Quantitation of the Number of Molecules of C and D on Normal Red Cells Using RadioiodinatedFab Fragments of Monoclonal

By Jon Smythe, Brigitte Gardner, andDavid J. Anstee

Two rat monoclonal antibodies (BRAC 1 and BRAC 1 1 ) cytes. Fabfragments of BRAC 1 1 and ERIC 10 gave values have been produced. BRAC 1 recognizes an epitope com- of 143,000 molecules GPC per red blood (RBC). Fab mon to the human erythrocyte membrane fragments of BRAC1 gave 225,000 molecules of GPC and C (GPC) and glycophorin D (GPD). BRAC 11 GPD per RBC. These results indicate that GPC and GPD is specific for GPC. Fabfragments of these antibodies and together are sufficiently abundantto provide membrane at- BRlC 10, a murine monoclonal anti-GPC,were radioiodin- tachment sites for all ofthe 4.1 in normal RBCs. ated and used in quantitative binding assays to measure 0 1994 by The American Societyof . the number of GPC and GPD molecules on normal erythro-

HE SHAPE AND deformability of the mature human (200,000)" and those reported for GPC (50,000).7 This nu- Downloaded from http://ashpublications.org/blood/article-pdf/83/6/1668/612763/1668.pdf by guest on 24 September 2021 T erythrocyte is controlled by a flexible two-dimensional merical differencehas led to the suggestion that a significant lattice of , which together comprise the membrane proportion of protein 4.1 in normal erythrocyte membranes skeleton.' The major components of the skeleton are spec- must be bound to sites other than GPC and GPD.3 The trin, , , and protein 4.1. The skeleton is at- value of 50,000 molecules of GPC per RBC was obtained tached to the cytoplasmic face of the through using radioiodinated murine monoclonal IgG anti-GPC.7 interactions involving ankyrin and protein 4. l. It is known This value is a minimum estimate, since accessibilityto the that ankyrin interacts with the N-terminal cytoplasmic do- relevant epitope on all GPC molecules may beimpeded by main of the integral , band 3. The way in steric hindrance resulting from the size of whole IgG mole- which protein 4.1 binds to the plasma membrane is less cules. In this report, we have measured the abundance of clearly understood. Evidence has been presentedto support GPC using Fab fragments of rat and murine monoclonal interaction of protein 4.1 with band 3,2-4glycophorin A,' anti-GPC antibodies. We also describe the use of Fab frag- (GPC) and glycophorin D (GPD),6s7 and ments from a rat monoclonal that recognizes an phospholipid.' The relative importance ofthese interactions epitope common to GPC and GPD. remains unclear. However, there is strong evidence in sup- MATERIALSAND METHODS port of a role for GPC and GPD in vivo. GPC andGPD are Human RBCs were obtained from Blood Services South West, present in skeletons preparedfrom normal erythrocytes,but Bristol, UK. Rat monoclonal antibodies (BRACs 1 and 11) were are absent from skeletons prepared from the erythrocytes produced in LOU rats in response to two intraperitoneal injections of individuals with hereditary elliptocytosis resulting from of human RBCs (0.2 ml of 50% suspension in phosphate-buffered homozygous protein 4. l defi~iency.~,~Erythrocyte mem- saline [PBS], pH 7) 24 days apart. Three days after the second im- branes of the Leach , which have a total defi- munization, splenocytes were fused with Y3.Ag. 1.2.3. myeloma ciency of GPC and GPD,7lack a membrane binding site for cells," in the ratio 4: I, using PEG 3500. Hybrid cells were selected protein 4.1, since incontrast to normal membranes, protein by culture in Dulbecco's modified Eagle's medium (DMEM; con- 4.1 is readily extracted from Leach phenotype membranes taininghypoxanthine, thymidine, aminopterin [HAT]) supple- mented with 20% fetal bovine serum (F'BS) and 10% Y3 culture under the low ionic strength conditions used to isolate spec- supernatant in the presence of irradiated 3T3 fibroblasts. BRICs 4 trin." and 10, murine monoclonal antibodies to GPC, were as previously One problem concerning interpretation of the in vivo sig- described.' , , and treatment of RBCs nificance of GPC and GPD as membrane binding sites for were as described el~ewhere,'~ was as the treatment of RBCs with protein 4.1 has been the apparent difference between the sialidasei4and with 2 aminoethylisothiouronium bromide (AET).'' number of protein 4.1 molecules per (RBC) Immunoblotting was performed as described el~ewhere,'~except that 5% wt/vol bovine milk powder was used as the blocking agent, Immobilon-P membranes (Millipore Ltd,Watford, UK) were used, From the International Blood Group Reference Laboratory, Bris- and rabbit anti-rat peroxidase conjugate (DAKO, High Wycombe, tol, UK; and the Department of Haematology, St Mary's Hospital UK) was used with the BRAC antibodies. The rat monoclonal anti- Medical School, London, UK. body isotypes were determined using a dot-blot method essentially Submitted May 25, 1993; accepted November 5, 1993. as described by McDougal et IgG was purified from cell-culture Presented in part at the Annual Meeting of the British Blood supernatant using Protein A Sepharose (Pharmacia, Milton Transfusion Society, York, UK, September 8-11, 1992 and pub- Keynes, UK) andIgM was purified using ion exchange on Q Seph- lished in Transf Med 2.69, 1992 (suppl l). arose fast-flow followed by gel filtration through Sephacryl S300 Address reprint requests to David J. Anstee, PhD, Director, ln- (Pharmacia) as recommended by the manufacturer. Methods for ternational Blood Group Reference Laboratory, Southmead Rd, the purification of Fab fragments, radiolabeling of whole immuno- Bristol BSI0 5ND,UK. globulin and Fab fragments, and their use in quantitative binding, The publication costs of this article were defrayed in part by page functional affinity, and competitive inhibition assays were as pre- charge payment. This article must therefore be hereby marked viously described." "advertisement" in accordance with 18 U.S.C.section I734 solely to RESULTS indicate this fact. 0 1994 by TheAmerican Society of Hematology. Characterization of rat monoclonal antibodies BRAC 1 0004-4971/94/8306-0002~3.00/0 and BRAC 11. BRAC 1 and BRAC agglutinated nor-

1668 Blood, Vol83, No 6 (March 15). 1994: pp 1668-1 672 WANTITATION OF GLYCOPHORINS C AND D ONRBCs

Origin -

MW 40K -

MW 32K - Downloaded from http://ashpublications.org/blood/article-pdf/83/6/1668/612763/1668.pdf by guest on 24 September 2021 MW 30K Fig 1. lmmunoblotting with BRAC 1 and BRAC - 11. Erythrocyte membrane components were sep- arated under nonreducing conditions by sodium do- decyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 10% (wt/vol) acrylamide gels with a 3% (wt/vol) overlay." Lanes a and 1, normal erythrocyte membranes.Lanes b and g, mem- branes from Leach phenotype erythrocytes. Lanes ! c and h, membranes from trypsin-treated normal erythrocytes. Lane d. membranes from Gerbich- negative erythrocytes of the Yus type. Lane e, a bcde fgh membranes from Gerbich-negative erythrocytes of the Ge phenotype. BRAC 1 BRAC 11

mal RBCs, but failed to react with RBCs ofthe Leach phe- ration (Scatchard analysis; Fig 2 and Table l). Values ob notype, which lack GPC and GPD. The reactivity of both tained with BRAC 1 gave site numbers for GPC plus GPD antibodies was unaffected by prior treatment of normal averaging 7,000 per RBC using IgMand 238,000 (corrected RBCs with chymotrypsin, sialidase, or AET. BRAC 1 re- to 225,000 assuming binding to Leach phenotype RBCs acted with pronase-treated RBCs and gave markedly re- [Table l] represents nonspecific binding) using Fab frag- duced reactions with trypsin-treated normal RBCs. BRAC 11 failed to react with either pronase-treated or trypsin- treated normal RBCs. Table 1. Number of Available Binding Sites on RBCs for Anti-GPC (BRAC 11, BRlC 4, BRlC 10) and Anti-GPC/D (BRAC 1) Analysis of BRAC 1 and BRAC I l by immunoblotting and Values of Affinity Constants showed that BRAC 1 bound to both GPC and GPD in nor- Sites per RBC mal RBC membranes, but failed to react with membranes Affinity of the Leach phenotype and gave very weak reactivity with Antibody Normal Leach K (x 10' mol/L") trypsin-treatednormal membranes (Fig la throughc). BRlC 4 IgG 62,525 ? 3,9613.000 (1) l .4 f 0.7 BRAC 1 alsoreacted with the abnormal GPC found in (16) (4) Gerbich-negative RBCs of the Yus type, but not with the BRlC 10 IgG 47,683 ? 4,4964.000 (2)4.9 f 1.8 abnormal GPC found in Gerbich-negativeRBCs of the Ge (4) (16) type (Fig Idand e). BRAC 1 l bound onlyto GPC in normal BRlC 10 Fab 142,463 f 13,660 ND 0.67 t 0.02 RBCs and failed to react with membranes ofthe Leach phe- (8) (7) notype or trypsin-treatednormal membranes (Fig If BRAC 1 1 IgG 29,456 f 1,065 ND 2.3 (2) through h). (5) Quantitative binding experiments using rat and murine BRAC11Fab 143,410t10.448 ND 0.12 (2) (4) IO monoclonalantibodies GPCand GPD. Rat monoclonal BRAC 1 IgM 7,374 ? 1,225 ND ND antibodies BRAC 1 and BRAC 1 1 and murine monoclonal (5) antibody BRIC 10 were purified and Fab fragments pre- BRAC 1 Fab 238,721 ? 22,38013,380 t 2.464 0.22 (2) pared from eachantibody. The purified antibodies and Fab (3) (4) fragments were radioiodinated and usedin quantitative Mean t SD values are givenfor three or more measurements (the num- binding assays. The number of molecules ofGPC and GPD ber of determinations is givenin parentheses). All other values are either per cell were calculated fromthe maximum number of an- the meanfrom two experiments or derived from a single experiment. tibody moleculesor Fab fragments bound perRBC at satu- Abbreviation: ND, not determined. 1670 SMYTHE, GARDNER, AND ANSTEE

" I 1 mal GPC resulting fromthe deletion of 3 of the normal GPC , which codes for residues36 to 63in normal GPC.'8.'9 In contrast, Gerbich-negative cells ofthe Yus type contain an abnormal GPC, which resultsfrom the deletion of exon 2 of the normal GPC gene, which codes for

XI \ I amino acid residues 17 to 35 of normal GPC.'' BRAC I \ bound to the abnormal GPC from Yus type cells, but not to A\ that from Ge type cells, indicating that the epitope recog- nized by BRAC 1 on normal GPC is located inthe region of residues 36 to 63.Trypsin treatment of normal RBCs fI cleaves GPC atArg-4g2' and since the reactivity of BRAC I is markedlyreduced after trypsin treatment of normal RBCs, it seems likely that the epitope recognized resides in the region of residues 36 to 48 of normal GPC. GPC and GPD are products of the same gene.2' GPD is a shortened form of GPC (corresponding to residues 22to 128 of GPC), Downloaded from http://ashpublications.org/blood/article-pdf/83/6/1668/612763/1668.pdf by guest on 24 September 2021 probably resultingfrom leaky initiation.22Amino acid resi- dues 15 to 27 of GPD correspond to residues 36 to 48 of 0 002 0.04 0.08 008 01 0.12 014 016 018 02 GPC, and this region would be expectedto contain the epi- Bound Fab fragments (us) tope recognized by BRAC I on GPD.A murine (NaM 19-3C4) ofthe same specificity as BRAC l Fig 2. Representative Scatchard plots obtained for the '261-la- beled Fab fragments of antibodies BRlC10, BRAC 11, and BRAC1. has recently been describedby Loirat et aLz3 The plot showsthe binding curves derivedfrom three experiments BRAC l 1 is specific for GPC (Fig If through h) and so performed at different times. An aliquot of RBCs was counted for must recognize an epitope in the region of residues 1 to 2 I each experiment.The maximum amount of antibody (re) bound (x- of GPC. Competitive inhibition experiments demonstrated axis intercept) was converted into molecules of antibody and ex- that BRAC l 1 sees a related epitope to that recognized by pressed as antibody-binding sites per RBC assuming1 a :l binding ratio; 1.35 X 10' RBC extrapolatedto 0.1 45 pg of bound Fab frag- BRIC 4. BRIC 10, unlike BRIC 4, requires the free amino ments for BRlC 10 l+).1.05 X 1O7 RBC extrapolatedto 0.12 pg of group at the N-terminus of GPC for rea~tivity.~~These re- bound Fab fragments for BRAC 11 (m), and 0.56 X 10' RBC ex- sults suggest the location of epitopes for BRAC1, BRAC 1 1, trapolated to 0.125 pg of bound Fab fragments for BRAC1 (A). BRIC 4, and BRIC 10 as depicted in Fig 3. Quantitative binding experiments gave results of 7,000, 29,500, and 48,000 GPC molecules per RBC for BRAC 1 ments. Values for BRAC1 1 gave site numbers for GPC av- (IgM), BRAC I1 (IgG), and BRIC 10 (IgG), respectively. eraging 29,500 withIgG and 143,000 withFab. Similar re- The value obtained for BRIC 10 agrees well with that ob- sults of 48,000with IgG and 142,000with Fab were tained previously.' When similar quantitative assays were obtained with BRIC 10. Size-exclusion chromatography of performed using radioiodinated Fab fragments of BRAC 1, radioiodinated BRIC 10 Fab on Sephadex G-IO0 demon- BRAC I I, and BRIC 10, the results (Fig2 and Table 1)dem- strated that the Fab fragments were monomeric (data not onstrated a dramatic increase inthe number of epitopes and shown). The functional affinity constants for immunoglob- hence GPC and GPD molecules bound. Fab fragments of ulin and Fab binding were also calculated. The Fab frag- the two anti-GPC antibodies (BRAC 1 1 and BRIC 10) gave ments from BRAC 11 and BRIC 10 gave affinity constants values in the region of 143,000 molecules bound per cell, approximately 10 times lower than those found with IgG while the anti-GPC plus GPD antibody BRAC l gave values (Table l). of approximately 225,000 molecules per cell. These results Competitive inhibition experimentsusing BRiC 4, BRiC suggest that there is considerable steric constraint on the 10, and BRAC 11. The ability of an approximate 10-fold binding of whole antibody molecules. We have previously molar excess of unlabeled IgGfrom BRIC 4, BRIC 10,and BRAC 1 1 to inhibit the binding of radioiodinated BRIC 4, BRIC 10, and BRAC 11 to normal RBCs was determined. Table 2. Inhibition of Binding of 'Z61-Labeled Anti-GPCto RBCs The results demonstrated that while BRIC 4 and BRAC 11 by Unlabeled Antibodies strongly inhibit the binding of each other, BRIC 10 defines Inhibition by Unlabeled Antibody (%) a separate epitope (Table 2). '251-Labeled Antibodv BRlC 4 BRlC 10 BRAC 11 DISCUSSION ~ ~ ~ BRlC 4 78 ? 15 (9) 31 11 (9) 71 ? 13 (8) The results of agglutination tests and immunoblotting BRlC 10 7 k 6 (6) 82 k 4 (6) 0.6 k 0.6 (4) clearly show that BRAC 1 recognizes an epitope common BRAC 11 78 k 14 (4) 24 f 4 (4) 90 f 4 (4) to GPC and GPD. The position of this epitope was further 'Z61-labeled IgG(0.1 pg) was incubated with RBCs (2 to 5 CL) and unla- defined by immunoblotting of membranes from Gerbich- beled IgG at 10-fold molar excess. Results are mean values & SD for the negative RBCs ofthe Ge type and the Yus type (Fig Id and percentage inhibition. The number of determinations is given in paren- e). Gerbich-negative cells ofthe Ge type contain an abnor- theses. QUANTITATION OF GLYCOPHORINS C AND D ON RBCS 1671

port, we have soughtto address directlythis particular ques- tion. Our results show that GPC, GPD, and protein 4.1 are present in similar amounts, and that it is therefore theoreti- cally possible for all protein 4.1 molecules to interact with GPC and GPD in a normal RBC. These resultsdo not allow us to distinguish the relative roles of GPC, GPD, band 3, , and phospholipid in binding 4.1 in vivo. Nevertheless, when taken together withother e~idence,~.~," it seems likelythat GPC, GPD, and protein 4.1 interactions play a major role in maintaining the functional integrityof the RBC membrane.

'[ Trypsin-)2/ REFERENCES 1. Palek J, Lambert S: Genetics of the red skele-

ton. Semin Hematol27:290, 1990 Downloaded from http://ashpublications.org/blood/article-pdf/83/6/1668/612763/1668.pdf by guest on 24 September 2021 2. Pasternack GR, Anderson RA, Let0 TL, Marchesi VT: In- teractions between protein 4.1 and band 3. J Biol Chem 260:3676, 1985 3. Lombard0 CR, Willardson BM, Low PS: Localisation of the protein 4.1 binding site on the cytoplasmic domain of erythrocyte membrane band 3. J Biol Chem 267:9540,1992 4. Jons T,Drenckhahn D: Identification ofthe binding interface involved in linkage of cytoskeletal protein 4.1 to the erythrocyte anion exchanger. EMBO J 11:2863,1992 5. Anderson RA, Marchesi VT Regulation of the association of membrane skeletal protein 4.1 with glycophorin by a polyphospho- inositide. Nature 318:295, 1985 6. Mueller TJ, Momson M: Glycoconnectin (PAS-2) a mem- brane attachment site for the human erythrocyte. Prog Clin Biol Res 56:95, 198 1 Fig 3. Diagrammatic representationof GPC and GPDshows the 7. Anstee DJ, Parsons SF, Ridgwell K, Tanner MJA, Merry AH, location of epitopes recognized by BRAC 1, BRAC 11, BRlC 4, and Thomson EE, Judson PA, Johnson P, Bates S, Fraser ID: Two indi- BRlC 10. viduals with elliptocytic red cells apparently lack three minor eryth- rocyte membrane . Biochem J 2 I8:615, 1984 8. Rybicki A, Heath R, Lubin B, Schwartz RS: Human erythro- cyte protein 4.1 is a phosphatidyl binding protein. J Clin In- observed a twofold increase thein binding of Fab fragments vest 8L255, 1988 over IgG with antibodies to CD44, CD58,25*26 acetylcholin- 9. Alloisio N, Morle B, Bachir D, Guetarni D, Colonna D, De- esterase,*' and some antibodies against glycophorinA," but launay J: Red cell membrane in homozygous and only in the case of some monoclonalantibodies to the Kell heterozygous 4.1 (-) hereditary elliptocytosis. Biochim Biophys have we seen Fab to IgG ratios of the magni- Acta 81657, 1985 tude observed in the present study.28 Curiously, there is 10. Pinder J: Red cell membrane and the control of some rather tentative evidence that GPC and GPD may be membrane properties. Biochem SOCTrans 19:1039, 1991 associated with the Kell glycoprotein, since RBCs of the 1 1. Tyler JM, Reinhardt BN, Branton D: Associations of eryth- Leach phenotype and Gerbich-negative cellsof the Ge type rocyte membrane proteins. Binding of purified band 2.1 and 4.1 to show weakened expression of Kell bloodgroup .29 . J Biol Chem 255:7034, 1980 12. Galfre G, Milstein C, Wright B: Rat x rat hybrid myelomas The marked degree steric hindrance observed when IgG of and a monoclonal anti-Fd portion of mouse IgG. Nature 277: 131, is used to quantitate GPC suggests that GPC molecules may 1979 be clustered in the RBC membrane. Such a concept would 13. Judson PA, Anstee DJ: Comparative effect of trypsin and be consistent with a model of the RBC skeleton in which chymotrypsin on blood group antigens. Med Lab Sci 34: 1,1977 several (twoto 12) spectrin dimers,and therefore band 4.1 14. Mallinson GM, Martin PG, Anstee DJ, TannerMJA, Merry molecules (one 4.1 molecule per spectrin dimer), bind to a AH, Tills D, Sonnebom H: Identification and partial characterisa- single actin core.3o tion of the humanerythrocyte components which express the anti- Although there is considerable indirect evidencein sup- gens ofthe LW blood group system. Biochem J 234:649, 1986 port of the hypothesis that GPC and GPD provide mem- 15. Advani H, Zamor J, Judd WJ, Johnson CL, Marsh WL: In- brane attachment sitesfor protein 4.1 in normal activation of Kell blood group antigens by 2-aminoethyl isothio- uronium bromide. Br J Haematol51: 107, 1982 RBCS,~,',~,'~others have suggesteda major role for band3,24 16. McDougal JS, Browning SW, Kennedy S, Moore DD Im- GPA,' and phospholipids.' The arguments in favorof these munodot assay for determining the isotype and chain type of other binding sites have been supported by evidence that murine monoclonal antibodies in unconcentrated hybridoma cul- insufficient GPC and GPD are present in normal RBCs to ture supernatants. J Immunol Methods 63:28 l, 1983 accommodate all of the protein 4.1 molecules.' In this re- 17. Gardner B, Parsons SF, Merry AH, Anstee DJ: Epitopes on 1672 SMYTHE,GARDNER, AND ANSTEE sialoglycoprotein a: Evidence for heterogeneity in the molecule. with the N-terminal domain of glycophorin C. Biol Chem Hoppe Immunology 68:283,1989 Seyler 370849, 1989 18.High S, Tanner MJA,MacDonald EB, AnsteeDJ: Re- 25. Anstee DJ, Gardner B, Spring FA, Holmes CH, Simpson arrangements of the red cell membrane glycophorin C (sialoglyco- KL, Parsons SF, MallinsonG, Yousaf SM,Judson PA: New mono- protein j3) gene. Biochem J 262:47, 1989 clonal antibodies in CD44 and CD58: Their use to quantify CD44 19. Colin Y, Le van Kim C, Tsapis A, Clerget M, d’Auriol L, and CD58 on normal human erythrocytes and to compare the dis- London J, Galibert F, Cartron J-P Human erythrocyteglycophorin tribution of CD44 and CD58 in human tissues. Immunology 74: C gene structure and rearrangement in genetic variants. J Biol 197,1991 Chem 261:229,1986 26. Fletcher A, Bryant JA, Gardner B, Judson PA, Spring FA, 20. Dahr W, Beyreuther K, Kordowicz M, Kruger J: N-terminal Parsons SF, Mallinson G, Anstee DJ: New monoclonal antibodies amino acid sequence of sialoglycoprotein D (glycophorin C) from in CD59: Use for the analysis of peripheral blood cells fromPNH human erythrocyte membranes. Eur J Biochem 125:57,1982 patients and for the quantitation of CD59 on normal and DAF- 2 1. Le van Kim C, Colin Y, Blanchard D, Dahr W, London J, deficient erythrocytes.Immunology 75:507, 1992 Cartron J-P: Gerbich group deficiency of the Ge:1-2-3 and Ge: 27. Spring FA,Gardner B, Anstee DJ: Evidencethat the antigens 1,-2,3 types.Eur J Biochem 165:571, 1987 of the Yt blood group system are located on human erythrocyte 22. Tanner MJA, High S, Martin PG, Anstee DJ, Judson PA, acetylcholinesterase. Blood8021 36, 1992

Jones TJ: Genetic variants of human red cell membranes sialogly- 28. Parsons SF, Gardner B, Anstee DJ: Monoclonal antibodies Downloaded from http://ashpublications.org/blood/article-pdf/83/6/1668/612763/1668.pdf by guest on 24 September 2021 coprotein j3: Study ofthe alterations occumng in the sialoglycopro- against Kell glycoprotein, serology,immunochemistry and quanti- tein @gene.Biochem J 250:407, 1988 fication of sites.Transfus Med 3: 137, 1993 23. Loirat MJ, Gourbil A, Frioux Y, Muller J, Blanchard D: A 29. Anstee DJ: Blood groupactive surface moleculesof the hu- murine monoclonal antibody directed againstthe Gerbich 3 blood man red blood cell.Vox Sang 58:1, 1990 group antigen. Vox Sang 62:45,1992 30. Cohen CM: The molecular organizationof the red cell mem- 24. Dahr W, Blanchard D, Kiedrowski S, Poschmann A, Car- brane skeleton. SeminHematol20: 14, 1983 tron J-P, Moulds J: High frequencyantigens of human erythrocyte 3 1. Laemlli UK Cleavage of structural proteins during the as- membrane sialoglycoproteins. VI Monoclonal antibodies reacting sembly of the head of bacteriophageT4. Nature 227:680, 1970