Comparative Oxidation of Hemoglobins A and S

THE MUTANT HEMOGLOBIN (Hb) that defines sickle cell anemia, HbS, is reported to exhibit an accelerated autooxidation rate during incubation in vitro.1,2Consistent with this, intact sickle red blood cells exhibit a correspondingly exaggerated generation of activated oxygen species during incubation in vitro.3 On the other hand, no structural or biochemical explanation for this enhanced oxidation tendency has been forthcoming. In fact, the two Hbs are isomorphic at the level of x-ray crystallography, have multiple identical functions, and show only minor differences at the N-terminal end of their respective β-globin chains.4 HbS does manifest one dramatic behavioral difference in the form of an abnormal surface denaturation tendency4 that is derived from the mutant Hb's hydrophobic surface substitution (β^6Glu→Val). While this probably underlies heme pocket disturbances during interfacial interactions between Hb and chemical denaturants5 or phospholipids,6 it comprises an uncertain proximate cause of abnormal Hb oxidation in solution.

Given this background, more recent observations require consideration of the possibility that observation of augmented HbS oxidation actually results from the presence of adventitious iron carried by the purified Hb. In first place, the cytosolic aspect of the membranes of sickle—but not of normal—red blood cells carries substantial amounts of molecular iron.7 Despite its association with the membrane with very high avidity,8 this Fe³⁺ is redox-active. For example, it can form a redox couple that, even while remaining membrane-bound, oxidizes soluble oxyHb to metHb.9This presumably corresponds to the known ability of Fe³⁺chelates to promote oxidation of oxyHb.10,11 Of particular relevance, binding studies demonstrated that Fe³⁺ binds to hemichrome,12 denatured low-spin ferric Hb, should this be present; HbS is more likely than HbA to form hemichrome from metHb as a result of its surface denaturation behavior.4 13 

Thus, the possibility exists that adventitious iron, whether physiologic (eg, derived from cytoplasm) or an outright contaminant (eg, derived from experimental buffers), could explain the previous impression that HbS has an abnormal tendency to autooxidize. This is important to resolve because a tendency for HbS to undergo enhanced oxidization (either because of intrinsic properties of the Hb or because of physiologic adventitious iron) has significant implications for the cellular pathobiology of sickle cell disease, in which red blood cells develop numerous membrane defects that are believed to be oxidative in etiology.14 

Chromatography supplies were obtained from BioRad (Hercules, CA), and reagents from Sigma Chemical (St Louis, MO). With the exception of Hb and iron itself, all reagents, solutions, and materials used in these experiments were treated for at least 3 hours before use to remove any contaminating iron using an iron-chelating resin, Chelex-100 (Sigma), at 1 g per 100 mL solution.

Fresh red blood cells were obtained from donors with sickle trait (ie, having both HbA and HbS), so that both Hbs studied would be from the same intracellular environment. The two Hbs were isolated, purified, and stripped (of organic phosphates and counter ions) using four sequential chromatographic steps (DEAE-cellulose, Sephadex G100, AG11A8, and AG501×8), as recommended by Caughey and Watkins.15 HbA and HbS were then dialyzed together against the same solution of experimental buffer: 140 mmol/L NaCl, 20 mmol/L HEPES, pH 7.0.

Hb at 1 mg/mL (15.5 μmol/L Hb tetramer, or 62 μmol/L in heme) and the various reactants, each formulated in experimental buffer (with pH adjusted as necessary to 7.0), were prewarmed separately and then admixed in a temperature-controlled spectrophotometer (37 ± 0.1°C). Absorbance at 577 nm was monitored every 30 seconds for 30 minutes to directly measure loss of oxyHb. Simultaneous monitoring of an isosbestic point for oxyHb and metHb (587 nm) showed no change in A₅₈₇ over the course of these experiments, confirming that no species other than oxyHb and metHb appeared in the system to introduce artifacts. Measurements verified that pH remained stable at 7.0 ± 0.01 U during all of these experiments. At this pH, some metHb would be in the form of ferrihemoglobinOH, but this would comprise a small percentage of the total metHb (<10%) at this pH, and its extinction coefficients indicate that its effect on results measured in this fashion would be to blunt, not augment, apparent oxidation rate.16 

To calculate results, the measured absorbance changes were expressed on a log scale and plotted against time, so that initial oxidation rates could be calculated from the slope of the resulting linear plots. Hb oxidation rates are expressed here as the disappearance of oxyHb in units of micromolar heme per minute (μmol/L/min).

Except as specifically noted later, all experiments were conducted with Hb at 1 mg/mL. To identify the inherent oxidation rate and show the presence of adventitious iron, the oxidation rate of the purified Hbs was compared with and without inclusion of 10 μmol/L DTPA, a hexadentate chelator that does not itself stimulate Hb oxidation (see Results).

To identify the oxidation-stimulating effect of iron chelates, we admixed Hb with iron chelates composed of 50 μmol/L Fe³⁺and variable chelator concentrations chosen so that greater than 99% binding of iron was assured in each case, based on reported chelator affinities. Expressed as log of the cumulative stability constant,8 these affinities are as follows: deferiprone (L1), 35.9; EDTA, 25.0; nitrilotriacetic (NTA), 24.3; citrate, 11.5; ATP, adenosine triphosphate, approximately 6; and adenosine diphosphate, approximately 4. Affinities for 2,3-disphosphoglycerate (23DPG) and glutathione (GSH) are not known. For these experiments, results are reported as the absolute increment in Hb oxidation rate measured in the presence of the iron chelate versus the presence of chelator without iron and thus eliminate any potential effects of the chelators themselves on absorbance measurements.

In other experiments, Hbs were exposed to enzymatically generated physiologic oxidants. Superoxide (·O₂⁻) was generated using 500 μmol/L xanthine, 1 × 10⁻⁴ U/mL xanthine oxidase, plus 10 μmol/L DTPA; the addition of catalase did not alter Hb oxidation rate in this system, so we did not routinely include it. H₂O₂ was generated using 500 μmol/L glucose, 4 × 10⁻³ U/mL glucose oxidase, plus 10 μmol/L DTPA. Hydroxyl radical (·OH) was generated using glucose, xanthine, and both enzymes as earlier, but with DTPA replaced by EDTA/Fe³⁺ at 10 μmol/L each. For each of these experiments, the negative control sample consisted of all relevant reagents except the enzyme(s).

To examine for any effect of Hb concentration, we compared oxidation rates for Hbs at several concentrations: 1 mg/mL (our standard condition), 10 mg/mL, and 100 mg/mL. For these experiments, we preserved the same Hb to DTPA ratio used for our standard conditions to ensure that all adventitious iron would be inactivated. Although incubations were performed at the higher Hb concentrations, for analysis, the samples were carefully diluted to [Hb] = 1 mg/mL for reading in the spectrophotometer. Therefore, the oxidation rate results (expressed in μmol/L/min) can be directly compared with those obtained in experiments at low Hb concentration. In similar fashion, we also examined the effect of one chelate, iron/ADP, at Hb and iron and ADP concentrations 10-fold higher than in our standard condition.

To determine if added iron would bind preferentially to HbS over HbA, purified preparations of both were treated with 10 μmol/L DTPA, followed by extensive dialysis (against 1.2 × 10⁵ vol of experimental buffer) to remove DTPA and all adventitious iron as iron/DTPA. Then, Hbs were exposed to iron by incubation with Fe³⁺/ADP (50 μmol/L/1 mmol/L) as earlier, since the poor solubility of iron in water precludes meaningful incubation with free iron per se. After 30 minutes, the Hbs were dialyzed together against 1.2 × 10⁵ vol of experimental buffer to remove ADP and any Fe³⁺ not bound to Hb, and the oxidation rate was then determined with and without 10 μmol/L DTPA.

The comparative Hb oxidation rates we observed here under our standard condition (Hb at 1 mg/mL) before removal of any adventitious iron showed an average 52% rate augmentation for HbS (Table1). The rates in the present report are expressed in the fashion in which the raw data were collected, in units of μmol/L/min. Conversion of these results to pseudo first-order rate constants (units of h⁻¹) yields corresponding values for HbA and HbS of 0.027 h⁻¹ and 0.041 h⁻¹, respectively, which are consistent with those previously reported.1 2 

Inclusion of 10 μmol/L DTPA resulted in a significant decrement in the observed oxidation rate for HbS (Table 1). However, even after any adventitious iron was thus entirely inactivated by DTPA, a significantly increased oxidation rate is still evident for HbS (Table1), substantiating the existence of an inherent increase in an autooxidation tendency for HbS. Interestingly, even though the two Hbs were derived from the same intracellular environment, addition of DTPA did not reduce the rate of HbA oxidation. These oxidation rates recorded in presence of DTPA represent the minimal rate observable under these conditions, as evidenced by the fact that higher DTPA concentrations exerted no additional protective benefit (data not shown). Also, oxidation rates dropped no further even after removal of all DTPA and iron/DTPA by extensive dialysis against iron-free buffer, confirming that the observed rate in the presence of DTPA really represents a minimal value (Table 1).

Since the degree of dimerization increases as the Hb concentration decreases, our standard experimental conditions would entail substantially greater dimerization than under physiologic Hb concentrations (∼320 mg/mL in heme). Therefore, we also determined oxidation rate at a 100-fold higher Hb concentration that would diminish the proportion of dimers by an order of magnitude.^* Results of this experiment (Table 1) were identical to those observed under our standard conditions, establishing that our data are indeed relevant to physiologic Hb concentrations.

Since the previous data suggest that adventitious iron associates preferentially with HbS, we designed an experiment to test this directly (Table 2). We obtained iron-free preparations of Hbs (by addition of DTPA and subsequent removal of iron/DTPA by dialysis) and determined their oxidation rates (Table 2). Then, we reexposed them to Fe³⁺/ADP, after which the two Hbs were extensively dialyzed together to remove ADP and any Fe³⁺ not bound to Hb. Subsequent examination of the oxidation rate showed reinstitution of a greater increase for HbS than for HbA (Table 2). Since direct exposure of HbA and HbS to the same concentration of iron/ADP exerts exactly the same effect on oxidation rate (as shown later), this experiment suggests preferential binding of iron to HbS. Consistent with this, final addition of DTPA restored the basal oxidation rate identified in the iron-free preparations in the first place (Table 2).

Any molecular iron in red blood cells that is not associated with Hb or the membrane is likely to be in the form of low–molecular-weight chelates. Since these are known to promote Hb oxidation,10 11 we tested whether HbS was abnormally susceptible to this effect. All of the tested Fe³⁺ chelates significantly stimulated Hb oxidation, but in each case this incremental effect was equivalent for HbA and HbS (Table3). This excludes an abnormal inherent susceptibility for HbS to be oxidized by Fe³⁺ chelates. However, the abnormal decompartmentalization of iron characteristic of the sickle red blood cell nevertheless predicts that this is a mechanism relevant to biology, because the sickle cell cytoplasm may have low–molecular-weight Fe³⁺ chelates not present in normal red blood cells.

The magnitude of oxidation effected by the different Fe³⁺chelates varied enormously in these experiments, although discerning the reasons why is far beyond the scope of this investigation. Clearly, our expectation, based on earlier and more limited data,10that oxidation rates would be simply related inversely to chelator affinity for Fe³⁺ is not substantiated by our data. The reason for the observed variation must be related to the multiplicity of factors that are known to influence rate of chelate-induced oxidation.10 Significantly, the incremental change in oxidation rate seen in presence of Fe³⁺/ADP was the same for Hbs examined at 1 mg/mL (Table 3) and at 10 mg/mL (.342 ± .020 for HbA and .332 ± .018 for HbS; n = 3).

All three physiologic oxidants (·O₂⁻, H₂O₂, and ·OH) significantly accelerated Hb oxidation, but no difference in susceptibility between HbA and HbS was noted in this regard (Table 4). Again, however, this mechanism remains relevant to biologic HbS oxidation, since the sickle red blood cell generates excessive amounts of these activated oxygen species compared with normal red blood cells.3 

Because our observation of minimal oxidation promotion by the Fe³⁺ chelate with GSH (Table 5)differs from the large GSH effect reported earlier,17 we examined this further. Our data demonstrate that GSH has a hemoglobin oxidizing effect even in the absence of iron (Fig1). However, this effect is dose-dependent, and is only seen at GSH to heme ratios of 2:1 and greater, ratios that far exceed physiologic conditions (ratio of ≃1:4). In fact, these unusual conditions actually were used in that earlier report,17 so this explains why our results differ. Additional experiments performed at two different Hb concentrations substantiate the importance of the GSH to Hb ratio (Table 5).

These results document that purified HbA and HbS preparations, even from the same cellular source (ie, genotypic HbAS donors), exhibit different autooxidation rates. The described data suggest that this is accounted for by differential association of adventitious molecular iron so that sufficient amounts to augment the Hb oxidation rate are carried on HbS, but not on HbA (although our data set does not include direct measurement of such adventitious iron). However, elimination of this effect still leaves an exaggerated oxidation rate for HbS, clarifying the earlier reports of this fact.1,2 In addition to this inherent autooxidation tendency, oxidation of HbS would also be influenced by exogenous stimuli, such as low–molecular-weight Fe³⁺ chelates and physiologic oxidants (ie, oxygen radicals), as shown here. Our data indicate that HbS has no exaggerated inherent susceptibility to such stimuli. However, these mechanisms are still relevant to sickle disease pathobiology, because sickle cells would be subjected to abnormal levels of such stimulation derived from the abnormal iron decompartmentalization7 and excessive oxidant generation3 characteristic of red blood cells in this disease.