Journal of Immunological Methods, 134 (1990) 121-128

121

Elsevier JIM 05741

Murine monoclonal anti-idiotype antibody (a) as a probe to detect human monoclonal antibody bound to human tumor tissues Hiroaki Saito, Duan-Ren Wen, Saburo Yamamoto, Toshiko Yamamoto, Romaine E. Saxton, Alistair J. Cochran and Reiko F. Irie Division of Surgical Oncology, UCLA School of Medicine, Los A ngeles, CA 90024, U.S.A.

(Received 30 March 1990, revised received 11 July 1990, accepted 12 July 1990)

A new immunohistochemical assay was developed for the detection of human monoclonal antibody (HuMAb) bound to human biopsied tumor tissues. A murine anti-idiotype monoclonal antibody, a type, 18C6 (IgG1), was raised against an IgM HuMAb, L612, defining a tumor-associated ganglioside antigen (G~a3) and used as a probe in a three step cell-binding assay ( H u M A b + anti-id + biotinylated anti-mouse Ig). Anti-id 18C6 has an exclusive binding specificity for H u M A b L612, but does not interfere with the binding of L612 to antigen positive melanoma cell lines or to a purified antigen, GM3. The applicability of 18C6 in the three step cell-binding assay was tested first using a melanoma cell line, UCLASO-M12. L612 bound to M12 cells was specifically detected by 18C6 without any background reactivity in ELISA. When this assay was compared with the standard two-step cell-binding assay ( H u M A b + peroxidase-conjugated anti-human IgM) using various cultured tumor cell lines, parallel reactivity was observed. The three-step cell-binding assay was then applied to various fresh-frozen human tumor sections. Positive reactivity was demonstrated on various histologic types of human tumor tissues: primary melanoma (10/10), metastatic melanoma (4/4), nevus (10/10), lung cancer (3/6), breast cancer (2/6), and colon cancer (1/1). Adjacent normal tissues were unstained. Control experiments included the cell-binding assay with L612 alone, 18C6 alone, L612 + unrelated mouse IgG, and unrelated IgM H u M A b (L72) + 18C6; but biotinylated anti-mouse IgG did not react with these control preparations. The results indicate that anti-id 18C6 is a highly specific probe to assess the expression of the ganglioside antigenic epitope recognized by the L612 H u M A b on biopsied human tumor tissues. Key words." Anti-idiotype; Monoclonal antibody, human; Melanoma; Ganglioside; Immunohistochemistry

Correspondence to: R.F. Irie, Division of Surgical Oncology,

9th Floor Louis Factor Building, UCLA School of Medicine, Los Angeles, CA 90024, U.S.A. Abbreviations: HuMAb, human monoclonal antibody; Anti-id, anti-idiotype antibody; PBS, phosphate-buffered saline; ELISA, enzymed-linked immuno sorbent assay; IA, immune adherence.

Introduction

Murine monoclonal antibodies have been employed to define and characterize many antigenic molecules on human cancer cells, and subsequent clinical studies have demonstrated their potential use in cancer treatment and diagnosis. Murine

0022-1759/90/$03.50 ~ 1990 Elsevier Science Publishers B.V. (Biomedical Division)

122 monoclonal antibodies have several advantages over human m0noclonal antibodies (HuMAb) including a strong affinity for tumor antigens, higher antibody secretion by hybridoma ascites, and high antigen density on tumor ceils. However, recent clinical trails with murine monoclonals, reported by many investigators, have clearly indicated that HuMAbs would be preferable for therapy since repeated injections of MuMAbs induce antimurine Ig antibodies in virtually all patients, leading to formation of immune complexes and immune reactions with potentially hazardous complications. In addition, HuMAbs may recognize epitopes that are overlooked by the murine immune system. We previously reported the development of HuMAbs that react with ganglioside antigens on human cancer cells and demonstrated their anti-tumor effect at the clinical level (Irie and Morton, 1986; Irie et al., 1989). Patients with recurrent melanoma received intratumor injections of HuMAb to ganglioside GD2 o r GM2 , and partial or complete regression was observed in about 70% of the patients. In those melanoma patients in whom the immunotherapy was ineffective, the target antigen G m o r GM2 , was not expressed on the tumor cells. Because the quantity and quality of gangliosides on human melanoma are widely heterogeneous between different cancer patients (Tsuchida et al., 1987), unnecessary administration of HuMAb can be avoided by examination of a pre-treatment biopsy to identify which gangliosides dominate on each patient's tumor cells. In our previous treatment studies, we utilized three different immunological assays; the immune adherence assay (IA), direct immunofluoresence with fluoresceinated microspheres, and IA absorption, and a biochemical assay. These assays each have certain limitations and advantages. The immunologic assays required single cell suspensions from the biopsied tumor tissues, but it was often difficult to obtain viable high yield tumor cell populations, and under a light microscope tumor cells were not readily distinguished from monocytes and macrophages. The biochemical assay did not require intact cells and a relatively larger volume of tumor was necessary for ganglioside extraction and measurement of sialic acid in the glycolipid preparation. The most commonly utilized im-

munologic technique for defining antigen expression on biopsy specimens using murine monoclonal antibodies is immunohistochemical staining of tissue sections. However, this sensitive method is not readily applicable to combinations of human monoclonal antibodies and human tissues. The indirect staining of human tumor tissues with the second antibody (anti-human Ig) usually results in high background from non-specific binding to abundant endogenous human Ig. Direct immunostaining using biotinylated human monoclonal antibodies may overcome this high background (Yano et al., 1988), but is usually less sensitive and is most effective when a high density antigen is present on the cell surface. An ideal solution to this problem would be to develop a sensitive indirect assay with a second antibody reacting only to human monoclonal antibodies. In this investigation we utilized the principles of Jerne's network theory to develop a highly specific anti-idiotype antibody (Ab-2) for the detection of HuMAb (Ab-1) bound to human tissues. In this original concept, Jerne (1974) divided anti-idiotype antibodies into Ab-2 ct and Ab-2 ft. Ab-2 t~ recognizes conventional idiotopes and may or may not interfere with the binding of the paratope of Ab-1. Ab-2 fl recognizes paratope-associated idiotype (antigen combining sites) of Ab-1 and contains the internal image of the original antigen. Those Ab-1 bound to antigens on the cell surface should retain functional combining sites for Ab-2 ~ molecules that bind to frame-work associated idiotopes of Ab-1. In this study we have developed an a type anti-idiotype murine MAb binding only to L612 HuMAb which itself reacts to the ganglioside G M3 on human cancer cells. The usefulness and specificity of the anti-id in a three-step indirect assay (HuMAb + murine anti-id + anti-murine IgG) was first determined by using cultured tumor cell lines. The assay was then successfully adapted to biopsied human tissues.

Materials and methods

Cell fines and antibodies

The human tumor cell lines used in this study include UCLASO-M12, UCLASO-M14, UCLA-

123 SO-M15, UCLASO-M25, UCLASO-M27, UCLASO-M101 (melanoma cell lines), Colo 38 (colon cancer cell line), M K N 45 (gastric cancer cell line), Cooper and MCF-7 (breast cancer cell lines), and C H P 100 (neuroblastoma cell line). The S P 2 / O murine cell line, derived from B A L B / c myeloma cells, was used for cell hybridization. Tumor cell lines and hybridomas were cultured in RPMI 1640 medium containing 2 mM glutamine supplemented with 10% fetal calf serum (FCS). An antibiotic mixture of penicillin (100 U / m l and streptomycin (100 /~g/ml) was added to the medium. HuMAbs L612 (IgM, x), L55 (IgM, r ) (Cahan et al., 1982), and L72 (IgM, K) (Tai et al., 1983) were established in our laboratory and purified using the method described previously (Katano et al., 1984). Aliquots of purified antibody were stored in a liquid nitrogen freezer until needed. Normal human IgM, peroxidase-conjugated goat anti-human IgM, and peroxidase-conjugated goat antimouse IgG + IgM, and IgG were purchased from Jackson Immuno Research, West Grove, PA. Pristane (Aldrich) primed B A L B / c mice were injected with hybridoma cells and ascitic fluids harvested to obtain monoclonal antibodies for these studies.

using ELISA. Hybridomas secreting anti-ids (Ab2) were identified by their strong binding reactivity to HuMAb L612 and absent reactivity to three other control human IgMs: L55, L72, and human serum IgM. Unrelated proteins used as antigens included fetal bovine serum and human serum albumin. 50 /~1 of IgMs or proteins (50 ~ g / m l ) were coated on a 96-well ELISA plate and served as antigens to detect Ab-2. Peroxidase-conjugated goat anti-mouse IgG + IgM was used as the Ab-2 detection probe followed by substrate and reading absorbancy at 490 nm as described previously (Tai et al., 1987). The isotypes of these monoclonal antibodies were determined by ELISA using a screening/isotyping kit for murine monoclonal antibodies (Boehringer Mannheim Biochemicals, Indianapolis, IN).

Biopsied tissues Fresh specimens of malignant tissue biopsies were obtained from surgically resected tissues. Specimens were embedded in OTC compound (Miles Laboratories, Napaville, IL), frozen in blocks in isopentane at liquid nitrogen temperature, and then stored at - 8 0 ° C until examination.

Three-step celI-ELISA Viable M12 cells (1 × 105) were plated onto a U-bottom 96-well microtiter plate (Immulon-1, Dynatec) after pre-blocking with 1% BSA-PBS. 50 /~1 of L612 (100/~g/ml) were added and incubated for 1 h at room temperature. After washing the mixture to remove unbound HuMAb L612, the cells were incubated with murine monoclonal antiid (100/~g/ml) for 1 h at room temperature. After washing, 50 /~1 of peroxidase-conjugated goat anti-mouse IgG antibody (1/10,000 diluted) (Jackson Immuno Research) were added and the plate was incubated for 30 min. After washing with PBS solution, the substrate for peroxidase was added and binding activity was deterimned as a function of absorbance at 490 nm with a Vmax kinetic microplate reader.

Generation of monoclonal anti-idiotype antibodies B A L B / c mice were immunized by a subcutaneous injection of 200/~g of purified L612 monoclonal antibody in complete Freund's adjuvant. After 2 weeks the animals were boosted by another s.c. injection of L612 in incomplete Freund's adjuvant. 11 days following the booster the mice were injected intraperitonealy with 200/~g of L612 in saline. After 3 days the spleens were removed and the splenocytes fused with myeloma cell line S P 2 / O using the standard procedure to produce hybridomas. After HAT medium selection, hybridoma culture wells were tested for antibody

Assays for HuMAb detection The immune adherence (Irie et al., 1982), cellELISA (Saito et al., 1988) and GM3-ELISA (Tai et al., 1987) assays were used to detect HuMAb L612 bound to the cell surface of tumor cell lines or to purified G M3. Procedures for these assays have been described in the references listed above.

Three-step immunoperoxidase staining of tissue sections Tissue sections 4 /~m thick were cut from tissues freshly frozen in OTC compound and im-

124

mediately fixed in cold formaldehyde buffer (12 g Tris buffer, 9 g sodium chloride, 40 ml 37% formaldehyde, pH 7.4) and air dried. Slides were dipped in Tris buffer for 5 min then treated with 3% hydrogen peroxide for 10 rain to quench endogenous peroxidase activity. After washing in running water for 5 min, sections were overlaid with 5% normal human serum for 20 min. HuMAb L612 (10/xg IgM in 200/~1) was then applied and incubated for 45 min. The slides were washed in Tris buffer for 5 min, the purified anti-id 18C6 (10 /~g IgG1 in 200/~1) was applied and incubated for 30 min. After washing the slides again, the third antibody, a biotinylated goat anti-mouse IgG (Vector Laboratories, Burlingame, CA) at 1/100 dilution was applied and incubated for 25 min. Proxidase-conjugated streptavidin (1/1000 dilution) (Zymed Laboratories, San Francisco) was added after washing and incubated for 20 min. After washing, the slides were immersed in substrate solution containing 6 ml amino-ethyl carbazole, 50 ml of 0.02 M sodium acetate buffer (pH 5.1), and 0.4 ml of freshly prepared 3% hydrogen oxide for 5 rain. The slides were washed once more in tap water, counterstained with hematoxylin, and cover-slips applied to the stained sections using glycerol-gelatin.

Results

Establishment of anti-idiotypic monoclonal antibodies to L612 H u M A b H A T selection of approximately 2500 hybridoma culture wells yielded 40 hybridomas secreting antibodies with distinct reactivity to L612 HuMAb, but no reactivity to three other control human IgMs and two unrelated serum protein antigens. To determine whether these anti-L612 antibodies were Ab-2 fl type directed against the antigen combining site of L612, or were Ab-2 antibodies bound to peptide regions outside the antigen-combining site of L612, the inhibitory activity of these anti-L612 antibodies against L612 binding t o G M 3 positive target cell lines or to the purified antigen, ganglioside G M 3 , w a s tested by using three assay systems: IA inhibition, cellELISA inhibition, and G~3-ELISA inhibition. Of the 40 antibodies tested, seven inhibited L612

TABLE I ELISA O F T W O M U R I N E M O N O C L O N A L A G A I N S T L612 HuMAb, 18C6 A N D 19A6 Coated antigen (5 #g/well)

612 IgM HuMAb (GM3) L55 IgM HuMAb (GM2) L72 IgM HuMAb (GD2) Norman human IgM Fetal calf serum Human albumin

Ab-2

A490nm a

19A6 b

18C6 b

1.041__+0.021 0.035 __+0.009 0.034 __+0.008 0.038 __+0.006 0.025 __+0.005 0.029__+0.007

0.968__+0.011 0.036 __+0.008 0.042 __+0.007 0.030 __+0.008 0.031 __0.004 0.029+0.011

a Values are mean_+ SD from which the absorbance of culture medium was subtracted. b 3 days of culture supernatants in the confluent condition were used.

binding to an antigen positive target melanoma cell line, M12, and t o G M 3 greater than 50% in the assays, while 12 others had weak or no inhibitory activity. Of the seven inhibitory anti-ids, 19A6 was cloned and selected as a potential Ab-2 fl candidate (Yamamoto et al., 1990). From the non-inhibitory group, 18C6 was chosen as a possible Ab-2 a candidate for subsequent studies (Table I). Both anti-ids, 19A6 and 18C6, were tested with isotype antiglobulins and found to be of the IgG1 class and contain x light chains. These cloned hybridoma cell lines were grown in FCS-containing RPMI 1640 medium and secreted 5 - 1 0 / ~ g / m l of antibody into culture supernatants. Titers of the anti-ids in these culture supernatants against L612 by ELISA ranged between 1 : 2 0 0 to 1 : 1000/106 hybridoma. The percent inhibition of L612 in the IA and GM3-ELISA assays are shown in Fig. 1. Anti-id 18C6 demonstrated low binding inhibition of HuMAb L612 to target cells in the IA assay (25%) and to ganglioside GM3 in ELISA (27.5%) whereas 19A6 at the same antibody concentration showed strong inhibition in both the ELISA assay (84.5%) and the IA assay (87.5%). As a control assay, 19A6 and 18C6 failed to inhibit the binding of an unrelated antigen system, HuMAb L72, to M14 target cells, or to G D2 antigen. The lack of binding inhibition of 18C6 indicates a binding location on L612 outside the GM3 antigen combining site, and the specific binding inhibition of 19A6 indicates its binding

125 l o c a t i o n to be within or n e a r the antigen c o m b i n ing site. I.O-

c

Use of anti-id a, 18C6, to detect H u M A b L612 bound to tumor cell fines Theoretically, when the antigen c o m b i n i n g sites of L612 are o c c u p i e d b y GM3 expressed on t u m o r cells, the cell b o u n d L612 should have r e d u c e d its b i n d i n g activity to a n t i - i d /3, yet still retain full b i n d i n g activity to a n t i - i d c~. W e tested this hypothesis first using c u l t u r e d M12 m e l a n o m a cells which express a high d e n s i t y of the c o r r e s p o n d i n g antigen. This cell b i n d i n g assay represents a m o d ified f o r m of the E L I S A technique as d e s c r i b e d in the m a t e r i a l s a n d m e t h o d s section. Several c o n t r o l assays were i n c l u d e d to establish the specificity of the positive reaction. C o n t r o l anti-ganglioside H u M A b s i n c l u d e d L55 ( I g M anti-Gr42) a n d L72 ( I g M anti-GD2 ), b o t h of which exhibit strong b i n d i n g a b i l i t y to the GM2 a n d GD2 rich M14 m e l a n o m a cell line ( C a h a n et al., 1982; Tai et al., 1983). A s shown in Fig. 2, anti-id 18C6 r e a c t e d strongly to M12 cells after p r e - i n c u b a t i o n with H u M A b L612, b u t d i d n o t react to M14 cells that were p r e - i n c u b a t e d with L55 or L72 H u M A b . T h e p e r o x i d a s e - c o n j u g a t e d a n t i - m o u s e Ig also failed to react with M12 cells in other c o n t r o l s including m u r i n e a n t i - i d alone, L612 alone, or L612 plus 19A6 (anti-id /3). T h e c e l l - E L I S A assay was then

100

% Inhibition

80

60

40

GM3

M12 20

0 19A6

1806

Ab-2 Clones

Fig. 1. Murine monoclonal anti-ids, 18C6 and 19A6, were tested for their inhibition of L612 binding to the M12 melanoma cell line in IA (solid bars), and to purified GM3 by ELISA assay (hatched bars), respectively. In ELISA, the OD value (490 nm) in the absence of both anti-id and L612 HuMAb was 0.053, and in the presence of L612 alone was 1.197. In IA the negative control (without L612) was '0' and the positive control (with L612) was '4'.

001..1o d

~ 0.5-

Target Cells Hu M A b

MI2

MI2

MI2

L612 L612

A n ti-~[d 18C6

+

-

+

HR P - A n t i - [ g

Mo

Mo

Mo

MI2

MI2

MI4

MI4

MI4

MI4

-

L612

L55

L72

L55

L?2

Mo

Hu

+

+

-

MO

MO

Hu

Hu

Fig. 2. Three-step cell binding ELISA using anti-id a, 18C6, to detect HuMAb L612 on M12 melanoma cells. Two other human monoclonal antibodies, L55 and L72, and M12 melanoma cells, were used as control antibodies and target cells, respectively. Abbrevations used are HRP, horseradish peroxidase conjugated; Mo, goat anti-mouse IgG; and Hu, goat anti-human IgM. OD values in the absence of anti-id are 0.097 to M12 cells, and 0.083 to M14 cells. These background values were subtracted from each test value.

a p p l i e d to several other h u m a n t u m o r cell lines. A two-step cell b i n d i n g assay ( H u M A b + p e r o x i d a s e - c o n j u g a t e d a n t i - h u m a n I g M ) was c o m p a r e d with the three-step c e l l - b i n d i n g assay to evaluate the validity of the three-step assay. A s shown in T a b l e II, the three-step assay h a d p a r a l l e l reactivity with the two-step assay a n d was slightly m o r e sensitive in almost every cell line. This d a t a indicates that the E L I S A a b s o r b a n c y value of the three-step assay a c c u r a t e l y reflects differences in the density of cell surface GM3 antigens a n d correlates closely with the two-step in vitro assays.

Three-step immunostaining of biopsied human tumor tissue T h e three-step assay e s t a b l i s h e d a b o v e was applied to surgically b i o p s i e d t u m o r tissues t h a t h a d been snap frozen. T h e results are shown in T a b l e I I I a n d Fig. 3. H u M A b L612 was f o u n d to react with 100% of nevus tissue ( 1 0 / 1 0 ) , 100% of m e l a n o m a ( 1 4 / 1 4 ) , 40% of lung c a r c i n o m a ( 2 / 5 ) , 33% of b r e a s t c a r c i n o m a ( 2 / 6 ) , a n d 1% of one c o l o n carcinoma. Positive reactivity was not seen in the s a r c o m a s ( 0 / 2 ) tested, n o r in the n o r m a l skin tissues e x a m i n e d ( 0 / 3 ) . N o n - c a n c e r o u s p o r tions of a n t i g e n - p o s i t i v e t u m o r sections (Figs. 3A,

126

3C and 3E) and a normal skin tissue (Fig. 3 F ) were clearly negative. N o n e of the tumor tissues were positive in the absence of L612 (Figs. 3B and

3D). Specificity was luther confirmed by an additional control experiment: positive tissues were not stained when they were pre-treated with unre-

Fig. 3. Three-step immunoperoxidase staining of biopsied melanoma and skin tissues. In A (primary melanoma), C (primary melanoma), E (metastatic melanoma), and F (normal skin), tissues were reacted first with L612 HuMAb, then with anti-id a 18C6, and finally with biotinylated anti-mouse IgG (photomicrograph 250X). In B and D, the same melanoma tissues (A and C, respectively) not treated with L612 but only with anti-id a 18C6 and biotinylated anti-mouse IgG (250 x). Details of the procedure are described in the materials and methods section.

127 TABLE II THREE-STEP CELL B I N D I N G ELISA U S I N G A N T I - I D a, 18C6, ON VARIOUS H U M A N T U M O R CELL LINES

TABLE III THREE-STEP I M M U N O S T A I N I N G OF BIOPSIED HUMAN TISSUE SECTIONS U S I N G A N T I - I D a, 18C6

Tumor cell lines

Tissues b (no. tested)

UCLASO -M12 (melanoma) -M14 (melanoma) -M15 (melanoma) -M24 (melanoma) -M25 (melanoma) -M27 (melanoma) M K N 45 (stomach) Colo 38 (colon) Cooper (breast) MCF (breast) CHP 100 (neuroblastoma)

A49on~ a L612 HuMAb + 18C6 MuMAb b

L612 HuMAb c

1.297 ± 0.034 0.744 + 0.051 1.583 ± 0.063 0.417 ± 0.035 0.219 ± 0.020 1.085 i 0.038 0.134+0.015 0.222 i 0.016 0.089 i 0.011 0.036 ± 0.005

0,958 ± 0.032 0.553 ± 0.048 1.277 ± 0.058 0.323 ± 0.024 0.154 ± 0.017 0.799 + 0.045 0.129 ±0.019 0.153 ± 0.019 0.074 i 0.009 0.031 + 0.007

0.412 ± 0.054

0.315 i 0.025

aValues are mean + SD corrected for background absorbance ( < 0.1) of culture medium. b Peroxidase-conjugated goat anti-mouse IgG (1/10,000 dilution) was used as 3rd antibody. c Two-step cell binding ELISA: peroxidase-conjugated goat anti-human IgM (1/10,000 dilution) was used as 2nd antibody.

lated human monoclonal antibodies (L72 or L55 HuMAbs). These results indicate that the murine anti-id a employed in the three-step assay is a sensitive and specific new probe to detect H u M A b s bound to biopsied tumor tissues and to identify expression of ganglioside antigens on biopsied human tissues.

Nevus Compound nevus (6) Dysplastic nevus (4) Primary melanoma Clark level I-III (2) Clark level III-V (8) Metastatic melanoma (4) Breast cancer (6) Lung cancer (5) Colon cancer (1) Sarcoma (2) Normal skin (3)

no. tissues with reactivity a +++

++

+

--

0 1

0 1

6 2

0 0

0 2 0 1 0 0 0 0

1 3 0 0 1 1 0 0

1 3 4 1 1 0 0 0

0 0 0 4 3 0 2 3

a Degree of reactivity was scored as follows: + + + , strongly positive; + + , moderately positive; + , weakly positive; - , negative. b Snap fresh-frozen surgical tissue specimens were used for each immunostaining.

Discussion In this study we established a new immunohistochemical assay utilizing murine monoclonal anti-idiotype alpha antibody as a probe to detect H u M A b bound to ganglioside antigens on human cancer tissues. The validity of the assay was demonstrated by strong staining of cancer tissues without staining of adjacent normal tissues, and by negative staining of tumor tissues in a variety of control experiments (Table IV). In addition, when cultured cell lines were tested, the assay was

TABLE IV SUMMARY OF THREE-STEP I M M U N O H I S T O C H E M I C A L S T A I N I N G OF H U M A N M E L A N O M A TISSUES U S I N G ANTI-IDS a Tissues

Melanoma Melanoma Melanoma Melanoma Melanoma Normal skin

Reacted With

no. positive/no, tested

HuMAb

Anti-id

Anti-mouse Ig

L612

18C6, anti-L612 (a) 18C6, anti-L612 ( a )

+ +

L612 L612 L72 L612

-

+

19A6, anti-L612 (fl) 18C6, anti-L612 (a) 18C6, anti-L612 (a)

+ + +

14/14 0/14 0/14 0/14 0/14 0/3

128

shown to be as sensitive as the standard two-step indirect assay. The clinical and biological implications of human monoclonal antibody L612 binding to various human cancer systems were not directly investigated in the present study. L612 has been found to react with the ganglioside GM3 o n human cancer cells. Whether or not the L612 antibody also detects sialoglycoproteins expressing this epitope on cancer cells is not known. However, the present immunohistochemical analysis and additional studies by an immune adherence absorption assay have revealed that L612 reacts strongly with cancer tissues. Non-cancerous tissues including erythrocytes, lymphocytes, and normal tissues adjacent to tumor ceils were negative for antigen expression on their cell surface. However, some binding was seen with nevus specimens, indicating G M3 expression on these benign melanocytic tissues. Of the limited number of cancer tissues examined in this study, primary melanomas expressed the highest amount of antigen. Dense staining was observed in primary melanomas of Clark levels III or greater more frequently than in metastatic melanoma. This observation is consistent with reports by other investigators (Herlyn et al., 1985), who have shown that as melanoma progression occurs, G M3 c o n v e r t s to G D3, reversing the ratio of GM3 t o GD3 in primary and metastatic melanoma. H u m a n monoclonal antibody L612 reacted with two out of six breast cancers in this three-step assay, but we have previously observed positive reactivity in five out of seven breast cancer tissues tested in the more sensitive IA absorption assay. More extensive tissue sampling is required to provide an accurate description of the antigen distribution in human cancer tissues.

Acknowledgements We would like to thank Dr. Donald L. Morton for his invaluable suggestions and support; Ms. Lan Sze, Mr. Jien Ming, and Miss A m y Lee for their technical assistance; and David DeNunzio for his editorial assistance.

This research was supported by Grants CA 12582, CA30647, and CA42396 awarded by the National Cancer Institute.

References Cahan, L.D., Irie, R.F., Singh, R., Cassidenti, A. and Paulson, J.C. (1982) Identification of a human neuroectodermal tumor antigen (OFA-I-2) as ganglioside GD2. Proc. Natl. Acad. Sci. U.S.A. 79, 7629. Herlyn, M., Thurin, J., Balaban, G., Bennicelli, J.L., Herlyn, D., Elder, D.E., Bondi, E., Guerry, D., Nowell, P., Clark, W.H. and Koprowski, H. (1985) Characteristics of cultured human melanocytes isolated from different stages of tumor progression. Cancer Res. 45,5670. Irie, R.F. and Morton, D.L. (1986) Regression of cutaneous metastatic melanoma by intralesional injection with human monoclonal antibody to ganglioside GD2. Proc. Natl. Acad. Sci. U.S.A. 83,8694. Irie, R.F., Sze, L.L. and Saxton, R.E. (1982) Human antibody to OFA-I, a tumor antigen, produced in vitro by EBVtransformed human B-lymphoblastoid cell lines. Proc. Natl. Acad. Sci. U.S.A. 79,5666. Irie, R.F., Matsuki, T. and Morton, D.L. (1989) Human monoclonal antibody to ganglioside GM2 for melanoma treatment. Lancet 1,786. Jerne, N.K. (1974) Towards a network theory of the immune system. Ann. Immunol. (Paris) 125C,373. Katano, T., Saxton, R.E. and Irie, R.F. (1984) Human monoclonal antibody to tumor associated ganglioside GD2. J. Clin. Lab. Immunol. 15,119. Saito, H., Uchiyama, K., Nakamura, I., Hiraoko, H., Yamaguchi, Y. and Taniguchi, M. (1988) Characterization of a human monoclonal antibody with broad reactivity to malignant tumor cells. J. Natl. Cancer Inst. 80,728. Tai, T., Paulson, J.C., Cahan, L.D. and Irie, R.F. (1983) Ganglioside GM2 as a human tumor antigen (OFA-I-1). Proc. Natl. Acad. Sci. U.S.A. 80,5392. Tai, T., Kawashima, I., Saxton, R.E. and Irie, R.F. (1987) Monoclonal antibody detects monosialoganglioside having sialic acid a 2-3 galactose residue. J. Biol. Chem. 262,6803. Tsuchida, T., Saxton, R.E., Morton, D.L. and Irie, R.F. (1987) Gangliosides of human melanoma. J. Natl. Cancer Inst. 78,45. Yamamoto, S., Yamamoto, T. and Irie, R.F. (1990) Anti-idiotype monoclonal antibody carrying the internal image of ganglioside GM3. J. Natl. Cancer Inst., in press. Yano, T., Yasumoto, K., Nagashima, A., Murakami, H., Hashizume, S. and Nomoto, K. (1988) Immunohistological characterization of human monoclonal antibody against lung cancer. J. Surg. Oncol. 39,108.