Aggregation of Deoxyhemoglobin S at Low Concentrations*

The self-association of deoxyhemoglobin S was measured in dilute solutions (0 to 5 g/dl) by Rayleigh light scattering at 630 nm and osmometry in 0.05 M potassium phosphate buffer (pH 7.35). Weight and number average molecular weights (M, and M,,, respectively) and the second or higher virial coefficients, B’ were determined. No experimentally significant differences were observed between oxy- and deoxy-Hb S up to the concentration of 2 gldl; their apparent average molecular weights were within experimental error. Above that concentration, both M, and M, of deoxy-Hb S were significantly different from that of oxy-Hb S. The negative second virial coefficient of deoxy-Hb S, observed by both techniques, is consistent with the self-association of this protein. The lack of effect of 0.4 M propylurea on the state of aggregation and the significant influence of 0.1 M NaCl suggests that polar interactions are involved in formation of these aggregates.

The sickling of erythrocytes associated with sickle cell anemia is caused by the polymerization of deoxyhemoglobin S (1). The polymers consist of microtubules of about 180 A diameter whose supermolecular structure is presently in the process of being defined (2,3). The understanding of the mechanism of polymerization of Hb S has also been the subject of active research in recent years (4,5)  Completion of deoxygenation inside the glove box was checked by observing the gelation of a 24 g/d1 solution of hemoglobin S in an open beaker located inside the box or spectrophotometrically in a tonometer (13). With this last method no more than 8% of oxy-Hb was observed.
Hemoglobin solutions were not exposed to the presence of sodium dithionate.
Osmometric pressure data were interpreted on the basis of the equation Where v is the measured osmotic pressure, M, is the number average molecular weight, C is the concentration of the protein (in grams per liter), R is the gas constant, 7' is the absolute temperature, and B' is the second virial coefficient.
Light where n is the refractive index of the solvent, (anlac), is the refractive index increment used, p is the depolarization ratio, and (6 + 6p)/ (6 -7~) is the Cabbanes factor appropriate for the instrument of Brice's design. The Rayleigh ratio R,, representing the ratio of scattered light at 90 to 0" angle, is the actually measured experimental quantity.

Differential
Refractometry-The refractive index increments at constant chemical potential (an/&), were determined using differential refractometer attachment in a Wood light scattering photometer at 630 nm as described by Elbaum and Herskovits (12). Table I    where (1 -a) represents the weight fraction of associated Hb tetramers of molecular weight (Hb,), (m times 64,450) and a is the fraction of undissociated tetramers, free in solution.
In order to fit the experimental data the required a values at any concentration C have to be estimated, employing the "best fit" estimates of the equilibrium constant.
Using the same notation as before (12) Very similar expressions were derived previously by Benerjee and Lauffer (16). Although this linear condensation model is an oversimplification when applied to the average aggregation of deoxy-Hb S, it may be of value during the initial stages of the process, or for purposes of a general quantification of the extent of the apparent association observed by us. Fig. 7 compares the fraction of total number of sites joined to form aggregates of deoxy-Hb S calculated from light scattering and osmometric data. If macromolecules of deoxy-Hb S would follow simple linear condensation at concentration of 3 g/dl, approximately 20% of their sites would participate in the aggregate formation. At higher protein concentration the agreement between the two techniques is poor, probably due to polydispersity of aggregated forms and the assumptions of the model. Whitin et al. (9) has reported on the basis of Rayleigh turbidity that deoxy-Hb S (at concentration of less than 5 g/ dl) underwent some sort of pregelation aggregation. Similar conclusion was obtained by Wilson et al. (8) studying a quasielastic light scattering of deoxy-Hb S solutions at higher concentrations.
Aggregation of deoxy-Hb S hemolysates was observed to occur at concentration greater than 11 g/dl. A linear association was proposed as the mode of interaction during the early stage of aggregation. Lindstrom et al. (10) observed that oxygenated sickle hemoglobin molecules have additional intermolecular interactions not found for normal hemoglobin, and from the magnetic properties of the solvent water proton (spin-lattice relaxation time) they concluded that oxygenated Hb S molecules form aggregates comprising several tetramers.
The presence of a sequential phase change in gelation of hemoglobins has been suggested by Briehl (17) who studied gelation of deoxyhemoglobin S by sedimentation of gels and determination of concentrations of the protein in the solution phase. Conflicting with this, is the data of Williams (6) who did not find significant amounts of higher weight molecular species in deoxy-Hb S solutions, when studied by equilibrium ultracentrifugation.
The data presented here are consistent with the presence of aggregates in low concentration solutions of deoxy-Hb S. Experiments with propylurea, changes in ionic strength, and the effect of inositol hexaphosphate suggest that these aggregates have different properties from the Hb S gel or polymer (18). The bonding is apparently mostly polar instead of hydrophobic and organic phosphate effecters do not seem to influence the extent of aggregation.
If is conceivable that pregelation aggregates could be stabilized by different interacting sites than the ones found in the polymer. It is critical to investigate the range of concentrations between 10 g/d1 and 25 g/d1 of deoxy-Hb S to establish if these aggregates are the precursors of the polymer or an additional property of the mutated hemoglobin.