Introduction

Fluorescence resonance energy transfer (FRET) is the most common spectroscopic technique to study association between two proteins [1, 2]. One of the best example of FRET is the elaboration of association of the chaperonin proteins GroEL and GroES [3]. Fluorescence anisotropy has also been used in studying association of proteins, e.g. peptide binding to calmodulin [1, 4]. We have used a simple fluorescence quenching technique to study interactions of hemoglobin and its derivatives with spectrin, the major membrane skeletal protein of erythrocytes.

Hemoglobin is a soluble globular tetrameric protein composed of two identical α-like globin chains and two identical β-like globin chains found within vertebrate erythrocytes, at a very high concentration of about 5 mM [5]. Spectrin, the major protein of red cell membrane skeleton, is composed of two large, worm-like subunits (α- and β-chain), which are associated into double stranded, fiber-like flexible heterodimers about 100 nm in contour length. In normal red cells, spectrin dimers are assembled, head to head, predominantly into tetramers imparting mechanical flexibility to the erythrocytes [6–10].

Hemoglobin has been found to be associated with erythrocyte membranes prepared by hypotonic lysis of erythrocytes [11–13]. Several reports indicated the association of spectrin with hemoglobin under different experimental and physiological conditions, e.g. in senescent red blood cells and under oxidative stress [14–16] The extent of peroxide-induced hemoglobin–spectrin complexation, the effects of bound hemoglobin on conversion of spectrin dimer to the tetramer, and the identification of two types of hemoglobin–spectrin complexes within the membrane skeleton has also been evaluated [16].

Hemoglobin E is the most common hemoglobin variant in the world [17]. It is generated by a point mutation (Glu26(B8)→Lys) in the β-globin gene [18]. The primary clinical importance of HbE trait arises when the βE allele interacts with a β-thalassaemia mutation leading to a severe anaemia known as HbEβ-thalassaemia [19]. Several blood diseases are associated with erythrocyte deformation and defects in spectrin, e.g. various types of hereditary haemolytic anaemia involve mutations in spectrin [20–22]. Various reports indicating the presence of cross-linked aggregates of hemoglobin and spectrin in thalassemic erythrocytes have been directly implicated in explaining pathophysiological symptoms like reduced deformability, enhanced rigidity, and enhanced phagocytosis by macrophages [23, 24].

Phosphate metabolites, ATP and DPG, are found to modulate different molecular and biochemical processes inside erythrocyte. It has been shown earlier that DPG affects the mobility of intrinsic membrane proteins and weakens the association between the components of the membrane skeleton [25, 26]. Both ATP and DPG bind to hemoglobin in a comparable manner [27]. Irreducible complexation of hemoglobin with spectrin is a natural phenomenon of erythrocyte aging and such complex, generated in vivo, was associated with increased red cell membrane rigidity [28, 29]. There were conflicting reports both for and against the binding of hemoglobin [30, 31] to erythroid spectrin before we measured the binding dissociation constant between them using this fluorescence technique [32, 33].

Studies from this laboratory have also shown that spectrin exhibits chaperone-like properties and could bind denatured heme proteins in an ATP-dependent manner [34–36]. This review elaborates on the fluorescence quenching method that we have adopted to study protein–protein interactions when one protein is a heme protein and the other being spectrin [32, 33]. We have also studied the spectrin binding properties of the individual globin subunits [37] and intact hemolysates with varying compositions of HbA and HbE. However, this work could be extended to any other protein that interacts with a particular heme protein for the estimation of binding constants. Results of binding studies of HbA, the variant HbE, the two globin chains, α-globin and β-globin of HbA, and mixture of HbE and HbA, purified from hemolysates of the blood samples of HbEβ-thalassemics, with spectrin in presence and absence of phosphate metabolites ATP, 2,3 DPG and high salt have been discussed. The binding affinity increased about 5-fold from HbA to HbE. The binding dissociation constant, K d decreased to 16.1 ± 1.4 μM for the α-globin and remained comparable at 21.8 ± 1.9 μM for the β-globin with HbA. Such binding affinities decreased in presence of the phosphate metabolites and somewhat increased in presence of high salt. The spectrin binding affinity of HbE and α-globin chain has been always higher than those of HbA and the β-globin chain.

Materials and Methods

Sephadex G-50, DEAE-cellulose, CM-cellulose, Sepharose 4B, PMB, sodium azide, DTT, PMSF, Tris, KCl, ATP, and DPG were obtained from Sigma-Aldrich (St. Louis, MO). BioGel-P2 was from BioRad and Sephadex G-100 was from Amersham. Fluorescein isothiocyanate (FITC) was obtained from Molecular Probes. All other chemicals were of analytical grade and obtained locally. De-ionized water was doubly distilled on quartz before using for the preparation of buffers and all other solutions.

Collection and Isolation of Hemoglobin from Human Blood Samples

Human blood samples, taken for diagnosis from patients suffering from Eβ-thalassemia, were collected from different Thalassemia Clinics in the city of Kolkata and were characterized by the BioRad Variant HPLC system with enriched levels of HbE [38]. Samples were collected with the proper consent of patients who did not have blood transfusions. The levels of different hemoglobin variants, mainly of the normal adult HbA, HbE, and HbF in the blood samples, were estimated from the HPLC system, Variant™ in the blood samples of normal and homozygous HbE individuals and Eβ-thalassemics.

Human erythrocytes, after removal of the buffy coat and plasma, were extensively washed with phosphate-buffered saline (5 mM phosphate, 0.15 M NaCl, pH 7.4). Hemoglobin was isolated from packed clean erythrocytes by osmotic lysis using three volumes of 1 mM Tris, pH 8.0, at 4°C for 1 h. The hemoglobin mixture was purified by gel filtration on Sephadex G-100 column (30×1 cm) in the same buffer containing 100 mM KCl. The hemoglobin samples were stored in oxy-form at –70°C for not more than 7 days and characterized by the measurements of absorption at 415 and 541 nm, respectively. The protein concentration was determined spectrophotometrically using the molar extinction coefficient of 125,000 M–1cm–1 at 415 nm and 13,500 M–1cm–1 at 541 nm, respectively [39, 40]. Both HbA and HbE were also purified in cyano-met form by ion-exchange chromatography on SP-Sephadex elaborated earlier [41]. ATP concentration was determined assuming a molar absorbance of 15,400 M–1cm–1 at 259 nm [42]. ATP was used in presence of 10 fold molar excess of MgCl2.6H20 (Mg/ATP).

Isolation and Purification of Spectrin

Clean, white ghosts both from ovine and human blood were prepared by hypotonic lysis in 5 mM phosphate, 1 mM EDTA containing 20 g/ml of PMSF at pH 8.0 (lysis buffer) and dimeric spectrin was isolated at 37C following the protocol of Gratzer [43, 44]. Spectrin concentration was determined spectrophotometrically using an absorbance of 10.7 at 280 nm for 1% spectrin solution [43].

Preparation of Human α- and β-Globin Chains

The PMB derivatives of HbA were prepared following the method of Bucci and Fronticelli [45]. The α-PMB and β-PMB chains were separated by following a method consisting of two-column selective ion-exchange chromatography as described earlier [46]. To obtain α-PMB, the splitting solution was equilibrated with 0.01 M phosphate buffer at pH 8.0 and passed through a DEAE-cellulose column equilibrated and eluted with the same buffer. To obtain β-PMB, the splitting solution was equilibrated with 0.01 M phosphate buffer at pH 6.6 and applied on a CM-cellulose column, equilibrated and eluted with the same buffer. The PMB was removed from the isolated α-PMB and β-PMB chains by the addition of 50 MM 2-mercaptoethanol in 0.1 M phosphate buffer, pH 7.5. The intact globin chain was purified from the mixture of globin chains and unreacted PMB by gel filtration on a BioGel P2 column. Immediately after separation, the subunits were dialyzed extensively against 0.1 M phosphate buffer, pH 7.5 [47]. The globin chains were not stored for more than 48 h at 4°C.

Fluorescein-Conjugated Spectrin

Spectrin was covalently labeled with FITC in a buffer of pH ∼9.0. About 1 mg spectrin dimer was reacted with 50–100 fold molar excess of FITC, taken in a small volume of dry acetone. The labeling was carried out in a buffer containing 20 mM NaHCO3 for about an hour at 4°C. The FITC-labeled spectrin (F-spectrin) was separated and purified from the reaction mixture by gel filtration on Sephadex G-50 column using the buffer containing 5 mM Tris–HCl, 50 mM KCl, pH 8 [32, 33]. The concentration of F-spectrin was determined spectrophotometrically from the absorbance at 495 nm and using the molar extinction coefficient of 76,000 M–1cm–1 [48]. The labeling ratio of fluorescein to spectrin in F-spectrin was determined to be 2 fluorescein per spectrin dimer.

Fluorescence Measurements and Quenching of F-spectrin by Hemoglobins

The steady state fluorescence measurements were performed using a FluoroMax 3 fluoremeter (Jobin-Yvon Edison, NJ) fluorescence spectrophotometer. The excitation and the emission bandpass were set to 5 nm each. F-spectrin (20–50 nM) was excited at 495 nm and the change in fluorescence emission intensity at 520 nm was monitored in presence of the hemoglobin derivatives in a buffer containing 5 mM Tris–HCl, 50 mM KCl, pH 8.0. At both the wavelengths of excitation and emission maxima, 495 and 520 nm respectively, hemoglobin absorbed minimally and corrections of intensities due to inner filter effects were not necessary up to 50 μM hemoglobin [32, 33].

The changes in the extent of fluorescence quenching as a function of increasing concentrations of hemoglobin derivatives were analyzed by a model independent method and the apparent dissociation constant of hemoglobin binding to spectrin (K d) was determined using non-linear curve fitting analysis following the equations below. All experimental points for binding isotherm were fitted by least-square analysis using Microcal Origin software package (Version 5.0) from Microcal Software Inc., Northampton, MA.

$$K_{\rm d}=[C_0-({\Delta}F/{\Delta}F_{\max})C_0] [C_L-({\Delta}F/{\Delta}F_{\max})C_0]/ ({\Delta}F/{\Delta}F_{\max})C_0,$$
((1))
$$C_0({\Delta}F/{\Delta}F_{\max})^2-(C_0+C_{\rm L}+K_{\rm d}) ({\Delta}F/{\Delta}F{\max})+C_{\rm L}=0,$$
((2))

In equations (1) and (2), ΔF is the change in fluorescence emission intensity at 520 nm (λ ex = 495 nm) for each point on the titration curve. ΔF max denotes the same when hemoglobin is completely bound to spectrin, C L is the concentration of the ligand, hemoglobin, and C 0is the initial concentration of spectrin. From the double reciprocal plots of 1/ΔF against 1/C L, the magnitude of ΔF max was determined using equation (3):

$$1/ {\Delta}F=1/{\Delta}F_{\max}+1/[K_{\rm app} {\Delta}F_{\max}(C_{\rm L}-C_0)].$$
((3))

Since in this case C L >> C 0, hence (C L– C 0) ≅ C L, the linear double reciprocal plot was obtained by plotting 1/ΔF against 1/C L and the plot is extrapolated to the ordinate to obtain the value of ΔF max from the intercept [32–35]. The above approach is based on assumptions that emission intensity is proportional to the concentration of the ligand and the ligand concentration was in large excess than spectrin.

We have studied the effects of high salt, 0.5 M and 1.0 M KCl, to elucidate the mechanism of hemoglobin binding to spectrin and the effects of two important phosphate metabolites – Mg/ATP and DPG, at two different concentrations 0.5 and 1.0 mM. All estimated K d values are represented as mean ± SEM employing a minimum of 4–5 sets of independent experiments from two sets of samples with nearly identical compositions in terms of HbE, HbF, and HbA. Also, to check out the statistical significance level, all the reported values were subjected to the two-tail Student’s t-test with equal variance.

Results

Heme is a potent inhibitor of the intrinsic protein fluorescence due to its strong absorption in the Soret region with a molar extinction coefficient of 125,000 M–1cm–1 at 415 nm causing inner filter effect [1]. We have bypassed this problem by covalently modifying spectrin with FITC, and the fluorescein-conjugated spectrin (F-spectrin) was used for the study of the interaction of hemoglobin with spectrin. Hemoglobin showed minimal absorption at 495 and 520 nm, the excitation and emission maxima of F-spectrin, respectively. Figure 1 shows the fluorescence emission spectra of F-spectrin indicating quenching of the fluorescence intensity with increasing concentrations of oxy-HbA. Figure 2 shows a representative binding isotherm as a plot of ΔF/ΔF max against the concentrations of hemoglobin from the fluorescence intensity data. The inset of Fig. 2 shows the linear double reciprocal plot of 1/ΔF against 1/[HbA], extrapolated to the ordinate for obtaining the value of ΔF max from the intercept. The binding data were analyzed by model independent curve fitting elaborated in the previous section. The apparent dissociation constant of HbA binding to spectrin was evaluated to be 22.5±1.6 μM. Due to no inner filter effect of hemoglobin even at a concentration as high as 50 μM, 80% saturation of hemoglobin binding to spectrin could be attained. Presence of a large number of moderately high affinity hemoglobin binding sites in spectrin is probable in view of the high concentrations of hemoglobin. Identical sets of experiments were performed with all hemoglobin samples, differing in the composition of HbA, HbF, and HbE, with enriched levels of HbE. Table 1 summarizes the apparent dissociation constants of binding of hemoglobin and spectrin.

Fig. 1
figure 15_1_978-0-387-88722-7

Fluorescence emission spectra of F-spectrin (a) in absence and presence of (b) 5 μM; (c) 10 μM; (d) 20 μM; (e) 30 μM; and (f) 40 μM Hemoglobin. The excitation wavelength was at 495 nm

Full size image
Fig. 2
figure 15_2_978-0-387-88722-7

The binding of hemoglobin with spectrin as reflected in the plot of ΔF/ΔF max against the hemoglobin concentrations [Hb], from the fluorescence intensity data. Inset shows the linear double reciprocal plot of 1/ΔF against 1/[Hemoglobin], extrapolated to the ordinate for obtaining the value of ΔF max from the intercept

Full size image
Table 1 Effect of Mg/ATP, DPG, and KCl on the binding of spectrin and hemoglobin mixtures (K d in μM) containing increasing levels of HbE
Full size table

The extent of fluorescence quenching was found to increase with increasing levels of HbE in the hemoglobin mixtures. Figure 3 shows the histogram representations of the extent of fluorescence quenching (ΔF/F0) where F 0 is the fluorescence intensity of F-spectrin in the absence of hemoglobin, against the elevated levels of HbE% in the hemoglobin mixtures at three fixed concentrations of total hemoglobin at 5, 10 and 20 μM. The K d values on the other hand decreased with increasing levels of HbE% in the hemoglobin mixtures indicating high affinity binding of HbE over HbA (Table 1).

Fig. 3
figure 15_3_978-0-387-88722-7

Histogram representation of the extent of fluorescence quenching ΔF/F 0 % at 520 nm against the level of HbE % indicated in the X-axis, three different concentrations of 5, 10, and 20 μM with respect to the total hemoglobin

Full size image

Spectrin binding experiments were also carried out with purified globin chains from HbA. K d decreased to 16.1±1.4 μM for α-globin and for β-globin chains remained comparable to that of HbA at 21.8±1.9 μM [37]. The binding dissociation constants for HbA and β-globin chains also remained comparable in the presence of both ATP and 2,3 DPG. However, the binding affinity decreased substantially for the α-globin chains with K d values increasing from 16.1±1.4 μM to 38.0±3.3 μM in presence of ATP and 28.9±2.1 μM in presence of 2,3 DPG (P<0.05).

The fluorescence quenching measurements of F-spectrin were then done with purified hemolysates having varying levels of HbE along with HbA and HbF from normal individuals with HbA ranging from 90–95% and HbE up to 90–92% from homozygous HbE patients. Only those hemolysates were used for study where the levels of HbF stayed within 5–8%. HbA (>95%) and HbE (>90%) stays predominantly in oxy-form, determined from the absorbance at 415 and 541 nm as the yardstick [40]. Three different sets of experiments were performed with hemoglobin mixtures having ∼30%, ∼50%, and ∼70% HbE with the rest of HbA characterized by the VariantTM system. Table 1 summarizes the apparent dissociation constants of spectrin binding of the hemoglobin mixtures in the presence and absence of Mg/ATP, DPG, and KCl. The apparent dissociation constant of HbA binding to spectrin was evaluated to be 22.5±1.6 μM which decreased to 17.1±1.5 μM in presence of 30% HbE and 5.4±0.5 μM in presence of 70% HbE. In presence of both Mg/ATP and DPG, the binding affinity decreased. For example, fin case of hemoglobin mixture containing ∼70% HbE the K d value increased to 8.4±0.6 μM and 7.7±0.6 μM from 5.4±0.5 μM in presence of 1 mM Mg/ATP and DPG, respectively (P<0.05). On the other hand, the K d values decreased to 4.2±0.3 μM in presence of 1 M KCl (Table 1).

Figure 4 summarizes the binding affinities in form of histogram representation of spectrin complexation with HbA, HbE, α-globin, and β-globin chains of the affinity constants (= 1/K d). The spectrin binding affinities are found to increase from HbA to HbE and α-globin to β-globin chains. It is also evident that ATP decreases the binding affinity substantially in HbE and in α-globin chains. Inset of Fig. 4 shows a representative 4% SDS–PAGE experiment showing peroxide-induced cross-linked complexes (HMWA), generated from 1 μM spectrin, 22.5 μM hemoglobin and 1 mM H2O2 in absence of DTT with hemolysates containing HbE starting from 20 to 90%, further indicating physical association of spectrin with globin chains of HbA [33].

Fig. 4
figure 15_4_978-0-387-88722-7

Histogram representation of the affinity constants of spectrin binding to HbA, HbE, α-globin chains, and β-globin chains in the presence (filled bar) and absence (empty bar) of 1 mM Mg/ATP. The error bars are SEM of 5–10 independent experiments. Inset shows a representative 4% SDS–PAGE showing high molecular weight aggregates generated from 1 μM spectrin, 22.5 μM Hemoglobin, and 1 mM H2O2 in absence of DTT with hemolysates containing HbE starting from 20 to 90%

Full size image

Discussion

Using this simple fluorescence technique, we have determined the binding constant between spectrin and hemoglobin, for the first time, to be around 105 M–1, two orders of magnitude higher than that between the cytoplasmic domain of band 3 and hemoglobin [32, 49]. This moderately high affinity binding of hemoglobin with spectrin could play an important role in the maintenance of structural integrity of the erythrocyte membrane. Moreover, in hemoglobinopathy, e.g. HbEβ-thalassemia, where the amount of HbE rises to more than 50% in acute cases, the stronger spectrin–hemoglobin interaction takes place as indicated by significant lowering of the dissociation constants from 22.51.6 μM to 5.40.5 μM. Implications of these results in hemoglobin disorders might go a long way in understanding the stability, deformability and cytoskeletal integrity of erythrocytes containing varying amounts of HbE. This decreasing trend in Kd with increasing HbE content was also associated with increase in the extent of fluorescence quenching of F-spectrin indicating its possible usefulness in the detection of differential spectrin interaction of HbA and HbE (Fig. 3). The maximum extent of hemoglobin-induced quenching of F-spectrin, in our opinion, could even be used in the detection of HbEβ-thalassemia and other hemoglobin disorders involving a particular hemoglobin variant out about 1000 known variants.

The spectrin–hemoglobin binding is favored in presence of high salt indicating possible role of hydrophobic interactions in such binding (Table 1). The presence of hydrophobic patches on human globin subunits could be responsible for the interaction with spectrin as predicted in an earlier report [50]. Kd values decreased to 4.2±0.3 μM in presence of 1 M KCl and the effect was more pronounced in the hemolysates with higher HbE content. The effects of Mg/ATP on the spectrin binding affinity of HbA, HbE and the globin chains along with the yield of oxidative cross-linked complexes were even more interesting [33, 37]. In presence of 1.0 mM Mg/ATP and DPG, the binding affinities were found to decrease significantly. The difference in the Kd values listed in Table 1, both in the absence and presence of ATP, DPG and KCl were found to be statistically significant (P<0.05). K d values for HbA and β-globin chains remained comparable both in the presence and absence ATP and 2,3 DPG. However, the K d increased substantially for the α-globin chains in presence of 1 mM Mg/ATP and 2,3 DPG (P<0.05). The histogram representations of the spectrin binding affinities between the two variants HbA and HbE and the two globin chains of HbA clearly indicates preference of HbE over HbA and α-globin chains over β-globin chains in an ATP-dependent manner.

Our earlier work revealed a large number of moderately high affinity hemoglobin binding sites in spectrin [32, 33, 37]. A rough estimate from fluorescence studies revealed ∼100 hemoglobins to bind to one spectrin dimer. Stoichiometric analysis of the soluble cross-linked complexes generated between HbA and spectrin in presence of H2O2, after separation on Sepharose 4B, indicated 8-10 globin chains to associate with one spectrin subunit [33, 37]. A representative SDS-PAGE is shown in the inset of Fig. 4 indicating increased formation of HMWA with increase in HbE% in the hemolysates. The band seen in the middle of HMWA and the α-spectrin subunit corresponding to a molecular weight of ∼400 kDa were analyzed for finding the stoichiometry of spectrin subunit with the globin chains.

Thermodynamic studies on the subunit assembly of hemoglobin showed the equilibrium constant associated with the conversion of hemoglobin from dimer to tetramer has been in the range of 1.0–1.2 × 106 M–1 between 20 and 27°C [51, 52]. We have chosen the hemoglobin concentration above 10–5 M in order to assure the presence of significant amount of hemoglobin tetramers preventing the formation of the hemoglobin dimers. The hemoglobin-induced quenching of F-spectrin is a static effect and enables one to obtain a quantitative estimate of the binding constants. Comparative experiments with different fluorescein or rhodamine labeled protein that interacts with hemoglobin can be used to measure the relative binding constant. Our measurements of the excited state lifetime of the fluorescein moiety in F-spectrin also showed no significant changes in the mean lifetime of spectrin in presence of different hemoglobin derivatives [44].

A possible mechanism involved in the interaction of HbE and β-thalassemia lies in the observation that HbE is oxidatively unstable in vitro [53]. When the level of HbE rises above ∼30% (HbE carrier) basal level to ∼70% in acute cases of HbEβ-thalassaemia, stronger spectrin–hemoglobin interactions take place as evidenced by significant lowering of the binding dissociation constants from 17.1±1.5 μM (∼30% HbE) to 5.4±0.5 μM (∼70% HbE). The preferential HbE interactions, within a mixture of HbA and HbE, with erythroid spectrin could contribute to the premature hemolysis of erythrocytes. The results of the present work are more significant in the light of erythroid spectrin exhibiting chaperone-like activity [34]. Moreover, both the binding of erythroid spectrin to a denatured heme protein and proteolysis of excess α-globin chains in β-thalassemic cells have been shown to be ATP-dependent, supporting our observations of the effects of Mg/ATP and DPG [35, 54, 55]. The susceptibility of HbE to oxidation could also play a major role governing the release of heme which strongly destabilizes lipid bilayer and eventually triggers the oxidative haemolytic effect in such hemolytic anemia [56, 57].