Proc. Natl. Acad. Sci. USAVol. 81, pp. 6752-6756, November 1984Cell Biology

Differentiation of human erythroid cells is associated with increasedO-glycosylation otthe major sialoglycoprotein, glycophorin A

(erythrocyte differentiation/mabria/erythroleukemia)

CARL G. GAHMBERG*, MARJA EKBLOM,t AND LEIF C. ANDERSSONt*Department of Biochemistry, University of Helsinki, 00170 Helsinki 17, Finland; and tDepartment of Pathology and Transplantation Laboratory, University ofHelsinki, 00290 Helsinki 29, Finland

Communicated by Robert L. Hill, July 13, 1984

ABSTRACT Glycophorin A, the major human erythro-cyte sialoglycoprotein, is found exclusively on cells of the ery-throid lineage. The amino acid sequence is known, and glyco-phorin A isolated from mature erythrocytes contains a singleN-glycosidic and 15 O-glycosidic oligosaecharides. Monoclonalantibodies against erythrocyte glycophorin A reacted weaklywith erythroid precursors while a monospecific rabbit antise-rum reacted strongly with immature and mature red cells.Glycophorin A was isolated from cells representing variousstAges of erythropoiesis in normal bone miarirow, from bloodcells of neonates with erythroblastosis fetalis, and from theerythroleukemic cell lines K562 and HEL before and after in-duced differentiation. Analysis of the oligosaccharides showedless O-glycosylation of glycophorin A in erythroid precursors.The degree of glycosylation increased concomitantly with dif-ferentiation.

The major sialoglycoprotein of human erythrocytes, glyco-phorin A (GPA), consists of 131 amino acids distributed inthree separate domains: at the cell surface, within the lipidbilayer, and in the cytoplasm (1). The external NH2-terminalportion is highly glycosylated, containing one N-glycosidicoligosaccharide at Asn-26 and 15 O-glycosidic oligosaccha-rides. The structure of the N-glycosidic oligosaccharide hasbeen determined (2). Most O-glycosidic oligosaccharideshave the structure Neu5Aca2-3Gall31-3(Neu5Aca2-6)Gal-NAc (ref. 3; Neu5Ac, N-acetylneuraminic acid; see ref. 4 forthe condensed symbolism for oligosaccharide chains).

It was shown that GPA is confined to the erythroid celllineage and appears at the basophilic normoblast stage oferythropoiesis (5, 6). GPA is also found on the erythroleuke-mia cell lines K562 (7) and HEL (8, 9). The biosynthesis ofthe protein has been extensively studied in K562 cells and itsN- and O-glycosylations have been elucidated (10-12).We have now isolated GPA from normal bone marrow

precursor cells and from the K562 and HEL cell lines beforeand after induction of differentiation, and we have studied itsoligosaccharides. Our results show that, during differentia-tion of red cells, the GPA molecules become increasingly 0-glycosylated. This change in structure of a major membranemolecule may be important to the understanding of cellularinteractions and of the relationship between cellular differen-tiation and membrane protein glycosylation.

MATERIALS AND METHODSCells. Normal and En(a-) erythrocytes were obtained

from the Red Cross Blood Transfusion Service, Helsinki.Bone marrow was recovered from pieces of ribs, which wereresected during open thorax surgery at the Helsinki Univer-sity Hospital. The cells were subjected to Ficoll-Isopaquecentrifugation, and the interphase cells were collected. Frac-

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

tionation according to cell size was accomplished using 1 x gvelocity sedimentation (13). Blood from neonatal patientswith erythroblastosis fetalis was obtained from the Depart-ment of Pediatrics, Helsinki University Hospital, and themononuclear cells were isolated. K562 cells were obtainedfrom G. Klein, Karolinska Institute, Stockholm, and HELcells from E. Papayannopoulou, University of Washington,Seattle.

Induction of Differentiation in K562 and HEL Cells. K562cells and HEL cells were grown in RPMI 1640 medium con-taining 10% fetal calf serum. Cells were induced to differenti-ate for 1-6 days with 1.5 mM sodium butyrate, 0.1 ,uM retin-oic acid, 10 nM phorbol i2-myristate 13-acetate (PMA), or25-50 ,uM hemin as described (14-16). The degree of differ-entiation was estimated from May-Grunwald-Giemsa-stained smears, and cultures containing differentiated cellswere harvested for further studies.

Antisera. Rabbit anti-GPA antisera were produced by im-munizing with purified GPA (17). The antisera were ad-sorbed with En(a-) red cell membranes, which lack GPA(18-20), as described (5). Monoclonal anti-GPA antisera R10and R18, and VIE-G4 were obtained from P. A. W. Edwards(21) and W. Knapp (22), respectively.

Binding of Protein A-Containing Staphylococci to Anti-GPA-Treated Bone Marrow Cells. The presence of GPA inbone marrow cells was assessed with a quantitative staphy-lococcal rosetting assay of anti-GPA antiserum-treated cells(5).

Radioactive Labeling. Cell surface glycoconjugates wereradioactively labeled using the periodate/NaB3H4 technique(23). Radiolabeled red cell membranes were isolated as de-scribed (24). Labeled nucleated cells and red cell membraneswere solubilized in 1% Triton X-100/0.01 M sodium phos-phate/0.15 M NaCl, pH 7.4, at 0°C and centrifuged at 5000 xg for 10 min, and the supernatants were recovered. For la-beling with [35S]methionine and 3H, 3 x 107 uninduced cellsor K562 cells induced with hemin for 3 days were incubatedfor 90 min with [35S]methionine (10), washed, and labeled bythe periodate/NaB3H4 method. After solubilizatioti in deter-gent, the extracts were passed through lentil lectin-Sepha-rose columns and the radioactive glycoproteins were elutedwith a-methylmannoside (10, 11).

Immunoprecipitation. Labeled cell extracts were subject-ed to immune precipitation using the staphylococcal proteinA technique (25). When monoclonal antibodies were used,rabbit anti-mouse IgG antiserun (Dako, Copenhagen) wasused as a second antibody.

Polyacrylamide Slab Gel Electrophoresis. Polyacrylamideslab gel electrophoresis in the presence of sodium dodecylsulfate was done using 8% acrylamide gels (26). The gelswere fixed with 5% sulfosalicyclic acid and treated for fluo-rography (27).

Abbreviations: GPA, glycophorin A; PMA, phorbol 12-myristate 13-acetate.

6752

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

Proc. Natl. Acad Sci USA 81 (1984) 6753

Table 1. Binding of protein A-containing Staphylococcus aureus cells to anti-GPA-treated bone marrow cells

RabbitmAb R10 mAb control Rabbit anti-GPA preimmune IgG

Cell type N Binding N Binding N Binding N Binding

Pronormoblasts 64 1.03 ± 3.98 56 0.13 ± 0.81 76 7.99 ± 10.39 50 2.88 + 7.9%Basophilic

normoblasts 66 2.30 ± 4.43 50 0.06 ± 0.31 76 15.43 ± 10.99 51 1.47 ± 2.16Polychromaticnormoblasts 53 11.83 ± 10.14 50 0.04 ± 0.20 63 21.60 ± 11.24 53 3.92 ± 7.73

Orthochromaticnormoblasts 59 17.80 ± 12.54 50 0.04 ± 0.20 63 21.27 ± 10.52 61 2.46 ± 2.71

Matureerythrocytes 76 16.84 ± 10.91 50 0 0 50 22.30 ± 8.99 50 1.48 ± 1.84

Binding of staphylococci to bone marrow cells that had been treated with the indicated antisera was determined by arosetting assay (5). Values are given as the mean and SD of the number of bacteria bound per bone marrow cell. N, numberof marrow cells examined; mAb R10, a monoclonal antibody specific for GPA; mAb control, a monoclonal antibody ofunrelated specificity.

Preparation and Analysis of Glycopeptides/Oligosaccha-rides. 3H-labeled GPAs isolated by immune precipitationwere treated with 5 mg of Streptomyces griseus protease(Pronase, Sigma) per ml of 0.15 M NaCl/0.01 M sodiumphosphate, pH 7.4/0.1% sodium dodecyl sulfate at 60°C for24 hr. After lyophilization, the samples were dissolved in0.25 ml of 0.05 M NaOH/1 M NaBH4 and incubated at 45°Cfor 16 hr to liberate O-glycosidic oligosaccharides (28). Onedrop of glacial acetic acid was then added and the sampleswere lyophilized. The samples were dissolved in 0.1 MNH4HCO3/0.1% sodium dodecyl sulfate and applied to a 1x 80 cm Bio-Gel P-6 column equilibrated in the same buffer.The void volume was determined each time using Blue Dex-tran 2000 (Pharmacia). Radioactivity in eluate fractions wasmeasured in a Triton X-114-based scintillation fluid using anLKB-Wallac 1210 Ultrobeta counter.

RESULTSReactivity of GPA from Normal Erythroid Cells and K562

Cells with Monoclonal Anti-GPA-Antibodies and Heteroanti-serum. Results obtained using the staphylococcal rosettingassay to detect reactivity with monoclonal (R10) and hetero-anti-GPA antiserum in normal bone marrow cells are shownin Table 1. The monoclonal antibody reacted poorly withpronormoblasts and basophilic normoblasts whereas poly-chromatic normoblasts and cells at later stages of differentia-tion showed a strong reaction. In contrast, rabbit anti-GPAantiserum showed a strong reactivity even with pronormo-blasts and basophilic normoblasts.

Surface-labeled blood erythrocyte membranes were sub-

A B

GPA-D- I

jected to immune precipitation using the R10 monoclonalanti-GPA antibody. Fractionation by polyacrylamide gelelectrophoresis revealed heavily labeled GPA monomer(GPA-M) and dimer bands (GPA-D) (Fig. 1, lane A). A simi-lar pattern was obtained using the rabbit antiserum (Fig. 1,lane B).The surface glycoprotein patterns of K562 cells before and

after hemin-induction are shown in Fig. 2 (lanes C and D).There was a relative increase in radioactivity in the positionof GPA from induced cells. Rabbit antiserum precipitatedGPA molecules from uninduced cells (Fig. 2, lane E) but noprecipitate was seen with the R10 antibody (Fig. 2, lane F).However, after hemin-induced differentiation, the monoclo-nal antibody also precipitated GPA (Fig. 2, lane H). Flowcytometry (FACS IV, Becton Dickinson) gave similar re-sults: K562 and HEL cells showed increased reactivity withthe monoclonal antibodies R10 and R18 after induced differ-entiation (results not shown).

Polyacrylamide Slab Gel Electrophoresis Patterns of GPAfrom Bone Marrow Cells and Blasts from Patients with Eryth-roblastosis Fetalis. Erythroid cells from bone marrow weresize-fractionated and surface-radiolabeled, and GPA wasisolated by immunoprecipitation with rabbit antiserum. The

A B C D E F G H

M-

PHb- "I SW1

GPA-D- kBS

-q

OA- *ErGPA-M--

CA-. GPB-

m * - GPA-M

GPA-M

FIG. 1. Fluorogram of a polyacrylamide slab gel after electro-phoresis of immunoprecipitates obtained from equal numbers of sur-face-labeled erythrocytes using monoclonal antibody R1O (lane A)or rabbit anti-GPA antiserum (lane B). GPA-D, GPA dimer; GPA-M, GPA monomer.

FIG. 2. Fluorogram of a polyacrylamide slab gel after electro-phoresis of extracts of surface-labeled uninduced and induced K562cells and of immunoprecipitates obtained with anti-GPA antisera.Lane A: "4C-labeled standard proteins (M, myosin; PHb, phosphory-lase b; BSA, bovine serum albumin; OA, ovalbumin; CA, carbonicanhydrase). Lanes B-D: extracts of surface-labeled erythrocytes(B), uninduced K562 cells (C), and hemin-induced K562 cells (D).Lanes E-H: immunoprecipitates obtained using rabbit anti-GPAantiserum and uninduced cells (E), monoclonal antibody R10 anduninduced cells (F), rabbit anti-GPA antiserum and induced cells(G), and monoclonal antibody R10 and induced cells (H). The cellswere allowed to differentiate for 3 days, and similar amounts of ra-dioactivity were used for the immunoprecipitations.

Cell Biology: Gahmberg et aL

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

6754 Cell Biology: Gahmberg et al.

Table 2. Erythroid cell composition of bone marrow fractions

% of total cells

Pro- Basophilic Poly- Ortho-Frac- normo- normo- chromatic chromatic Erythro-tion* blasts blasts normoblasts normoblasts cytes

1 20 45 20 15 02 5 15 17 63 03 3 2 5 55 35

*Cells were fractionated according to size by unit-gravity velocitysedimentation (13).

erythroid cell compositions of the three cell fractions isolat-ed are shown in Table 2. Fraction 1 was enriched in the earlyprecursor cells, fraction 2 consisted of a mixed cell popula-tion, and fraction 3 contained erythrocytes and late normo-blasts. From fraction 1, two weakly labeled bands were ob-served after polyacrylamide gel electrophoresis, one in theposition of the GPA monomer and the other, designated GP-26, with an apparent molecular weight of 26,000 (Fig. 3, laneB). The GPA monomer was the major species precipitatedfrom fraction 2 cells, but GP-26 was also obtained (Fig. 3,lane C). Only the species corresponding to GPA monomersand dimers were precipitated from fraction 3 (Fig. 3, lane D).GPA-M and GP-26 were both recovered from nucleatedblood cells of patients with erythroblastosis fetalis (Fig. 3,lane G).

Electrophoretic Mobilities of GPA Molecules Obtained fromK562 and HEL Cells Before and After Induction. Induction ofdifferentiation of K562 cells with sodium butyrate or hemindecreased the electrophoretic mobilities of the GPA mole-cules (Fig. 4, lanes A, C, and E and lanes G-J, respectively).Treatment of HEL cells with retinoic acid or PMA did notresult in any major change in the apparent molecular weightsof the GPA molecules (results not shown).

Analysis of Glycopeptides/Oligosaccharides of GPA Mole-cules. Gel filtration of 3H-labeled Pronase/alkaline borohy-dride-treated GPA molecules obtained by immune precipita-tion was used to determine relative degrees of glycosylation.We know (29) that the N-glycosidic glycopeptide appears inthe void volume (peak 1) of Bio-Gel P-6 columns, followedby the O-glycosidic tetrasaccharide (peak 2) and the O-gly-cosidic trisaccharides (peak 3). Fig. 5A shows that GPA fromblasts of patients with erythroblastosis fetalis were labeledmainly in the N-glycosidic oligosaccharide (peak 1). In GPA

A B C

GPA-D--sb

D E F G

from the most immature bone marrow cells (fraction 1, seeTable 2), there was more label in the O-glycosidic tetrasac-charide and some in the trisaccharide region (Fig. SB). GPAfrom fraction 2 was highly O-glycosylated (Fig. 5C) and hada labeling pattern similar to that of GPA from fraction 3, thefraction which contained late normoblasts and erythrocytes(Fig. 5D).GPA isolated from uninduced K562 cells contained a rela-

tively small amount of O-glycosidic oligosaccharides, andthe tetrasaccharide/trisaccharide ratio was lower than inGPA from bone marrow cells (Fig. 6A). A very small amountof radioactivity was precipitated with the monoclonal anti-body R10 (Fig. 6B) After differentiation induced by heminthe relative amount of tetrasaccharide increased (Fig. 6C).The GPA molecules from induced cells reacted with themonoclonal antibodies to give a glycopeptide/oligosacchar-ide pattern (Fig. 6D) similar to that obtained using the rabbitantiserum. Treatment with sodium butyrate gave a small rel-ative increase in tetrasaccharides (results not shown).To get a semiquantitative value for the change in glycosy-

lation, uninduced and hemin-induced K562 cells were la-beled with both [35S]methionine and periodate/NaB3H4. TheGPA molecules were isolated and the 3H/35S ratios were de-termined. The ratios were 1.51 for uninduced cells and 2.01for induced cells.The glycosylation of HEL cell GPA molecules also

changed after treatment with inducing agents. After cultiva-tion in the presence of retinoic acid, the relative level of tet-rasaccharides increased (Fig. 7B); PMA treatment had theopposite effect (Fig. 7C).

DISCUSSIONThere are few examples of polypeptides whose carbohydratestructures vary depending on the tissue localization or thedevelopmental stage of the cells of origin. Best known arethe ABO and Ii blood-group antigens, which in the red cellare associated predominantly with the band 3 and band 4.5proteins (30, 31). Fetal cells contain simple, essentially un-branched i-active oligosaccharides whereas erythrocytesfrom adults contain high molecular weight branched I-activeoligosaccharides (32). The rodent Thy-1 glycoproteins frombrain, thymocytes, and T-lymphocytes are also differentlyglycosylated (33-35).

A B C D E F9.

:--ss, GP100--

3, GPA-D--6M.-.

GPA-M - 4' GPA-MJ-GPA-26:!!- ;

GPB* I GPB-*l1

-GPA-MGPA-M -- -

GPA-D -"

G H I J

..

GPA-M b*-

-GPA-26

FIG. 3. Fluorograms of polyacrylamide slab gels after electro-phoresis of extracts of surface-labeled erythrocytes (lanes A and E)and of immunoprecipitates obtained using rabbit anti-GPA antise-rum and bone marrow blast cells (fraction 1) (lane B), bone marrowcells (fraction 2) (lane C), bone marrow cells (fraction 3) (lane D),and nucleated precursor cells from a patient with erythroblastosisfetalis (lane G). Also shown is the extract of the surface-labeled cellsof the patient (lane F). GPA-26, GPA molecule with an apparentmolecular weight of 26 000; GPB, glycophorin B; GP-100, glycopro-tein with an apparent molecular weight of 100,000. The GPA-26bands are barely seen in lanes B and G.

FIG. 4. Fluorograms of polyacrylamide slab gels after electro-phoresis of immunoprecipitates obtained using rabbit anti-GPA anti-serum and K562 cells before and after induction of differentiationwith sodium butyrate or hemin. Lanes A-F: patterns obtained fromuniriduced cells with antiserum (A) and preimmune serum (B); pat-terns obtained from cells induced with sodium butyrate for 3 dayswith antiserum (C) and preimmune serum (D); patterns obtainedfrom cells induced with sodium butyrate for 6 days with antiserum(E) and preimmune serum (F). Note the decreased mobility of theGPA monomer from induced cells. Lanes G-J: patterns obtainedwith antiserum and cells that were uninduced (G) or grown in thepresence of hemin for 1 day (H), 3 days (I), or 6 days (J). The mobil-ities of the GPA monomer (GPA-M) and dimer (GPA-D) bands showa decrease with increased time of induction.

Proc. NatL Acad Sci. USA 81 (1984)

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

Proc. Natl. Acad. Sci. USA 81 (1984) 6755

FIG. 5. Bio-Gel P-6 chromatography of3H-labeled glycopeptides/oligosacchar-ides from GPA molecules immunoprecipi-

_ tated with rabbit anti-GPA antiserum. (A)Pattern obtained from GPA of erythroidprecursor cells from a patient with eryth-roblastosis fetalis. (B) Pattern from GPAof bone marrow cells (fraction 1, Table 2).(C) Pattern from GPA of bone marrowcells (fraction 2). (D) Pattern from GPA ofbone marrow cells (fraction 3). Peak 1 cor-responds to the N-glycosidic glycopep-tide, peak 2 to the O-glycosidic tetrasac-charide, and peak 3 to the O-glycosidic tri-saccharide. All cells were surface-labeled

80 using periodate/NaB3H4. Arrow indicatesthe void volume.

The type of variation in the glycosylation of GPA is quitedifferent. GPA acquires an increased number of O-glycosid-ic chains when the erythroid cells differentiate. This wastrue for GPA both from normal precursor cells and fromerythroleukemia cell lines induced to differentiate. Thechanges in glycosylation were detected after labeling cellsurface sialoglycoconjugates by the periodate/NaB3H4method, which is specific for sialic acids (23). The resultswere essentially the same for cells labeled using the galac-tose oxidase/NaB3H4 technique (24) to detect terminal ga-lactose/N-acetyl galactosaminyl residues. The increased3H/35S ratio of [3 S]methionine/periodate/NaB3H4 labeledGPA molecules from hemin-induced cells also indicates thatthe number of O-glycosidic oligosaccharides increased dur-ing differentiation. The GP-26 band seen in some precursorcell preparations apparently represents GPA molecules witha very low level of O-glycosylation because a similar mole-cule was obtained when erythrocyte GPA was partially de-glycosylated with endo-N-acetylgalactosaminidase (12).GPA is not important for the mature red cell because

En(a-) individuals, lacking glycophorin A (18-20), do not

04

0D

x

Fraction

show any signs of erythrocyte malfunction. On the otherhand, it is possible that GPA is needed at earlier stages oferythrocyte differentiation. GPA and its incompletely glyco-sylated precursor molecules could function as receptors incellular recognition; f3-galactosyl-binding lectins have beenfound in several vertebrates (36).GPA has recently been shown to act as a receptor for the

malarial parasite Plasmodium falciparum (37, 38). The dif-ferentiation-related structural changes in the O-glycosidicoligosaccharide composition of GPA reported here could ex-plain the well-known restriction in infectibility of the P. fal-ciparum merozoites to mature red cells (39) and the inhibi-tion of merozoite binding by carbohydrate (40).The change in GPA structure during erythroid differentia-

tion is also reflected in its reaction with monoclonal anti-GPA antibodies. The R10 antibody reacts with an epitope inthe middle part of the polypeptide chain, the R18 antibodyreacts with a region close to the lipid bilayer (21), and theVIE-G4 antibody needs sialic acid for reactivity (22). All ofthese antibodies reacted weakly with the GPA moleculesfrom immature cells. This indicates that the carbohydrate

FIG. 6. Bio-Gel P-6 chromatographypatterns of 3H-labeled glycopeptides/oli-gosaccharides of immunoprecipitatedGPA from K562 cells. Shown are resultsobtained using uninduced cells and rabbitanti-GPA antiserum (A), uninduced cellsand monoclonal antibody R10 (B), cellstreated with hemin for 2 days and rabbitanti-GPA antiserum (C), and cells treatedwith hemin for 2 days and monoclonalantibody R10 (D). Arrow indicates thevoid volume.

IlJ0I

x

Fraction

Cell Biology: Gahmberg et aL

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

6756 Cell Biology: Gahmberg et al.

Fraction

FIG. 7. Bio-Gel P-6 chromatography patterns of 3H-labeled gly-copeptides/oligosaccharides of GPA molecules isolated from HELcells using rabbit anti-GPA antiserum. Patterns were obtained fromdigests of GPA from uninduced cells (A), from cells treated withretinoic acid for 3 days (B), and from cells treated with PMA for 3days (C). Arrow indicates the void volume.

contributes to the conformation of the antigenic determi-nants of GPA.We have earlier reported that the malignant blasts in a sig-

nificant proportion (610%) of undifferentiated acute leuke-mias carry surface structures that react with rabbit anti-GPAantiserum (41). Using monoclonal antibodies to GPA for thephenotyping of leukemic cells, only occasional reactivity isfound (22, 42). This discrepancy might be explained by thedifferentiation-related structural changes in GPA.

We thank Drs. P. A. W. Edwards and W. Knapp for monoclonalantibodies, U. Katajarinne for assistance, and B. Bj6rnberg for sec-

retarial help. This research was supported by National Cancer Insti-tute Grant 2 R01 CA26294-04, the Sigrid Juselius Foundation, andthe Academy of Finland.

1. Tomita, M. & Marchesi, V. T. (1975) Proc. Natl. Acad. Sci.USA 72, 2964-2968.

2. Yoshima, H., Furthmayr, K. & Kobata, A. (1980) J. Biol.Chem. 255, 9713-9718.

3. Thomas, D. B. & Winzler, R. J. (1969) J. Biol. Chem. 244,5943-5946.

4. IUB-IUPAC Joint Commission on Biochemical Nomenclature(JCBN) (1982) J. Biol. Chem. 257, 3347-3351.

5. Gahmberg, C. G., Jokinen, M. & Andersson, L. C. (1978)Blood 52, 379-387.

6. Robinson, J., Sieff, C., Delia, D., Edwards, P. A. W. &Greaves, M. (1981) Nature (London) 289, 68-71.

7. Gahmberg, C. G., Jokinen, M. & Andersson, L. C. (1979) J.Biol. Chem. 254, 7442-7448.

8. Martin, P. & Papayannopoulou, T. (1982) Science 216, 1233-1234.

9. Papayannopoulou, T., Yohochi, T., Nakamoto, B. & Martin,P. (1983) Globin Gene Expression and Hematopoietic Differen-tiation (Liss, New York), pp. 277-292.

10. Jokinen, M., Gahmberg, C. G. & Andersson, L. C. (1979) Na-ture (London) 279, 604-607.

11. Gahmberg, C. G., Jokinen, M., Karhi, K. K. & Andersson,L. C. (1980) J. Biol. Chem. 255, 2169-2175.

12. Gahmberg, C. G., Jokinen, M. & Andersson, L. C. (1983) RedCell Membrane Glycoconjugates and Related Genetic Mark-ers, eds. Cartron, J.-P., Rouger, P. & Salmon, C. (Librarie Ar-nette, Paris), pp. 51-63.

13. Hayry, P. & Andersson, L. C. (1976) Scand. J. Immunol. 5,31-44.

14. Andersson, L. C., Jokinen, M. & Gahmberg, C. G. (1979) Na-ture (London) 278, 364-365.

15. Rutherford, T. R., Clegg, J. B. & Weatherall, D. J. (1979) Na-ture (London) 280, 164-165.

16. Benz, E. J., Murnane, M. J., Tonkonow, B. L., Berman,B. W., Mazur, E. M., Cavallesco, C., Jenko, T., Snyder,E. L., Forget, B. G. & Hoffman, R. (1980) Proc. Nati. Acad.Sci. USA 77, 3509-3513.

17. Hamaguchi, H. & Cleve, H. (1972) Biochem. Biophys. Res.Commun. 47, 459-464.

18. Gahmberg, C. G., Myllyla, G., Leikola, J., Pirkola, A. &Nordling, S. (1976) J. Biol. Chem. 251, 6108-6116.

19. Dahr, W., Uhlenbruck, G., Leikola, J., Wagstaff, W. & Land-fried, K. (1976) J. Immunogenet. 3, 329-346.

20. Tanner, M. J. A. & Anstee, D. J. (1976) Biochem. J. 155, 701-703.

21. Anstee, D. J. & Edwards, P. A. W. (1982) Eur. J. Immunol.12, 228-232.

22. Liszka, K., Majdic, O., Bettelheim, P. & Knapp, W. (1983)Am. J. Hematol. 15, 219-226.

23. Gahmberg, C. G. & Andersson, L. C. (1977) J. Biol. Chem.252, 5888-5894.

24. Gahmberg, C. G. & Hakomori, S. (1973) J. Biol. Chem. 248,4311-4317.

25. Gahmberg, C. G. & Andersson, L. C. (1978) J. Exp. Med. 148,507-521.

26. Laemmli, U. K. (1970) Nature (London) 227, 680-685.27. Bonner, W. M. & Laskey, R. A. (1974) Eur. J. Biochem. 46,

83-88.28. Carlson, D. (1966) J. Biol. Chem. 241, 2984-2986.29. Gahmberg, C. G. & Andersson, L. C. (1982) Eur. J. Biochem.

122, 581-586.30. Karhi, K. K. & Gahmberg, C. G. (1980) Biochim. Biophys.

Acta 622, 344-354.31. Finne, J. (1980) Eur. J. Biochem. 104, 181-189.32. Hakomori, S. (1981) Semin. Hematol. 18, 39-62.33. Barclay, A. N., Letarte-Muirhead, M., Williams, A. F. &

Faulkes, R. A. (1976) Nature (London) 263, 563-567.34. Hoessli, D., Bron, C. & Pink, J. R. L. (1980) Nature (London)

283, 576-578.35. Carlsson, S. R. & Stigbrand, T. I. (1983) J. Immunol. 130,

1837-1842.36. Barondes, S. H. (1981) Annu. Rev. Biochem. 50, 207-231.37. Perkins, M. (1981) J. Cell Biol. 90, 563-567.38. Pasvol, G., Wainscoat, J. S. & Weatherall, D. J. (1981) Nature

(London) 297, 64-67.39. Tanner, M. J. A. (1982) Trends Biochem. Sci. 7, 231.40. Jungery, M., Boyle, D., Patel, T., Pasvol, G. & Weatherall,

D. J. (1983) Nature (London) 301, 704-705.41. Andersson, L. C., Gahmberg, C. G., Teerenhovi, L. & Vuo-

pio, P. (1979) Int. J. Cancer 24, 717-720.42. Greaves, M. F., Sieff, C. & Edwards, P. A. W. (1983) Blood

61, 645-651.

Proc. NatL Acad Sci. USA 81 (1984)

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0