HISTO-BLOOD GROUP A/B ANTIGEN DELETION/ REDUCTIONVS. CONTINUOUSEXPRESSION IN HUMAN TUMOR CELLS AS CORRELATEDWITH THEIR MALIGNANCYDaisuke ICHIKAWA 1,2, Kazuko HANDA1 and Sen-itiroh HAKOMORI1*1Pacific Northwest Research Foundation, Seattle, WA, USA, and Department of Pathobiology,University of Washington, Seattle, WA, USA2First Department of Surgery, Kyoto Prefectural University of Medicine, Kyoto , Japan

Deletion or reduction of histo-blood group A or B antigenin tumors of A or B individuals is clearly correlated with thedegree of malignancy and metastatic potential in many typesof human cancer. Haptotactic motility of A1H2 or B1H2

colonic or gastric tumor cell lines produced by transfection ofA or B gene was significantly lower than that of parental A2H1

or B2H1 cells. This is ascribable to reduced function of a3 ora6/b1 integrin receptor as we have recently shown. However,phenotypic changes resulting from gene transfection may notreflect physiological states associated with deletion or reduc-tion vs. continuous expression of A or B antigen in tumors.We now describe the separation and phenotype characteriza-tion of A2 cells from A1 tumor cell lines derived originallyfrom colonic tumors of patients with histo-blood group A. A1

and A2 populations were detected in originally A1 tumor celllines SW480 and HT29. A2 separated from A1 populationsisolated from SW480 and HT29 were characterized by greatlyenhanced haptotactic motility associated with reduced ordeleted A expression at a3, a6, and b1 integrin receptorswhich control cell motility. Nevertheless, expression of inte-grin receptors at the surface of A2 populations is the same asthat for A1 populations for both SW480 and HT29 cells. Thus,A vs. H glycosylation in integrin receptors may alter theirhaptotactic function. Cell proliferation as reflected by 3H-thymidine incorporation was also reduced significantly in A1

as compared to A2 populations. Our findings indicate that thedegree of haptotactic motility and proliferation of colonictumor cells are physiologically associated with the deletion orreduction vs. continuous expression of the histo-blood groupA antigen. Int. J. Cancer 76:284–289, 1998.r 1998 Wiley-Liss, Inc.

Deletion or reduction of histo-blood group A or B epitopes intumors of patients with histo-blood group A or B status wasoriginally observed in gastric cancer by Masamuneet al. (1952,1958). Extensive immunoadherence studies on lung and cervicalcancer showed that invasive and metastatic properties are corre-lated with the degree of A/B deletion (Davidsohnet al., 1969;Davidsohn and Ni, 1969). Studies on A/B deletion in oral(Dabelsteenet al., 1975) and bladder carcinoma (Ørntoft, 1992)showed a similar correlation. Subsequently, the effect of deletion ofA and B determinants in primary lung carcinoma on 5- or 10-yearsurvival of patients was studied. In one study, continued Aexpression was correlated with a higher survival rate, while Areduction or deletion was strongly correlated with reduced survival,but there was no such correlation between B expression andsurvival rate (Leeet al., 1991). In contrast, in an independent studyof the same type of carcinoma, both A and B expressionvs.reduction/deletion were correlated with survival rate (Matsumotoet al., 1993). These observations indicate that addition or deletionof a single sugar residue (i.e. a1 r B genes and observed resultingphenotypic changes (Ichikawaet al., 1997). Results of experimentswith such a transfection approach were clear. Nevertheless, resultsof this approach may not reflect the physiological phenotypicchanges associated with spontaneous A/B deletion in tumors, sincetransfectionper secauses artificial overexpression ofA or B genes.In contrast, phenotypic changes of tumor cells (e.g., migration,proliferation) associated with spontaneous deletion or reduction ofA/B antigen in tumors more closely reflect physiological processes,i.e., spontaneous deletion or persistent expression of A/B epitopes

observed in human cancer development, which strikingly affect theprognostic outcome of the patient. In view of these considerations,we separated A2 variants from A1 tumor cell lines, and comparedhaptotactic motility and cell proliferation of the variantsvs.parental cells.

MATERIAL AND METHODS

Cell lines, antibodies and reagents

Histo-blood group A1 cell lines SW480 and HT29 werepurchased from the ATCC (Rockville, MD). According to theATCC catalog description, the host blood group phenotype of thesecell lines was A, though the genotype is unknown. The followingcell lines were used as controls. HRT18-A9 was one of the clonesestablished after transfection of histo-blood group A gene (pAAAAincluded in pcDNA 3 vector) into HRT18 cells (H phenotype)(Ichikawaet al., 1997). HRT18-vec was obtained after transfectionof pcDNA 3 vector without inclusion of A gene (Ichikawaet al.,1997). Anti-A monoclonal antibody (MAb) 81FR2.2, anti-H MAb92FR-A2, and anti-human integrin subunita3 mouse MAb P1B5were from Dako (Carpinteria, CA). Anti-human integrin subunita6rat MAb GoH3 was from Pharmingen (San Diego, CA). Anti-human integrin subunitb1 mAb ZH1 was prepared as describedpreviously (Zhenget al., 1994). Matrigel was purchased fromBecton Dickinson, Bedford, MA.

Separation of A2 and A1 cell populations by panningNon-adherent plastic petri dishes (Fisher Scientific, McLean,

VA) were coated by incubating goat anti-mouse IgM (Sigma, StLouis, MO) with 50 µg/mL PBS at 4°C overnight. After washingwith PBS 33 and blocking with 1% BSA in PBS at room temp for1 hr, the dishes were incubated with anti-A MAb 81FR2.2 (50µg/mL) at room temp for 2 hr, and washed several times with PBS.Tumor cells were added to the dish and incubated at 4°C for 45min. Unbound cells were collected as A2 populations. The plateswere washed with PBS 33, and bound cells were collected as A1

populations by vigorous agitation through repeated pipetting. BothA2 and A1 populations were cultured and expanded. This panningprocedure was repeated 3 times for each A2 and A1 populationfrom SW480 and HT29 cell lines.

*Correspondence to: Pacific Northwest Research Foundation, 720 Broad-way, Seattle, WA 98122, USA. Fax: (206)726-1212. E-mail: hakomori@u.washington.edu

Abbreviations used: PBS, phosphate-buffered saline (9.2 mM sodiumphosphate buffer, pH 7.4, containing 138 mM NaCl, 0.9 mM CaCl2, 2.7mM KCl, and 0.5 mM MgCl2); SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis.

Contract Grant Sponsor: National Cancer Institute; Contract GrantNumber: CA42505.

Received 27 October 1997; Revised 26 November 1997

Int. J. Cancer:76,284–289 (1998)

r 1998 Wiley-Liss, Inc.

Publication of the International Union Against CancerPublication de l’Union Internationale Contre le Cancer

FIGURE 1 – Flow cytometric patterns of SW480 (upper panel) and HT29 cells (lower panel) with anti-A MAbs. Left: A1 populations. Center:parental cells. Right: A2 populations. Abscissa, log fluorescence. Ordinate, cell number.

FIGURE 2 – Matrigel-dependent haptotactic motility of A1 (leftcolumn) and A2 populations (right column) of SW480 (panela) andHT29 cells (panelb). Cell numbers were counted under magnification3100 and340 for panelsa andb, respectively. Values are means oftriplicate experiments. Bar: standard variation. Statistical significanceof the difference between A1 and A2 values is indicated byp valuefrom Student’st-test.

FIGURE 3 – 3H-Thymidine incorporation in A1 (left column) and A2

populations (right column) of SW480 (panela) and HT29 cells (panelb). Values are means of triplicate experiments. Data expression andstatistical significance as in Figure 2.

3H-thymidine incorporation assayCells were seeded on 96-well plates (23 104 cells/well),

incubated at 37°C for 24 hr, 1 µCi3H-thymidine was added, andcells were harvested 4 hr later onto glass fiber paper using a cellharvester. Thymidine incorporation was measured by liquid scintil-lation countering. Assays were performed in triplicate. Statisticalsignificance of the difference between A1 and A2 populations isindicated byp value from Student’st-test.

Haptotactic activity assayHaptotactic assay was performed based on the previously

published method using 6.5 mm Transwell chamber (8 µm poresize; Costar, Cambridge, MA) (Albiniet al., 1987; Sadahiraet al.,1992), with slight modification as follows. Briefly, the lowersurface of the filter was coated with 25 µL Tris-HCl buffercontaining various quantities of Matrigel (0.5, 1.5, or 5.0 µg perfilter), and the filter was dried at room temp. Cells were harvestedand placed in the upper compartment of a Transwell chamber. Cellsthat had not migrated from the upper surface of the filter wereremoved using a cotton swab, and cells that had migrated to thelower surface were fixed in methanol, stained with Toluidine blue,and counted under a microscope. Assays were performed intriplicate. Statistical significance of the difference between A1 andA2 populations is indicated byp value from Student’st-test.

Determination of antigen expressionExpression of A, H, and integrins was determined using flow

cytometry with specific primary and secondary antibodies asdescribed previously (Handaet al., 1995; Itoet al., 1994).

Western blotting analysisCells were harvested and solubilized in lysis buffer (140 mM

NaCl, 10 mM Tris-HCl, 5 mM EDTA, pH 7.8) containing 1%Triton X-100, 2 mM phenylmethyl-sulfonyl fluoride, and 0.05—0.1 trypsin inhibition units (TIU)/mL aprotinin. Cell lysates wereprepared by centrifugation at 10,000g for 15 min. Protein contentof the lysates was determined with a BCA kit (Pierce, Rockford,IL). For immuno-adsorption, cell lysates (50 µg protein) wereprecleared with protein G-agarose beads (Santa Cruz Biotechnol-ogy, Santa Cruz, CA) for 2 hr at 4°C. After centrifugation, thesupernatant was incubated with anti-integrin-a3, -a6, -b1 subunitantibody for 2 hr at 4°C, and then incubated overnight at 4°C withprotein G-agarose beads to collect the immunocomplex. Thefractions immunoadsorbed on beads were washed 53 with lysisbuffer, and bound protein was eluted by boiling in electrophoresissample buffer containing 5% 2-mercaptoethanol and subjected to8% SDS-PAGE.

Proteins were electrophoretically transferred to Immobilon-P(Millipore, Bedford, MA), blocked overnight in PBS containing5% defatted milk (4°C), and washed in PBS containing 0.05%Tween-20 (PBS/Tween). Membranes were incubated withanti-A MAb for 2 hr at 4°C, washed in PBS/Tween, then incubatedwith horseradish peroxidase-labeled secondary antibody, andblotted bands were detected by a chemiluminescence system(Pierce).

RESULTS

Flow cytometric profile of A2 and A1 populationsas compared to parental cells

Flow cytometric analysis revealed the presence of A2 popula-tions in parental SW480 and HT29 cells (Fig. 1, center panel inupper and lower rows), although both cell lines were originallyisolated from colonic carcinoma of patients with blood groupphenotype A. Repeated panning procedures led to successfulseparation of A2 (Fig. 1, upper and lower right panels), and A1

populations (Fig. 1, upper and lower left panels). H-antigenexpression detected with MAb 92FR-A2 was highly positive forboth A2 HT29 and A2 SW480 populations, as expected. A1 HT29

FIGURE 4 – Expression of integrin receptors (a6 andb1) in A1 and A2

populations of SW480 (upper) and HT29 cells (lower), as determined byflow cytometry. Vertical lines indicate average control staining with normalmouse IgG.Abscissa, log fluorescence; Ordinate, cell number.

FIGURE 5 – Western immunoblotting pattern of A antigen ona3, a6,andb1 integrin receptors from A1 and A2 cell lines and their controls.Immuno-adsorbed fraction of cell lysate in the presence of anti-integrinreceptor antibodies was separated on SDS-PAGE, followed by Westernblotting. Lane 1, HRT18-vec (negative control). Lane 2, HRT18-A9(positive control). Lane 3, A2 SW480. Lane 4, A1 SW480. Lane 5, A2HT29. Lane 6, A1 HT29.

286 ICHIKAWA ET AL.

FIGURE 6 – Two possible processes (A and B) to explain spontaneous deletionvs.persistence of A antigen in tumors from A individuals asrelated to tumor malignancy. Process A: an unidentified factor (e.g., transcription factor) that down-regulates A/B glycosyltransferase genetranscription or that promotes an unknown mechanism to enhance cell proliferation is produced during multi-step tumor progression. Commontranscription factors which controls glycosyltransferases and other cellular phenotypes are known. For example, transcription factorets-1positiveregulates expression of metalloprotease gene which affects angiogenesis and tumor cell invasiveness (Vandenbunderet al., 1994).ets-1alsoregulates expression of GlcNAc transferase V (Kanget al., 1996) and a consensusets-1binding site is needed for full FucTIV promoter activity(Withers and Hakomori, 1997). Process B: an unknown factor produced during multi-step tumor progression down-regulates A/Bglycosyltransferase gene transcription, resulting in A/B glycosylation of (i) integrin receptor as described here; (ii) other receptors includinggrowth factor receptors, leading to enhancement of cell motility and cell proliferation.

287A/B ANTIGEN DELETION DEFINING MALIGNANCY

population did not show detectable H-antigen expression, but A1

SW480 population showed a level of H-antigen expression similarto that of A2 SW480 (data not shown). The reason for thisphenomenon is not clear, but it may be related to expression ofA-associated H, which is assumed to be A2 expression. A1 H1

SW480 may have branched structure, with A at one chain and H atanother chain.

Haptotactic motility of A2 and A1 populationsMatrigel-dependent haptotactic motility was much lower in A1

as compared to A2 populations in isolates from both SW480 (Fig.2a) and HT29 cells (Fig. 2b). The motility of A1 SW480 was,30% that of A2 SW480 (Fig. 2a); motility of A 1 HT29 was,15% that of A2 HT29 (Fig. 2b). The difference in motilitybetween A1 and A2 populations was highly significant (p , 0.005,t-test). This reduction of motility was much greater than thatobserved forA gene transfectants in our previous study (seeDiscussion).

Proliferation of A2 and A1 populationsCell proliferation in terms of 3H-thymidine incorporation

for A2 and A1 SW480 and HT29 populations is shown inFigures 3a and 3b, respectively. A significant reduction of prolifera-tion in A1 relative to A2 populations was observed (p , 0.01 forSW480;p , 0.005 for HT29). However, the difference between A1

and A2 populations was not as great as that observed forhaptotactic motility. No such reduction of proliferation was ob-served forA gene transfectants in our previous study (see Discus-sion).

Western blotting assay of integrin receptorsfrom A2 and A1 populations

Integrin receptorsa3, a6, and b1, which control haptotacticmotility of colonic carcinoma SW480 and HT29 cells, wereexpressed to the same extent in A2 and A1 populations. Cytofluoro-metric analysis profiles ofa6 andb1 expression of these cells areshown in Figure 4. The receptors from A1 populations have Adeterminant, as indicated by strong blotting by anti-A MAbs inWestern blot analysis of fractions immunoprecipitated by antibod-ies to the respective integrin receptors (Fig. 5, lanes 2, 4, 6),although staining was weak in A1 HT29 populations compared toA1 SW480 populations. Receptors from A2 populations, immuno-precipitated under the same conditions, showed very weak or noA-positive band (Fig. 5, lanes 1, 3, 5).

DISCUSSION

There is increasing evidence that reduction or deletion of A/Bdeterminant in various types of human cancer is correlated withdegree of malignancy (i.e., invasiveness and metastatic potential)and with post-operative survival (Leeet al., 1991; Matsumotoetal., 1993). However, there have been few studies on the underlyingmolecular mechanism. The mRNA level forA andB transcriptionin urothelial bladder carcinoma cells is down-regulated in closecorrelation with cell proliferation (Ørntoftet al., 1996). Thisfinding suggests thatA/B gene transcription is inhibited by ayet-unknown mechanism, possibly through change of transcriptionfactor, which also affects cell proliferation.A/B gene transfection toH-positive colonic carcinoma cells resulted in isolation of variouspopulations expressing A or B antigen, with simultaneous deletionof H antigen. All theseA/ B gene transfectants showed significantreduction of Matrigel-dependent haptotactic motility, to levelssimilar to those observed when cells were incubated withMAbs directed toa3, a6, and b1 receptors, and were char-acterized by the presence of A-bearing or B-bearinga3, a6,

andb1 receptors (Ichikawaet al., 1997). However, there was nodifference in proliferation between parental cells andA/B transfec-tants.

Various phenotypes expressed inA/B gene transfectants ascompared to H1 parental cells may not be identical to those seen invariants from originally A1 and B1 tumor cells, having spontane-ous deletion of A or B determinant.A/B gene transfectantstend to overexpress A or B antigen, and the transfection pro-cedure per se is ‘‘unphysiological.’’ Separation of A2 or B2

variants out of A1 or B1 tumor cell lines and comparison of theirphenotypes are desirable in order to (i) confirm results of transfec-tion experiments; (ii) study phenotypic differences related tonaturally-occurringA/ B gene deletion in tumors of A or B patients;(iii) study the factors or mechanism involved in deletion/reductionof A/B gene transcription associated with tumor development. Wetherefore isolated A2 variants from A1 colonic tumor cell lines andcompared their phenotypes. The presence of A2 or B2 populationsfrom primary lung, urothelial, or cervical tumors from bloodgroup A or B individuals is indicated by various studies ascited in the Introduction. This is presumably due to spontaneousreduction or deletion of A or B determinant through suppression ofA or B glycosyltransferase gene expression, by an unknownmechanism.

Our present results indicate a marked enhancement of hap-totactic motility in A2 populations as compared to persistent A1

populations isolated from originally A1 tumor cell lines. This wasobserved for A2 populations isolated from 2 independent A1 celllines, SW408 and HT29. The motility difference between spontane-ously-occurring A1 and A2 populations is much greater than thatbetweenA gene transfectants and H1 parental tumor cells. Interest-ingly, we observed a significant difference in cell proliferationbetween spontaneously-occurring A1 and A2 variants isolatedfrom both SW408 and HT29 cell lines. In contrast, such cellproliferation difference betweenA gene transfectants and H1

parental tumor cells was minimal.The haptotactic motility difference between A1 and A2 variants

is ascribable to A-glycosylation ofa3, a6, and b1 integrinreceptors, since A-glycosylation was present in receptors from A1

variants and absent in receptors from A2 variants, although thequantity of integrin receptors expressed in the variants wasidentical. The significantly greater cell proliferation observed in A2

relative to A1 cells is surprising, and the underlying mechanism forthis is unknown. One possibility is that a common factor (e.g.,transcription factor) that controls both cell proliferation andA/Bgene expression is induced as one of the multiple steps of tumorprogression (Fig. 6a). Another possibility is that A-glycosylationenhances function of certain growth factor receptors (e.g., EGF,TGF-b) in the same way that it modifiesa3, a6, andb1 integrinreceptor function (Fig. 6b). The type of N-linked glycosylationgreatly affects EGF receptor function through changes of receptor-receptor interactions. EGF receptor function is strongly inhibitedby introduction of bisecting GlcNAc to N-linked multiantennarystructure by transfection of GlcNAc transferase III (Rebbaaet al.,1997).

There must be numerous other phenotypic changes asso-ciated with A/B deletion in A tumor cells, since A/B glycosylationin these cells occurs in various enzymes and receptors that controlproliferation, adhesion, motility, and other cell social functions.Dissection of various functional mechanisms affected by A or Bglycosylation is important for understanding tumor cell malig-nancy.

ACKNOWLEDGEMENT

This work was supported by National Cancer Institute Outstand-ing Investigator Grant CA42505 (to SH). We thank Dr. S. Andersonfor scientific editing and preparation of the manuscript.

288 ICHIKAWA ET AL.

REFERENCES

ALBINI , A., IWAMOTO, Y., KLEINMAN , H.K., MARTIN, G.R., AARONSON, S.A.,KOZLOWSKI, J.M. and MCEWAN, R.N., A rapidin vitro assay for quantitatingthe invasive potential of tumor cells.Cancer Res., 47,3239–3245 (1987).DABELSTEEN, E., ROED-PETERSEN, B. and PINDBORG, J.J., Loss of epithelialblood group antigens A and B in oral premalignant lesions.Acta pathol.microbiol. Immunol. Scand., 83,292–300 (1975).DAVIDSOHN, I., KOVARIK, S. and NI, Y., Isoantigens A, B, and H in benignand malignant lesions of the cervix.Arch. Pathol., 87,306–314 (1969).DAVIDSOHN, I. and NI, L.Y., Loss of isoantigens A, B, and H in carcinoma ofthe lung.Amer. J. Pathol., 57,307–334 (1969).HANDA, K., WHITE, T., ITO, K., FANG, H., WANG, S. and HAKOMORI, S.,P-selectin-dependent adhesion of human cancer cells: requirement forco-expression of a ‘‘PSGL-1-like’’ core protein and the glycosylationprocess for sialosyl-Lex or sialosyl-Lea. Int. J. Oncol., 6, 773–781 (1995).ICHIKAWA , D., HANDA, K., WITHERS, D.A. and HAKOMORI, S., Histo-bloodgroup A/B versusH status of human carcinoma cells as correlated withhaptotactic cell motility: approach withA andB gene transfection.CancerRes., 57,3092–3096 (1997).ITO, K., HANDA, K. and HAKOMORI, S., Species-specific expression ofsialosyl-Lex on polymorphonuclear leukocytes (PMN), in relation toselectin-dependent PMN responses.Glycoconj. J., 11,232–237 (1994).KANG, R., SAITO, H., IHARA, Y., MIYOSHI, E., KOYAMA , N., SHENG, Y. andTANIGUCHI, N., Transcriptional regulation of theN-acetylglucosaminyltrans-ferase V gene in human bile duct carcinoma cells (HuCC-T1) is mediatedby Ets-1.J. biol. Chem., 271,26706–26712 (1996).LEE, J.S., RO, J.Y., SAHIN, A.A., HONG, W.K., BROWN, B.W., MOUNTAIN,C.F. and HITTELMAN , W.N., Expression of blood-group antigen A: afavorable prognostic factor in non-small-cell lung cancer.N. Engl. J. Med.,324,1084–1090 (1991).MASAMUNE, H., KAWASAKI , H., ABE, S., OYAMA , K. and YAMAGUCHI, Y.,Molisch-positive mucopolysaccharides of gastric cancers as compared withthe corresponding components of gastric mucosae. First report: fraction-ation procedure of gastric cancer and gastric mucosa.Tohoku J. exp. Med.,68,81–91 (1958).

MASAMUNE, H., YOSIZAWA, Z., OH-UTI, K., MATSUDA, J. and MASUKAWA,A., Biochemical studies on carbohydrates: on sugar components ofhexosamine-containing carbohydrates from gastric cancer, normal humangastric mucosa, and human liver, and of glacial acetic acid-soluble proteinsfrom those tissues, as well as liver metastasis from gastric cancer.Tohoku J.exp. Med., 56,37–42 (1952).

MATSUMOTO, H., MURAMATSU, H., SHIMOTAKAHARA , T., YANAGI, M.,NISHIJIMA, H., MITANI , N., BABA, K., MURAMATSU, T. and SHIMAZU , H.,Correlation of expression of ABH blood group carbohydrate antigens withmetastatic potential in human lung carcinomas.Cancer, 72,75–81 (1993).

ØRNTOFT, T.F., Carbohydrate changes in bladder carcinomas.Acta pathol.microbiol. Immunol. Scand., 100 (Suppl. 27),181–187 (1992).

ØRNTOFT, T.F., MELDGAARD, P., PEDERSEN, B. and WOLF, H., The bloodgroup ABO gene transcript is down-regulated in human bladder tumors andgrowth-stimulated urothelial cell lines.Cancer Res., 56,1031–1036.

REBBAA, A., YAMAMOTO, H., SAITO, T., MEUILLET, E., KIM, P., KERSEY, D.S.,BREMER, E.G., TANIGUCHI, N. and MOSKAL, J.R., Gene transfection-mediated overexpression ofb1,4 GlcNAc bisecting oligosaccharide struc-ture in a glioma cell line, U373MG, inhibits EGF receptor function.J. biol.Chem., 272,9275–9280 (1997).

SADAHIRA , Y., RUAN, F., HAKOMORI, S. and IGARASHI, Y., Sphingosine1-phosphate, a specific endogenous signaling molecule controlling cellmotility and tumor cell invasiveness.Proc. nat. Acad. Sci. (Wash.), 89,9686–9690 (1992).

VANDENBUNDER, B., WERNERT, N., QUEVA, C., DESBIENS, X. and STEHELIN,D., Does the transcription factor c-ets1 take part in the regulation ofangiogenesis and tumor invasion?Folia Biol. (Prague), 40, 301–313(1994).

WITHERS, D.A. and HAKOMORI, S., Characterization of the human FucTIVgene promoter [Abstract P120].Glycoconj. J., 14,764 (1997).

ZHENG, M., FANG, H. and HAKOMORI, S., Functional role ofN-glycosylationin a5b1 integrin recrptor: De-N-glycosylation induces dissociation oraltered association ofa5 and b1 subunits and concomitant loss offibronectin binding activity.J. biol. Chem., 269,12325–12331 (1994).

289A/B ANTIGEN DELETION DEFINING MALIGNANCY