Nonuniform loss of membrane glycoconjugates during in vivo aging of human erythrocytes: Studies of normal and diabetic red cell saccharides

ARCHIVES

Vol.

OF BIOCHEMISTRY

232, No. 1, July,

AND

BIOPHYSICS

pp. 310-322,

1984

Nonuniform Loss of Membrane Glycoconjugates during in Vivo Aging of Human Erythrocytes: Studies of Normal and Diabetic Red Cell Saccharides KIYOSHI

MIYAHARA

Departments of Biological the Elliott P. Joslin Received

January

AND

MARY

JANE

SPIRO’

Chemistry and Medicine, Harvard Medical Research Laboratory, Boston, Massachusetts 17, 1984, and in revised

form

March

School 02215

and

20, 1934

Changes occurring in membrane saccharides during the in vivo aging of normal human erythrocytes have been evaluated after the fractionation of the red cells into five age groups by density gradient centrifugation. The glycoconjugate fractions studied included sialoglycoproteins, macroglycolipids, low-molecular-weight glycolipids, and Band 3 glycoproteins. All of the carbohydrate constituents of the membrane were found to decrease relative to the total ghost protein as a function of cell age, with the most substantial losses occurring in the macroglycolipids (50%) and Band 3 glycoprotein (30%); the smallest changes were observed in the sialoglycoproteins (13%). No preferential loss of sialic acid or other peripheral sugars was found, making unlikely the importance of glycosidase action in the removal of sugars from the membrane. It is suggested that the changes observed in the composition of the ghosts during aging are best explained by a loss of membrane segments enriched in glycoproteins and glycolipids and deficient in internally located molecules such as spectrin. Analyses were also performed on the glycoconjugate fractions from diabetic erythrocytes separated according to cell age. These erythrocytes, which had glycosylated hemoglobin values twice those of normals, had somewhat smaller amounts of membrane-bound carbohydrate. The difference between diabetic and normal erythrocytes was greatest when young cells were examined (diabetic to normal = 0.93), suggesting that the known increased turnover of red cells in diabetes leads to an early loss of membrane constituents. During its 120-day life span in the circulation, the erythrocyte undergoes a number of changes, including reductions in the level of some intracellular metabolites and enzymes (1) and decreases in membrane constituents such as phospholipids (l), cholesterol (l), and carbohydrates (l-5). In particular, loss of sialic acid residues concomitant with red cell aging has attracted considerable attention, 1 To whom correspondence should be addressed at the Jo&n Research Laboratory, One Joslin Place, Boston, Mass. 02215. This work was supported by Grant AM 23009 from the National Institutes of Health. 0003-9861/84 Copyright All rights

$3.00

0 1984 by Academic Press, Inc. of reproduction in any form reserved.

310

since it has been suggested that enzymatic release of this sugar may play a role in targeting the cells for destruction (6). The finding of decreased sialic acid in erythrocytes from animals with experimental diabetes mellitus has, moreover, raised the possibility that abnormalities in surface saccharides may account for the decreased half-life of red cells in this disease (7). Since the carbohydrate of the erythrocyte membrane is distributed among a number of glycoproteins and glycolipids, measurement of changes in monosaccharides in the whole ghost does not indicate whether the losses are generalized or whether specific glycoconjugates are being removed.

LOSS OF ERYTHROCYTE

MEMBRANE

In the present investigation after centrifugal separation of normal and diabetic erythrocytes into five groups on the basis of cell age, several carbohydrate-containing fractions were studied; these included sialoglycoproteins, Band 3 glycoprotein, macroglycolipids, and low-molecularweight glycolipids. Changes which occur as a function of cell age or the diabetic state were evaluated by polyacrylamide gel electrophoresis, monosaccharide composition, and gel filtration of glycopeptides. The results of these investigations indicated that all of the carbohydrate-containing components of normal erythrocyte membranes decrease during in vivo aging of the red cell, with the greatest change being observed in the Band 3 glycoprotein and the macroglycolipids. Diabetic erythrocyte ghosts contained smaller amounts of total carbohydrate than normals, and this was most apparent in the young cells.

SACCHARIDES

DURING

CELL AGING

311

phosphate dehydrogenase. After separation of the ghosts from an underlying dense white pellet which was most prominent in the uppermost fraction, they were washed four times with the lysis buffer at 40,OOOg for 30 min at 4°C. Analyses perfwmed on hewzolysate. Glucose-6phosphate dehydrogenase activity was assayed at 37°C in 0.1 M Tris-chloride, pH 8.0, containing 0.1 mM EDTA, 10 mM MgCIr, 0.6 mM glucose g-phosphate, and 0.2 mM NADP by measuring the change in absorbance at 340 nm over a IO-min period using a Gilford recording spectrophotometer (10). Hemoglobin was measured as hemoglobincyanide at 540 nm (11) using a commercial standard (Diagnostic Technology, Inc.). Glycosylated hemoglobin was determined using the Fast Hemoglobin Kit (Isolab, Inc.); prior to analysis the hemolysates were kept at 2°C for 24 h to permit decomposition of unstable glucosylated compounds. Preparation of glycoumjugatefractions Erythrocyte membranes suspended in 10 mM Tris-chloride, pH 7.4,0.1 mM EDTA at a concentration of 2.5 mg protein/ ml were mixed with 9 vol of chloroform/methanol (2/l) and shaken vigorously for 40 min at room temperature (12). After centrifugation at 600g for 20 min, the upper fraction (aqueous phase) and lower fraction EXPERIMENTAL PROCEDURES Fractionation of er@hrocytes and preparation of (organic phase) were carefully removed and the residual interphase was again extracted in the same ghosts Fresh blood was collected in EDTA-containing manner. The combined aqueous phases were lyophVacutainer tubes (Becton, Dickinson & Co.) from norilized, resuspended in distilled water, and centrifuged mal volunteers and from insulin-dependent diabetic at 100,OOOgfor 30 min; the supernatant contained the individuals. The bloods were centrifuged at 1000~ for 20 min at 4°C and, after removal of the plasma and sialoglycoproteins (12) and the macroglycolipids (13, 14). The organic phases, in which most of the membuffy coat by aspiration, the erythrocytes were rebrane glycolipids were found, were concentrated on suspended in 0.9% NaCl, 5 mru Tris-chloride, pH 7.4, a rotary evaporator at 3O’C. The interphase material, 0.1 mM EDTA, and washed four times by centrifuwhich contained all of the ghost proteins with the gation. A sample of the washed cells was retained exception of the sialoglycoproteins, had as its primary for ABO typing, and the remainder was fractionated according to cell age by the method of Rigas and glycoconjugate the Band 3 glycoprotein. In order to facilitate aliquoting of this insoluble material, it was Koler (8) with the modification that the ultracentridigested with Pronase (Calbiochem) for 72 h at 37°C fugation was performed in buffered saline (9). The in 50 mM Tris-acetate, pH 7.8,15 mM calcium acetate, washed red cells were resuspended to a hematocrit of 90% in the NaCl/Tris-chloride/EDTA solution and in the presence of toluene. The enzyme was added in an amount equal to approximately 2% of the protein centrifuged at 4°C for 120 min at 52,OOOgin a Beckman to be digested. Recovery of red cell membrane glyL5-65 centrifuge using a SW 27.1 rotor which was coconjugates in these three fractions, as monitored allowed to stop without braking; the tube size was 15 X 100 mm. The supernatant and any visible buffy by sialic acid analyses, was essentially complete. Separation of wuxmglycolipids from sialoglycqnw layer were removed by aspiration, and the column of reins. An aliquot of the aqueous phase from the chloerythrocytes was divided into five equal portions, with roform/methanol extraction containing 0.6 mg protein the top layer representing fraction 1 and the bottom was dissolved in 50 mM Tris-chloride, pH 8,0.2% Trifraction 6. The cells were removed by pipet, resuspended in the buffered saline, and sedimented at 1OOOg ton X-100 and loaded on a small (0.3 ml bed volume) column of DE-52 cellulose (Whatman) which was for 20 min. Ghosts were prepared from total or age-fractionpreequilibrated with 50 mM Tris-chloride, pH 8. The ated erythrocytes by lysis with 9 vol of 10 mM Trismacroglycolipids were obtained by washing with 10 chloride, pH 7.4, 0.1 mM EDTA, 2 mM phenylmethml of this buffer, and the sialoglycoproteins were ylsulfonyl fluoride by centrifugation at 40,OOOgfor 30 eluted with 10 ml 50 mM Tris-chloride, pH 8.0, conmin. The hemolysate was saved for measurement of taining 0.2% Triton X-100 and 0.5 M NaCI. Both frachemoglobin, glycosylated hemoglobin, and glucose-& tions were dialyzed extensively and then lyophilized.

312

MIYAHARA

Electrophoresia Examination of the ghost proteins and the components of the glycoconjugate fractions was performed by polyacrylamide slab gel electrophoresis in SDS’ according to the procedure of Laemmli (15) using a 3-mm gel, with 3% acrylamide in the stacking gel and a linear 7 to 12% acrylamide gradient in the separating gel. The samples in 80 mM Tris-chloride buffer, pH 6.8, containing 4% SDS and 5% 2-mercaptoethanol were heated for 2 min at 100°C and incubated for 1 h at 37°C before electrophoresis. The gels were stained with Coomassie blue for protein and with the PAS reagent for carbohydrate according to the method of Fairbanks et al (16). Anulyses. Sialic acid was released from ghost membranes and the glycoconjugate fractions by hydrolysis with 0.1 N sulfuric acid at 80’ for 1 h, and was determined by the thiobarbituric acid assay of Warren (17). The sialic acid of the glycolipids, after release by the mild acid hydrolysis, was adsorbed on and eluted from Dowex 1 formate (18) prior to its colorimetric determination. For analysis of the sugar composition of the glycoconjugate fractions, hydrolysis was performed in 1 N HCl for 6 h at 100°C under nitrogen, after which the hydrolysates were passed through coupled columns of Dowex 50 (H+ form) and Dowex 1 (formate form) (18). The amino sugars eluted from the Dowex 50 columns were measured on the Technicon NC-2 analyzer with a pH 5.0 gradient (19), while the neutral sugars were determined by borate-complex anion-exchange chromatography by a modification of a previously published method (20). In order to enhance the sensitivity of the technique, the ion-exchange column containing DA-X4-20 resin (Dionex) and maintained at 60°C was reduced in size to 0.6 X 36 cm. A three-chambered gradient was used, consisting of 73 ml 0.4 M sodium borate, pH 8.0; 71 ml 0.4 M borate containing 50 mM sodium sulfate; and 69 ml 0.4 M borate containing 0.3 M sodium sulfate in chambers 1,2, and 3, respectively. Together with each group of samples analyzed, a mixture of standard sugars was carried through the entire procedure so that corrections could be made for recoveries, which averaged 85% for the hexosamines and 88% for the neutral hexoses. Amino acids were determined on a Technicon NC2 amino acid analyzer after hydrolysis of the samples in constant-boiling HCl in sealed tubes under nitrogen at 105°C for 24 h. These analyses were routinely utilized for determining the protein content of the interphase fraction, since its solubilization with Pronase precluded the use of more conventional protein assays. Protein contents of the ghosts and the sialoglycoprotein fraction were measured by the method of

’ Abbreviations PAS, periodic

used: SDS, sodium acid-Schiff.

dodecyl

sulfate;

AND

SPIRO

Lowry et al (21) using bovine serum albumin as a standard. The ghosts were solubilized prior to this analysis by incubation with 0.1 N NaOH at 37’C for 2 h. Radiolabeling of glpxonjugates. Intact ghosts representing 6 mg protein were labeled by reduction with tritium-labeled sodium borohydride (New England Nuclear; sp act, 330 mCi/mmol) after treatment with sodium metaperiodate (22) or with galactose oxidase (23). Prior to either treatment, membrane samples were reduced at pH 8.0 with unlabeled sodium borohydride (5 mM) and washed three times by ultracentrifugation. The galactose oxidase (Sigma Chemical Co.) was treated with 2 mM phenylmethylsulfonyl fluoride at 22°C for 20 min before being added to ghosts; incubations were performed in 0.15 M NaCl, 50 mM sodium phosphate, pH 7.8, with 7 units of the enzyme for 60 min at 37°C. After centrifugation, the ghosts were resuspended in the buffer and reacted with 6 qmol radiolabeled sodium borohydride at room temperature for 30 min, followed by 2 mg unlabeled sodium borohydride for 10 min. Periodate oxidation was carried out by treating the ghosts with 5 mM sodium metaperiodate in 0.1 M sodium acetate, 0.15 M NaCl, pH 5.0, for 10 min at 0°C in the dark. The remaining periodate was removed by addition of an excess of ethylene glycol. After centrifugation and washing with 0.15 M NaCl, 50 mM sodium phosphate, pH 7.8, the ghosts were resuspended in this buffer and reduced with radiolabeled sodium borohydride as described above. Radiolabeled ghosts were examined directly by SDS-polyacrylamide electrophoresis; in addition, their glycoconjugate fractions were prepared by the chloroform/methanol extraction procedure. Thinlayer chromatography of the glycolipids was performed on silica gel 60 (0.2 mm thickness, Merck) in chloroform/methanol/water (100/42/6). The radioactivity in individual components separated by thinlayer chromatography and detected by fluorography was measured by scraping the silica gel from the appropriate segments directly into scintillation vials and solubilizing the lipids in chloroform/methanol/ water (10/10/3) prior to addition of scintillation fluid. Dried polyacrylamide gels were cut into segments and allowed to hydrate in 0.1 ml distilled water in scintillation vials; after addition of 10 ml 3% Protosol in Econofluor (New England Nuclear), incubation was carried out at 37” for 24 h. Radiolabeled glycopeptides were prepared by application of the galactose oxidase-tritium-labeled borohydride procedure to the Pronase digests of the interphase proteins. Gel filtration of these compounds was then performed on Bio-Gel P-10 (1.8 X 120 cm) in 0.1 M pyridine acetate (pH 5.0). Dekrmination of radioactivity. Radioactive components on thin-layer sheets or dried polyacrylamide slab gels were detected by fluorography by exposure

LOSS

OF

ERYTHROCYTE

MEMBRANE

at -70°C after treatment with Enhance (New England Nuclear) using X-Omat AR film (Eastman-Kodak). Scintillation counting was performed in a Beckman LS 7500 instrument with Ultrafluor (National Diagnostics). Statistics. The change in content of erythrocyte constituents from young cells (fraction 1 of the gradient) to old cells (fraction 5) was determined by linear regression analysis, with the t value being determined from the correlation coefficient (r), using the formula t* = r’(N-2)/l-r2 (24). as is discussed in greater detail under Results. When data from unfractionated ghosts were compared, the values were expressed as the mean + standard error of the mean, and Student’s t test was used. RESULTS

Fractionation of erythrocytes to age. Since previous studies

according

have indicated that, during centrifugal fractionation, the youngest red cells have the lowest density and are found at the top of the gradient (8, 25), the designations young cells and old cells have been used in this report to indicate the top (fraction 1) and bottom (fraction 5) of the gradient. Glucose-6-phosphate dehydrogenase, an enzyme known to decline during erythrocyte aging (l), was used as an indicator of the N r 1234512345

D

SACCHARIDES

DURING

CELL

AGING

313

success of the ultracentrifugal fractionation technique used in the present study. For normal cells (n = 9) the average activity of this enzyme was 11.2 pmol substrate min-’ mg hemoglobin-’ in the youngest cells and 7.5 pmol min-’ mg hemoglobin-’ in the oldest cells, representing a decrease of 33%. Similarly, in the diabetic red cells studied (n = 5) a decline of 41% occurred, from an activity of 10.7 in the young to 6.4 pmol min-’ mg hemoglobin-’ in the old erythrocytes. Ekctrophoretic anal~ti. The components of the ghosts from the five age groups of erythrocytes were examined on SDS-polyacrylamide gels, and the pattern of the proteins stained with Coomassie blue and the glycoproteins detected with the PAS reagent are shown in Fig. 1. The main sialoglycoprotein, PAS-l (glycophorin A) (16), is clearly evident in all of the fractions after staining with the PAS reagent, while the band 3 glycoprotein, which does not react with the PAS stain, is seen as a diffuse Coomassie blue-staining band (26). The membranes from all of the fractions of both normal and diabetic erythrocytes had similar patterns; no new components N 123451

D 2345

,

FIG. 1. SDS-polyacrylamide electrophoresis of the membranes of age-fractionated normal (N) and diabetic (D) erythrocytes. The gels were stained either with Coomassie blue (left panel) or with the PAS reagent (right panel). The numbers at the top refer to the gradient fraction of the erythrocytes, with 1 representing the youngest cells. The designations for the protein components (at the left) and the glycoproteins (at the right) are those of Fairbanks et al. (16). Aliquots of ghosts equivalent to 60 pg protein were applied to each slot; the stacking gel was 3% acrylamide while the running gel was a 7 to 12% acrylamide gradient. Electrophoresis was performed at 10 mA/gel for 17 h.

314

MIYAHARA

AND

were detected with aging, and the migration of the major glycoproteins, PAS-l, PAS-2, PAS-3, as well as Band 3, remained the same with increasing cell age. When the gels were scanned and the amounts of the components present were compared by densitometry, no change in the ratios was observed with aging or between normal and diabetic cells. Composition of glycownjugates obtained from ghosts by chloroform/methanol extraction. The procedure of Hamaguchi and Cleve (12) proved effective in the separation of three groups of glycoconjugates. In agreement with these investigators, we observed that the aqueous phase contained approximately 5% of the ghost protein and 85% of its sialic acid (Table I). Electrophoresis of this fraction (data not shown) indicated that it contained primarily PAS1, with much smaller amounts of PAS-2, 3, and 4 also being evident; the only protein bands visualized by Coomassie blue were

TABLE CHANGEIN~ARBOHYDRATE

CONSTITUENTSOF

SPIRO

those corresponding to the glycoproteins. However, as will be discussed later, about 10% of the total carbohydrate of the aqueous phase was found to be present in the compounds which have been designated as macroglycolipids or polyglycosylceramides (13, 14). The remainder of the red cell protein remained at the interphase during the extraction procedure. When this material was analyzed by SDS-polyacrylamide electrophoresis (data not shown), it was found to contain all of the protein bands seen in the intact ghosts by Coomassie blue staining. Only about 10% of the ghost sialic acid was present in this fraction, and its sugar composition (primarily galactose and N-acetylglucosamine) was consistent with the presence of the Band 3 glycoprotein (27-29). The components of the organic phase are the gangliosides and low-molecular-weight neutral glycosylceramides of the red cell

I

NORMALERYTHROCYTE

GHOSTSDURING

Monosaccharide/ghost Aqueous-phase glycoconjugates

Sugar Gala&se N-acetylglucosamine N-acetylgalactosamine Mannose Fucose Glucose Sialic acid

Interphase

Young” cells

Old cells

Old/ young

Young cells

63.1 30.0 33.6 9.32 9.31 c 73.4

52.4 24.4 29.0 8.49 7.01

0.83* 031** 0.86 0.91 0.75

68.4

0.93

52.2 52.1 b 20.5 10.7 d 11.3

a After separation of erythrocytes into five fractions by top layer (fraction 1) represented the young cells and the reported are derived by linear regression from analyses individuals, five of blood type 0 and two of blood type A; coefficient as described under Materials and Methods. *Analyses for this constituent not performed. c These constituents were present in amounts below the d Glucose values for the interphase proteins were highly * P < 0.05. ** P = 0.05. *** P < 0.01.

protein

CELL AGING

(nmol/mg) Organic-phase glycolipids

glycoproteins Old cells

Old/

young

37.1 33.5 b 13.0 6.85

0.71* 0.74*** b 0.63* 0.64*

10.0

0.33

in Viwo

Young cells

Old cells

Old/ young

59.4 4.36 22.2 c

47.1 3.25 17.6

0.79** 0.74 0.792

2si 2.34

22.5 1.5

0.76 0.78

ultracentrifugation as described in the text, the bottom layer (fraction 5) the old cells. The data performed on the red cells from seven normal statistics were performed using the correlation

limits of detection. variable and are not reported.

LOSS OF ERYTHROCYTE

MEMBRANE

membrane; approximately 2% of the sialic acid was found in this fraction. Changes in membrane carbohydrates during cell aging. Although the erythrocytes from each blood were separated by centrifugation into five age groups, it was not feasible on a routine basis to perform compositional studies for the separated glycoconjugates from each of these five cell fractions. In order to determine whether the changes in sugar content could be described by a linear regression, as has been shown for other red cell constituents (l), analyses were performed on all five age groups from two normal and two diabetic samples. In Fig. 2 the average total saccharide content of the aqueous-phase glycoconjugates of each gradient fraction is plotted along with the linear regression curve calculated from the individual values obtained for each blood; the correlation coefficient determined from these data was 0.66 (P < 0.005). The sugars of the glycolipids present in the organic phase and of the Band 3 glycoprotein of the interphase were also found to decrease linearly with cell age, with the correlation coefficients being 0.59 and 0.60, respectively (P < 0.01). Since the upper gradient fraction representing the youngest cells was the most difficult to collect quantitatively because of the need to separate its ghosts from an underlying dense pellet which appeared fibrous under the microscope and contained proteolytic enzymes typical of polymorphonuclear leukocytes,3 analyses were routinely performed on fractions 2 and 5 of the gradient. Regression analysis was performed using these values, and the data presented for “young cells” (fraction 1) and “old cells” (fraction 5) were determined in this manner; statistical significance of the decrease was obtained with the use of the correlation coefficient. From the data presented in Table I for normal erythrocytes it is clear that, during in viva red cell aging, all of the carbohydrate-containing molecules undergo changes and that losses occur in each of the sugar constituents. Sialic acid is the a K. Miyahara vations.

and M. J. Spiro, unpublished

obser-

SACCHARIDES

DURING

CELL

AGING

315

FRACTION

FIG. 2. Change with cell age of aqueous-phase carbohydrate. A summation was made of the monosaccharides present in the aqueous-phase glycoconjugates in each of five age groups from four blood samples, and the average of the values for each fraction is plotted, with fraction 1 representing the youngest cells. The values were calculated relative to the milligrams of original ghost protein from which the aqueous phase was derived. The dashed line represents the linear regression curve calculated from the individual values; the correlation coefficient obtained was 0.66 and the P value derived from this was <0.095. The average values plotted (+ the standard error of the mean) were 1, 262 (k8.8); 2, 196 (k3.2); 3, 186 (k8.6); 4, 171 (214.5); and 5, 163 (k9.9).

least affected of the monosaccharides, particularly in the components present in the aqueous phase, while the galactose and Nacetylglucosamine in this fraction show a more substantial decline. Greater losses of monosaccharides were observed from both the interphase protein and the organicphase glycolipids, with the major constituents of the Band 3 glycoprotein decreasing by about 30% and the loss of the glycolipids being about 20%. Compariscm of changes in sialogljpoproteins and mwrogl~col&ids. The greater loss of galactose and N-acetylglucosamine than sialic acid in the aqueous phase suggested that the macroglycolipids, which are also present in this fraction, might be removed from the membrane at a faster rate than the glycophorins. Since these neutral hydrophilic glycolipids can be separated from the negatively charged sialoglycoproteins by anion-exchange chromatography (13),

316

MIYAHARA

AND

DE-52 cellulose fractionation was used for this purpose. The material not adsorbed on the DE-52 column contained galactose and N-acetylglucosamine in equimolar amounts (Table II); small amounts of fucase and N-acetylgalactosamine were also present. All of the sialic acid and most of the N-acetylgalactosamine were retained by the column, and the composition of the material which eluted with 0.5 M NaCl (Table II) is characteristic of glycophorin A, the principal sialoglycoprotein (30). The relative changes in the macroglycolipids and sialoglycoproteins were found to be quite different; while a decrease of only about 15% was observed in the glycoproteins, a much greater proportion (50%) of the macroglycolipid was lost (Table II). Glycopeptides

from

pmg

quences (27,28). One explanation for such heterogeneity could be the removal during aging of portions of the polysaccharide units, which would manifest itself in different-sized glycopeptides being obtained from young and old erythrocytes. The elution patterns obtained upon gel filtration of the Pronase digests of interphase proteins from fractions 1 and 5 from normal erythrocytes is shown in Fig. 3; little decrease in size appears to occur during aging, suggesting that the loss of Band 3 carbohydrate probably occurs through removal of the intact glycoprotein rather than by glycosidase action. Glycopeptides from fractions 1 and 5 from four bloods were examined (2 normal and 2 diabetic); patterns similar to those shown in Fig. 3 were obtained from all.

and old cells.

Comparison of glgcoconjugates mal and diabetic eqthrocytes.

The substantial decline in the carbohydrate content of the interphase protein made this fraction a good choice for a study of the mechanism of this loss. The carbohydrate units of the Band 3 glycoprotein, the primary glycoconjugate of the fraction, have been reported to be of high molecular weight, with as many as 70 sugar residues, and to show size heterogeneity due to the presence of chains of varying length containing repeating N-acetyllactosamine se-

OF Loss OF SUGARS PROTEINS DURING

FROM

in

MACROGLYCOLIPIDS AND viva RED CELL AGING’

protein

Macroglycolipids

Galactose iV-acetylglucosamine iV-acetylgalactosamine Sialic acid

nor-

II

Monosaccharide/ghost

Sugar

from

The sialic acid content of the intact erythrocyte membranes was found to be slightly higher (97.2 nmol/mg ghost protein; n = 18) for normals than for diabetics (91.1 nmol/mg protein; n = 12). When the monosaccharide composition of the three glycoconjugate fractions from young and old erythrocytes from diabetic blood was examined (data not shown), it was observed that the diabetic cells had somewhat lower amounts

TABLE COMPARISON

SPIRO

SIALOGLYCO-

(nmol/mg) Sialoglycoproteins

Young cells

Old cells

Old/ young

Young cells

Old cells

Old/ young

14.4 16.7

7.5 8.4

0.52* 0.50*

42.2 15.0 31.1 75.2

35.9 11.4 25.5 67.5

0.85 0.76 0.82 0.90

o Macroglycolipids and sialoglycoproteins were separated by DE-52 cellulose chromatography phases obtained by chloroform/methanol extraction as described under Materials and were performed on five erythrocyte fractions separated by density gradient centrifugation of four individuals. * P -C 0.01, determined from the correlation coefficient as explained in the text.

of the aqueous Methods; analyses from the bloods

LOSS

OF

ERYTHROCYTE

10 TUBE

20

30

NUMBER

MEMBRANE

SACCHARIDES

40

interphase 5 (A) of the reacted with reduced with act, 330 mCi/ then filtered cm), with 0.1 of 12 ml/h. by filtration

not only of sialic acid but of other monosaccharides as well, and that this was true particularly in the younger cells. However, TABLE Loss

OF MEMBRANE

GLYCOCONJUGATES

DURING

OF NORMAL

Monosaccharides/ghost

AND DIABETIC protein

Normal

Fraction Aqueous-phase glycoconjugates’ Interphase glycoproteins Organic-phase glycolipids Total

317

AGING

III

in Viva AGING

Total

CELL

the differences were too small to be statistically significant. A summation was made of the total sugars present in each fraction of both normal and diabetic erythrocytes (see Table III). While the youngest diabetic ghosts were found to have only 93% as much carbohydrate as normals (449 nmol total saccharides/mg ghost protein), the oldest cells contained 372 nmol sugars/mg ghost protein, which was 96% of normal. The loss of sugars during aging of the diabetic erythrocytes was less than that observed for normals and, since the content of sugars was almost equivalent in the oldest cells of both types, the smaller decrease may indicate that the youngest cells of the diabetics have already undergone a loss of membrane constituents. Radiolabeling of normal and diabetic er@hroc@e rnewzbranes. Another manner of comparing normal and diabetic red cell glycoconjugates which was performed was the assessment of their ability to incorporate radioactivity from labeled sodium borohydride after oxidation of peripheral sialic acid or galactose. Normal and dia-

(5 ML)

FIG. 3. Glycopeptides prepared from proteins (5 mg) of fractions 1 (0) and erythrocytes from a normal blood were gala&se oxidase and were subsequently tritium-labeled sodium borohydride (sp mmol). The labeled glycopeptides were on a column of Bio-Gel P-10 (1.8 X 120 M pyridine acetate, pH 5, at a flow rate The void volume of this column (measured of bovine serum albumin) was tube 21.

DURING

RED CELLS

(nmol/mg)* Diabetic

Young” cells

Old cells

Old/ young

Young cells

Old cells

Old/ young

219 147 118 484

190 104 92 387

0.87” 0.71** 0.78*** 0.80

212 127 110 449

177 102 93 372

0.83* 0.80* 0.84 0.83

’ The designations, young and old cells, refer to the top and bottom fifths of density gradient separation of erythrocytes, as described under Materials and Methods. *The data given represent the summation of the individual monosaccharides present in each fraction; statistics for the loss of total sugars during aging were determined by use of the correlation coefficient, as described in the text. Normal values are derived from Table I, while those of the diabetics are based on analyses of fractions from five patients with insulin-dependent diabetes, four of blood type 0 and one of blood type A. The glycosylated hemoglobin value for the normals was 6.6 + 0.3, while that of diabetics was 13.4 f 1.5 (P < 0.001). Differences between normal and diabetic glycoconjugates were not statistically significant. ‘The aqueous-phase glycoconjugates include, as shown in Table II, both sialoglycoproteins and macroglycolipids. * P < 0.05. ** P < 0.02. ***p = 0.02.

318

MIYAHARA

AND

betic ghosts treated with sodium periodate acquired similar amounts of radioactivity after borohydride reduction (Table IV). Analyses by SDS-polyacrylamide electrophoresis of both the intact ghost and the extracted sialoglycoproteins (data not shown) indicated a similar distribution of the tritium label among the glycoprotein components, with PAS-l having 63% of the radioactivity, and 6, 15, and 13% being present in PAS-4, PAS-2, and PAS-3, respectively. Examination of the glycolipid extracts from the labeled ghosts by thinlayer chromatography, followed by counting of the individual gangliosides detected by fluorography after elution from the thin layer segments, showed no difference between the normal and diabetic patterns (data not shown). Reduction with sodium borotritiide after galactose oxidase treatment, however, led to the incorporation of more radioactivity

into the intact diabetic membranes as well as into the three glycoconjugate fractions isolated after chloroform/methanol extraction. Using the radiolabeling after periodate oxidation as the baseline, the activity incorporated into the diabetic carbohydrates was substantially enhanced (Table IV). The nature of the increased radioactivity was explored by SDS-polyacrylamide electrophoresis of the intact ghosts as well as by thin-layer chromatography of the glycolipid fraction. Counting of gel segments after electrophoresis indicated that, while approximately one-third of the radioactivity was recovered in the area of Band 3 in the case of both normal and diabetic membranes, more label was found in the region of macroglycolipids and glycolipids migrating with the indicator dye in the case of the diabetic cells than in the normals (45% compared to 35%). The data in Table

TABLE RADIOLABELING

SPIRO

IV

OF ERYTHROCYTE GHOST GWPONENTS WITH SODIUM BOROTRITIIDE TREATMENT WITH PERIODATE OR GALACTOSE OXIDASE Radioactivityb

Fraction Ghost

membranes

Aqueous-phase

Interphase

Glycolipids

glycoconjugates

glycoproteins

Treatment” A. Periodate B. Galactose A/B” A. Periodate B. Galactose A/B A. Periodate B. Galactose A/B A. Periodate B. Galactose A/B

Normal

oxidase

oxidase

oxidase

oxidase

44.7 8.5 5.8 33.4 1.1 34.6 9.7 4.2 2.5 4.7 1.8 3.1

+ + + + zk + + f k + k +-

AFTER

(dpm Diabetic

1.7 1.7 1.0 2.0 0.2 5.5 0.5 0.5 0.3 0.2 0.4 0.6

44.4 11.1 4.5 31.0 1.4 24.3 9.5 4.7 1.9 5.1 2.6 2.2

+ + * + f + + * + zk + f

X 10m6) D/N

1.6 1.5 0.7 1.6 0.3 4.0 0.8 0.8 0.2 0.3 0.9 0.4

0.99 1.30 0.78 0.93 1.27 0.70 0.98 1.12 0.78 1.09 1.44 0.71

“Ghost membranes equivalent to 5 mg protein were reacted with tritiated sodium borohydride after treatment with periodic acid or galactose oxidase; aliquots were then extracted with chloroform/methanol to obtain the three glycoconjugate fractions, as described in the text. Red cells from four normal and six diabetic individuals were studied; the glycosylated hemoglobin values were 5.6 +- 0.2 for the normals and 11.4 + 0.8 for the diabetics (P -C 0.01). The sialic acid content of the normal and diabetic ghosts was 0.57 and 0.51 pg/mg hemoglobin, while the protein was 20.1 and 17.1 pg/mg hemoglobin, respectively. b The values were expressed as the means + standard error of the mean for 5 mg of original ghost protein. Recovery of radioactivity in the three glycoconjugate fractions after periodate treatment was 107% for normal cells and 103% for diabetics, while for the galactose oxidase-treated ceils it was 84% for normals and 78% for diabetics. c These values represent the averages of the ratios for the individual ghosts.

LOSS

OF

ERYTHROCYTE

MEMBRANE

IV also suggests a greater incorporation of activity into the lipid fraction in the diabetic erythrocytes. Examination of the glycolipid pattern by thin-layer chromatography indicated a similar distribution of components between normals and diabetics, with none being preferentially elevated. DISCUSSION

The development of centrifugal techniques (8, 9, 25) for the separation of erythrocytes on the basis of their age has made possible the study of changes occurring in these cells during their extended life span in the circulation. As these cells age there is an increase in their density coupled with a diminution in their size; a number of intracellular enzymes and metabolites decrease as do certain membrane constituents, including protein, sialic acid, and cholesterol (1). The loss of peripheral sialic acids, thereby exposing galactose and N-acetylgalactosamine residues which may be involved in recognition phenomena, has been considered of potential importance to removal of red cells from the circulation (6, 30). As the effect of aging on red cell carbohydrate has been studied in more detail, a decrease of not only sialic acid but of the other membrane-bound monosaccharide constituents has been observed as well (25). However, previous investigations failed to indicate which macromolecules were affected by this process. The present study, in which erythrocytes were separated into five age groups by centrifugation, focused attention on the changes occurring in several distinct groups of carbohydrate-containing molecules, the sialoglycoproteins, the macroglycolipids, the Band 3 glycoprotein, and the low-molecular-weight glycolipids. The carbohydrate of all of these molecules was found to decrease with age, with the macroglycolipids and Band 3 glycoprotein showing the largest change and the sialoglycoproteins the smallest. In this study we have referred to the interphase protein as Band 3 although a component designated as Band 4.5, which is detected by labeling with sodium borotritiide after

SACCHARIDES

DURING

CELL

AGING

319

galactose oxidase treatment of intact cells or ghosts, has been shown to have similar carbohydrate units (31,32). To evaluate the contribution of Band 4.5 to the interphase carbohydrate, ghosts labeled in this manner were separated into the three glycoconjugate fractions by the usual procedure and the interphase protein was examined by SDS-polyacrylamide electrophoresis and fluorography. The distribution of galactose oxidase-sensitive groups between Bands 3 and 4.5, determined by densitometric scanning of the fluorogram, indicated that 88% of the radioactivity was with the former and 12% with the latter. Because of the similarity of the carbohydrate units of the two proteins, it is felt that the labeling after galactose oxidase treatment also reflects the distribution of total carbohydrate, and that Band 4.5 therefore makes only a small contribution to the interphase glycoproteins. There was no indication of removal of saccharides by glycosidase action; the sialic acid, which as a peripheral sugar might be considered a prime target for such enzyme action, showed the smallest decline of any monosaccharide. This is in agreement with a study of glycophorin from young and old red cells which demonstrated that the ratio of sialic acid to protein measured by iodination remained constant regardless of the age of the cell (33). Moreover, the similar size of the glycopeptides obtained from the Band 3 glycoprotein from young and old erythrocytes argues against the importance of glycosidase action in causing a loss of carbohydrate from this fraction under in vivo conditions. The susceptibility of intact erythrocytes and ghosts to in vitro treatment with an endo-&galactosidase from Escherichia freundii has been demonstrated (31, 32). The components of the red cell membrane from which saccharide was removed by this enzyme were Band 3 and 4.5 proteins, and the macroglycolipids and a considerable sharpening of the electrophoretic profile of the proteins was observed. In the present study no change in the breadth of the Band 3 staining on polyacrylamide gels was observed, indicating a lack of enzymatic modification of the

320

MIYAHARA

carbohydrate of this protein during cell aging. The possibility of a small amount of proteolytic cleavage of Band 3 during aging is suggested by a recent study on the senescent antigen of red cells, which has indicated that it is immunologically related to Band 3. Since the antigen has a molecular weight of only 62,000, it may be derived from Band 3 by peptide cleavage; however, the amount of the antigen present in red cell membranes represents only about 0.01% of the amount of the parent protein (34). Examination of ghost proteins from the five erythrocyte age groups by SDS-polyacrylamide electrophoresis revealed a similar pattern by both Coomassie blue and PAS staining. There was no indication of new components such as might be expected if proteolytic cleavage preceded loss of a glycoprotein segment from the membrane. The concomitant loss of both glycoproteins and glycolipids suggests that an explanation for the decreases seen may be the budding-off of membrane segments from the red cell, leaving behind ghosts deficient in glycoproteins and glycolipids and enriched in spectrin and other structural elements associated with the cytosolic surface. The loss of spectrin-free vesicles containing lipids and glycoproteins has been demonstrated during in vitro incubations carried out with both sheep and human red cells (35,36). The conditions employed for the human cells were those of energy depletion (36); a decrease of erythrocyte ATP levels during aging has been reported (l), and it is possible that the age-related loss of a number of enzymes may compromise the ability of the cell to produce adequate energy to maintain membrane structure intact. Such a removal of membrane segments would also explain the decrease with age of other membrane constituents, such as the insulin receptor (37). Loss of red cell membrane has also been cited as an explanation for the fact that, although the erythrocyte loses sialic acid during aging, the electrophoretic mobility of young and old cells is not different. Membrane removal concomitant with the decrease in sialic acid might be compen-

AND

SPIRO

satory and result in the same surfacecharge density, the factor important for electrophoretic mobility (38). The differential loss of membrane components observed in the present study probably reflects the arrangement of the glycoconjugates in the membrane and the tightness of their binding to the red cell cytoskeleton. Rearrangements of spectrin on the internal surface of the membrane with the use of specific antibodies was found to be accompanied by a migration and clumping of sialoglycoproteins, suggesting their linkage to spectrin (39). Band 3, on the other hand, appears to move freely in the membrane unrelated to spectrin (40); its linkage is to ankyrin (41), and it has been calculated (42) that there is only 1 molecule of ankyrin for every 3 Band 3 tetramers or for 6 dimers, suggesting that only a portion of the Band 3 can be linked to the cytoskeleton. The relative freedom of Band 3 from the structural elements is also supported by recent studies which have shown that the bulk of this protein can be extracted with Triton (43) and that, conversely, only 10% of the ankyrin and spectrin are resistant to extraction with alkaline solutions from ghosts treated with DIDS (diisothiocyanostilbenedisulfonic acid), an anion transport inhibitor which reacts specifically with Band 3 (44). The substantial loss of the Band 3 glycoprotein and the macroglycolipids is of particular note because of the similarity of their carbohydrate (13,14,27-29); both contain large polysaccharides made up of repeating units of galactose and N-acetylglucosamine with varying amounts of branching, and neither has any substantial amount of sialic acid. Thus, the aspect presented on the outside of the erythrocyte by these compounds is quite different from the highly negative profile produced by the sialoglycoproteins. The importance of the removal of sialic acid itself to the process of specifying red cells for clearance from the circulation is made less likely by the finding in the present study that this sugar declines less with age than the other membrane monosaccharides, and that there was no evidence

LOSS

OF

ERYTHROCYTE

MEMBRANE

for its specific removal. If the release of sialic acid and the consequent exposure of galactose or N-acetylgalactosamine plays a role in targeting red cells for destruction, a small number of highly specific components must be involved. In previous studies of diabetic erythrocyte ghosts, a reduced sialic acid content has been demonstrated for rat (7) and human (45) red cells, although another study of human erythrocytes found the decrease in this sugar in diabetes too small to be significant (46). In the present investigation the decrease in sialic acid in diabetic erythrocyte membranes (to 0.94 of normal) also lacked statistical significance. Moreover, while decreased levels of sialic acid and the other monosaccharides were observed in diabetic erythrocyte membranes when young cells were compared, only minimal differences were found between aged normal and diabetic erythrocytes. Since an increased red cell turnover has been shown to occur in diabetes (47), the lower saccharide content in the youngest cells may indicate that they have already undergone some loss of membrane glycoconjugates as part of an accelerated aging process. Whether this increased rate of removal of glycoprotein and macroglycolipidenriched membrane segments could account for some of the erythrocyte defects observed in this disease, such as increased membrane microviscosity (48) and decreased red cell deformability (49), is not known at the present time. Another difference between normal and diabetic erythrocytes found in the present study was a relative increase in peripheral galactose and/or N-acetylgalactosamine residues, primarily associated with glycolipids, which reacted with galactose oxidase and were subsequently reduced by tritiated borohydride. The degree to which the low-molecular-weight glycolipids are radiolabeled after galactose oxidase treatment has been found to reflect their accessibility to the enzyme (50, 51), which appears to be hindered by a shield of glycoproteins projecting out further from the cell surface (51). In the case of the diabetic cells, exposure of the lipid molecules may

SACCHARIDES

DURING

CELL

AGING

321

be promoted by the early loss of membrane constituents. ACKNOWLEDGMENT The authors are grateful for the excellent assistance of W. C. Fisher, Jr., in the performance of the sugar and amino acid analyses. REFERENCES 1. COHEN, N. S., EKHOLM, J. E., LUTHRA, M. D., AND HANAHAN, D. J. (1976) B&him Biophys. Acta 419, 229-242. 2. GATTEGNA, L., BLADIER, D., GARNIER, M., AND CORNILLOT, P. (1976) Carbohydr. Res. 52, 197208. 3. BAXTER, A., AND BEELEY, J. G. (1978) Biochem Biophys. Res. Commun 83, 466-471. 4. CHOY, Y. M., WONG, S. L., AND LEE, C. Y. (1979) Biochem. Biophys. Res. Commun 91,410-415. 5. BLADIER, O., GATPEGNA, L., FABIA, F., PERRET, G., AND CORNILLOT, P. (1980) Carbohydr. Res. 83, 371-376. 6. AMINOFF, D., VORDER BRUEGGE, W. F., BELL, W. C., SARPOLIS, K., AND WILLIAMS, R. (1977) Proc. Nat1 Acad Sci USA 74, 1521-1524. 7. CHANDRAMOULI, V., AND CARTER, J. R., JR. (1975) Diabetes 24, 257-262. 8. RIGAS, D. A., AND KOLER, R. D. (1961) J. Lab. Clin Med 58.242-246. 9. PRENTICE, T. C., AND BISHOP, C. (1965) Cell. Camp Physiol. 65, 113-126. 10. BEUTLER, E. (1977) in Hematology (Williams, W. J., Beutler, E., Ersley, A. J., and Rundles, R. W., eds.), pp. 1605-1607, McGraw-Hill, New York. 11. HENRY, J. B. (1979) in Clinical Diagnosis and Management by Laboratory Measurements, pp. 363-868, Saunders, Philadelphia. 12. HAMAGLJCHI, H., AND CLEVE, H. (1972) B&hem Biophys. Res. Ccnnmux 47,459-464. 13. DEJTER-JUSZYNSKI, M., HARPAZ, N., FLOWERS, H. M., AND SHARON, N. (1978) Eur. J. Biochem. 83, 363-373. 14. KOSCIELAK, J., MILLER-P• DRAZA, H., KRAUZE, R., AND PIASEK, A. (1976) Eur. J. B&hem. 71, 918. 15. LAEMMLI, V. K. (1970) Nature (London) 227,680685. 16. FAIRBANKS, G., STECK, T. L., AND WALLACH, D. F. H. (1971) Biochemistry 10,2606-2617. 17. WARREN, L. E. (1959) J. Bid Chem 234, 19711975. 18. SPIRO, R. G. (1966) in Methods in Enzymology (Neufield, E. F., and Ginsburg, V., eds.), Vol. 5, pp. 3-26, Academic Press, New York.

322

MIYAHARA

19. SPIRO, R. G., SPIRO, M. J., AND BHOYROO, V. D. (1976) J. Bid Chem 251,6409-6419. 20. SPIRO, R. G. (1972) in Methods in Enzymology (Ginsburg, V., ed.), Vol. 28, pp. 3-43, Academic Press, New York. 21. LOWRY, 0. H., ROSEBROUGH, N. J., FARR, A. L., AND RANDALL, R. J. (1951) .I. BioL Chem 234, 265-275. 22. VAN LENTEN, L., AND ASHWELL, G. (1972) in Methods in Enzymology (Ginsburg, V., ed.), Vol. 28, pp. 209-211, Academic Press, New York. 23. GAHMBERG, C. G. (1976) J. BioL Chem 251,510515. 24. HILL, A. B. (1977) A Short Textbook of Medical Statistics, pp. 161-180, Hodden and Stoughton, London. 25. PIOMELLI, S., LURINSKY, G., AND WASSERMAN, L. R. (1967) J. Lab. Clin. Mea! 69, 659-674. 26. Yu, J., AND STECK, T. L. (1975) J. BioL Chem 250. 9170-9175. 27. JARNEFELT, J., RUSH, J., LI, Y.-T., AND LAINE, R. A. (1978) J. BioL Chmn. 253, 8006-8009. 28. KRUSIUS, T., FINNE, J., AND RAUVALA, H. (1978)

Eur. J. B&hem

92.289-300.

29. STECK, T. L. (1978) J. Supramol Struct. 8, 311324. 30. BELL, W. C., LEVY, G. N., WILIAMS, R., AND AurNOFF, D. (1977) Proc Nat1 Acud Sci USA 74,

4205-4209. 31. FUKUDA, M. N., FUKUDA, M., AND HAKOMORI, S.-I. (1979) J. BioL Chem 254, 5458-5465. 32. MUELLER, T. J., LI, Y.-T., AND MORRISON, M. (1979)

J. BioL Chem 254.8103-8106. 33. LUTZ, H. U., AND FEHR, J. (1979) J. BioL Chew. 254,11177-11180. 34. KAY, M. M. B., GOODMAN, S. R., SORENSEN, K., WHITFIELD, C. F., WONG, P., ZAKI, L., AND RUD-

AND

SPIRO

LOFF, V. (1983) Proc Nati Acad 1631-1635. 35. LUTZ, H. U., BARBER, R., AND MCGUIRE, J. Biol Chem 251,3500-3510. 36. LUTZ, H. U., LIU, S.-C., AND PALEK,

Sci USA 80, R. F. (1976) J. (1977)

J.

Cell Biol 73, 548-558. 37. DONS, R. F., CORASH, L. M., AND GORDON,

P. (1981)

J. Biol Chem. 256.2982-2987. 38. SEAMAN, G. V. F., KNOX, R. J., NORDT, F. J., AND REGAN, D. H. (1977) Blood 50,1001-1011. 39. NICOLSON, G. L., AND PAINTER, R. G. (1973) J. CeU

Bid

59.395-406.

40. CHERRY, R. J., BURKLI, A., BUSSLINGER, M., SCHNEIDER, G., AND PARISH, G. P. (1976) Nature

(London) 263, 389-393. 41. BENNET,

V., AND STENBUCK,

P. J. (1979)

J.

BioL

Chem 254, 2533-2541. 42. GRATZER, W. B. (1981) Biochem. J. 198.1-8. 43. SHEETZ, M. P. (1979) B&him. Biwphhys. Acta 557, 122-134. 44. Hsu, L., AND MORRISON, M. (1983) Arch. B&hem.

Biophys. 227,31-38. 45. GANDHI,

C. R., AND CHOWDHURY, D. R. (1979) Ind. 17, 585-587. 46. GERBAUT, L., DELAUTURE, D., OLIVE, A. G., AND PEQUIGNOT, H. (1978) Clin Chem 24, X277-1278. 47. PETERSON, C. M., JONES, R. L., KOENIG, R. J., MELVIN, E. T., AND LEHRMAN, M. L. (1977) Ann Int.

J. Exp, Biol.

Meal 86.425-429. 48. BABA, Y., KAI, M., KAMADA, T., SETOYAMA, S., AND OTSUJI, S. (1979) Diabetes 28,1138-1140. 49. MCMILLAN, D. E., U’ITERBACK, N. G., AND LAPUMA, J. (1978) Diabetes 27, 895-901. 50. GAHMBERG, C. G., AND HAKOMORI, S.-I. (1973) J. BioL Chem 248, 4311-4317. 51. GAHMBERG, C. G., MYLLYLA, G., LEIKOLA, J., PERKOLA, A., AND NORDBERG, S. (1976) J. Bid Chem 251, 6108-6116.