Glycosylated Minor Components of Human Adult Hemoglobin PURIFICATION, IDENTIFICATION, AND PARTIAL STRUCTURAL ANALYSIS*

Human hemolysate contains several minor components designated Hb A1a, Hb A1b, Hb A1c, which are post-translational modifications of the major hemoglobin component A0. Individuals with diabetes mellitus have elevated levels of Hb A1c, a hemoglobin modified with a glucose moiety at the NH2 terminus of each beta chain. A new chromatographic technique using Bio-Rex 70 is described which not only allows complete separation of Hb A1a from Hb A1b but also resolution of Hb A1a into two components, designated Hb A1a1 and Hb A1a2. Carbohydrate determinations with the thiobarbituric acid procedure revealed that Hb A1a1, Hb A1a2, and Hb A1b as well as Hb A1c were glycosylated. Total phosphate analysis revealed 2.06 and 1.01 mol of phosphorus/alphabeta dimer for Hb A1a1 and Hb A1a2 respectively; Hb A1b and Hb A1c contained no detectable phosphate. Hemoglobin incubated with D-[14C]glucose-6-P co-chromatographs precisely with Hb A1a2, strongly suggesting that Hb A1a2 is glucose-6-P hemoglobin. Levels of Hb A1a1 and Hb A1a2 are normal in individuals with diabetes mellitus. Furthermore, diabetic red cells contain normal levels of glucose-6-P. Therefore, glucose-6-P hemoglobin does not serve as a significant precursor to Hb A1c. Instead Hb A1c is formed by the direct reaction of hemoglobin with glucose. This suggests that hemoglobin can serve as a model system for nonenzymatic glycosylation of protein.


Hemoglobins
A,,, A,,, and A,, are negatively charged minor components found in normal human red cells (1) contains other sugar phosphate derivatives of hemoglobin, such as fructose-6-P-hemoglobin as well, cannot be ruled out. In like manner, the more negatively charged Hb A,,, is likely to be an adduct of a sugar moiety containing two phosphates on the p chain of hemoglobin.
Among the phosphorylated intermediates within the red cell, the only two candidates are glucose 1,6-diphosphate and fructose 1,6-diphosphate. The former compound does not form a stable adduct when incubated with Hb AO, while the latter does (8). Unlike synthetic glucose-6-P hemoglobin, our preparations of n-[Wlfructose 1,6-diphosphate hemoglobin have not produced a single major monodisperse chromatographic peak. Only a portion of this synthetic hemoglobin co-chromatographed with Hb A,,, . Thus are all shown. Nineto ten-milliliter fractions were collected. A linear salt gradient of 0 to 0.1 M NaCl (500 ml total) was started at Fraction 120 for the "Hb A,; column and at Fraction 121 for the "Hb A 1e " column.
we have less convincing evidence concerning the structure of Hb -%a,.
Our results are in direct opposition to those of Stevens et al. (8) who concluded that Hb A,, is an adduct between hemoglobin and glucose-6-P. They obtained a crude mixture of nonhemoglobin protein and Hbs A,,, and A,,, and Al,, and rechromatographed it by carboxymethylcellulose chromatography using a stepwise elution. A small heme-containing peak was obtained in the void volume and second peak (Cfold larger) was obtained immediately following the step to a buffer of higher pH. They interpreted these peaks as being Hb A,, and A,,, respectively. In our opinion, their first peak is probably non-hemoglobin protein, which has strong absorp- shown in Fig. 6s. The most notable difference between the profiles of normal (Fig. 1) and diabetic hemolysate is the increase in the quantity of Hb A,,. This increase is in agreement with the results found by several groups (11, 14, 1.5). It has also been reported (16)(17)(18) that the quantity of Hb A ,o,+bj was also elevated in diabetics. Table II shows the quantitation of minor components from six chromatographic runs. It is clear that the amount of Hb A,,, and Hb A,,, (the sugar phosphate derivatives of native hemoglobin) is the same in normal and diabetic individuals.
We suggest that the reported increase in Hb Al(a,bj seen in diabetic hemolysate, may be due to an increase in levels of Hb A,,, or possibly to an increase in the amount of "non-hemoglobin" protein. The amount of heme absorbance in the "non-heme" peak is always elevated in the diabetic. Spectral monitoring of this peak in the visible and Soret regions and, in the presence and absence of dithionite, revealed that the heme absorbance of this peak is not due to ferrous hemoglobin but rather is derived from a spectrally distinct heme protein or proteins.   Fig. 7S, no differences in the levels of glucose-6-P and fructose-6-P were observed in the red cell of normal uersus diabetic individuals. Our values for normal individuals, as well as those of Stevens et al. (8) are very close to those previously reported (19). In contrast, Stevens et al. (8) reported about a l.Bfold increase in these intermediates within diabetic red cells. The reason for this discrepancy is not clear. Theoretically, diabetic red cells would be expected to have normal levels of glucose-6-P since even in normal red cells, hexokinase is operating at its V,,,, and thus the rate of formation of glucose-6-P should be no higher in hyperglycemic individuals.
The normal levels of Hb A,,, and glucose-6-P and fructose-6-P in diabetic red cells provide strong evidence rebutting the proposal (7, 81, that glucose 6-phosphate Hb is the precursor of Hb A,,. Although our studies have not ruled out the possibility that some Hb A,, is derived from glucose-6-P Hb (Hb A,,,), clearly the bulk of Hb A,, does not come from this source. The concentration of intracellular glucose is about 200-fold that of glucose-6-P, while the rate at which glucose-6-P reacts with Hb A,, is at least lo-fold greater than that observed with glucose (7). It is apparent that glycosylation is a direct reflection of levels of different metabolites within the red cells as well as the rate at which these metabolites react with hemoglobin.
Thus, the simplest and most plausible mecha- here is that they are synthesized nonenzymatically by condensation between hemoglobin and intracellular sugars.