Mouse monoclonal antibodies with specificity for the melanoma-associated ganglioside disialyllactosylceramide (GD3) also react with the structural analogue disialylparagloboside

Btochimtu Elsevier

349

er Bioph,vsico Actu 837 (1985) 349-353

BBA Report

BBA 50119

Mouse monoclonal antibodies with specificity for the melanoma-associated ganglioside disialyllactosylceramide

(G D3) also react with the structural

analogue disialylparagloboside Thomas

Brodin”, Ingegerd Hellstriimb, Karl Erik Hellstriimh, Karl-Anders Karlsson’**, Hans-Olov Sjiigren”, Nicklas Striimberg and Jan Thurin’

“The Wallenherg Laboratory, Unruersrty ofLund. S-220 07 Lund, (Sweden), hOncogen. NO_5 Ftrst Avenue, Scuttle. WA 98121 und Deportments o/Mm-obiolog,: and Immunolo~ and Puthologv, Unioersrty OJ Washmgton, Seattle. WA 98195. (U.S.A.), and’ Department of Medical Biochemistry, Untuersrty oJGotehorg, P.O. Box 33031, S-400 33 Giiteborg (Sweden) (Received

Key words:

Monoclonal

antibody;

Glycolipid;

August

Ganglioside

9th. 1985)

G L)l; Ganglioside

disialylparagloboside:

(Melanoma

cell)

A mouse monoclonal IgM antibody, 4.2, has previously been shown to bind preferentially to the surface of human malignant melanoma cells and to have specificity for the G o3 ganglioside (NettAco2 + 8NeuAca2 + 3Galpl -+ 4GlcCer). Using overlay of antibodies on thin-layer chromatograms with glycolipids of various sources, it was shown that antibody 4.2, a further IgM and two IgG, mouse monoclonal antibodies, selected on the basis of reactivity with GD3, also bound with similar strength to the structural analogue NeuAccr2 + The SK-MEL 28 8NeuAca2 + 3Gal/ll + 4GlcNaq31 + 3Gal/31 + 4GlcCer or disialylparagloboside. melanoma cell line used for immunization was shown to contain a large amount of Go, but to lack disialylparagloboside. The demonstrated cross-reactivity may be of importance when considering the use of these antibodies for biochemical and medical purposes.

Several groups have produced mouse monoclonal antibodies to cell-surface antigens primarily expressed in human melanomas (for covering references, see Refs. 1 and 2). One of the better characterized of these tumor-associated antigens is (see Ref. 2 and Table I), the ganglioside G,, which is also present on several normal cells but at a lower density or in a cryptic state [3], which may explain the relative specificity for melanoma cells. In the early studies antibodies RI4 [4] and 4.2 [5,6] were shown to be specific for G,, using several melanoma and brain gangliosides as test substances. During our structural studies of glycolipids or rat colon carcinoma [7] we have used Authors appear in alphabetical order. * To whom correspondence should be addressed

0005-2760/85/$03.30

0 1985 Elsevier Science Publishers

monoclonal antibodies as assay reagents and found that several anti-G,,, antibodies also react with disialylparagloboside [8]. which may be considered a structural analogue of G o3 (Table I). The preparation and cell specificity and reactivity data for the IgM antibody 4.2 [5,6] and IgG, antibodies 2B2 and lF4 [2] have been described. In addition the similarly prepared IgM antibody lC9 was used for comparison. A solid-phase radioimmunoassay was used to evaluate antibody binding to purified glycolipids [9,10], and also the overlay technique for detection of glycolipid antigens on a thin-layer plate, done exactly as described elsewhere [ll]. Gangliosides were prepared from human brain obtained at autopsy, from rat colon carcinoma (DMH-W49) grown as syngeneic tumor grafts in

B.V. (Biomedical

Division)

350

TABLE

1

GANGLIOSIDES

USED FOR BINDING

G ,)3 and disialylparagloboside Abbreviation

were isolated

STUDIES from human

erythrocytes;

G,,,;,

and Golh were isolated

from human

brain

Sequence

PI G 07 Disialylparagloboside G T1.l GVlh

NeuAca2 NeuAcu2 NeuAca2 NeuAca2

+ + + +

8NeuAcor2 8NeuAccu2 8NeuAca2 8NeuAca2

4 3GalPl+ + 3GalPl+ --* 3GalPl+ + 3GalPl +

4GlcCer 4GlcNAc/31+ 3GalNAcPl* 3GalNAcPl+

the strictly inbred strain WF [7], and from erythrocyte membranes obtained from outdated human transfusion blood by toluene flotation [12]. Also, subfractions of gangliosides were prepared from SK-MEL 28 cells, whose cells have been used for the immunizations to obtain the antibodies [2].

9000

9000

I

_GD3

M

282 (A) DPG

P

-0

M

IF4 (B) DPG

/

1

-GD3

D----o

4.2 DPG

(Cl 9000

!

-GE

M

4GlcCer

The isolation of the gangliosides was made as described [13,14] starting with a mild alkaline degradation of the lipid extract [15] and using a combination of ion-exchange and absorption chromatography. The identity of the isolated gangliosides-(Table I) with earlier known structures

_GD3 9000

3GalPl+ 4GlcCer 4(NeuAcot2 ---t 3)Gal/31 --t 4GlcCer 4(NeuAca2 + RNeuAca2 + 3)Galbl+

IC9

/

(D)

DPG

Fig. 1. Binding of antibodies to glycolipids coated in microtiter wells. 50 PI/well of antigen dilutions in methanol were pipetted to microtiter plates (Cookes M24, Nutacon, The Netherlands) and allowed to evaporate slowly. Following incubation with 100 ~l/well of solution A (2% bovine serum albumin in phosphate-buffered saline containing 0.1% NaN,. pH 6.4). the wells were rinsed twice with 100 ~1 of solution A. This was followed by incubation for 4 h with antibody in the form of ascites fluid diluted 1 : 500. The plates were washed five times with solution A, and this was followed by incubation overnight with 100 pi/well of rabbit anti-mouse Fab (‘251-labelled F(ab’),, Amersham International, Amersham, U.K.). The wells were subsequently washed five times with 100 ~1 each of solution A, dried at room temperature, cut out and counted in a gamma counter. The antigens were gangliosides Go, and disialylparagloboside (DPG) of Table I (see curve explanations in the Figure), and the antibodies were 2B2 (A), 1 F4 (B). 4.2 (C) and lC9 (D).

351

[6,16,17] was confirmed by mass spectrometry [16,1X] and NMR spectroscopy [19]. Thin-layer chromatograms were developed on alumina sheets coated with silica gel 60 (Merck, Darmstadt, F.R.G.), and chloroform/methanol/2.5 M NH, (60 : 40: 9, by vol.) was used as solvent and anisaldehyde for detection [15].

2B2

Antibody 4.2, which was earlier shown to carry specificity for G,? [5,6], bound equally well to G,,, and disialylparagloboside as shown by solidphase radioimmunoassay (Fig. 1C). A similar situation existed for antibody lC9 (Fig. 1D) while, antibody 2B2 (Fig. 1A) and antibody 7F4 (Fig. 1 B) appeared to bind disialylparagloboside even 1F4

(B)

lC9

(D)

(A)

Y

123456 4.2

123456

123456 (C)

123456

Fig. 2. Autoradiograms from binding of antibodies to gangliosides on thin-layer chromatograms. The developed plate was treated with plastic and bovine serum albumin and overlayered with antibody followed by radioactive anti-antibody as described [ll] and autoradiographed for 32 h. The following ganglioside fractions were used: total gangliosides from DMH-W49 rat colon adenocarcinema (lane l), di- plus trisialogangliosides from DMH-W49 (lane 2). and mono-, di-, tri- and tetrasialogangliosides of human brain (lanes 3, 4, 5 and 6. respectively). The following antibodies were used: 282 (plate A), lF4 (plate B), 4.2 (plate C), and 1C9 (plate D). Arrows indicate the level of Go, (upper) and disialylparagloboside (lower).

352

better than G,,. In the chromatogram assay (Fig. 2) the 4.2. antibody gave a high background staining (C). There was a binding to disialogangliosides of rat carcinoma (81 in the level of G,,, and disialylparagloboside (lanes 1 and 2) shown to exist in a ratio of 50: 1 (Brodin, T., Karlsson, K.-A., Sjiigren, H.-O., Stromberg, N. and Thurin, J, unpublished results). With brain subfractions binding to G,, (1ane 4) Grla (main band of lane Purified 5) and Go,, ( lane 6) was demonstrated. GTla and Go,,, (Table I) analysed with solid-phase radioimmunoassay (data no shown) showed reactivities with 4.2 antibody which were about lOO-times weaker than for G,,, in agreement with previous findings [6]. Antibodies 2B2 and lF4 were similar to 4.2 in the chromatogram assay (Fig. 2A,B), binding to Go, (lanes 1, 2 and 4) Gr,;, (lane 5) and Golb ( lane 6) but with much less In solid-phase radioimbackground staining. munoassay (not shown) antibody 2B2 gave the same results as 4.2, binding about 100-times better to G,, than to GTla. Antibody lC9 was shown to bind to additional bands in the chromatogram assay compared to 4.2, 2B2 and lF4 (Fig. 2D). According to radioimmunoassay (data not shown) antibody lC9 bound a biequally well to Go, and GTla, indicating nding epitope restricted to the terminal three sugars (Table I). Compared to antibody 4.2 there was less background staining and no binding to rat carcinoma gangliosides in a region between Go, and disialylparagloboside (lanes 1 and 2). The avidity of the two IgG antibodies 2B2 and lF4 is probably somewhat lower than for the IgM antibodies 4.2 and lC9 as indicated by both Figs. 1 and 2. Disialogangliosides were isolated from SK-MEL 28 melanoma cells and tested for binding in the chromatogram assay (Fig. 3). Although a strong binding was found for the major Go, ganglioside there was not detectable binding in the interval for disialylparagloboside. antibodies analysed here (4.2, All four anti-G,, 2B2, lF4, lC9) therefore also bind to disialylparagloboside, a molecule present normally in human erythrocytes [16] and human kidney [20], and in rat colon carcinoma (Brodin, T., Karlsson, K.-A., Sjbgren, H.-O., Stromberg, N. and Thurin, J. unpublished results). The crossreactivity be-

Fig. 3. Gangliosides of SK-MEL 28 cells. Total lipids of SK-MEL 28 cells were eluted from a DEAE-Sepharose column and a fraction containing the disialogangliosides was separated by thin-layer chromatography and detected by anisaldehyde (lane A) or autoradiography (lane B) for 72 h after overlayering with antibody 2B2 and radioactive anti-antibody. Arrows indicate the level of Go, (right) and disialylparagloboside (not found). Fractions eluted before and after this subfraction were also shown by overlayering with antibody to lack disialylparagloboside.

tween the two gangliosides may be explained by the similarity of their terminal 4-sugar sequence (Table I). This binding is interesting in view of the almost total absence of the larger ganglioside from the melanoma cells used for immunization. Therefore, disialylparagloboside may not be the immunogen. It is therefore likely that the abundant G D3 was the immunogen producing antibodies that by coincidence react also with disialylparagloboside possessing an identical partial structure. An alternative possibility is that peptide-linked NeuAccu2 + 8NeuAccu2 + 3Gal/31 + 4GlcNAc is the antigen on the melanoma cells, because this sequence has been proposed to exist in fetal but not in adult human erythrocyte lactosaminoglycan

ml. Interestingly, antibody 2B2 but not lF4, both being IgG, and with similar specificity and affinity as shown here, was recently shown to inhibit the outgrowth of a human melanoma in nude mice [2]. As the activity in the antibody-dependent cellular cytotoxicity test was greater in the presence of human than mouse lymphocytes it was proposed that this antibody may be of value in the treatment of human cases of melanoma [2]. The crossreactivity of the well-studied 4.2 and antibodies with disialylparagthe other anti-Go,, loboside may exist also for antibody R,, [4]. It may be of interest when applying these antibodies as reagents for biochemical and medical purpose because the two antigens detected have different biosynthetic pathways and may therefore reflect separate regulatory mechanisms.

353

The work was supported by grants from the Swedish Medical Research Council (Nos. 3976 and 6810). References Cheresh, D.A., Varki, A.P., Varki, N.M., Stallcup, W.B., Levine, J. and Reisfeld. A.A. (1984) J. Biol. Chem. 259, 7453-7459 Hellstrom, I., Brankovan, V. and Hellstrom, K.E. (1985) Proc. Natl. Acad. Sci. USA 82, 149991502 Hakomori, S.-i. and Kannagi, R. (1983) J. Nat. Cancer Inst. 71, 231-251 Dippold. W.G.. Lloyd, K.O., Li, L.T.C., Ikeda. H., Oettgen. H.F. and Old, L.J. (1980) Proc. Natl. Acad. Sci. USA 77, 6114-6118 Yeh, M.-Y., Hellstrom, K.E. and Hellstrom, I. (1982) Int. J. Cancer 29, 269-275 Nudelman, E., Hakomori, S.-i., Kannagi, R., Levery, S., Yeh, M.-Y.. Hellstrom, K.E. and Hellstrom, 1. (1982) J. Biol. Chem. 257, 12752-12756 Sjligren, H.-O. (1980) Cancer 45, 122991233 Hansson, G.C.. Karlsson, K.-A., Larson, G., Stromberg, Y., Teneberg, S., Thurin, J., Brodin, T., Sjogren, H.-C)., Hellstrom, 1. and Hellstrom, K.E. (1983) in Proceedings of the 7th International Symposium on Glycoconjugates (Chester, M.A., HeinegLd, D., Lundblad, A. and Svensson S., eds.), pp. 854-855, Rahms i Lund, Lund, Sweden Young, W.W., Jr., MacDonald, D.M.S., Nowinsky, R.C.

and Hakomori, S.-i. (1979) J. Exp. Med. 150, 1008-1019 10 Brockhaus, M., Magnani, J.L., Blaszczyk, M., Steplewski 2.. Koprowski, H.. Karlsson, K.-A., Larson, G. and Ginsburg, V. (1981) J. Biol. Chem. 256, 13223-13225 11 Hansson, G.C.. Karlsson, K.-A., Larsson, G., McKibbin J.M., Blaszczyk, M., Herly, M., Steplewski, Z. and Koprowski, H. (1983) J. Biol. Chem. 258, 4091-4197 12 Falk, K.-E., Karlsson, K.-A. and Samuelsson, B.E. (1981) FEBS Lett. 124. 173-177 13 Iwamori, M. and Nagai, Y. (1978) Biochim. Biophys. Acta 528. 257-267 14 Nagai, Y. and Iwamori, M. (1980) Mol. Cell. Biochem. 29, 81-90 15 Karlsson, K.-A., Samuelsson, B.E. and Steen, G.O. (1973) Biochim. Biophys. Acta 316, 317-335 16 Kundu. S.K.. Samuelsson, B.E., Pascher, I. and Marcus, D.M. (1983) J. Biol. Chem. 258, 13857-13866 17 Ledeen, R.W. and Yu. R.K. (1982) Methods Enzymol. 83, 139-191 18 Fredman, P., Mansson, J.-E., Svennerholm, L., Karlsson, K.-A., Pascher, I. and Samuelsson, B.E. (1980) FEBS Lett. 110, 80-84 19 Falk, K.-E., Karlsson, K.-A. and Samuelsson, B.E. (1979) Arch. Biochem. Biophys. 192, 164-202 20 Rauvala. H.. Krusius, T. and Finne, J. (1978) Biochim. Biophys. Acta 531. 266-274 21 Fukuda, M., Dell, A. and Fukuda, M.N. (1984) J. Biol. Chem. 259, 4782-4791 22 IUB Document of Lipid Nomenclature (1977) Eur. J. Biothem. 79, 11-21