Glycophorin Expression in Murine Erythroleukaemia Cells

Glycophorin expression in murine erythroleukaemia cells

JEFFREY B. ULMER*, ELIZABETH D. DOLCI and GEORGE E. PALADE

Department of Cell Biology, Yale University School of Medicine, P.O. Box 3333, Nezv Haven, CT 06510, USA

•Author for correspondence.

Summary

We have identified mature and putative precursor synthesis was induced by dimethyl sulphoxide coin- forms of glycophorins expressed in a virus-trans- cidentally with that of gp-3 and gp-2. They do not formed murine erythroleukaemia (MEL) cell line appear to be glycoproteins, as evidenced by their and compared them with their normal erythroblast lack of incorporation of [3H]galactose, [3H]glucosa- counterparts. The following differences were mine or [3H]mannose. In contrast, gp-3 and gp-2 found: (1) the two major MEL cell glycophorins incorporated [3H]galactose and [3H]glucosamine 3 3 (apparent Mr values 29-30 and 43(xlO )) have but not [ H]mannose. Partial characterization of greater mobility on polyacrylamide gels than their the glycan moieties of MEL cell glycophorins indi- normal gp-3 and gp-2 counterparts, due at least in cates that they consist mostly of tri- and tetrasac- part to differences in their oligosaccharide side- charides, with no indication of any N-linked chains. chains; (2) MEL cell gp-3 consists of two discrete Hence, the glycans of MEL cell glycophorins are proteins; and (3) there are more potential glyco- mostly (if not all) O-linked. Furthermore, treatment phorin precursors in MEL cells than in normal with AT-glycanase did not change their electrophor- mouse erythroblasts. Four proteins, with apparent etic mobility on polyacrylamide gels. MEL cell 3 Mr values of 21, 23, 26 and 27(xlO ), have tenta- glycophorins were also shown to be modified by tively been identified as glycophorin precursors, phosphoryl and fatty acyl groups. based on the following findings: (1) they are immu- nologically related to the glycophorins; and (2) their Key words: glycophorin, erythroleukaemia cells.

Introduction are fatty-acylated (Dolci & Palade, 1985). The Friend murine erythroleukaemia (MEL) cell is a The glycophorins are extensively studied, transmem- virus-transformed erythroid precursor cell that rapidly brane sialoglycoproteins of the plasmalemma of human proliferates in culture (Marks & Rifkind, 1978). Differen- red blood cells. Two of the proteins (glycophorins A and tiation along the erythrocyte pathway can be initiated by a B) are closely related to one another (Furthmayr, 1978; number of agents and results in decreased cell size, Siebert & Fukuda, 1986). The function of human glyco- changes in cell shape (Malik & Langzam, 1982), restric- phorins has not been established. It has been shown, ted proliferative capacity, expression of erythrocyte anti- however, that they may play a role in the maintenance of gens including glycophorins (Kasturi & Harrison, 1985), cell shape (Anstee et al. 1984) via interactions with band production of haemoglobin and, in some cases, extrusion 4.1 protein (Anderson & Lovrien, 1984) and polyphos- of nuclei to yield reticulocyte-like cells (Tsiftsoglou et al. phoinositides (Anderson & Marchesi, 1985). Other func- tions have not been excluded but remain to be defined. 1979; Volloch & Housman, 1982). Commitment of MEL The known amino acid sequences of human (Tomita et cells to differentiate appears to be a stochastic process al. 1978; Tomita & Marchesi, 1975), pig (Honma et al. (Gusella et al. 1976) but can be synchronized (Levenson 1980) and horse (Murayama et al. 1982) glycophorins are & Housman, 1979). Because of these characteristics, the known and their comparison shows that primary struc- MEL cell line appears to be an excellent system for the ture is not conserved to a high degree across species. Two study of glycophorin biosynthesis as affected by neoplas- murine glycophorin equivalents, termed gp-2 and gp-3 tic transformation and induced differentiation. We report (Sarris & Palade, 1979, 1982a,6; Dolci & Palade, 1985), here that two major mature forms of glycophorin, corre- have been identified in erythrocytes and erythroblasts. sponding to normal mouse erythroblast gp-2 and gp-3, Like their human counterparts, they are heavily sial- and four putative glycophorin precursors are expressed in ylated (Sarris & Palade, 1979, 1982a). In addition, they MEL cells.

Journal of Cell Science 92, 163-171 (1989) Printed in Great Britain © The Company of Biologists Limited 1989 163 Materials and methods To explore the fatty acylation of MEL cell glycophorins, 2xl07cellsmF' were biosynthetically labelled with [3H]palmi- Materials tate (200^CimF') for 60min at 37°C and then processed through immunoprecipitation, gel electrophoresis and fluor- Friend virus-transformed MEL cells (clone 745) were kindly ography. provided by Dr Edward Benz, Jr and endoglycosidase F was a For phosphorylation experiments, suspended MEL cells generous gift from Dr Steven Rosenzweig. All cell culture were incubated for 60 min at 37°C in 11-5 mM-A?-2-hydroxyeth- reagents, including RPMI 1640, foetal bovine serum, penicillin, ylpiperazine-Ar-2-ethanesulphonic acid (pH7-4), 136 mM- streptomycin and L-glutamine, were obtained from GIBCO NaCl, 5-8mM-KCl, l-5mM-CaCl , l-2mM-MgSO , 11-5 mM- Laboratories (Grand Island, NY). Dimethyl sulphoxide 2 4 glucose containing lOOjuCimF1 H332PO4 and then processed (Me2SO), hexamethylbisacetamide, all protease inhibitors and for autoradiography. Staphlococcus aureus V-8 protease were purchased from Sigma (St Louis, MO). iV-glycanase (peptide: iV-glycosidase F, peptide-;V4[/V-acetyl-/3-glucosaminyl]asparagine amidase) was ob- Immunoprecipitation tained from Genzyme Corporation (Boston, MA). [35S]methio- Glycophorins were immunoprecipitated from lysed MEL cells nine (1200Cimmol~ ) and NCSIM were obtained from either by the procedure of Dolci & Palade (1985) or (most often) Amersham (Arlington Heights, IL). [l,6-3H(yV)]glucosamine by a modification thereof, in which each batch of washed 1x10 hydrochloride (40-50 CimmoF1) and [9,10-3H(AO]palmitate to 2x 107 cells was treated for 15 min on ice with 1 ml of 1 % NP- (23-5 CimmoF ) were obtained from New England Nuclear 40 in 50mM-tris(hydroxymethyl)aminomethane (Tris-HCl), 32 1 3 (Boston, MA). H3 PO4 (285 Ci nig" ), [6- H]galactose pH7-5. The ensuing lysates were processed as described by (25 Ci mmoF ) and D-[2-3H]mannose (2-30CimmoF1) were Dolci & Palade (1985) up to immunoprecipitation, which was purchased from 1CN (Irvine, CA). carried out by adding 25 jA of polyclonal anti-mouse glycophorin antiserum (1:125 dilution) followed by vortexing and Cell culture incubation on ice for 15 min. Immune complexes were recovered from these mixtures by incubation for 45 min at 4°C MEL cells were grown at 37 °C to densities ranging from 1X104 (with gentle agitation) with 10 mg of protein A-Sepharose CL- to 1 XlO6 cells ml"' in RPMI 1640 medium containing foetal 4B beads (Pharmacia). The beads were collected by centrifu- bovine serum (10%, v/v), penicillin (100 units ml"1), strepto- gation (10 min at 2000 g), washed three times according to mycin (lOO^tgmF1) and L-glutamine (0-3mgmF'), while Dolci & Palade (1985), then boiled in a water bath for 5 min in being maintained in a humidified atmosphere of 95 % air/5 % 15% sucrose, 5% SDS, 0-1% dithiothreitol and 0-1 M- CO2. Differentiation was initiated by seeding MEL cells at Tris-HCl, pH7-5 (to release the immune complexes). 5Xl04mF' ingrowth medium containing Me2SO (2%, v/v) or 5mM-hexamethylbisacetamide. At day 2 after the addition of In some experiments, the lysis (1% NP-40 in 50IYIM- the inducer, the cell culture was diluted 1: 1 with fresh growth Tris-HCl) and solubilization (4% SDS in 20mM-dithiothrei- medium and at day 4 the cells were harvested for experimen- tol, 50mM-Tris-HCl) buffers contained the following protease inhibitors (final concentration): antipain (0'ljUgmF ), aproti- tation. Differentiation of MEL cells was monitored by follow- 1 1 ing haemoglobin production via benzidine staining. To this nin (0-lmgmF ), benzamidine (01 jUgmF ), chymostatin intent, 1X106 cells were suspended in 05ml phosphate- (0-1 jUgmF ), diisopropyl fluorophosphate (O'l HIM), ethylene- glycol-bis-(/3-aminoethyl ether)-A',Ar'-tetraacetic acid (0'5ITIM) buffered saline containing 2 mM-sodium azide and bovine serum 1 albumin (002%, w/v); 50^1 of 1% benzidine in methanol and pepstatin (O'l jUgmF ). (w/v) and 10;«l of 30% hydrogen peroxide were added to the suspension and mixed by inversion. The presence of haemo- Alkaline borohydride treatment globin in MEL cells was indicated by a bluish green reaction Alkali-labile ohgosacchandes were cleaved according to Carlson product observed after lOmin by light microscopy. Cell vi- (1968) and separated by gel filtration according to Cummingse/ ability was monitored by Trypan Blue exclusion. Cell number al. (1983). [3H]galactose labelled immunoprecipitates from and percentage of cells positive for benzidine or Trypan Blue splenic erythroblasts and MEL cells were eluted from the 5. were calculated by counting cells in a haemocytometer by light aureus complexes with 0-5 % sodium dodecyl sulphate (SDS). microscopy. An equal volume of 2M-sodium borohydride in lOOmM-NaOH was added and the SDS concentration was further reduced to Metabolic labelling of MEL cells 01% by the addition of 50mM-NaOH, 1 M-sodium borohydride. The mixture was incubated for 14-20 h at 45 °C, Cells, collected by centrifugation (10 min at 2000g) and washed placed on ice and diluted with 2 vols of distilled water. The pH three times with methionine-free RPMI medium (by repeated was adjusted to 50 by the addition of 1 M-acetic acid and resuspension and sedimentation), were incubated in the same borohydride was removed by rotary evaporation in methanol. medium with the following reagents: 200 /.id ml" [3sS]methio- 1 1 The dried samples were resuspended in 0-1% SDS, 50mM- nine, 100 units mF penicillin, lOOjUgmF streptomycin, NH4HCO3, pH80, and the alkali-labile material was size- 0-3mgmF' L-glutamine, and under the following conditions: 7 1 fractionated on a column of Biogel P-6 (1-6 cm X 97 cm) IXW to 2xl0 cellsmF , 37°C in an atmosphere of 95% (BioRad). One ml fractions were collected and their radioac- O2/5 % CO2 with gentle agitation. To end the labelling, a tivity determined. Alkali-labile fetuin oligosaccharides were tenfold excess of ice-cold complete RPMI medium was added, used as standards (Spiro, 1962). then the cells were collected by centrifugation and washed three times with ice-cold phosphate-buffered saline. Splenic erythroid cells were prepared and labelled with Glycosidase treatment [3H]galactose as described (Dolci & Palade, 1985). MEL cells For N-glycanase treatment, [35S]methionine-labelled MEL cell (1 XlO6 cells mF1) in 50 ml of RPMI 1640 medium were incu- glycophorins were eluted from the protein A-Sepharose beads bated in the presence of [3H]galactose, [3H]mannose and by boiling in 20fil of 0-5% SDS for 3 min. The following [3H]glucosamine (ISftCimF1) for 4h or 24h at 37°C, then solutions were added in the given order: 21-6 jt*l of 0'55 M- recovered (by sedimentation), washed and processed for NaH2PO4 (pH8-6), 6;tl of 100mM-l,10-phenanthroline hy- immunoprecipitation. drate and lOjul of NP-40 (7-5%, w/v). The digestion was

164 J. B. Ulmer et al. carried out in the presence of 0-10 units ml ' of iV-glycanase at 1978), a variety of compounds including Me2SO and 37°C for 15 h. The treated samples were analysed by gel hexamethylbisacetamide induce MEL cell differentiation electrophoresis and fluorography. characterized by a reduction in the rate of cell replication, decrease in cell size and rapid production of haemoglobin Gel electrophoresis and other erythroid proteins. In our experiments, MEL Solubilized proteins were separated by sodium dodecyl sul- cell line 745 responded to Me SO treatment by: (1) a loss phate-polyacrylamide gel electrophoresis (SDS-PAGE) on 2 vertical slab gels (20cmXl5 cmXl-5 mm). The resolving gels of proliferative capacity after approximately five cell contained 15% acrylamide, 0-4% bisacrylamide, 0-1% SDS, divisions (Fig. 1A); (2) a striking increase in the rate of 0-05 % /V,iV,./V',./V'-tetramethylethylenediamine (v/v), 0-375 M- globin synthesis at day 3-4 (Fig. IB), followed by Tris-HCl (pH8-8) and 0-1% ammonium persulphate. The progressive increases for the duration of MezSO treat- stacking gels were the same as the resolving gels except that they ment (not shown); and (3) a rapid rise in the rate of contained 4% acrylamide, 0-11% bisacrylamide and 0-125M- synthesis of a group of proteins immunoprecipitable by Tris-HCl (pH6-8). The running buffer consisted of 0-1% anti-glycophorin sera and, therefore, assumed to be SDS, 0-05M-Tris-HCl (pH8-8) and 0-38M-glycine. Electro- glycophorins or glycophorin precursors (Fig. 2). In their phoresis was carried out for 16-18 h at 75 V (constant voltage) case, the rate peaked at 3-4 days when it reached a level in an apparatus cooled via a water jacket. Gels were either fixed 10- to 30-fold higher than in uninduced cells; by day 7 the and stained (25% isopropyl alcohol, 10% acetic acid, 0-1% rate had decreased to preinduction levels. These coordi- Coomassie Blue) or processed directly for fluorography. nated changes in rates of labelling indicate that the Determination of radioactivity syntheses of glycophorins and related proteins are tightly Incorporation of 3SS and 3H into MEL cell proteins was coupled. Hexamethylbisacetamide treatment gave similar determined by fluorography and, in some instances, liquid results (not shown). scintillation counting. Fluorography was performed according Given these findings, all subsequent experiments to Bonner & Laskey (1974). P-labelled proteins were visual- designed to characterize MEL cell glycophorins were ized by direct autoradiography of dried gels. The processed gels carried out on Me2SO-treated cells 4 days after the were placed on Whatman 3MM paper, covered with a protective beginning of the induction process. layer of cellulose membrane backing (Bio-Rad) and dried using a Bio-Rad gel drying apparatus. Radioactivity was visualized by MEL cell glycophorins exposure of Kodak XAR-5 X-ray film to dried gels at —80°C. 3s For quantitative determinations, fluorographs were aligned Metabolic labelling with [ S]methionine, followed by with the corresponding dried gels and individual protein bands were excised from the latter. The excised slices were hydrated with 200 /il of water in scintillation vials and their paper backing and protective membrane covers were removed. After addition of 0-5 ml NCS, incubation at 37°C (until the gel slices were clear, 4-12 h) and cooling to room temperature, each vial received 5 ml of Ecoscint (National Diagnostics) and was allowed to equilibrate for at least 6h before counting in an Beckman L5-200 scintillation counter.

One-dimensional peptide mapping The procedure for peptide mapping was essentially that of. Cleveland et al. (1977). Briefly, gels containing proteins to be mapped were dried without fixation or fluorography processing on Whatman 3MM paper. Radioactive protein bands, visualized by direct autoradiography onto Hyperfilm /8-max X-ray film (Amersham) at —80°C were excised from the gels as previously described, then hydrated in 0-1% SDS, 1 mM- EDTA, 0-125M-Tris-HCl (pH6-8) for lOmin and finally transferred to the stacking gel wells of a second acrylamide gel. The stacking gel, resolving gel and running buffer contained 1 mM-EDTA in addition to the previously described ingredi- ents. Each gel slice was overlaid first with the hydration buffer containing 20% glycerol and 0-1% Bromophenol Blue (10/Ulweir1) and then with V-8 protease (0-2 to 20jtg/l0/ul) in hydration buffer containing 10% glycerol. The gels were 2 3 4 5 6 electrophoresed at room temperature until the dye was less than Days in DMSO

5 mm from the resolving gel. At that point, electrophoresis was Fig. 1. Induction of MEL cell differentiation by Me2SO. discontinued for 30min (while V-8 digestion took place) and MEL cells were seeded at day 0 with RPMI medium then recommenced for 16—18 h at 75 V under constant voltage. containing MejSO (2%, v/v). Dilution 1: 10 with fresh RPMI medium was carried out at days 2 and 4. Each day the cultures were sampled and cell concentration (expressed Results as cells ml"1 of the original inoculum) (A), percentage of benzidine positive cells ((O) B) and percentage of Trypan Glycophorin induction in MEL cells Blue positive cells ((•) B) were determined. The data As is well documented in the literature (Marks & Rifkind, represent the averages of two experiments.

Biosynthesis of marine glycophorins 165 0 1 2 3 4 5 6 7

gp-2-

gp-3-

gp-2- -gp-2

gp-3- 19P-3

gp-3pr- 12 3 4 5 Days in DMSO Fig. 2. Glycophorin biosynthesis in MEL cells during differentiation. Glycophorins were immunoprecipitated from Fig. 3. Comparison of MEL cell and normal erythroblast lysates of cells (day 0 to 7) previously labelled for 30 min with glycophorins. Glycophorins were immunoprecipitated from 35 [ S]methionine. The immune precipitates were processed lysates of MEL cells or mouse splenic erythroblasts labelled through SDS-PAGE and fluorography (A). The positions of for 60 min with [35S]methionine; the immunoprecipitates the putative precursors (21K, 23K, 26K and 27K) and were processed through SDS-PAGE and fluorography. mature forms (gp-2 and gp-3) are indicated (A). B. The Indicated above are gp-2, gp-3 and the precursor of gp-3 (gp- amount of radioactivity incorporated in each glycophorin, i.e. 3 pr) of splenic erythroblasts (lane 1) and gp-2, gp-3 of MEL 21K (•), 23K (O), 26K (T), 27K (V), gp-2 (•) and gp-3 cells (lane 2). Putative precursors of apparent MT 21K, 23K, (D) was determined and plotted versus days in Me2SO. 26K, 27K and the 34K mature glycophorin are indicated by Representative example from a series of three experiments. arrowheads. lysis, immunoprecipitation of lysates with polyclonal anti-glycophorin sera (Sarris & Palade, 19826; Dolci & V-8 protease. The structural relatedness of normal and Palade, 1985), SDS-PAGE and fluorography, revealed transformed erythroblast glycophorins was explored by that Me2SO-induced MEL cells synthesized proteins of one-dimensional peptide mapping. Common sets of peptides as well as proteolytic fragments of similar size were slightly lower apparent molecular weights (Mr) than the gp-2 and gp-3 of normal mouse erythroblasts (Fig. 3). generated from MEL cell and splenic erythroblast gp-2 gp-3, which is normally a single entity, appeared as a by V-8 protease (Fig. 4, lanes 4,5, respectively). Splenic doublet in MEL cells. In addition, MEL cells syn- erythroblast gp-3 (Fig. 4, lane 3) was most similar to the higher M doublet of MEL cell gp-3 (Fig. 4, lane 2). It is thesized four lower Mr peptides (21K, 23K, 26K and r 3 interesting to note that the endpoint of digestion of 27K; K= 10 vV/r), compared to two peptides (23K and 27K) found in normal erythroblasts. The cells also normal erythroblast gp-2 and gp-3 resulted in a 16K produced a 34K protein, which probably corresponds to peptide, whereas MEL cell gp-2 and gp-3 were digested the 'intermediate component' detected in normal murine by a further 3K. These results suggest that one of the V-8 erythroblasts (Dolci & Palade, 1985). Only gp-2 and gp-3 cleavage sites (Glu or Asp) in MEL cell glycophorins is were revealed by immunoblotting of electrophoretogram either cryptic or absent in their normal counterparts. transfers of MEL cell lysates (not shown); the detection Like their mature counterparts in normal erythrocytes of the other peptides depended strictly on metabolic and MEL cells, the lower Mr proteins yield the same labelling and immunoprecipitation, presumably because peptides upon proteolysis with V-8 protease (Ulmer & of their low intracellular concentration. Palade, 1989), suggesting that they are structurally related to the mature forms. Peptide mapping All of these proteins from MEL cells and normal mouse Multiplicity of glycophorin precursors erythroblasts are related to one another; they are recog- By analogy to normal mouse erythroblasts (Dolci & nized and immunoprecipitated by the same antisera and Palade, 1985), we assume that the four low Mr MEL cell yield common, low Mr peptides upon proteolysis with proteins are precursors of the glycophorin mature forms.

166 J. B. Ulmer et al. 12 3 4 5 12 3 4 5 1

• • 43- gp-3 25- t t: gp-2- 18- # • 14- If" A B gp-3- Fig. 4. One-dimensional peptide mapping of MEL cell and mouse splenic erythroblast glycophorins. Cells were labelled and processed through immunoprecipitation, SDS—PAGE, V-8 protease treatment (0-2/.igml"', A; 20figm\~l, B) and one-dimensional peptide mapping as given in Materials and methods. MEL cell glycophorins: gp-3, lower and upper band of the doublet, lanes 1,2, respectively; gp-2, lane 4. Splenic erythroblast glycophorins: gp-3, lane 3; gp-2, lane 5. Aggregates are detected in lanes 2,3. The position of Mr markers is shown (in K) on the left side of A and the positions of undigested gp-2 and gp-3 are indicated between A and B.

Their unexpected multiplicity, however, raises questions about their nature; some of them could be preparation Fig. 5. The 21K, 23K, 26K and 27K MEL cell glycophorin- artifacts or true, partially processed precursors. First, related proteins are not experimental artifacts. MEL cells, MEL cell lysis in the presence of the battery of protease labelled with [35S]methionine for 30min, were boiled inhibitors given in Materials and methods did not lead to immediately in 4% SDS followed by immunoprecipitation a change in number, quantity or mobility of putative (lane 4) or lysed and immunoprecipitated by standard precursors, and direct solubilization of .labelled MEL procedures (see Materials and methods) in the presence (lane cells in boiling SDS gave identical results (Fig. 5). 2) or absence (lanes 1,3) of protease inhibitors. The Therefore, partial proteolysis during cell lysis or lysate immunoprecipitates were processed through SDS-PAGE and fluorography. gp-2 and gp-3 are labelled and 21K, 23K, 26K processing cannot explain the multiplicity of precursors. and 27K are indicated by arrowheads. Second, as indicated by their lack of incorporation of isotopically labelled sugars (Fig. 6) and the lack of effect of ./V-glycanase on their electrophoretic mobility (Fig. 7), none of the four peptides appears to be glycosylated. effect on their electrophoretic mobility (Fig. 7). Similar Moreover, four primary cell-free translation products of results were obtained with endoglycosidase F (not MEL cell mRNA, very close in apparent MT to 21K, shown). These results, taken together, suggest that 23K, 26K and 27K, were immunoprecipitated from the glycosylation of murine glycophorins is of the O-linked total translation products in the absence of endoplasmic type. MEL cell glycophorins, however, contain a greater reticulum membranes (Ulmer & Palade, 1989). Hence, proportion of trisaccharides than those of splenic eryth- these four peptides do not appear to be temporally related roblasts, which may contribute to the apparent differ- to one another, either by cotranslocational processing ences in their Mr (see Fig. 3). (e.g. removal of signal sequences) or modification (e.g, glycosylation). Post-translational modification of MEL cell glycophorins Glycosylation of MEL cell glycophorins The modification of MEL cell glycophorins by phos- Incubation of MEL cells with [3H]galactose and [3H]glu- phoryl and fatty acyl groups was investigated by meta- cosamine resulted in isotope incorporation into gp-2 and bolic labelling. Incorporation of 32P was detected in gp-2 gp-3 but not the putative precursors (Fig. 6), whereas and gp-3 but not in any of the putative precursors [ HJmannose was not incorporated into any of them. (Fig. 9). Similar results were obtained in normal mouse Partial characterization of the glycan moieties of splenic erythroblasts (not shown). MEL cell glycophorins were erythroblast and MEL cell glycophorins indicated that shown to undergo fatty acylation, as well, as demon- the oligosaccharide side-chains were mostly tn- and strated by their incorporation of [3H]palmitate (Fig. 10). tetrasaccharides with no indication of the presence of any In this case, however, at least one of the putative N-linked chains (Fig. 8). Moreover, /V-glycanase had no precursors (21K) was also modified.

Biosynthesis of murine glycophorins 167 123 456789

97- 68- I 43- I — —-gp-2 26- ~ — =gp-3 I

Fig. 6. Glycosylation of MEL cell glycophorins. MEL cells were labelled with [3H]mannose (lanes 1,4,5), [3H]galactose (lanes 2,6,7) or [3H]glucosamine (lanes 3,8,9) as described in Materials and methods. Total cell lysates (lanes 1-3), immune precipitates (lanes 5,7,9) and non-immune precipitates (lanes 4,6,8) were processed through SDS-PAGE and fluorography. The mobilities of [35S]methionine labelled glycophorins (gp-2 and gp-3) and putative precursors (arrowheads) are shown on the right side of the figure; those of the MT markers (in K) are indicated on the left side of the figure. Fig. 7. Treatment of MEL cell glycophorins with A'-glycanase. MEL cells, labelled with [3SS]methionine for 30min, were lysed and processed through A third glycophorin in MEL cells immunoprecipitation. Immune precipitates were incubated in the presence (lane 3) or absence (lane 2) of Ar-glycanase The 34K protein detected in our experiments appears to (10 units ml" ) and the reaction products were processed be a mature glycophorin; it has structural homology to through SDS-PAGE and fluorography. A control gp-2 and gp-3 (Ulmer & Palade, 1989), it accumulated immunoprecipitate is shown in lane 1. gp-2 and gp-3 are label in pulse-chase experiments only after a chase of indicated by large arrowheads and 21K, 23K, 26K and 27K lOmin (not shown), and it was synthesized with kinetics by small arrowheads. The broken lines mark the mobility of comparable to those of the major glycophorins during the 50K subunit of IgG. induced differentiation (Fig. 2). The 34K protein was not detected in the plasmalemma of mature erythrocytes (Sarris & Palade, 1982a). Hence, it may be a glycophorin between normal and transformed erythroblast glycophor- transiently expressed during erythroblast differentiation. ins. Partial characterization of their glycan moieties Its presence in MEL cells, though, clearly establishes its indicates that these differences may lie, at least in part, in erythroid nature. their oligosaccharide side-chains. Peptide mapping suggests that there may be differences in their primary structures as well. A second major difference in glyco- Discussion phorin expression is that gp-3, which is normally seen as a single entity, appears as two discrete proteins in MEL Two major sialoglycoproteins, thought to be the murine cells. We have also provided evidence for the existence of equivalents of human glycophorins, have been identified a third mature glycophorin molecule in MEL cells (34K). in mouse erythrocytes (Sarris & Palade, 1982a) and The degree of relatedness of this protein to the other erythroblasts (Dolci & Palade, 1985). Glycophorins have glycophorins remains to be established by direct or also been detected in MEL cells by immunocytochem- indirect sequencing. istry (Sarris & Palade, 19826) and metabolic labelling The presence of major and minor forms of glycophorin (Kasturi & Harrison, 1985). As judged by [3sS]methio- in MEL cells is analogous to mouse splenic erythroblasts nine incorporation, our results show that in these cells the (Dolci & Palade, 1985), human K562 erythroleukaemia predominant form (gp-3) and the minor form (gp-2) have cells (Jokkinen et al. 1985), human erythrocytes (Furth- apparent Mr values of 29-30K and 43K, respectively. mayr & Marchesi, 1983), horse erythrocytes (Murayama These values correlate reasonably well with those of et al. 1981) and chimpanzee erythrocytes (Blumenfeld et Kasturi & Harrison (1985), who reported major and al. 1983) (for review, see Krotkiewski, 1988). The major minor MEL cell glycophorins with MT values of 31K and human glycophorin (glycophorin A) and a minor form 46K, respectively. MEL cell gp-2 and gp-3 are slightly (glycophorin B) have extensive structural homology, lower in apparent Mv than their normal counterparts (see both at the amino acid (Blanchard et al. 1987) and cDNA Fig. 3), suggesting that there are structural differences (Siebert & Fukuda, 1987) sequence level. A third human

168 J. B. Ulmer et al. 500-1

-gp-2

-9P-3

Fig. 9. Phosphorylation of MEL cell glycophorins. MEL cells, labelled with [32P]orthophosphate, were precipitated with non-immune (lane 1) and immune (lane 2) sera and processed through SDS-PAGE and autoradiography. The autoradiograph shows that only gp-2 and gp-3 are labelled. The relative mobilities of 21K, 23K, 26K and 27K are 100 125 150 indicated by arrowheads. Fraction number

Fig. 8. Partial characterization of the oligosaccharide chains ping of the putative precursors revealed extensive hom- of splenic erythroblast and MEL cell glycophorins. The [3H]- ology amongst themselves and the mature glycophorins labelled products of sodium borohydride reduction of splenic (Ulmer & Palade, 1989). One other published study on erythroblast (A) and MEL cell (B) glycophorins were glycophorin biosynthesis in MEL cells reported only two separated by gel filtration as given in Materials and methods. (instead of four) putative precursors (21-5K and 24-5K) The radioactivity in the eluted fractions was plotted as a (Kasturi & Harrison, 1985). In that study, a monoclonal function of fraction number. The position of fetuin oligosaccharide chains used as standards is shown by anti-mouse glycophorin antibody and a different MEL arrowheads. cell clone (M707T) were used, which may explain the disparate results. Murine glycophorins, like human glycophorin B (Blan- glycophorin molecule (glycophorin C) has limited struc- chard et al. 1987) and other glycophorins (Krotkiewski, tural homology to glycophorins A and B (Blanchard et al. 1988), do not appear to contain N-linked oligosacchar- 1987; High & Tanner, 1987). ides, as suggested by their lack of mannose incorporation An unexpected observation in MEL cells was the and the lack of effect of glycosidases on their electrophor- multiplicity of putative glycophorin precursors. In ad- etic mobility. Furthermore, partial characterization of dition to the 23K and 27K proteins, for which there are the glycan moieties yielded no sugars of expected size for counterparts in normal mouse erythroblasts, MEL cells N-linked chains. Rather, MEL cell glycophorins have express glycophorin-related proteins of MT 21K and 26K. predominantly tri- and tetrasaccharide units, presumably On the basis of (1) synchronous induction of synthesis in an O-ester linkage. This type of glycosylation also during differentiation, (2) apparent lack of glycosylation, predominates in their normal counterparts. Other modi- and (3) previous results on the biosynthesis of glycophor- fications that MEL cell glycophorins were shown to ins in mouse splenic erythroblasts, the four low Mr undergo include phosphorylation, as is the case for peptides (21K, 23K, 26K and 27K) are assumed to be human glycophorins (Shapiro & Marchesi, 1977; Wax- precursors of MEL cell glycophorins. The extra putative man, 1979; Dzandu et al. 1985), and fatty acylation, glycophorin precursors in MEL cells are not likely to be which was previously demonstrated in normal mouse overexpressed forms of minor, normal erythroblast pro- splenic erythroblasts (Dolci & Palade, 1985). Apparently, teins because translation of splenic erythroblast mRNAs only the mature forms of murine glycophorin are phos- in a cell-free system does not yield the 21K and 26K phorylated and, interestingly, only the higher Mr form of products (Dolci & Palade, unpublished). Peptide map- MEL cell gp-3 appears to be so modified. The signifi-

Biosynthesis of murine glycophorins 169 association of membrane skeleton protein 4.1 with glycophorin by polyphosphoinositides. Nature, Loud. 318, 295-299. ANSTEE, D. J., PARSONS, S. F., RIDGWELL, K., TANNER, M. J. A., MERRY, A. H., THOMSON, E. E., JUDSON, P. A., JOHNSON, P., BATES, A. & FRASER, I. D. (1984). Two individuals with elliptocytic red cells apparently lack three minor erythrocyte membrane sialoglycoproteins. Biochem.J. 218, 615-619. BLANCHARD, D., DAHR, W., HUMMEL, M., LATRON, F., BEYREUTHER, K. & CARTRON, J.-P. (1987). Glycophorins B and C from human erythrocyte membranes. Purification and sequence analysis. J. biol. Chew. 262, 5808-5811. BLUMENFELD, O. O., ADAMANY, A. M., PUGLIA, K. V. & SOCHA, W. W. (1983). The chimpanzee M blood group antigen is a variant gp-2- - of the human M-N glycophorins. Biochem. Genet. 2l, 333-348. BONNER, W. M. & LASKEY, R. A. (1974). A film detection method for tritium-labelled proteins and nucleic acids in polyacrylamide gels. Eur.J. Biochem. 46, 83-88. CARLSON, D. M. (1968). Structures and immunochemical properties gp-3 = of oligosaccharides isolated from pig submaxillary mucins. J. biol. Chem. 243, 616-626. CLEVELAND, D. W., FISCHER, S. G., KIRSCHNER, M. W. & LAEMMLI, U. K. (1977). Peptide mapping by limited proteolysis in sodium dodecyl sulfate and analysis by gel electrophoresis. J. biol. Chem. 252, 1102-1106. CUMM1NGS, R. D., KORNFELD, S., SCHNEIDER, W. J., HOBGOOD, K. K., TOLLESHAUG, H., BROWN, M. S. & GOLDSTEIN, J. L. (1983). Biosynthesis of N- and O-linked oligosaccharides of the low density lipoprotein receptor. J. biol. Chem. 258, 15 261-15 273. Fig. 10. Fatty acylation of MEL cell glycophorins. MEL DOLCI, E. D. & PALADE, G. E. (1985). The biosynthesis and fatty 3 acylation of the murine erythrocyte sialoglycoproteins. J. biol. cells were labelled with [ H]palmitate and processed through Chem. 260, 10728-10735. lysis, precipitation with immune (lane 2) and non-immune DZANDU, J. K., DEH, M. E. & KIENER, P. (1985). Phosphorylation (lane 3) sera, SDS-PAGE and fluorography. An immune of glycophorin A in membranes of intact human erythrocytes. 3S precipitate of [ S]methionine-labelled MEL cells is shown in Biochem. biophys. Res. Commun. 127, 878-884. lane 1 for reference. Arrowheads mark the 21K, 23K, 26K FURTHMAYR, H. (1978). Structural comparison of glycophorins and and 27K proteins and the asterisk indicates that 21K is immunochemical analysis of genetic variants. Nature, Loud. 271, slightly labelled with [3H]palmitate. 519-524. FURTHMAYR, H. & MARCHESI, V. T. (1983). Glycophorins: Isolation, orientation and localization of specific domains. Meth. Eiixvm. 96, cance of this observation is being investigated. In con- 268-280. trast, both forms of gp-3 and at least one of the putative GUSELLA, J., GELLER, R., CLARKE, B., WEEKS, V. & HOUSMAN, D. (1976). Commitment to erythroid differentiation by Friend precursors (21K) is fatty acylated. Therefore, fatty erythroleukemia cells: A stochastic analysis. Cell 9, 221-229. acylation of murine glycophorins may take place earlier in HIGH, S. & TANNER, M. J. A. (1987). Human erythrocyte their biosynthesis than does phosphorylation. membrane sialoglycoprotein ji. The cDNA sequence suggests the The results presented here demonstrate that neoplastic absence of a cleaved N-terminal signal sequence. Biochem. J. 243, 277-280. transformation or induced differentiation of murine HONMA, K., TOMITA, M. & HAMADA, A. (1980). Amino acid erythroblasts has altered glycophorin expression in two sequence and attachment sites of oligosaccharide units of porcine ways. First, there are structural differences between erythrocyte glycophorin. J. Biochem. 88, 167-169. normal and transformed erythroblast glycophorins, due JOKKINEN, M., ANDERSSON, L. C. & GAHMBERG, C. G. (1985). at least in part to differential processing of oligosacchar- Biosynthesis of the major human red cell sialoglycoprotein, glycophorin A. J. biol. Chem. 260, 11 314-11 321. ide side-chains. Second, extra glycophorins are expressed KASTURI, K. & HARRISON, P. (1985). The cell specificity and in transformed erythroblasts. Whether this is due to the biosynthesis of mouse glycophorins studied with monoclonal expression of extra glycophorin genes or alternative antibodies. Expl Cell Res. 157, 253-264. mRNA splicing, as a result of the integration of the KROTKIEWSKI, H. (1988). The stucture of glycophorins of animal retroviral genome into erythroblast DNA, remains to be erythrocytes. Glycocoiijugate J. 5, 35-48. LEVENSON, R. & HOUSMAN, D. (1979). Synchronization of MEL cell explored. Isolation of cDNA clones of murine glycophor- commitment with cordycepin. Cell 18, 1073-1078. in(s) is currently under way in our laboratory. With the MALIK, Z. & LANGZAM, Y. (1982). Cell membrane maturation of cloned cDNAs we will be able to address this issue. Friend erythroleukemia cells and tocopherol-dependent erythropoietin effect: A scanning electron microscope study. Cell This work was supported by NIH grant Differ. 11, 161-167. 5-RO1-GM-27303-08. MARKS, P. A. & RIFKIND, R. A. (1978). Erthroleukemic differentiation. A. Rev. Biochem. 47, 419-448. MURAYAMA, J.-I., TAKESHITA, K., TOMITA, M. & HAMADA, A. (1981). Isolation and characterization of two glycophorins from References horse erythrocyte membranes. J'. Biochem. 89, 1593-1598. MURAYAMA, J.-I., TOMITA, M. & HAMADA, A. (1982). Primary ANDERSON, R. A. & LOVRIEN, R. E. (1984). Glycophorin is linked structure of horse erythrocyte glycophorin HA. Its amino acid by band 4.1 protein to the human erythrocyte cytoskeleton. sequence has a unique homology with those of human and porcine Nature, Ijond. 307, 655-658. erythrocyte glycophorins. J'. Membr. Biol. 64, 205-215. ANDERSON, R. A. & MARCHESI, V. T. (1985). Regulation of the SARRIS, A. H. & PALADE, G. E. (1979). The sialoglycoproteins of

170 J.B.Ulmeretal. murine erythrocyte ghosts. J. biol. Chem. 254, 6724-6731. structure of human erythrocyte glycophorin A. Isolation and SARRIS, A. H. & PALADE, G. E. (1982a). Isolation and partial characterization of peptides and complete amino acid sequence. characterization of the sialoglycoprotein fraction of murine Biochemistry 17, 4756- 4770. erythrocyte ghosts. J. Cell Biol. 93, 583-590. TOMITA, M. & MARCHESI, V. T. (1975). Amino acid sequence and SARRIS, A. H. & PALADE, G. E. (19826). Immunofluorescence oligosaccharide attachment sites of human erythrocyte glycophorin. detection of erythrocyte sialoglycoprotein antigens on murine Proc. natn. Acad. Sci. U.S.A. 72, 2964-2968.' erythroid cells." J. Cell Biol. 93, 591-603. TSIFTSOGLOU, A. S., BARNETT, R. J. & SARTORELLI, A. C. (1979). SHAPIRO, D. L. & MARCHESI, V. T. (1977). Phosphorylation in Enucleation of differentiated murine erythroleukemia cells in membranes of intact human erythrocytes. J. biol. Chem. 252, culture. Proc. natn. Acad. Sci U.S.A. 76, 6381-6385. 508-517. ULMER, J. B. & PALADE, G. E. (1989). Anomalies in the SIEBERT, P. D. & FUKUDA, M. (1986). Human glycophorins A and B translocation and processing of glycophorin precursors in murine are encoded by separate single copy genes coordinately regulated erythroleukemia cells. J. biol. Chem. (in press). by a tumor-promoter phorbol ester, j'. biol. Chem. 261, VOLLOCH, V. & HOUSMAN, D. (1982). Terminal differentiation of 12433-12436. murine erythroleukemia cells: physical stabilization of end-stage SlEBERT, P. D. & FUKUDA, M. (1987). Molecular cloning of a human cells. J. Cell Biol. 93,390-394. glycophrin B cDNA: Nucleotide sequence and genomic WAXMAN, L. (1979). The phosphorylation of the major proteins of relationship to glycophorin A. Proc. natn. Acad. Sci. U.S.A. 84, the human erythrocyte. Archs Biochem. Biophys. 195, 300-314. 6735-6739. SPIRO, R. G. (1962). Studies on the monosaccharide sequence of the serum protein fetuin. J. biol. Chem. 237, 646-652. TOMITA, M., FURTHMAYR, H. & MARCHESI, V. T. (1978). Primary (Received 19 September 19S8 - Accepted 3 Xovember I9SS)

Biosynthesis of murine glycophorins 171