The Schiff base adducts of glyceraldehyde with hemoglobin undergo Amadori rearrangement to form stable ketoamine structures; this reaction is similar to the nonenzymic glucosylation of proteins. In the present studies the analogous rearrangement of the Schiff base adducts of glycolaldehyde with proteins has been demonstrated. However, the Amadori rearrangement of this Schiff base adduct produces a new aldehyde function, an aldoamine, which is generated in situ and is capable of forming Schiff base linkages with another amino group, leading to covalent crosslinking of proteins. Sodium dodecyl sulfate gel electrophoresis of the glycolaldehyde-RNase A adduct showed the presence of dimers, trimers, and tetramers of RNase A, demonstrating the crosslinking potential of this a-hydroxyaldehyde. The crosslinked products exhibited an absorption band with a maximum around 325 nm and fluorescence around 400 nm when excited at 325 nm. The crosslinking reaction, the formation of a 325-nm absorption band, and the development of fluorescence were prevented when the incubation was carried out in the presence of sodium cyanoborohydride. This finding indicates that the Amadori rearrangement that generates a new carbonyl function is a crucial step in this covalent crosslinking. Glycolaldehyde could be a bifunctional reagent of unique utility because its crosslinking potential is latent, expressed only upon completion of the primary reaction.We have been interested in the reaction of a-hydroxyaldehydes with proteins, especially in view of our observation that glyceraldehyde is an efficient antisickling agent in vitro that acts primarily at the stage of aggregation of sickle cell deoxyhemoglobin (1). Only five of the 24 amino groups per ap dimer of hemoglobin are capable of forming stable ketoamine linkages with glyceraldehyde (2, 3), in a reaction analogous to the Amadori rearrangement of glucosylated hemoglobin (4). Recently, we have found that even though valine-1 of the a-chain readily forms a Schiff base with glyceraldehyde, it is refractory to subsequent Amadori rearrangement (5). On the other hand, valine-1 of the f3 chain does form stable ketoamine adducts with glyceraldehyde or with glucose by rearrangement. The lysine E-amino groups of hemoglobin also display selectivity in their pattern of reactivity with glyceraldehyde (2) or glucose (6). Therefore, it is clear that our knowledge of the rearrangement-process as it applies to proteins is limited with respect to both the mechanistic and physiological consequences of the process. The present manuscript addresses the latter question by using glycolaldehyde, a simple a-hydroxyaldehyde. We have now investigated whether the Amadori rearrangement reaction also would occur with the Schiff base adducts of glycolaldehyde with proteins. The rearrangement of the glycolaldehyde-Schiff base adduct would generate a new aldehyde function, which should have the potential to react with another free amino group of the protein and result in covalent crosslinking (Fig. 1). The present...