Nonenzymatic Galactosylation of Human Serum Albumin . In Vitro Preparation

Incubation of purified human serum albumin with D-[1-14C]galactose (5 mM) or D-[1-14C]glucose (5 mM) in vitro for 7 days under physiological conditions resulted in the time-dependent accumulation of radioactivity into trichloroacetic acid-precipitable material. Comparative studies indicated that the rate of sugar incorporation into albumin increased with increasing pH and temperature of incubation and followed a first order dependence with regard to monosaccharide and albumin concentrations. The extent of nonenzymatic galactosylation of human albumin was approximately 300% greater than the extent of nonenzymatic glucosylation under equivalent experimental conditions. Prolonged dialysis of the modified albumins against a large excess of the unlabeled monosaccharides failed to alter the amount of protein-bound radiolabeled carbohydrate, suggesting that the linkage between sugar and albumin is covalent in nature. The post-translational modification of proteins by nonenzymatic galactosylation may be of physiological significance in individuals with reduced galactokinase or galactose-1-phosphate uridyl transferase activities.

parative studies indicated that the rate of sugar incorporation into albumin increased with increasing pH and temperature of incubation and followed a first order dependence with regard to monosaccharide and albumin concentrations. The extent of nonenzymatic galactosylation of human albumin was approximately 300% greater than t h e extent of nonenzymatic glucosylation under equivalent experimental conditions. Prolonged dialysis of the modified albumins against a large excess of the unlabeled monosaccharides failed to alter the amount of protein-bound radiolabeled carbohydrate, suggesting that the linkage between sugar and albumin is covalent in nature. The post-translational modification of proteins by nonenzymatic galactosylation may be of physiological significance in individuals with reduced galactokinase or galactose-1-phosphate uridyl transferase activities.
It has been suggested that the clinical sequelae of diabetes may accrue from alterations in protein function occurring as a result of glucose-dependent chemical modification (3). At the present time, the relationship between elevated tissue galactose concentrations and observed pathological complications in galactosemic patients remains obscure. However, we suggest that the hypothesis offered for diabetes may also be applied regarding galactosemia.
* This work was supported in part by United States Public Health Service Research Grant NS05104. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
* To whom all correspondence should be addressed.
We present here a preliminary report of the nonenzymatic galactosylation of human serum albumin in vitro, and in addition, we demonstrate that human albumin incorporates galactose both faster and more extensively than glucose under various experimental conditions. A partial description has already appeared (30).

EXPERIMENTAL PROCEDURES
Materials-Human serum was obtained fresh from the University of Rhode Island Health Center (Kingston, RI). Human serum albumin was purified from fresh human serum by affinity chromatography on AB-Gel blue (Bio-Rad) (31), followed by gel filtration on Bio-Gel P-150 (Bio-Rad) prior to recrystallization (32). D-Galactose and D-glucose were purchased from Sigma. ~-[l-'~C]Galactose (6.2 mCi/mmol) was obtained from New England Nuclear, ~-[l-~~C]Glucose (5.8 mCi/ mmol) was purchased from International Chemical and Nuclear. Both chemical and radiochemical punty of the labeled sugars was assured at greater than 99% by the manufacturers; subsequent chromatographic and radiographic analysis in our laboratory confirmed the integrity of the tracers.
Incubations-All solutions and materials were sterilized by autoclaving or by ultrafitration prior to incubation at 37 "C in a shaking water bath. Sodium azide (0.0270) was added to prevent bacterial growth. Unless indicated otherwise, human albumin solutions (10 mg/ m l ) were prepared in Dulbecco's phosphate-buffered saline, pH 7.4 (33). containing either galactose or glucose at a concentration of 5 mM. Trace amounts of the radiolabeled sugars were added to the mixtures to obtain desired specific activity.
Assays-Protein-bound radioactivity was measured by removing 0.04-ml aliquots of incubation mixtures to 0.05 ml of a bovine serum albumin solution (10 mg/ml) and precipitating with 1.0 ml of cold 10% trichloroacetic acid. The mixtures were then centrifuged in an Eppendorf model 5412 Microfuge, and the pellets were washed twice with 1.0 ml of cold 10% trichloroacetic acid. Precipitates were dissolved in 0.5 ml of Protosol (New England Nuclear) and counted for radioactivity in Aquasol-2 (New England Nuclear). Protein was assessed by the method of Lowry et al. (34) with human serum albumin serving as the standard. Fig. 1 compares the incorporation of radionuclide into acidprecipitable material upon reaction of purified human serum albumin with either D-[l-14C]gdaCtOSe or D-[1-14C]glUCOSe. Protein accumulation of radioactivity followed a rectangular hyperbola with increasing incubation for 7 days, at which time approximately 1.6 mol of galactose or 0.5 mol of glucose was incorporated per mol of albumin. Despite the equivalence of reaction conditions, both the initial velocity and the degree of association of radionuclide with acid-precipitable material were significantly greater when galactose was employed relative to when glucose was substituted in the incubations. It should be noted that chromatography of a hydrolysate of albumin, obtained at 7 days from the incubation with galactose, described in Fig. 1, resulted in the recovery of radiolabel in both galactose and talose in a ratio of 3:l. The observation is not unexpected since the proposed reaction mechanism  We decided next to compare the nonenzymatic galactosylation and the nonenzymatic glucosylation of human albumin in vitro with regard to both the extent and the rate of protein incorporation of monosaccharide as a function of various reaction conditions. Figs. 2 through 7 inclusive depict data concerning several of the kinetic and thermodynamic properties of albumin glycosylation in vitro. For example, the rate of incorporation of both galactose (Fig. 2) and glucose (Fig. 3) increased linearly with increasing concentration of the sugar; however, despite the equivalence of reaction conditions, nonenzymatic galactosylation occurred significantly faster than nonenzymatic glucosylation. Similarly, when protein glycosylation occurred in the presence of augmented concentrations of albumin, the linear increase in the rate of nonenzymatic galactosylation (Fig. 4) was again substantially greater than any comparative increase in the rate of glucose incorporation (Fig. 5 ) even though identical reaction conditions were employed. The experimental results described in Figs. 2 to 5 are not surprising, since the reaction mechanism would suggest that the initial rate of albumin glycosylation in vitro is first order with regard to both protein and sugar concentrations separately and second order overall.

RESULTS
Additionally, we attempted to make a comparison of the rates of nonenzymatic galactosylation and nonenzymatic glu- spectively. Once again, the modification of human albumin by galactose exceeded the protein alteration occurring when glucose was substituted in the reaction mixtures.
Finally, it must be emphasized that in all of the experimental circumstances described in Figs. 2 to 7 inclusive, the rate of nonenzymatic galactosylation of human albumin in vitro was consistently 4-fold greater than the rate of nonenzymatic glucosylation despite the apparent identity of reaction environments.

DISCUSSION
Evidence for the nonenzymatic incorporation of glucose into proteins, both in uiuo and in uitro, is overwhelming (1-20), and it has been postulated that the enhancement of this phenomenon in diabetes may contribute to the pathophysiology of the disorder (e.g. atherosclerotic and microangiopathic vascular disease and neuropathy) (3,5,16,(21)(22)(23)(24). Working from this hypothesis we suggested that the elevated tissue galactose concentrations and the accompanying pathological complications in galactosemia may be partially related through the nonenzymatic galactosylation of proteins in uiuo. Accordingly, we decided to examine the nonenzymatic incorporation of galactose into serum proteins in uitro, and the results of this study have been presented here.
Additional experiments enabled us to demonstrate that the rate of protein glycosylation in uitro followed a fist order dependence with regard to monosaccharide (Figs. 2 and 3) and albumin (Figs. 4 and 5) concentrations and increased with increasing pH (Fig. 6) and temperature (Fig. 7) of reaction.
The major finding of this report, however, consists of the observation that galactose nonenzymatically glycosylates human serum albumin in uitro, both faster and more extensively than glucose, under apparently equivalent incubation conditions (Figs. 1-7). Reasons for this behavior are not readily evident. Nevertheless, we can speculate that the greater general instability of galactose relative to glucose might render the former more vulnerable to nucleophilic attack by appropriately reactive groups on the protein. In fact, Kelly (36) has documented the approximately 2-fold greater instability of the preferred conformation of galactose, compared to the preferred conformation of glucose, as resulting from an axial orientation of the hydroxyl group on the fourth carbon of galactose. We may suppose that the ensuing steric hindrance could cause the generation of a higher percentage of the chemically more reactive straight-chain form of the sugar (37). Our conjecture is an adequate interpretation for the observed differences in the initial rates of nonenzymatic galactosylation and glucosylation of human serum albumin. However, the disparity in the overall extent of protein incorporation of galactose and glucose remains to be explained. In spite of this, we believe it is possible that, just as certain sugars are more chemically reactive relative to one another, certain groups on the protein may possess various levels of reactivity, especially where specific monosaccharides are concerned. In other words, it is plausible that there may exist some measure of distinctness between protein galactosylation sites and protein glucosylation sites, which might account for the observed disparity regarding incorporation of certain sugars. It should be mentioned that Dolhofer and Wieland (5) also noticed a difference in both the initial rates and the final extent of nonenzymatic incorporation of mannose and glucose into insulin, with the former combining both faster and to a greater degree. Again, according to Kelly (36), the preferred conformation of mannose is significantly more unstable in comparison to that of glucose, due to the presence of an axial hydroxyl group on the second carbon atom of the mannopyranose ring.
It is conceivable that the galactose-dependent chemical modification of proteins in uiuo might alter both their rate and their path of metabolism, in addition to their proper functioning in the body. For example, the nonenzymatic galactosylation of albumin may affect its ability to transport metabolites such as bilirubin. Indeed, further investigation concerning this phenomenon may provide insight regarding the pathophysiology of certain disorders of carbohydrate metabolism.