The reversibility of the ketoamine linkages of aldoses with proteins.

The reaction of glyceraldehyde (aldotriose) with hemoglobin A is analogous to the nonenzymic glycosylation of the protein with glucose in that the initial reversible Schiff base adduct (aldimine) of aldotriose undergoes Amadori rearrangement as does that of aldohexose to form the more stable ketoamine adduct. The modification of the alpha-amino group on Val-1(beta) of hemoglobin A as a ketoamine (2-oxo-3-hydroxypropyl group) apparently lowers the pKa of the alpha-amino group of the protein, since this derivative of hemoglobin elutes earlier on carboxymethylcellulose columns than the derivatives containing 2-oxo-3-hydroxypropyl groups on the epsilon-amino groups, and unmodified hemoglobin A. Similar chromatographic behavior has been reported for hemoglobin A1c which contains glucose at its Val-1(beta) as the ketoamine adduct. This suggests the similarity in the chemical consequences of having the ketoamine adduct of an aldose at Val-1(beta) of hemoglobin A under physiological conditions. The formation of the 2-oxo-3-hydroxypropyl groups on Val-1(beta) is nearly irreversible as has been suggested for similar adducts of glucose. On the other hand, the 2-oxo-3-hydroxypropyl groups on the epsilon-amino groups appear to be labile. The buffer conditions considerably influence the reversibility of the ketoamine adducts of aldotriose on the epsilon-amino groups; the reversibility is significantly higher in Tris buffers as compared with that in phosphate buffers. It is suggested that under physiological conditions the ketoamine adducts of aldotriose on the epsilon-amino groups exist in equilibrium with the aldimine, the equilibrium being favored toward the ketoamine. The enhanced release of the 2-oxo-3-hydroxypropyl groups in Tris buffers is probably a reflection of the trans-Schiff base reaction of aldimine with Tris. In support of this hypothesis, sodium cyanoborohydride, a reagent selective for the reduction of the aldimine linkages, inhibited the labilizing influence of Tris. A similar reversibility appears to be true of the ketoamine adducts of aldohexoses at the epsilon-amino groups of hemoglobin A, and ribonuclease A as well. This information on the reversibility of the ketoamine adducts of glucose on the epsilon-amino groups of proteins should be useful in understanding the details of the mechanism of nonenzymic glycosylation. It is also expected to influence the interpretation of the measurements of nonenzymically glucosylated hemoglobin A in diabetic patients.