Proc. Nati. Acad. Sci. USAVol. 84, pp. 2416-2420, April 1987Immunology

Cell-surface antigens of melanoma recognized by humanmonoclonal antibodies

(gangliosides/Epstein-Barr virus transformation/cancer immunology)

HIROSHI YAMAGUCHI, KOICHI FURUKAWA, SHEILA R. FORTUNATO, PHILIP 0. LIVINGSTON,KENNETH 0. LLOYD, HERBERT F. OETTGEN, AND LLOYD J. OLDMemorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021

Contributed by Lloyd J. Old, October 10, 1986

ABSTRACT Human monoclonal antibodies (mAbs) werederived from lymph node lymphocytes and peripheral bloodlymphocytes (PBL) from patients with melanoma. Four meth-ods for generating human mAbs were compared: fusion withhuman [LICR-LON-HMy-2 (LICR-2)] or mouse (NS-1) cells;transformation by Epstein-Barr virus (EBV); and EBV trans-formation followed by NS-1 fusion. NS-1 fusion with lymphnode lymphocytes resulted in a higher number of growinghybrids than LICR-2 fusion. Virtually no hybrids were ob-tained from NS-1 or LICR-2 fusions with PBL. EBV trans-formed lymphocytes from lymph node and peripheral bloodwith equal efficiency, and the yield of proliferating cultures forantibody screening was more than 10- to 30-fold greater thanthat obtained by fusion techniques. However, once antibody-producing cultures had been identified, stability and clonabilityofEBV-transformed cells were poorer than that ofNS-1 hybridcells. To combine the strengths of both methods, cultures ofEBV-transformed cells were fused with NS-1, and hybridclones were isolated that showed vigorous growth, clonability,and stable antibody secretion. Detailed specificity analysis ofthe mAbs produced by six of these clones indicated detection ofa class 1 (unique) melanoma antigen, a class 3 melanomaantigen, and four ganglioside antigens (GD3, GM3, and twoother, as yet uncharacterized, heterophile antigens).

Several groups have reported specific humoral and cellularimmune reactions to melanoma cell-surface antigens (1-9). Inour laboratory, the reactivity of serum with surface antigensof cultured melanoma cells from the same patient (termedautologous typing) has been analyzed in >200 patients. Threeclasses of antigens have been defined in this way (2). Class 1(unique) melanoma antigens are restricted to the autologousmelanoma; six examples of class 1 antigens have beendetected (10-15). Class 2 melanoma antigens are detected onthe autologous melanoma, on a subset of allogeneic melano-ma cells, and on other neuroectodermally derived tumors,and one ofthe best-analyzed class 2 melanoma antigens is theganglioside GD2 (16). Class 3 melanoma antigens are notrestricted to any differentiation lineage and are more widelydistributed.The advent of methods for producing human monoclonal

antibodies (mAbs) through immortalizing human lympho-cytes with Epstein-Barr virus (EBV) (17) or by fusion withhuman or mouse lymphoblastoid/myeloma partners (18)provides a new level of precision in the analysis of theimmune response to melanoma. Houghton et al. (19), Irie andco-workers (20-22), and other groups (23, 24) have isolatedantibodies from melanoma patients reacting with melanomacell surfaces or intracellular antigens. In the present study,we compared the efficiency of EBV immortalization and

hybridization with mouse or human partners in the capture ofhuman lymphocytes producing melanoma-reactive antibody.By combining the two techniques-i.e., EBV transformationfollowed by fusion with mouse NS-1 myeloma-we havebeen successful in generating a panel of human mAbsreactive with cell-surface antigens of melanoma.

MATERIALS AND METHODSTissue Culture. Tumor cell lines were established and

maintained as described (10). Several melanoma cell lines(see Table 1) were adapted to growth in serum-free insulin (5,ug/ml)/transferrin (5 ,ug/ml)/selenium (5 Ag/ml) (ITS) medi-um.

Fusion Procedure. Lymph node lymphocytes (LNL) andperipheral blood lymphocytes (PBL) were isolated by Ficoll/Hypaque centrifugation and fused with human LICR-LON-HMy-2 (LICR-2) lymphoblastoid cells or mouse NS-1 mye-loma cells as described (19, 25). NS-1 fusions with EBV-transformed cells (after two limiting dilution passages of1-104 cells per passage) or LICR-2 hybrids were performedin the same way as direct lymphocyte fusions, but hybridswere selected by 0.2 mM hypoxanthine/0.4 ,uM aminopterin/32 ,4M thymidine (HAT) and 10 ,M ouabain (Sigma) for thefirst week and HAT for the second week; this was followedthereafter by culture in RPMI 1640 medium containing 15%fetal bovine serum, 0.2 mM hypoxanthine, and 32 utMthymidine.EBV Transformation. EBV-containing supernatants from

the B95-8 marmoset lymphoblastoid cell line (26) werecollected and passed through a 0.45-,um filter and stored at-800C. LNL or PBL were suspended in RPMI 1640 mediumcontaining 10o fetal bovine serum and adjusted to 2 x 106cells per ml. Five volumes oflymphocyte cell suspension wasmixed with 1 vol of B95-8 supernatant and placed in T-30flasks (Falcon). After incubation overnight in a 5% CO2 (inair) incubator, cells were washed once, resuspended at 1-4 X105 cells per ml, and plated (2-8 x 104 cells per well) in 96-welltissue culture plates. Wells containing antibody-producingcells were expanded in 24-well tissue culture plates andsubsequently subcultured by limiting dilution in 96-wellplates.

Serological Assays. Techniques for the detection of cell-surface antigens (11, 14) and intracellular antigens (19, 25)have been described. Antigen characterization by heat treat-ment and neuraminidase treatment was performed as de-scribed (27, 28). Antibody inhibition tests were carried out bymixing the cell extract or purified gangliosides with culturesupernatant (diluted three doubling dilutions below the end

Abbreviations: mAb, monoclonal antibody; LICR-2, LICR-LON-HMy-2; EBV, Epstein-Barr virus; ITS, insulin/transferrin/seleni-um; LNL, lymph node lymphocytes; PBL, peripheral blood lym-phocytes; HAT, hypoxanthine/aminopterin/thymidine; IA, immuneadherence; PA, protein A; IF, immunofluorescence.

2416

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.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

2, 2

021

Proc. Natl. Acad. Sci. USA 84 (1987) 2417

point), incubating for 30 min at room temperature, and testingfor residual antibody reactivity on SK-MEL-23 target cells.

Glycolipids. Cells were extracted by chloroform/methanol,and neutral and acidic glycolipids were isolated as described(28). GM3 and GD3 were purified in this laboratory. GM2 was

prepared as described (29). GM1, GD1a, and GT1 were

purchased from Supelco (Bellefonte, PA). GD2 was a gen-erous gift of H. Wiegandt (University Marburg, F.R.G.).GD1b was kindly given by R. K. Yu (Yale University, NewHaven, CT).ELISA for glycolipids and immunostaining afterTLC were

performed as described (28-30).

RESULTS

Generation of Human mAbs. Four approaches to generat-ing human mAbs from the lymphocytes of patients withmelanoma were compared (Table 1). Fusion of LNL withmouse NS-1 myeloma cells resulted in a higher yield ofclonesthan fusion with LICR-2. NS-1 hybrids grew more vigorouslythan LICR-2 hybrids, and this facilitated clonal selection andexpansion of NS-1-derived clones. No LICR-2 hybrids wereobtained after fusion with PBL, and the frequency of NS-1hybrids was extremely low with lymphocytes from thissource. In contrast to fusion techniques, EBV transformationresulted in an equally high frequency of proliferating cellsfrom LNL and PBL. Testing individual wells for immuno-globulin secretion after initial plating of fused or EBV-transformed lymphocytes indicated that NS-1 or LICR-2hybrids were clonally derived, producing a single heavy chainclass, whereas EBV transformation resulted in polyclonalexpansion of B cells with >95% wells containing cellssecreting IgM, IgG, and IgA. To identify cells producingantibodies with cell-surface reactivity, supernatants contain-ing .500 ng of immunoglobulin per ml were allowed to reactwith a screening panel of 20 different cell lines, including 10melanomas, 5 leukemias, and 5 epithelial cancers. In 22cases, autologous melanoma cells from the lymphocytedonor were also available for screening.As shown in Table 1, supernatants from 1.1% of wells

(10/895) from LICR-2 fusions with LNL contained surface-reactive antibodies; in the case of NS-1 fusions, the figurewas 1% (5/509). Three stable antibody-secreting clones werederived from LICR-2 fusions and three were derived fromNS-1 fusions. Primary screening of supernatant from EBV-transformed cells identified 2-3% of wells with surface-reactive antibody against allogeneic cells and 0.3-0.4%against autologous cells. Expansion and limiting dilution

plating of EBV-transformed cells resulted in a loss >80% ofantibody-secreting cultures. However, 17 stable cultures ofEBV-transformed cells secreting cell-surface-reactive anti-body were derived. To overcome problems of instability andlow cloning efficiency of EBV-transformed cultures, weattempted to develop stable NS-1 hybrids of antibody-secreting EBV-transformed cells. The fusion frequency wasgenerally low (5.0 per 107 EBV-transformed cells), buthybrids secreting antibody with the same reactivity as theparental EBV-transformed cultures were obtained in 15/35attempts. With vigorous subcloning, 8/15 clones retainedstable antibody-producing capacity, a frequency similar toour experience in deriving stable NS-1 hybrids in this andpast studies (25).

Specificity Analysis ofHuman mAbs. Six human mAbs werechosen for detailed specificity analysis. The pattern ofcell-surface reactivity in direct tests and absorption analysiswith a panel of cultured cells is shown in Table 2. Five of theantibodies were produced by NS-1 hybrids with EBV-transformed cells; the other antibody (2.39M) came from anNS-1-LICR-2 cloned hybrid.GXMI. GXM1 is an IgM antibody derived from PBL of

patient GX that reacts with the patient's own melanoma cellline, SK-MEL-177 (titer, 1/2048). With the exception ofautologous melanoma reactivity and low titer reactivity with1 of the 13 other melanoma cell lines, no other cell type,including autologous PBL or autologous EBV-transformed Bcells, reacted in direct tests or absorption analysis. Theantigen is heat labile but has not been further characterized.HJMJ. HJM1 is an IgM antibody derived from PBL of

patient HJ. In direct tests this antibody reacts with 13/14melanoma cell lines but not with 50 other cell types. Quali-tative absorption tests with cells collected by mechanicalscraping showed a positive absorption pattern with all cellsexcept erythrocytes. Indirect immunofluorescence (IF) testswith fixed cells to identify cytoplasmic antigens indicated thatall nucleated human cells were antigen-positive. To test thepossibility that antigen-absorbing cell-surface reactivity wasderived from intracellular antigen released by cell damageduring collection, we compared the absorbing capacity ofcells harvested by scraping (cell viability, <30%) and cellsharvested by trypsinization (cell viability, >95%) (Fig. 1). Incontrast to the results with cells harvested mechanically,absorption tests with trypsinized cells correlated with surfacereactivity-i.e., antibody was absorbed with cells havingsurface expression of the antigen and not absorbed with cellslacking surface expression. The antigen detected by HJM1 isheat-stable and neuraminidase-sensitive, suggesting that it is

Table 1. Generation of human mAbs from lymphocytes of melanoma patientsWells with growing cells, Wells screened Positive Stable

Method/ Trials, no. per 107 lymphocytes % IgC wells* for cell-surface wells, cultureslymphocyte source no. Median (Range) Median (Range) reactivity,t no. no. or clones, no.

LICR-2 fusionLNL 20 11.2 (1.1-200.0) 76 (21-94) Allo, 895 10 3PBL 10 0 (0)

NS-1 fusionLNL 12 25.2 (6.3-37.3) 19 (3-32) Allo, 509 5 3PBL 18 0 (0-27.6) 0 (0-8) Allo, 21 0

EBV transformation

LNL 13 270.0 (31.0-500.0) 100 (56-100) f Allo, 613 18 61Auto, 1,817 8 0

PBL 38 300.0 (33.3-480.0) 100 (85-100) I Allo, 1,899 38 5Auto, 15,288 46 6

NS-1 fusiont 35 5.0 (0-76.7) 1,505 258 8

*% of wells with growing cells having Ig in culture supernatants of .500 ng/ml.tAllo, reactivity with allogenic target cells. Auto, reactivity with autologous melanoma cells.tFusion with EBV-transformed B cells or LICR-2 clones producing antibody to cell-surface antigens.

Immunology: Yamaguchi et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

2, 2

021

2418 Immunology: Yamaguchi et al.

Table 2. Serological characterization of six human mAbs derived from lymphocytes of melanoma patients

Cells

MelanomaSK-MEL-13SK-MEL-23SK-MEL-28SK-MEL-28*SK-MEL-29SK-MEL-31SK-MEL-37SK-MEL-61SK-MEL-93-IISK-MEL-93-II*SK-MEL-94

SK-MEL-130SK-MEL-173SK-MEL-177

SK-MEL-177*SK-MEL-189MeWo

AstrocytomaSK-MG-1SK-MG-4SK-MG-14U-251-MGSK-MG-3, -11SK-MG-21, -23

NeuroblastomaSK-NMCSK-NSH, IMR-32

LeukemiaHL-60K-562CCRF-HSB-2CCRF-CEM, T-45NALL-1NALM-1ARA-10BALL-1DAUDI, SK-LY-16

EBV-transformed 13cellAH, DS, HJFCGX

GXM1 HJM1 FCM1 DSM1 32-27M

T A T A T A T A T A

- - 64 + 512 +- -1024 + 1024 +- - 64 + 128 +

64 25616 + 32 + 32 +- - + + 4+- - 4 + 64 +- - 128 + 256 +- - ± + 128 +

- - 256 + 4096 +

- - 32 + T21 +- - 64 + 256 +.048+ - + - +

10,000

40,00040,00040,0002,560

2,560640640

10,00010,0002,560

1+ 256 + 2,560- - + 4096 + 40,000

_ _ - + - +_ _ _ + - +

_ - - + 8+

6402,560640

6,40010,00

+ 256 +- 512 ++ 256 ++ -_

+ 128 ++ 1024 ++ 32 ++ 32 ++ 512 ++ -_

- 512 +

+ 16+ 1024 ++ 128

+ 64+ 512 +

+ 128 ++ 256 +++ 128 ++

- - + 32 + 40,000 + 256 +

- - + - + 2,560+ 4+_ _ + _ + - + + +

- - + - + - - + +

+ + _-

_ _ + _ + _ + -_ - + - + _-

- + - +

+ -+ - +-

+ - + -

+ - +_

Cells

Renal cancerSK-RC-6SK-RC-7SK-RC-9SK-RC-54SK-RC-45, -48

Bladder cancerT-24235-J5637, Scaber

Lung cancerSK-LC-8SK-LC-6, -12SK-LC-7

Breast cancerMCF-7MDA-MB-361MDA-MB-231CAMASK-BR-5

Colon cancerHT-29, SW-480SW-403, SW-837

Other cancersCAPAN-2ME-180SK-OV-3GCC-SV

PBL (n = 4)

FibroblastsWI-38Adult skin

Melanocytes (n = 3)

+ -_

+ - - Normal kidney- - - epithelium (n = 3)

ErythrocytesA, B, AB, 0

- - - Xenogenic cells+- - JB-RH

+ - - Sheep erythrocytes

GXM1 HJM1 FCM1 DSM1 32-27MT A T A T A T A T A

- +

_ _ - +

_ _ - +

32128

4096+

+ 2,560 ++ 40,000 ++ 40,000 +

10,000 ++ _ _

- - - + - + 10,000 +40,000 +

40,000 +- + - +

- +

_ _ - +

- + _-8+ - _8 - _

_ - +

- + - + _-

+- + _-_ + _-_ + _-_ + -

or+

- - - + 64 + 10,000 + 256 +2,560 +

- - 4096 - -or or2,560 +

- - 128 - or40,000

- + 4+ - -512 +

2.39M was tested by the same cell panel used here, and it only reacted with sheep erythrocytes (titer, 1:128). T, titer: dilution (reciprocal)showing 50% erythrocyte-rosetted target cells. -, No reaction in direct tests at a dilution of 1:2; ±, <50o reactivity at a dilution of 1:2; o, resultswith autologous combinations of antibody and target cells. IgM antibodies were detected by immune adherence assays. DSM1 was tested byPA assays. A, absorption test. Supernatant (diluted according to end point) was absorbed with the indicated cell type and tested for residualactivity for SK-MEL-177 (GXM1), SK-MEL-23 (HJM1, FCM1), SK-MEL-13 (DSM1), or MeWo (32-27M) target cells. +, Complete absorption;±, partial absorption; -, no absorption.*Cells cultured in serum-free medium containing ITS.

an acidic glycolipid. Antibody inhibition tests with purifiedgangliosides showed that HJM1 was strongly inhibited byGD3 and GD2, inhibited less by GD1b and GM3, and notinhibited by GM2, GM1, and GD1a (Fig. 2). Immunostainingof TLC plates confirmed the strong reactivity of HJM1 withGD3 (Fig. 3).FCMI. FCM1 is an IgM antibody derived from PBL of

patient FC. The reactivity of this antibody in direct tests isbroader than HJM1, extending to renal cancer cell lines andmelanocytes. Similar to- findings with HJM1, absorption andindirect IF tests with fixed zells indicated that all nucleatedcells were positive for the FCM1-detected antigen. However,in the case of FCM1, trypsinization did not abolish theabsorbing capacity of cells that were not reactive in direct

tests, indicating that low levels of surface antigenic expres-sion were detectable by the more sensitive absorption test butnot by direct tests. The antigen is heat-stable and neur-aminidase-sensitive, suggesting that it is an acidic glycolipid.ELISAs with purified gangliosides showed that FCM1 reactsmost strongly with GM3 and GD1a (Fig. 4). Immunostainingof TLC plates confirmed this strong reactivity of FCM1 forGM3 and GD1a (Fig. 3).DSMI. DSM1 is an IgG antibody derived from PBL of

patient DS that reacts with 42/85 cell lines in this panel.Patient DS had been vaccinated with repeated injections ofanallogeneic melanoma cell line, SK-MEL-13, followed byinjections of the autologous melanoma cell line (SK-MEL-94). The patient developed protein A (PA) reactivity for the

Proc. Natl. Acad. Sci. USA 84 (1987)

E

Dow

nloa

ded

by g

uest

on

Aug

ust 1

2, 2

021

Proc. NatL. Acad. Sci. USA 84 (1987) 2419

.f)M~1 0w

yC 100

50

0

A

0.0012 0.02 0.3 5 80

Cells for absorption, no. x 10-6

FIG. 1. Quantitative absorption analysis of human mAb HJM1.Comparative absorption capacity of cells harvested by mechanicalscraping (cell viability, <30%o) (A) or trypsinization (cell viability,>95%) (B). In direct serological tests, SK-MEL-13, -23, and -29showed titers of 1:32 to 1:1024, whereas T-24 and ME-180 were notreactive. IA, immune adherence. Dashed line, unabsorbed HJM1supernatant.

allogeneic melanoma cell line (titer, 1/40,000) but not for theautologous melanoma cell line. A comparison of the absorp-tion analysis ofDSM1 and high-titered serum from patient DSshowed identical results (data not shown), indicating that themAb has the same specificity as serum antibody. EBV-transformed B cells from the donor of the SK-MEL-13 targetcell (AH) did not absorb reactivity from DSM1 or DS serum,suggesting that alloantigenic systems such as HLA class 1 orclass 2 antigens were not involved. The antigen detected byDSM1 is heat-labile, hydrophobic, and binds to Con A (H.Y.and P. Srivastava, unpublished observations).32-27M. 32-27M is an IgM antibody derived from LNL of

a melanoma patient. It reacts with all melanoma, astrocy-toma, and neuroblastoma cell lines, WI-38 fibroblasts, andsheep erythrocytes but not with cancer lines of epithelialorigin. Despite this clear relation between neuroectodermal

AbHJMI

GM3-GM2,GD3-GMI-GDloSGD27GDb-GTl

AbFCM 1

(0100' - -'7s--~

o GM 7 \G

M

Cf) GMI

-j GD2WL GM-b.50 "GM3 ~ D

a)

0

0.

2x10-9 2x101-2 2x1-15 2 x0616 2x10-21

Ganglioside, mol

FIG. 2. Antibody inhibition tests of human mAb HJM1 withpurified gangliosides. Dashed line, uninhibited HJM1 supernatant.

origin and antigen expression, the reactivity of 32-27Mdisappeared when cells were cultured in serum-free ITSmedium in the absence of fetal bovine serum (Table 2). Whenfetal bovine serum was returned to the medium, partialrecovery of 32-27M reactivity occurred after 6 hr of incuba-tion and full recovery of reactivity was observed after 24 hrof incubation. The antigen is heat-stable and neuraminidase-sensitive, suggesting an acidic glycolipid. In immunostaining,32-27M reacted with a number of acidic glycolipids in sheeperythrocyte extracts (Fig. 3). It appears that neuroectoder-mally derived cells, but not most other cell types, incorporatethis antigen from fetal bovine serum.2-39M. 2-39M is an IgM antibody derived from LNL of a

melanoma patient. It does not react with any of the humancultured cell lines or human erythrocytes (Table 2). Itagglutinates sheep erythrocytes and reacted with an acidicglycolipid isolated from these cells in immunostaining (Fig.3).

DISCUSSIONEBV transformation and cell-fusion techniques representdistinct approaches to the isolation of human mAbs. Eachmethod, as currently applied, combines certain advantagesand certain limitations. For instance, easy clonability is anadvantage with methods depending on hybrid formation(particularly NS-1 hybrids) that is not achievable with EBV-transformed cells at this time. However, there is a higherlikelihood that antibody-secreting populations will be iden-tified after EBV infection, because of the polyclonal prolif-eration of EBV-infected cells, than after the more stringentand inherently more random process of hybrid formation. Byusing these two approaches sequentially (31, 32), the superiorsampling characteristics ofthe EBV method can be combined

Ab32-27M Ab2-39M

a

1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 10 1 2 3 4 5 6 7 8 11 1 2 3 4 5 6 7 8 11

FIG. 3. Reactivity of four human mAbs with gangliosides as determined by immunostaining after TLC. Lanes 1-8, purified N-acetyl-typegangliosides. Lanes 1, GM3; lanes 2, GM2; lanes 3, GD3; lanes 4, GM1; lanes 5, GD1a; lanes 6, GD2; lanes 7, GD1b; lanes 8, GT1; lane 9,ganglioside mixture from cat erythrocytes; lane 10, ganglioside mixture from horse erythrocytes; lanes 11, ganglioside mixture from sheeperythrocytes. Lanes 1-8 contained 1 mmol of ganglioside. The plates were developed with chloroform/methanol/3.5 M NH40H, 60:35:8,vol/vol. mAbs: hybridoma supernatants (diluted 1:5 to 1:10) were used, except in the case of HJM1, where a 1:5 dilution of EBV-transformedsupernatants was used.

Immunology: Yamaguchi et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

2, 2

021

2420 Immunology: Yamaguchi et al.

FIG. 4.

ELISA.

10 I 0.1

Dilution of ganglioside, pmol per well

Reactivity of human mAb FCM1 with glycolipids in the

with the superior clonability and growth characteristics of theNS-1 hybrid. However, the major problem in the use of thisstrategy is the loss of antibody-secreting cells during theinitial expansion of the EBV-transformed cultures to obtainsufficient number of cells for NS-1 fusion. Whether this is dueto instability and loss of antibody secretion or to outgrowthof unrelated cell populations is unknown. The ability to cloneEBV-transformed cells and maintain antibody secretion willbe a critical next step in this technology. In addition, thesearch needs to continue for a human fusion partner that is as

suitable as mouse NS-1 in generating human mAbs.Several of the mAbs derived in this study identify antigens

that exemplify the features of the three classes of antigensidentified by autologous and allogeneic typing of humanserum using conventional serological methods (2). The anti-gen recognized by GXM1 has the characteristic of a class 1(unique) melanoma antigen-i.e., highly restricted expres-

sion to autologous melanoma cells. The availability of a mAbto a human class 1 antigen should facilitate study of this classof antigen and determine whether they are a related family ofantigens and whether homologous molecules are expressedby normal cells. The antigen detected by HJM1 has thecharacteristics of a class 2 melanoma antigen-i.e., expres-

sion by the autologous melanoma, and a subset of allogeneicmelanomas, but not by cells of nonneuroectodermal origin.HJM1 reacts strongly with GD3 and reacts less with GD2,two gangliosides that are widely expressed by neuroecto-dermal cells. Because individuals can develop humoral an-

tibody against these normally occurring cell-surface compo-

nents, we have termed this class of tumor antigen autoim-munogenic differentiation antigens (16). Irie and co-workers(21) have also isolated a human mAb produced by EBV-transformed cells that reacts with GD2, but its reactivity isrestricted to GD2 and it does not show crossreactivity withother gangliosides, as does HJM1. The antigens detected byFCM1 and DSM1 are classic examples of class 3 antigens.They are expressed by cells derived from diverse differenti-ation lineages but are not found on all cells from these sources

or on all human cells.We do not know the source of the immunogenic signal that

elicited the antibodies identified in this study and are there-fore unable to interpret their significance in terms of specificimmune recognition of cancer. This will become clear as

comparable surveys are performed with lymphocytes fromnormal individuals. It may be significant that autologousmelanoma cells express low levels of antigen detected byHJM1 or FCM1. Does this indicate that there has been in vivo

immunoselection for low-expression variants in these pa-tients or have these variants been fortuitously selected invitro? Immunoselection of stable antigen-negative or low-expressing tumor cell variants has been obtained in mice (33)and in humans (34) following passive administration ofanti-ganglioside antibodies. It is not difficult to imagine thata similar process could occur as a consequence of anautologous humoral immune response.

This work was supported in part by grants from the NationalCancer Institute (CA-33049, CA-08748) and the American CancerSociety (IM-429) and by the Oliver S. and Jennie R. DonaldsonCharitable Trust, Inc.

1. Irie, R. F., Irie, K. & Morton, D. L. (1976) Cancer Res. 36, 3510-3517.2. Old, L. J. (1981) Cancer Res. 41, 361-375.3. Livingston, P. O., Shiku, H., Bean, M. A., Pinsky, C. M., Oettgen,

H. F. & Old, L. J. (1979) Int. J. Cancer 24, 34-44.4. Knuth, A., Danowsky, B., Oettgen, H. F. & Old, L. J. (1984) Proc.

Nati. Acad. Sci. USA 81, 3511-3515.5. Mukherji, B. & MacAlister, T. J. (1983) J. Exp. Med. 158, 240-245.6. Anichini, A., Fossati, G. & Parmiani, G. (1985) Int. J. Cancer 35,

683-689.7. Livingston, P. O., Old, L. J. & Oettgen, H. F. (1985) Monoclonal

Antibodies and Cancer Therapy (Liss, New York), pp. 537-548.8. Tai, T., Cahan, L. D., Tsuchida, T., Saxton, R. E., Irie, R. F. &

Morton, D. L. (1985) Int. J. Cancer 35, 607-612.9. Bystryn, J.-C., Jacobsen, S., Harris, M., Roses, D., Speyer, J. & Levin,

M. (1986) J. Biol. Response Modif. 5, 211-214.10. Carey, T. E., Takahashi, T., Resnick, L. A., Oettgen, H. F. & Old,

L. J. (1976) Proc. Nati. Acad. Sci. USA 73, 3278-3282.11. Shiku, H., Takahashi, T., Oettgen, H. F. & Old, L. J. (1976) J. Exp.

Med. 144, 873-881.12. Takeyama, H., Shiku, H., Resnick, L. A., Houghton, A. N., Albino,

A. P., Oettgen, H. F. & Old, L. J. (1981) Proc. Am. Assoc. Cancer Res.21, 300 (abstr.).

13. Albino, A. P., Lloyd, K. O., Houghton, A. N., Oettgen, H. F. & Old,L. J. (1981) J. Exp. Med. 154, 1764-1778.

14. Real, F. X., Mattes, M. J., Houghton, A. N., Oettgen, H. F., Lloyd,K. 0. & Old, L. J. (1984) J. Exp. Med. 160, 1219-1233.

15. Carey, T. E., Lloyd, K. O., Takahashi, T., Travassos, L. & Old, L. J.(1979) Proc. Nati. Acad. Sci. USA 76, 2898-2902.

16. Watanabe, T., Pukel, C. S., Takeyama, H., Lloyd, K. O., Shiku, H., Li,L. T. C., Travassos, L. R., Oettgen, H. F. & Old, L. J. (1982) J. Exp.Med. 156, 1884-1889.

17. Steinmetz, M., Klein, G., Koskimies, S. & Makel, 0. (1977) Nature(London) 269, 420-422.

18. Engelman, E. G., Foung, S. K. H., Larrick, J. & Raubitschek, A. (1985)Human Hybridomas and Monoclonal Antibodies (Plenum, New York).

19. Houghton, A. N., Brooks, H., Cote, R. J., Taormina, M. C., Oettgen,H. F. & Old, L. J. (1983) J. Exp. Med. 158, 53-65.

20. Irie, R. F., Sze, L. L. & Saxton, R. E. (1982) Proc. Natl. Acad. Sci.USA 79, 5666-5670.

21. Cahan, L. D., Irie, R. F., Singh, R., Cassidenti, A. & Paulson, J. C.(1982) Proc. Nati. Acad. Sci. USA 79, 7629-7633.

22. Tai, T., Paulson, J. C., Cahan, L. D. & Irie, R. F. (1983) Proc. Nati.Acad. Sci. USA 80, 5392-5396.

23. Watson, D. B., Bums, G. F. & Mackay, I. R. (1983) J. Immunol. 130,2442-2447.

24. Kan-Mitchell, J., Imam, A., Kempt, R. A., Taylor, C. R. & Mitchell,M. S. (1986) Cancer Res. 46, 2490-2496.

25. Cote, R. J., Morrissey, D. M., Houghton, A. N., Beattie, E. J., Jr.,Oettgen, H. F. & Old, L. J. (1983) Proc. Nati. Acad. Sci. USA 80,2026-2030.

26. Miller, G. & Lipman, M. (1983) Proc. Nati. Acad. Sci. USA 70, 190-194.27. Dippold, W. G., Lloyd, K. O., Li, L. T. C., Oettgen, H. F. & Old, L. J.

(1980) Proc. Nati. Acad. Sci. USA 77, 6114-6118.28. Pukel, C. S., Lloyd, K. O., Travassos, L. R., Dippold, W. G., Oettgen,

H. F. & Old, L. J. (1982) J. Exp. Med. 155, 1133-1147.29. Natoli, E. J., Livingston, P. O., Pukel, C. S., Lloyd, K. O., Weigandt,

J., Szalay, J., Oettgen, H. F. & Old, L. J. (1986) Cancer Res. 46,4116-4120.

30. Furukawa, K., Clausen, H., Hakomori, S., Sakamoto, J., Look, K.,Lundblad, A., Mattes, M. J. & Lloyd, K. 0. (1985) Biochemistry 24,7820-7826.

31. Kozber, D., Roder, J. C., Chang, T. H., Steplewski, Z. & Koprowski,H. (1982) Hybridoma 1, 323-328.

32. Thompson, K. M., Melamed, K., Eagle, B. D., Gorick, T., Gibson, T.,Holburn, A. H. & Hughes-Jones, N. C. (1986) Immunology 58, 157-160.

33. Young, W. W. & Hakomori, S. (1981) Science 211, 487-489.34. Houghton, A. N., Mintzer, D., Cordon-Cardo, C., Welt, S., Fliegel, B.,

Vadhan, S., Carswell, E., Melamed, M. R., Oettgen, H. F. & Old, L. J.(1985) Proc. Natl. Acad. Sci. USA 82, 1242-1246.

Proc. Natl. Acad. Sci. USA 84 (1987)

Dow

nloa

ded

by g

uest

on

Aug

ust 1

2, 2

021