~ole~~l~r ~m~una~ogy, Vol. 24, No. 2, pp. 171-176, 1987 Printed in Great Britain.

0141-5s~/g7 $3.00 f 0.00 0 1987 Pergamon Journals Ltd

CHARACTERIZATION OF A MURINE MONOCLONAL ANTIBODY SPECIFIC FOR THE HUMAN Pl BLOOD

GROUP ANTIGEN

PASCAL BAILLY*, JEAN CHEVALEYRE~, DANIELLE SONDAG:, CHRISTIANE FRANCOIS-G~RARD$ YVES PIcp_m,f GJ?RARD VEzoNt and JEAN-PIERRE CARTRON*§

*Unit& INSERM U76, Centre National de Transfusion Sanguine Institut, 6, rue Alexandre Cabanel, 75015 Paris, France; ?Centre de Transfusion Sanguine, Place Amelie Raba Lion, 33000 Bordeaux, France,

and fCentre de Transfusion Sanguine, 41 rue DOS Fauchon, 4020 Liege, Belgique

(First Received 23 June 1986; accepted 15 July 1986)

Abstract-Four monoclonal antibodies (MAbs) directed against the Pl blood group antigen were produced by hybridomas obtained from mouse immunized with turtle-dove avomucoid. One of the MAb (154 IX B6) selected as a blood typing reagent agglutinated native Pl and Pk 1 red cells with a high titer but was inactive against native P2, Pk2 and p erythrocytes. After papain treatment the reactivity towards Pl and Pkl erythrocytes was enhanced whereas p erythrocytes remained unreactive. A weak cross- reactivity of the MAb with the Pk antigen was suspected since enzyme-treated Pk2 erythrocytes became significantly agglutinated. Further analysis of the antibody specificity was established by binding studies using neutral glycolipids prepared from Pl and P2 erythrocytes, affinity immunoabsorbents carrying known oligosaccharide structures and hapten inhibition with synthetic oligosaccharides. The MAb bound weakly to the Gal CL I-4Gal structure co&mon to PI and Pk antigens but had a marked preference for the Pl determinant (Gal GI 1-4 Gal /? l-4 GlcNAc) and the bindine was abolished bv nrior treatment of . . oligosaccharide antigens by a(not p)- galactosidase, which supports evidence that a terminal a-galactose residue is involved in the blood group Pl and Pk specificities. The MAb has a slightly broader specificity than the human anti-P1 counterpart but can be used safely for routine blood typing.

The human P and Pl blood group systems consist of two common phenotypes, Pl and P2 and of three rare phenotypes Pk 1, Pk 2 and p, which were first identified with human antisera (Race and Sanger, 1975) and more recently with murine monoclonal antibodies (MAbs) directed against the Pk (Nudelman et al., 1983; Fellous et al., 1985) and P antigens (von dem Borne er af., 1986).

The P antigen is present in all normal erythrocytes and the Pl antigen is present in about 75% of individuals. The Pk antigen occurs only in two rare phenotypes, Pkl and Pk2, which lack the P antigen, but it is detectable in small quantities both chemically (Marcus et al., 1976; Fletcher et al., 1979) and immunologically (Naiki and Kato, 1979) on all nor- mal erythrocytes. Erythrocytes from p individuals Iack all the three antigens P, Pl and Pk. During the last several years these antigens have been identified on human red cells as glycosphingolipids (for review see Marcus et uf. 1981) and the structures are as follows:

Pk: Calcr lAGal@ 1-4Glc-ceramide (ceramide tri- hexoside)

§Author to whom correspondence should be addressed.

P: GalNA@ l-3Gala 1-4GalP l~Glc-ceramide (Globoside)

Pl : Galor l -4Gal/ll-4GlcNAcP 1 - 3Gal/3 l -4Glc - ceramide.

The Pk and P antigens are not only present on red cells and erythroblasts but also on platelets, mega- karyocytes, fibroblasts and endothelial cells (Fellous et al., 1977 and 1985; von dem Borne et al., 1986). The P antigen however is absent from immature red cell precursors as well as from neutrophils, granu- locytes, monocytes and peripheral blood lymphocytes and appears therefore as a marker of red cell differentiation (von dem Borne et al., 1986). Glob- oside (P antigen) has been also characterized on human and mammalian tissues (Martensson, 1969) but not on lymphocytes (Stein and Marcus, 1977). The Pk antigen (ceramide trihexoside) is present on large amounts on red cells and lymphoid cell lines from Pk individuals (Fellous et al., 1985) and accumu- lates on Burkitt lymphoma cells (Wiels et al., 1981; Nudelman et al., 1983). The tissue distribution of the Pl antigen is largely unknown since this glycolipid is present in minute amounts on cell membranes and good quality reagents with this specificity are not frequent. In this report we have serologically and biochemically characterized a murine MAb which can be useful for blood typing and celluar expression of the Pl antigen.

I71

172 PASCAL BAILLY et al.

MATERIALS AND METHODS

Materials

Ovomucoid was purified from turtle-dove egg white (Streptopelia risoria) as described previously (Francois-Gerard et af., 1979).

The neutral glycolipids were extracted from eryth- rocytes according to Kundu et al. (1979) Standard glycolipids were donated by Prof. D. M. Marcus (Houston, TX). The blood group P-active tet- rasaccharide (GaINA@ I-3Galcc 1-4Gal/J 1-4Glc) was a generous gift from Prof. H. Paulsen (Ham- burg, F.R.G.). Pl and Pk active trisaccharides

(Gala 1-4Gal/3 1-4GlcNA@0-propyl and Galcc l- 4Galb 14Glc) were synthesized and donated by Prof. P. Sinaj; and Dr J. C. Jacquinet (Orleans, France). The disaccharides Gal/l 14GlcNAc, Fuccr l-6Gal and the trisaccharide Gal/I 14GlcNAcP l-3 Gal were donated by Dr S. David and Dr A. Veyritres (Orsay, France).

Bovine serum albumin (BSA), galactose (Gal), glucose (Glc), N-acetyl-galactosamine (GalNAc), N-acetylglucosamine (GlcNAc), methyl-a-D-galacto- side, methyl-fi-o-galactoside, b-galactosidase from jack beans E.C. 3.2.1.2.3., cr-galactosidase from coffee beans E.C. 3.2.1.22, a-galactosidase from As- pergillus niger E.C. 3.2.1.22 were purchased from Sigma, St. Louis, MO. Oligosaccharides coupled to an insolubilized matrix (Synsorb) were obtained from Chembiomed Ltd, Edmonton, Alberta, Canada (structures given in Table 2). ‘*‘INa (555 MBq ‘*jI/pg iodine) was supplied by Amersham, U.K. Iodogen was purchased from Pierce Chem. Co. (Rockford, IL). Red cells from individuals typed in the P blood group system were from the Centre National de Transfusion Sanguine, Paris.

Hybridoma and antibody labelling

The monoclonal antibody designated 154 IX B6 was produced by the hybridoma technique after fusion of mouse splenocytes (Biozzi mouse from the Foundation Curie, Institut du Radium, Paris) sensi- tized with turtle-dove ovomucoid with the murine myeloma cells P3 x 63 ag-8-653.1 as described in a preliminary report (Vezon et al., 1986). Hybridoma cells were grown in BALB/c mice primed with 0.5 ml pristane (2,6,10,14-tetramethylpentadecane, Aldrich chemicals, U.S.A.). Ascitic fluid was purified by gel filtration on an Ultrogel AcA34 column (IBF Reac- tifs, Villeneuve-La-Garenne, France) equilibrated in 10 mM sodium phosphate buffer, pH 7.2, 150 mM NaCl (PBS). The IgM fraction with blood group Pl reactivity was radioiodinated by the iodogen pro- cedure (Fraker and Speck, 1978) and stored in PBS containing 1% (w/v) BSA at - 20°C. The ‘251-label in the MAb (sp. act. 2 x lo5 cpm/pg protein) was 95% trichloracetic acid precipitable.

Enzyme treatments and agglutination assays

Agglutination assays using native and protease-

treated human Pl, P2, Pkl (Let. F.), Pk2 (Let. H.) and p (Val.) erythrocytes were performed with serial dilutions of MAb in PBS-BSA at 4 and 22°C as described before (Fellous et al., 1985). Digestion of ovomucoid by cr-galactosidase from Aspergillus niger was performed according to FranGois-Gerard et al. (1980). Hydrolysis of neutral glycolipids bound to microtiter plates (see below) with 0.1 unit (per well) of c(- and /J-galactosidase (one unit = amount of enzyme which releases 1 pmole of galactose from appropriate p-nitrophenyl-galactoside/min under op- timal conditions) was carried out in 50mM sodium citrate, pH 4.5, containing 1% (w/w) BSA, 2 mM EDTA for the jack beans B-galactosidase or in 50 mM phosphate citrate buffer, pH 5.4, containing 1% (w/v) BSA for the coffee beans cc-galactosidase. After 18 hr at 37°C the excess of enzyme was removed and the microtiter plates were washed, dried and processed for antibody binding.

Solid-phase binding assays

The binding of ‘2SI-labelled antibody to neutral glycolipids extracted from Pl or P2 erythrocytes was measured by solid-phase radioassay before and after treatment of the glycolipids with CI- and /3-galactosidases. Each assay was performed in tripli- cate. Glycolipids in 20~1 of methanol were added to each well of a polyvinylchloride microtiter plate (Microtest III, Becton-Dickinson) and the solutions dried by evaporation. After 30min, the wells were filled with PBS-BSA and left for 1 hr at room temp (saturation step). One hour later the wells were emptied and filled with the ‘251-labelled monoclonal antibody solution. The plates were covered with parafilm, incubated for 2 hr at 4°C and washed 6 times with cold PBS. Each well was then cut from the plate and assayed individually for “‘1 in a LKB gamma counter (model 1274). The solid-phase bind- ing assay with ovomucoid was conducted using Dy- natech microtiter plates (Dynatech Laboratories Inc).

In other assays, different oligosaccharides coupled

to Synsorb immunoadsorbents (see Table 2) were aliquoted in 0.5 mg portions in polypropylene tubes and saturated for 2 hr at room temp with PBS-BSA. The tubes were then emptied and filled with the radioiodinated MAb solution in 20~1 of PBSBSA (lo5 cpm/well). After 2 hr at 4°C with gentle shaking, the supernatants were removed and the Synsorbs washed 6 times with cold PBS and assayed individu- ally for “‘1 in the LKB gamma counter.

Immunostaining on thin -layer plates

Binding of the 154 IX B6 MAb to neutral gly- colipids on TLC plate was performed according to Brockhaus et al. (1982) with some modifications. Purified neutral glycolipids were chromatographed on aluminium-backed high-performance thin-layer chromatography plates (Merck, F.R.G.) in chloroform/methanol/water (60: 35 : 8, by vol) and the dried chromatogram was dipped in n-hexane

Monoclonal anti-P1 antibody 173

containing 0.075% (w/v) polyisobutylmethacrylate (Polyscience, Inc., Warrington, PA) for 1 min at room temp, followed by a 2 set spray with PBS and a saturation step of 45 min at 4°C in PBS-BSA. The plate was overlayed with the binding solution (6.5 ,ul/cm* area) containing 10 pg of 154 IX B6 MAb per ml and incubated for 3 hr at 4°C in a humidified atmosphere.

The chromatogram was then washed by dipping in four successive changes of cold PBS. After incubation with ‘251-labelled rabbit anti-mouse IgM (Fc) (2.5 x 106cpm/ml) at 4°C for 3 hr, the plate was finally washed as before, dried and exposed to X-ray film (Fuji RX, Japan).

RESULTS AND DISCUSSlON

The ovomucoid from the egg white of turtle-dove carries penta-antennary glycans with terminal galac- tose residues c( 14 linked to N-acetyllactosaminyl groups (FranGo&Gerard et al., 1979). This oli- gosacharide structure correponds to the terminal hapten of the Pl blood group antigen (Naiki et al.,

1975). Four murine IgM MAbs reacting with PI but not

P2 erythrocytes have been produced by hybridoma technique after fusion of the murine myeloma cells P3 x 63 ag-8-653.1 with the splenocytes from mice immunized with ovomucoid (Vezon et al., 1986). One of these MAbs (clone 154 IX B6) reacting with a high titre, was selected and further characterized by immu- nochemical and biochemical investigations.

At 4°C the purified IgM antibody agglutinated Pl and P’ 1 native erythrocytes at 22 and 5 ng/ml concn, respectively, whereas P2 and p red cells were still not recognized at concns even higher than 200hg/ml (Table 1). The MAb however reacted weakly with Pk2 erythrocytes (at an antibody concn of 180 ng/ml) which have no PI antigen but accumulate the cer- amide trihexoside (Pk antigen), a glycolipid sharing the terminal Galcl1-4Gal disaccharide with the PI antigen.

After papain-treatment, both Pl and Pk 1 red cells were agglutinated to a similar extent (0.35 ng/ml of MAb) but p erythrocytes remained unreactive. In contrast, papain-treated Pk2 and P2 red cells were agglutinated, respectively, by the MAb at 6 and 180 ng/ml. It is not yet established, but it is likely that these reactions are related to the ceramide trihexoside

Table 1. Reactivity of the MAb 154 IX 86 with native and papain-treated human erythrocytes

Minimum agglutinating concn (ng/ml) 4°C 22°C

- _I_____.___ Erythrocytes U P u P

~.... Pi 22 0.35 150 38 P2 > 200,000 180 z 180,000 > 180,000 Pkl 5 0.35 150 38 P’2 180 6 39,000 300

P > 200,000 > 200,000 > 180,000 > 180,000

U = untreated; P = papain-treated.

content of PK 1 and Pk2 red cells. It is well known that P2 red cells carry a small amount of ceramide tri- hexoside (Naiki and Kato, 1979; Fletcher et al. 1979) which may become available to the MAb after reor- ganisation of the membrane components following enzyme treatment (Hakomori and Kannagi, 1983). The MAb is a cold reacting antibody since it reacts more weakly at room temp than at 4°C with native and papain-treated Pl red cells, respectively (Table 1).

The specificity of the MAb was further investigated by a solid-phase binding assay with neutral glycolipid extracts prepared from PI and P2 erythrocytes. Fig- ure l(A) shows that the “51-labelled MAb binds to the neutral glycolipids of Pl red cells absorbed to microtiter plates but not to the P2 neutral glycolipids. The antibody also binds strongly to the turtle-dove ovomucoid which was used for the immunization of mice [Fig. l(B)]. The binding is abolished after a-galactosidase digestion of both the glycolipids and the ovomucoid bound to the plates [Fig. l(A) and (B)], but the treatment by a p-galactosidase prepara- tion had no effect [Fig. l(A)]. This is indicative that the MAb binds specifically to non-reducing cc-galactose containing components present in Pl glycolipids and in ovomucoid, respectively.

The specificity of the MAb was further studied by antibody binding to the neutral glycolipids from Pl and P2 erythrocytes separated by TLC followed by autoradiography (Fig. 2). In the lipid extract from PI red cells, the antigen specific of the MAb 154 IX B6 was found to be a neutral glycolipid, presumably the PI glycolipid, with the mobility of As-GM, (Gal/l 1 - 3GalNAcfi 1 - 4Gal@ l -4Glc/ll - lceramide) under the conditions of thin-layer chromatography described by Naiki et al. (1975). As expected from agglutination studies, no binding to the neutral gly-

colipids from P2 red cells could be detected. In addition, there was no binding to other purified glycolipids such as: CMH, CDH; CTH, globoside, paragloboside, Forssman glycolipid and As-GM I used as standards (for abbreviations see legend of Fig. 2). The MAb does not react with CTH on TLC plates (Fig. 2) although agglutination studies with papain-treated Pk2 and P2 red cells have suggested that the MAb might react with this glycolipid in situ.

This is not contradictory since several examples indicate that the expression of a glycoliplid antigen is modulated by the composition and organization of the components at the cell surface (Hakumori and Kannagi, 1983; Kannagi et al., 1983). A clustering of the CTH molecules on the red cell surface might also enhance the affinity and the agglutination efficiency of the MAb (Singer and Nicolson, 1972).

In another experiment, we examined the binding properties of the MAb 154 IX B6 to well defined oligosac~haride structures immobilized on an inert matrix. Under the conditions used, the MAb binds in a similar manner to the Galcc I-4GalP 14GlcNAc (Pl) t~sa~haride and to the Gala 14GaI structure which is common to the PI and Pk determinants

174 PASCAL BAILLY et AI.

Ovomucoid (pg)

Fig. 1. Binding of the ‘251-labelled MAb 154 IX B6 in solid-phase radioassays against glycoliplid and glycoprotein preparations and effect of galactosidases. (A) Binding of the MAb to intact neutral glycolipid extracts from PI(@) and P2(W) erythrocytes coated on microtiter plates as described under MateriaIs and Methods. About 1 x 105cpm of MAb added to each well in 20 ~1 PBS-BSG. Symbols (0) and ( x ) show the binding of the MAb to PI glycoliplids treated, respectively, by the LX-galactosidase from coffee beans and by the /I-galactosidase from jack beans. (B) Binding of the MAb to the turtle-dove ovomucoid coated to microtiter plates before (A) and after (A) treatment by the a-galactosidase from Aspergdhs niger.

About 2.7 x 10scpm of MAb added to each well in 20~1 of PBS-BSA.

(Table 2). Unfortunately, the trisaccharide Gala1-4Galfi 1AGlc (true P”) coupled to the inert

matrix was not available for comparison. The MAb binds to the structurally related Gala l-3Gal deter- minant as previously noted by inhibition of hemag-

Table 2. Binding reactivity of the ‘2sI-labelled MAb 154 IX B6 to oligosaccharide structures covalently coupled to inorganic carrier

Per cent of ‘251-labelled Structures MAb bound

Gala 1-4Gala 14GlcNAcj?-R 100 Gala IHGalB-R 93.6 Gala I-3Galp-R 35.0 Gala I-3Galp-R 0

Tell-2 LFuc

GalP 1-4GlcNAc~-R 3.0 Galb I-3GlcNAc/3-R 1.7 Galp 14Glc&R 1.9 Galfi l-3GalNAca-R 2.6

lO’cpm of ‘ZSf-labelled MAb added to a fixed amount of Synsorb reagents (see Material and Methods) in triplicate experiments. All data corrected for the sugar hapten content of each type of Synsorb. (R = Synsorb inert matrix, from Chemibiomed, Ed- monton, Alberta.)

glutination with human anti-P1 anti-sera (Watkins and Morgan, 1976) but it does not bind significantly to the blood group B trisaccharide which contains

this disaccharide or to other disaccharides with a terminal b-galactose (Table 2).

In order to delineate more precisely the specificity of the MAb, inhibition binding assays with oli- gosaccharides of known structures have been carried out (Fig. 3). Binding of the MAb to ovomucoid is inhibited 50% by 0.008mM of PI trisaccharide and 1.8mM of Pk trisaccharide but not by the other structurally related oligosaccharides: GalNAc, Glc- NAc, Gal, Glc, Gala-O-methyl, Galp-O-methyl, lactose, N-acetyllactosamine, Fuccr l-6Ga1, Galj? l- 4GlcNAcP 1-3 Gal. From these data, it is clear that the MAb binds about 200 times better to the Pl than to the Pk structure, a value in close agreement with the results obtained previously by inhibition of hu- man anti-P1 with oligosaccharides isolated from hy- datic cyst fluid (Watkins and Morgan, 1976) and with glycolipids from human red cells (Naiki et al., 1975;

Watkins and Morgan, 1976). By comparing the Pk

Monoclonal anti-P1 antibody 175

S-f-D 1 2 STD 1 2 Fig. 2. Binding of the ‘z51-labelled MAb 154 IX B6 to thin-layer chromatograms containing neutral glycolipid extracts from PI and P2 red cells as detected by autoradiography following the procedure described under Materials and Methods. About 0.3 nmole of glycolipids were applied and developed in chlorofo~/methanol/water, ~/35/~ by vol. Panel A, detection by orcinol spray; Panel B, autoradiogram. Lanes 1 and 2: neutral glycosphingolip~ds from Pl and P2 red cells, respectively. Lane STD: standard mixture of neutral glycosphingoliplids. The abbreviations used are: CMH, Glcfi I-I-ceramide; CDH, Gal/? l~GI@l-lceramide; CTH, Gala 1-4Gal~I-4Glc~ I-1-ceramide; Globoside (Glob), GalNAc@l- 3GaIa lAGal@ 14Glcfl I-l-ceramide; Paragloboside (PG), Galp 1-4GlcNAc~ I-3GalP 14Glc/3 l- lceramide; Forssman ~ycolipid (Forss), GalNAcu I-3GalNAcfi I-3Gala l-QGal@ 14Glcp I- I-ceramide

and As-G,, , Gal@ l-3GdlNAcjI l-4Gal,!IlWGlc~ l-I-ceramide.

and Pl trisaccharides we have shown (Fig. 1) that the hand, the reducing end of the Pl trisaccharide used N-acetyl group of the Pl trisaccharide is blocked by a /3-U-propyl linkage, and this group Ga~al~Gal~l-4 GlcNAc probably plays a critical might also contribute to enhance the inhibitory po- role in the blood group Pl specificity. On the other tency of the PI antigen.

OSOOl oml 0.01 0.1

Ofigosaccharide (mM)

1 10

Fig. 3. Inhibition of ‘2sI-labelled MAb 154 IX B6 to turtle-dove ovomucoid by synthetic oligosaccharides. Solid-phase radioimmunoassays were carried out as described under Materials and Methods using 0.05 pg of turtle-dove ovomucoid coated to microtiter plates and 104cpm of L251-labelled MAb per well. The symbols related to each inhibitor tested are: (0) GalNAc, GlcNAc, Gal, Glc, Gala-U-methyl, Galfi-O-methyl, Gal/? 1_4GIc, Gal@ 1-4GlcNAc, Fuca I -6Ga1, Galfl 14GlcNAc~ I-3Gal;

( x )Gaia 1-4CalP 14Glc, specificity Pk; and (0) Gala 14Gall-4GlcNAc/?-O-propyl, specificity Pi.

176 PASCAL BAILLY et al.

In conclusion, we have reported the first example of murine monoclonal antibody against the blood group Pl structure and our immunochemical and biochemical investigations demonstrate that this MAb which reacts predominantly with the PI deter- minant can also recognize the Pk determinant to some extent, although 200 times less well. This is not unexpected from the structures of the two antigens and taken together our studies show that this MAb can be safely used for routine PI typing by using native or papain-treated erythrocytes at room temp (Table 1). This antibody will be also useful for tissue typing.

Acknowledgements-The authors wish to thank Prof. D. M. Marcus (Houston, TX) far the gift of glycolipids and Prof. H. Paulsen (Hamburg, F.R.G.), Prof. P. Sinaj; and Dr J. C. Jacquinet (Orleans, France) for the donation of P. Pl and Pk oligosaccharides. They are also indebted to Prof. S. David and A. Veyrieres (Orsay, France) for the gift of synthetic oligosaccharides.

REFERENCES

Brokhaus M., Magnani S. L., Herlyn M. D., Blaszczyk M., Steplewski Z., Koprowski H. and Ginsburg V. (1982) Monoclonal antibodies directed against the sugar se- quence of lacto-N-fucopantaose III are obtained from mice immunized with human tumors. Archs Biochem. Biophys. 217, 647-65 I.

Fellous M., Cartron J. P., Wiels J. and Trusz T. (1985) A monoctonal antibody against a Burkitt lymphoma associ- ated antigen has an anti-Pk red blood cell specificity. Br. J. Haematoi. 60, 559-565.

Fellous M., Gerbal A., Nobillot G. and Wiels J. (1977) Studies on the biosynthetic pathway of human erythro- cyte antigen using genetic complementation tests between fibroblasts from rare p and Pk phenotype donors. Pox Sang. 32, 2622268.

Fellous M., Gerbal A., Tessier C., Frezal J., Dausset J. and Salmon C. (1974) Studies on the biosynthetic pathway of human P erythrocyte antigens using somatic cells in culture. VOX Sang. 26, 518-536.

Fletcher K. S., Bremer E. G. and Schwarting G. A. (1979) P blood group regulation of glycosphingolipid levels in human erythrocytes. J. biol. Chem. 254, 11196-l 1198.

Fraker P. M. and Speck J. C., Jr. (1978) Protein and cell membrane iodinations with a sparingly soluble chloro- amide I ,3,4,6-tetrachloro-3a, 6~-diphenylglycolu~l. Bio- them. biophys. Res. Commun. 80, 849-857.

Franqois-Gerard C., Brocteur J., Andre A., Gerday C., Pierce-Cretel, A., Montreuil J. and Spik G. (1980) Demonstration of the existence of a specific blood group Pl antigenic determinant in turtle-dove ovomucoid. Bfood fI’ran.$ Immunohemat. 23, 579-588.

Franqois-Gerard C., Gerday C. and Beeley J. G. (1979) Turtle-dove ovomucoid, a glycoprotein proteinase in-

hibitor with Pl-blood-group antigen activity. Biochem. J. 177, 679-685.

Hakomori S. and Kannagi R. (1983) Glyosphingolipids as tumor-a~ociat~ and differentiation markers. .I. natn. Cancer Inst. 71, 231-251.

Kannagi R., Stroup R., Cochran N. A., Urdal D., Young W. W., Jr and Hakomori S. (19831 Factors affecting expression of glycolipid tumor antigens: influence 07 ceramide composition and coexisting glycolipid on anti- genicity of gangliotriaosylceramide in murine lymphoma cells. Cancer Res. 43, 4997-5005.

Kundu S. K., Steane S. M., Bloom J. E. C. and Marcus D. M. (1978) Abnormal glucolipid composition of eryth- rocytes with a weak P antigen. VOX Sang. 35, 161f-167.

Marcus D. M., Kundu S. K. and Suzuki A. (1981) The P blood group system: recent progress in immunochemistry and genetics. Sem. Hematol. 18, 63-71.

Marcus D. M., Naiki M. and Kundu S. K. (1976) Abnor- malities in the glycosphingolipid content of Pk and p erythrocytes. Proc. natn. Acad. Sci. U.S.A. 73,2362-2367.

Miirtensson E. (1969) Gly~osphingolipids of animal tissue. In Progress in the Chemistry of Fats and Other Lipids, Vol. 10, p. 367. Pergamon Press, Oxford.

Naiki M., Fong J., Ledeen R. and Marcus D. M. (1975) Structure of the human erythrocyte blood group Pl glycolipid. Biochemistry 14, 43314837.

Naiki M. and Kato M. (1979) Immunolo8ical identification of blood group P” antigen on normal human erythrocytes and isolation of anti-PL with different affinity. I’ox Sang. 37, 3C-38.

Race R. and Sanger R. (1975) BIooaGroups in Man, 6th Edn Blackwell Scientific Publications, Oxford, pp. 139-I 77.

Singer S. J. and Nicolson G. L. (1972) The fluid mosaic model of the structure of cell membranes. Science 175, 720-73 I.

Naiki M. and Marcus D. M. (1975) An immunochemical study of the human blood group PI, P and Pk gly- cosphingolipid antigens. Biochemistry 14, 48374841.

Nudelman E., Kannagi R., Hakomori S. I., Parson M., Lipinski M., Wiels J., Fellous M. and Tursz T. (1983) A glycolipid antigen associated with Burkitt lymphoma defined by a monoclonal antibody. Science 220.5099511.

Stein K. I?. and Marcus D. M. (1977) Glycosphingolipids of purified human lymphocytes. Biochemistry 16, 5285-5291.

Vezon G., Piquet Y., Chevaleyre J., Sondag-Thull D. and Cartron J. P. (1986) Obtention et caracterisation d’anticorps monoclonaux de specificitb anti-PI. C.r. Sot. Biol. 180, 57-63.

von dem Borne A. E. G. Kr., Bos M. J. E., Joustra-Maas N., Tromp 3. F., van? Veer M. B., van Wijngaarden-du” Bois R. and Tetteroo P. A. T. (19861 A murine mono- _ clonal IgM antibody specific for blood group P antigen (globoside). Br. J. Haematol. 63, 3546.

Watkins W. M. and Morgan W. T. J. (1976) Immuno- chemical observations on the human blood group P system. J. Immunogenet. 3, 15-27.

Wiels J., Fellous M. and Tursz T. (1981) Monoclonal antibody against a Burkitt-l~phoma-associated antigen. Proc. natn. Acad. Sci. U.S.A. 78. 6485.-6488.