority of DM sera (19/39) showed strong binding to ROS-Hb over native Hb at 1:100 serum dilution (p<0.05). No appreciable binding was observed with the normal subjects. The average absorbance at 410 nm (±SD) of 19 DM sera binding to native and ROS-damaged Hb was 0.47±0.06 and 0.83±0.07, respectively. Whereas, the absorbance at 410 nm (±SD) of 22 normal human sera binding to both protein antigens was 0.24±0.04. The binding specificity of antibodies from 19 selected sera of DM patients was evaluated by competition ELISA using native and ROS-Hb as inhibitors. Mircrotitre plates were coated with ROS-Hb. Results point to a higher reactivity of autoantibodies in patient`s sera towards ROS damaged Hb over native Hb. Similarly, rest of the sera showed high percent inhibitions with modified Hb over native Hb. The average percent inhibition (±SD) in the binding of 19 type 1 DM sera to native and ROS-Hb was 27.6±5.6 and 49.3±9.9, respectively. The data reveals striking differences in the recognition of native and oxidized Hb by DM autoantibodies (p<0.001). Table 1 summarizes complete demographics, clinical, biochemical and immunological characteristics of all 19 selected DM sera with their controls. Among 19 DM sera, 11 sera were from those male patients which are having smoking history of 22.4±10.4 years. Direct binding and competition ELISA results point to a strong reactivity of all DM smokers towards ROS-Hb over native Hb. Sera from aging patients with high degree of diabetes showed substantially higher binding to ROS-Hb as compare to native Hb (Table 1). Present results are in full agreement to the view that oxidative damage of protein presents unique epitopes, which help to initiate autoimmunity (Scofield et al., 2005; Rasheed and Ali, 2006; Rasheed et al., 2006, 2007a, b).

Fig. 5:	(A) Inhibition of anti-ROS-Hb IgG binding to ROS-Hb. The inhibitors were ROS-Hb (■) and native Hb (∼). The microtitre plates were coated with the ROS-Hb (10 μg mL-1). (B) Inhibition of anti-Hb IgG binding to native Hb. The inhibitors were native Hb (▲) and ROS-Hb (Δ). The microtitre plates were coated with the native Hb (10 μg mL-1). Each point represents a mean±SD of three independent assays