Biochimica et Biophysica Acta, 490 (1977) 462470
© Elsevier/North-HollandBiomedicalPress BBA 37576 GLYCOSYLATION OF HEMOGLOBIN S BY REDUCING SUGARS AND ITS EFFECT ON GELATION
PETER M. ABDELLA,JAMES M. RITCHEY, JOSEPH W. O. TAM and IRVING M. KLOTZ Department of Biochemistry and Molecular Biology and Department of Chemistry, Northwestern University, Evanston, 1ll. 60201 (U.S.A.)
(Received August 3rd, 1976)
SUMMARY The binding of various reducing mono- and disaccharides to hemoglobin S has been measured both before and after treatment of the sugar-protein adducts with NaBH4. Incubation of 0.3 M solutions of o-glucose, o-galactose, o-maltose, and lactose, with 2 ~ hemoglobin for 2 h at 37 °C, pH 7.2, leads to the incorporation of 1.1, 1.8, 1.8, and 3.3 mol of sugar, respectively, into I mol of hemoglobin tetramer (either A or S). Exposure of these aldose-protein adducts to NaBH4 for an additional hour at 10 °C increases the binding to 2.0, 3.3, 2.5, and 4.1 mol per mol tetramer, as would be expected if Schiff base linkages were involved in this protein modification reaction. The data suggest a stereochemical requirement for enhanced binding. The dependence of the pre-reduction binding of glucose on the sugar concentration, and on the oxygenation state of hemoglobin has also been examined. Glycosylation of hemoglobin significantly increases the minimum gelling concentration of the deoxy conformation, as measured by sedimentation equilibrium ultracentrifugation. Of the sugar derivatives of hemoglobin S examined by this method, those modified by D-galactose or lactose have minimum gelling concentrations (in the absence of 2,3-diphosphoglycerate) which are comparable to, or greater than, that of fully carbamylated hemoglobin S.
INTRODUCTION It has been known for some time that a hexose adduct of hemoglobin appears in normal bloods, manifested in the form of hemoglobin A~c [1, 2]. The increased concentration of hemoglobin A~c in diabetics has likewise been noted [3, 4]. A likely pathway of linkage of hexose to the protein would be through Schiff base formation, followed by possible reduction to a more stable substituted amine bond. It has been shown in this regard that glucose couples with a Val-His peptide , the N-terminal sequence in hemoglobin. Furthermore, extensive studies [6-8], have demonstrated that aldehydic pyridoxal compounds can be also be so coupled to hemoglobin in vitro. It has seemed appropriate, therefore, to examine the reaction of glucose and other sugars with hemoglobin [9-I1], in vitro, for the formation of carbohydrate adducts
463 should confer increased solubility on hemoglobin and hence countervail the aggregation leading to sickling in hemoglobin S. The technique of equilibrium ultracentrifugation has recently been used to examine various aspects of hemoglobin S gelation [12-14]. An estimate of the effect of each bound sugar on the gelation of hemoglobin S has been obtained through an application of this sedimentation method. MATERIALS AND METHODS Hemoglobin A was prepared from residual normal blood obtained from the hematology laboratories of Evanston Hospital. Dr. George Honig of the Pediatrics Department of the University of Illinois Hospitals (Chicago) supplied samples of sickle-type blood. Plasma-freed cells were washed and lysed, and the membranes removed, as described by Benesch et al. . After centrifugation of the hemolysate at 40 000 × g for about 1.5 h at 5 °C, the supernatant was decanted and dialyzed against 0.05 M Bis-Tris buffer (Grand Island Biological Co.) at pH 7.2, and freed of phosphates as described by Berman et al. . Purified hemoglobin solutions were concentrated by Diaflo ultrafiltration (Amicon Corp., PM30 membranes) and stored in a liquid nitrogen refrigerator (Cryogenics Service Division, Union Carbide). No change in functional or physical properties was noticed during such storage for several months. Hemoglobin concentrations were determined from the absorbance at 540 nm of the cyanomethemoglobin derivative or by the method of Benesch et al. . Solutions containing more than 5 % methemoglobin were routinely discarded. Stock solutions in 0.05 M Bis-Tris buffer (pH 7.2) were prepared which contained radioactively labeled forms of one of the following sugars: D-glucose, methyla-D-glucoside, D-galactose, D-fructose, lactose, D-maltose, and sucrose. These solutions, each containing approx. 6.5/~Ci/ml of radioactivity, were filtered (Millipore Corp., Swinnex Sterilized Filter Unit, Type GS, 0.22/~m) and kept at 4 °C until used. Unlabeled sugars were purchased from Aldrich Chemical Co. Amersham/Searle Corp. supplied the following labeled sugars, all of which were of 98-99 % warranted radiochemical purity: D-[U-~4C]glucose, specific activity 268-297 Ci/mol; methyl a-D- [U-14C]glucopyranoside, 3.0 Ci/mol; D- [UJ4C]galactose, 43 Ci/mol; D - [ U - 1 4 C ] fructose, 296 Ci/mol; [1-14C]lactose, 20 Ci/mol; [U-14C]maltose, 10 Ci/mol; and [U-14C]sucrose, 556 Ci/mol. Other solutions of glucose were prepared in a similar manner but in 0.05 M Tris/chloride buffer, pH 7.2, to assess the effect of buffer nature on the extent of incorporation of labeled compound. Finally, glucose stock solutions, prepared in 0.05 M Bis-Tris/chloride under conditions that would hydrolyze o-glucono-6-1actone (stirring for 1 h under N2 at pH 9-9.5), were brought to pH 7.2 and used for reactions with deoxyhemoglobin. Reaction mixtures generally contained 2 % "stripped" hemoglobin (A or S) and 0.30 M sugar in 0.05 M Bis-Tris/chloride (or Tris/chloride) buffer at pH 7.2. These were incubated at 37 °C from 0.5 to 2.0 h, and then placed in an ice bath for temporary storage. As a control, identical experiments were carried out with hemoglobinfree reaction mixtures. A modification  of the general procedure described by Benesch et al.  was used to reduce Schiff bases on hemoglobin. Reaction mixtures that contained
464 approx. 1.2- 10 -a M hemoglobin monomer, previously incubated for 2 h with 0.30 M labeled sugar, were cooled to l0 °C. To this was added NaBH4 (Ventron Co.), always freshly dissolved in l0 -a M NaOH, to a final concentration of 0.15-0.16 M. The reaction was allowed to proceed for an additional hour at l0 °C. Each of these samples was then immediately dialyzed (in dialysis tubing of Union Carbide Corp.) with stirring against two 500-ml changes of 0.05 M Bis-Tris/chloride (or Tris/chloride), pH 7.2, at 4 °C for an least 8 h each. Solutions prepared in this manner were immediately analyzed for radioactive content. Columns of Sephadex G-50 Fine were employed to isolate the labeled hemoglobin. These were eluted with 0.05 M Bis-Tris buffer, pH 7.2, which had also been used to equilibrate the columns prior to application of sample. Aliquots (50/A each) of the eluted fractions were mixed with 10 ml of Handifluor scintillation fluid (Mallinckrodt Chemical Works) and then assayed in a Tri-Carb scintillation spectrometer (Packard Instrument Co., 3000 Series). To correct for quenching, internal standards were used. Phenol in the presence of H2SO4 was also used to quantitate [17, 18] the amounts of glucose or of a-methylglucoside bound to hemoglobin after exposure to borohydride. For hemoglobin, interfering absorbance from the heme necessitated removal of this group  prior to spectrophotometric analysis. High-speed, meniscus-depletion sedimentation equilibrium experiments were carried out in a Beckman Model E analytical ultracentrifuge, with the Schlieren optical system, according to procedures described by Briehl and Ewert . A suitable concentration distribution is achieved with initial concentrations of about 10~ hemoglobin. All runs were carried out at 20 000-26 000 rev./min at a temperature of 20 °C. For each experiment, hemoglobin concentrations were calculated from the area under the Schlieren curve with optical factors determined for the ultracentrifuge either by standard procedures or from technical information supplied by the manufacturer. The final photograph of each run was analyzed with a Gaertner microcomparator and numerically integrated to give concentration as a function of radial position. Schlieren baselines were obtained from photographs of the plateau region at the start of a run, or by the use of paired reference cells containing the appropriate buffer. Sedimentation velocity experiments were also carried out, using the scanner optical system, at 60 000 rev./min and at 20 °C. Values of S20,w were measured for solutions of increasing hemoglobin concentration, from which plots of ~20,w against hemoglobin concentration were constructed. RESULTS Columns of Sephadex G-50 Fine (exclusion limit 30 000 daltons) were routinely employed to remove unreacted sugar from that bound by hemoglobin. A typical elution pattern is shown in Fig. 1 for galactose. All of the sugars exhibited identical elution profiles. Fig. 1 indicates that the elution volume for the coupled radioactivity corresponds exactly with that for the hemoglobin. Clear separation between the early fractions and the position at which the uncoupled label appears is large enough to allow the hemoglobin to be totally eluted from the column before the free sugar appears. The decrease in hemoglobin concentration to zero before the elution of
Froction number Fig. 1. F r a c t i o n a t i o n (after i n c u b a t i o n at 37 °C) of a reaction m i x t u r e containing h e m o g l o b i n b o u n d o-[U-14C]galactose a n d n o n - i n c o r p o r a t e d D-[UJ4C]galactose. C o l u m n c o n t a i n e d Sephadex G-50 Fine in 0.05 M Bis-Tris buffer at p H 7.2, 4 °C. - . . . . . . . . , c o n c e n t r a t i o n o f h e m o g l o b i n (Hb); - - , c o u n t s per m i n per 0.05 m l ; . . . . . . , c o u n t s per rain in control experiment, incubation without hemoglobin.
unreacted label is also clearly shown in Fig. 1. The same result is obtained for both hemoglobins A and S. With non-reduced samples containing glucose or galactose, a small quantity of high molecular weight polymer (possibly polysaccharide) was frequently eluted with, or slightly before, the hemoglobin. For a 2 h incubation, the actual number of counts due to this impurity amounted to at most about 10~ of that found in the labeled protein, and this was corrected for in the calculation of bound sugar. Samples of labeled fructose and of the labeled non-reducing sugars, a-methylglucoside and sucrose, appeared to be free of this contaminant, and labeled lactose and maltose samples contained smaller amounts of this impurity than did glucose or galactose. The quantity of this impurity increased with longer periods of incubation at 37 °C, but the mechanism of formation has remained obscure. The dependence of incorporation of sugar into hemoglobin on time is shown in Fig. 2 for samples containing either D-glucose or D-galactose. As illustrated, the quantity of D-galactose incorporated into hemoglobin is greater than that observed for D-glucose. The extent of incorporation was the same for hemoglobin S or A.
1.0 1.5 Incubation time (hrs)
Fig. 2. Incorporation of labeled sugars into hemoglobin A or S as a function of time of exposure to the sugar. Incubations were at 37 °C with 2% hemoglobin (Hb) in 0.05 M Bis-Tris buffer, pH 7.2. ©, 0.3 M D-[U-14C]galactose; ~2, 0.3 M D-[U-14C]glucose. Reduction with NaBH4 was performed for an additional hour at 10 °C. O, o-[U-14C]galactose; •, D-[U-14C]glucose. Reaction mixtures were generally not incubated for longer times because of the noticeable rise in methemoglobin content of the samples. For both hemoglobin A and hemoglobin S incorporation of labeled sugar is increased upon exposure of the incubated sample to NaBH4 (for an additional hour at 10 °C). The values of incorporated D-glucose and o-galactose after treatment with borohydride (Fig. 2) are 2.0 and 3.3 mol per mol hemoglobin tetramer, respectively, for samples previously incubated for 2 h in the absence of borohydride. Experiments were carried out to measure the binding of glucose under conditions in which the concentration of unlabeled sugar was varied, while the amount of radioactively labeled material was fixed. The calculated binding of glucose was found to increase linearly with increasing sugar concentration. The binding of glucose to hemoglobin prior to reduction could have been due to the presence of a small amount of o-glucono-6-1actone, an oxidation product of glucose. This lactone might form a covalent adduct to hemoglobin by reaction with a lysine side chain. Therefore, the extent of labeling after alkaline hydrolysis of any of this lactone that might have been present was measured with deoxygenated hemoglobin. Under these conditions, 1.4 mol glucose per mol deoxyhemoglobin tetramer was bound, as compared to 1.5 mol per mol oxyhemoglobin. Clearly the lactone, if present, does not react with hemoglobin. The binding of a series of different (labeled) mono- and disaccharide sugars to hemoglobin, before and after exposure to NaBH4, is summarized in Table I. All tabulated values represent averages of at least two experiments; uncertainty is estimated to be less than ! 0.15 mol sugar per tool of hemoglobin tetramer. As indicated in Table I, galactose, lactose, and maltose all appear to be more strongly bound to hemoglobin than glucose, both before and after treatment of the incubated samples with borohydride. Fructose is only weakly bound to hemoglobin,
467 TABLE I INCORPORATION OF SUGARS BY HEMOGLOBIN Sugar (0.3 M)
Mol sugar bound/mol hemoglobin tetramer After incubation After NaBH4 (37 °C, 2 h) reduction
Glucose" 1.1 a-Methylglucoside* 0.6 Galactose 1.8 Fructose 0.5 Lactose 3.3 Maltose 1.8 Sucrose 10.6
2.0 0.7 3.3 0.8 4.1 2.6 11.2
" Data taken from Ritchey et al. . even after exposure to NaBH4, and appears to behave much like the monosaccharide control for glucose, a-methylglucoside, a non-reducing sugar. This methylglycoside does exhibit some binding to the protein, as verified by the phenol-HzSO4 colorimetric assay, but the extent of labeling is not increased upon exposure to the reducing agent. Sucrose, a non-reducing sugar which only crudely approximates a control nonreducing compound for the other disaccharides examined, is clearly the most strongly bound sugar, even without treatment with NaBFI 4. Only a trivial increase in the extent of sucrose labeling is observed upon exposure to borohydride. Binding experiments were also carried out in 0.05 M Tris buffer, pH 7.2, to assess the effect of buffer nature on the glycosylation reaction. After exposure to NaBH4, no significant difference in the amount of label incorporated was observed between reactions carried out in Bis-Tris or Tris. This experiments shows that the presence of NH2 groups from the buffer does not affect the incorporation reaction. Table II lists values of the minimum gelling concentration (cget) for deoxygenated samples of hemoglobin S and various hemoglobin S derivatives, obtained TABLE II MINIMUM GELLING CONCENTRATIONS OF NATIVE AND MODIFIED DEOXYHEMOGLOBIN S Values determined by equilibrium ultracentrifugation. Modifier
None (hemoglobin S control) Cyanate Glucose (non-reduced) Glucose (reduced) Galactose (non-reduced) Galactose (reduced) Lactose (non-reduced) Lactose (reduced)
17.7 (4- 0.3) 19.5; (21 ") 19.1 19.8 20.2 20.3 21.1 21.8
"Sedimentation equilibrium value for a2c/~2~, estimated from data of Nigen et al. .
468 from sedimentation equilibrium measurements at 22 000 rev./min carried out at 20 °C in 0.05 M Bis-Tris buffer, pH 7.2. Computerized approximations of the area under the Schlieren curve by the method of overlapping parabolas, or by a cubic spline procedure incorporating Simpson's rule, yielded results identical to that obtained by simple rectangular approximation as described by Chervenka . The estimated uncertainty in cge~, from duplicate determinations of the minimum gelling concentration of unmodified deoxyhemoglobin S, was 2 0 . 3 g/dl. Oxyhemoglobins A and S and deoxyhemoglobin A samples did not exhibit gelation. Deoxyhemoglobin S showed a Cge~of 17.7 g/dl (Table II). Glycosylation of hemoglobin appears to significantly increase the concentration of deoxyhemoglobin S required for gelation, with every sugar studied (Table II). The largest increase is observed for NaBH4-treated lactosyl-hemoglobin S, but the minimum gelling concentration for even non-reduced glucosyl-hemoglobin S is comparable to that of the carbamylated protein. A similar constraint on the aggregation tendency of hemoglobin S was also observed in sedimentation velocity experiments carried out with glucosylated protein. The profiles of g20.w versus hemoglobin concentration were identical for samples of deoxyhemoglobin A and glucosylated deoxyhemoglobin S up to 19 g/dl, whereas unmodified deoxyhemoglobin S was found to gel at about 14.5 g/dl. DISCUSSION The present work demonstrates that reducing sugars can be linked to hemoglobin in vitro, presumably through Schiff base linkages. For an aldohexose the reaction may be written as H
HOCHz(CHOH)4C=O + H2N-Hb ~- HOCH2(CHOH)4C=N-Hb + HzO
Upon reaction with NaBH4 the Schiff base should be converted to the more stable secondary amine: H ] NaBH4 HOCH2(CHOH)4C=N-Hb -+ HOCHz(CHOH)4CH2-NH-Hb
Treatment with NaBH4 was carried out under mild conditions so that the hemoglobin molecule should not become denatured. The increase in incorporation of reducing sugars after reduction was not large (Table I). Evidently the equilibrium of reaction 1 must be substantially toward the left side and the rate of formation of Schiff base must not be fast. Very good evidence has been reported recently  that the carbohydrate linked to hemoglobin in vivo passes from an aldimine (Schiff base) to a rearranged ketoimine linkage. Such an Amadori rearrangement may also occur in our experiments in vitro and may be the reason for the strong binding of the reducing sugars even before reduction.
469 Table I also reveals some structural preferences in the incorporation of the reducing sugars. Galactose is bound more strongly than glucose. The disaccharide lactose is clearly the most strongly bound of the reducing sugars. On the other hand the disaccharide maltose is bound only about as well as the monosaccharide galactose prior to reduction, and the increase in binding after reduction is less for the disaccharide. It is not obvious whether a particular disposition of OH groups facilitates binding, presumably by some hydrogen-bonding interactions, or whether the arrangement of axial H atoms strengthens apolar interactions. A combination of both structural constraints may provide the steric disposition necessary for complementary interactions with the specifically arranged protein side chains of hemoglobin. Fructose stands out among the reducing sugars in its weak coupling to hemoglobin. This may be a reflection of a steric effect in the formation of a Schiff base with the ketohexose, for the carbon atom needed for the imine linkage is flanked by two neighboring carbons, instead of by just one as it would be in an aldohexose. The binding of a-methylglucoside (Table I) demonstrates that an interaction of carbohydrate with hemoglobin is possible even without formation of a Schiff-base linkage. This phenomenon is even more strikingly illustrated by the relatively extensive binding of sucrose, far higher than that of any of the sugars examined (Table I). For sucrose, the incorporation is very much greater than that with a sum of equimolar glucose and equimolar fructose. The disposition of combined axial and equatorial groups must be particularly favorable for interaction at a large number of regions in the protein tetramer. All of the hemoglobin S carbohydrate adducts examined showed increased solubilities as compared to untreated protein (Table II). It has been previously reported that solutions of 0.3 M glucose inhibit gelation of hemoglobin S  but this is in the presence of the high concentration of sugar (and 1.5 M hydroxyl groups), whereas in our experiments all unbound sugar was removed by extensive dialysis followed by column chromatography on Sephadex, as in the labeled sugar binding experiments. The increases in minimum gelling concentration in the list in Table II roughly parallel the extent of incorporation of the sugar adduct (Table I). This trend is also reflected by the slightly enhanced gel points of the adducts treated with NaBH+, each reduced sugar showing a higher gel point than the corresponding non-reduced sample (Table I1), in conformance with the higher degree of incorporation (Table 1). All of the sugars examined for their effects on gelling raised the minimum gelling concentration above that for the carbamylated hemoglobin S. In our experiments all of the protein may not have been carbamylated. For fully carbamy|ated hemoglobin S, a2e~2 ¢, the data of Nigen et al.  lead to an estimate of 21 g/dl for the minimum gelling concentration obtained from sedimentation equilibrium experiments. Reduced galactosyl-hemoglobin S and both lactose adducts give values in the same range or slightly higher (Table II). In these in vitro experiments high concentrations of sugar (0.05-1 M) were necessary to obtain substantial extents of incorporation into hemoglobin S. In vivo, plasma concentrations of glucose are much lower. On the other hand the protein is constantly exposed to glucose. In any event it is evident that glycosylation occurs in vivo [1, 2]. It would be of interest to know whether clinical manifestations of sickling are less common in diabetics (with homozygous S genes) where the concentration of glycosylated hemoglobin is higher in blood [3, 4]. Similarly it might be of
470 interest to determine whether galactose, lactose or sucrose in vivo might lead to higher contents of sugar adducts of hemoglobin. ACKNOWLEDGMENT
This investigation was supported in part by Grant No. GM-09280 from the National Institute of General Medical Sciences, U.S. Public Health Service. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Holmquist, W. R. and Schroeder, W. A. (1966) Biochemistry 5, 2489-2503 Bookchin, R. M. and Gallop, P. M. (1968) Biochem. Biophys. Res. Commun. 32, 86-93 Rahbar, S. (1966) Clin. Chim. Acta, 22, 296-298 Trivelli, L. A., Ranney, H. M. and Lai, H.-T. (1970) Blood 36, 852 Dixon, H. B. F. (1972) Biochem. J. 129, 203-208 Benesch, R. E., Benesch, R., Renthal, R. D. and Maeda, N. (1972) Biochemistry 11, 3576-3582 Benesch, R. E., Yung, S., Suzuki, T., Bauer, C. and Benesch, R. (1973) Proc. Natl. Acad. Sci. U.S. 70, 2595-2599 Benesch, R., Benesch, R. E., Yung, S. and Edalji, R. (1975) Biochem. Biophys. Res. Commun. 63 1123-1129 Klotz, I. M. and Tam, J. W-O. (1973) Proc. Natl. Acad. Sci. U.S. 70, 1313-1315 Ritchey, J. M. (1974) Ph.D. Dissertation, Northwestern University, Evanston, I11. Ritchey, J. M., Tam, J. W-O. and Klotz, I. M. (1974) Fed. Proc. 33, 1451 Briehl, R. W. and Ewert, S. (1973) J. Mol. Biol. 80, 445-458 Briehl, R. W. and Ewert, S. (1974) J. Mol. Biol. 89, 759-766 Williams, Jr., R. C. (1973) Proc. Natl. Acad. Sci. U.S., 70, 1506-1508 Berman, M., Benesch, R. and Benesch, R. E. (1971) Arch. Biochem. Biophys. 145, 236-239 Benesch, R., MacDuff, G. and Benesch, R. E. (1965) Anal. Biochem. 11, 81-87 Dubois, M., Gilles,~K. A., Hamilton, J. K., Rebers, P. A. and Smith, F. (1956) Anal. Chem. 28, 350-356 Gray, G. R. (1974) Arch. Biochem. Biophys. 163,426-428 Rossi-Fanelli, A., Antonini, E. and Caputo, A. (1958) Biochim. Biophys. Acta 30, 608-615 Chervenka, C. H. (1973) A Manual of Methods for the Analytical Ultracentrifuge, Spinco Division, Beckman Instruments Inc., Palo Alto, Calif. Bunn, H. F., Haney, D. N., Gabbay, K. H. and Gallop, P. M. (1975) Biochem. Biophys. Res. Commun. 67, 103-109 Freedman, M. L., Weissmann, G., Gorman, B. D. and Cunningham-Rundles,W. (1973) Biochem. Pharmacol. 22, 667-674 Nigen, A. M., Njikam, N., Lee, C. K. and Manning, J. M. (1974) J. Biol. Chem. 249, 6611-6616