Background of the Invention

This application is a continuation-in-part of U.S. Serial No. 860,663, filed May 7, 1986, the contents of which are incorporated by reference intδ the present application.

This invention was made with government support under Grant Numbers CA-36120 and CA-08748 from the National Cancer Institute. The U.S. Government has certain rights in the invention.

Throughout this application, various publications are referenced by arabic numerals within parentheses. Full citations for these references may be found at the end of the specification immediately preceding the claims. The disclosures of these publications in their entire- ties are hereby incorporated by reference into this application in order to more fully describe the state of the art as known to those skilled therein as of the date of the invention described and claimed herein.

Gangliosides are prominent cell-surface constituents of melanoma and other tumors of neuroectodermal origin.

Three gangliosides, the monosialoganglioside GM2 and the di si aloganglio sides GD2 and GD3, are of particular interest to tumor i munologists because of their po¬ tential as targets for passive immunization with monoclonal antibodies (mAbs) and for active immuniza¬ tion with cancer vaccines. Despite the presence of GM2, GD2, and GD3 in normal brain and other tissues (1) , these gangliosides are immunogenic in mice and

humans; mouse mAbs have been generated against GM2 (2) , GD2 (3,4) , and GD3 (5-8) , and human sera and human mAbs with reactivity for GM2 (9-12), GD2 (11, 13-, 14) and GD3 (15) have been identified.

Over the past decade we have immunized sequential groups of melanoma patients with a variety of melanoma cell vaccines (16-20). These vaccine trials were based on our serological analysis of the humoral immune re¬ sponse of melanoma patients to cell-surface antigens of autologous and allogeneic melanomas (21) , and each vaccine was constructed to contain melanoma surface antigens known to be immunogenic in humans. Although vaccinated patients readily produced antibody to HLA- related alloantigens and heterologous serum components in the vaccine, only rarely was antibody elicited to more restricted melanoma antigens, such as class 1 (unique) , or GD2 or other class 2 (shared) melanoma antigens. In parallel vaccine studies in the mouse, we have identified immmunizing procedures that facilitate the serological response to tumor antigens (22-24) . In the case of GM2, immunization with GM2-expressing tumor cells or purified GM2 only infrequently induced GM2 antibody in mice, whereas vaccines containing GM2 with adjuvants such as bacillus Calmette-Guerin (BCG) or Salmonella minnesota R595 were far more effective (24) .

Livingston, et al. (22, 23) teach BCG and Salmonella useful as adjuvants in stimulating an immunogenic re¬ action. However, Livingston, et al. neither teach nor suggest the use of such adjuvants to stimulate or enhance antibodies against GM2. Further, Livingston, et al. do not disclose the use of the Salmonella minnesota R595 as an adjuvant.

Tai, et al. identified the immunogenicity of GM2 in studies of melanoma patients (12). However Tai, et al. do not teach or suggest the use of BCG or Salmonell minnesota R595 to stimulate or enhance antibodies di¬ rected against GM2.

Irie, et al., U.S. Patent No. 4,557,931, issued Decem¬ ber 10, 1985, disclose the use of GM2 in combination with an adjuvant to raise antibodies against GM2. However, Irie, et al. use as adjuvants liposomes or serum albumin. Moreover, Irie, et al. do not teach or suggest the use of BCG or Salmonella minnesota R595 to stimulate or enhance antibodies directed against GM2.

In the present study, we have examined the immunogenicity of GM2-containing vaccines in stage III melanoma patients. Two types of vaccines were used : a whol e-cell vaccine containing high level s of GM2 and vaccines containing purif ied GM2 with or without micro¬ bi al adj uvants.

Sug-aary of the Invention

This invention provides a vaccine for stimulating or enhancing in a subject to whom the vaccine is adminis¬ tered, production of antibodies directed against GM2. The vaccine comprises an amount of purified GM2 effec¬ tive to stimulate or enhance antibody production in the subject, an effective amount of a microbial adjuvant and a pharmaceutically acceptable carrier. The micro¬ bial adjuvant may be bacillus Calmette-Guerin or Salmo- nella minnesota R595.

This invention also provides a method for stimulating or enhancing in a subject production of antibodies directed against GM2. The method comprises adminis- tering to the subject an effective dose of the vaccine of this invention.

This invention further provides a method for treating cancer in a subject affected with cancer. The method comprises administering to the subject an effective dose of the vaccine of this invention.

Finally, this invention provides a method for prevent¬ ing cancer in a subject affected with cancer. The method comprises administering to the subject an ef¬ fective dose of the vaccine of this invention.

Brief Description of Figures

Figure 1 shows the GM2 antibody response of stage III melanoma patients after immunization with a whole-cell vaccine or purified GM2 ganglioside vaccines. Each curve represents the response of an individual patent. Arrows indicate time of Cy injection or vaccine injec¬ tion. The adjuvant used in the booster vaccine was the same as the adjuvant used in the initial vaccines, except in cases in which BCG was replaced by R595.

Figure 2 shows the detection of GM2 antibody in sera from vaccinated melanoma patients by immunostaining. Identical TLC plates were stained with resorcinol (far left) or were allowed to react with patients vaccinated with GM2/BCG: GM2 antibody titers by EL ISA 1:160 and 1:320. D and E, sera from two patients treated with Cy and vaccinated with GM2/BCG: GM2 antibody titers by ELISA-1:80 and 1:160. F, serum from a patient vacci¬ nated with GM2/R595: GM2 antibody titer by ELISA 1:20.

De tailed Description of the Invention

This invention provides a vaccine for stimulating or enhancing in a subj ect to whom the vaccine is adminis¬ tered, production of antibodies directed against GM2 . The vaccine comprises an amount of purif ied GM2 effec¬ tive to stimulate or enhance antibody production in the subj ect, an effective amount of a microbial adj uvant and a pharmaceutically acceptable carrier. Prefera¬ bly, the subj ect is a human being and the GM2 is bound to the microbi al adj uvant by a hydrophobic bond be¬ tween the l ipid portion of the GM2 and the cell mem¬ brane of the microbial adj uvant.

The pharmaceutically acceptable carrier may be any well known carrier. However, the presently pr eferred carri¬ er is a phosphate buffered saline solution.

The preferred microbial adj uvant is bacillus Calmette- Guerin. However, the microbial adj uvant may also be Salmonella minnesota R595.

A range of the amount of purif ied GM2 may be emopl oyed. However, the preferred amount present in the vaccine i s an amount between about 50 micrograms and about 300 micrograms. Additionally, a range of thee amount of Salmonella minnesota R595 may be empl oyed. However, the preferred amount present in the vaccine is an amount between about .2 mg and about 1 .5 g. Further, a range of the amount of bacillus Calmette-Guerin may be empl oyed. However, the pref erred amount present in the vaccine is an amount between about 10 ³ viable uni ts and about 3 x 10 ⁷ viable uni ts.

This invention also provides for administering a vac¬ cine of this invention to a subject afflicted with cancer. The antibody produced in the subject upon administration of the vaccine effectively treats the cancer.

This invention still further provides for administering a vaccine of this invention to a subject susceptible to cancer . The antibody produced in the subj ect upon administration of the vaccine effectively prevents the cancer. A subj ect susceptible to cancer incl ude subjects predisposed to cancer as well as subjects who have previously had cancer.

The cancer treated or prevented by the vaccine of this invention may be of neuroectodermal origin. Specif i¬ cally, the cancer of neuroectode rmal origin may be a melanoma.

This invention also discloses a method for stimulating or enhancing in a subject production of antibodies directed against GM2. The method comprises adminis¬ tering to the subject an effective dose of the vaccine of this invention.

This invention further discloses a method for treating cancer in a subject affected with cancer. The method comprises administering to the subject an effective dose of a vaccine of this invention.

This invention still further discloses a method for preventing cancer in a subject affected with cancer. The method comprises administering to the subject an effective dcse of a vaccine of this invention.

The vaccine may be administered subcutaneously, intradermally or intramuscularly. Further, the vaccine may be administered in a single dose or by. a single dose followed by a booster dose. The booster dose is preferably administered 12-16 weeks after the inital dose.

Preferably, the methods of this invention employ the GM2 bound to the microbial adjuvant by -a hydrophobic bond between the lipid portion of the GM2 and the cell membrane of the microbial adjuvant. Additionally, the microbial adjuvant may be Salmonella minnesota R595 or bacillus Calmette-Guerin.

The methods of treating or preventing cancer may in- volve a cancer of neuroectodermal origin. Specifical¬ ly, the cancer of neuroectodermal origin may be a mela¬ noma.

Finally, an effective amount of cyclophosphamide may be administered to the subject prior to administering the vaccine. A range of time prior to administering the vaccine may be employed. However, the perferred time for administering the cyclophosphamide is between about 3 days and about 7 days prior ^' to administering the vaccine. Further, a range of the amount of c clophosphamide may be employed. However, the pre- f erred amount is between about 1 mg/πr and about 500 mg/m .

Materials and Methods

Animal Trials

Aniaals _* Female BALB/c-C57BL/6 F- (B6) mice, 2 to 5 mo of age, were obtained from The Jackson Laboratory, Bar Harbor, ME.

Timor cells. The derivation by Berkelhammer et al. (26) of the JB-RH cell line, and by ourselves of the JB-RH-16 subclone used for serological analysis, as well as the serological and biochemical analysis char¬ acterizing this subclone as an exceptional expressor of GM2, have been described (2, 26). All references to JB-RH refer to this JB-RH-16 subclone.

Serological assays. Mice were bled from the retro- orbital sinus at 2 wk intervals after vaccination, and serum samples for serological testing (approximately 0.1 ml) were stored at -20°C. The immune adherence (IA) and complement-dependent cytotoxicity assays (with the use of rabbit complement) that detect primarily immunoglobulin M (IgM) and the protein A (PA) , and anti-IgG assays that detect primarily IgG, were per¬ formed as described (22, 37, 38). The results of the PA, anti-IgG, and IA assays were analyzed microscopi¬ cally and were expressed as the highest antibody titer resulting in 20% of the target cells having indicator red cells attached to 50% or more of their cell perime¬ ters.

Inhibition tests with the use of purified gangliosides to inhibit IA reactivity were also performed as de¬ scribed (2) . Various quantities of ganglioside were dissolved in chloroform :methanol 2:1, were aliquoted into dilution tubes (6 x 50 mm) , an were dried in a desiccator. Thirty microliters of serum were allowed to react with the gangliosides at 4°C for 1 hr and then were tested in an IA assay on JB-RH cells.

Binding capacity of R595 and bacillus Calmette-Guerin (BCG) for G 2. Fifty micrograms of GM2 were mixed with various quantities of BCG and R595 and were lyoph- ilized. The resulting powder was resuspended as de¬ scribed for vaccine production, was sedimented by cen- trifugation at 15,600 x G, and thin layer chromato- graphy (TLC) was performed to detect GM2 in supernatant and pellet.

10 Preparation of vaccines: vaccine containing JB-RH cells. JB-RH was grown in tissue culture with medium containing 5% fetal calf serum. On the day of vaccina¬ tion, cells were dislodged mechanically, irradiated with 10,000 rad from a Cobalt 60 source, and counted. ι~ ^α Cells (5 x 10') were mixed with 50 micrograms of MPLA in 0.2 ml normal saline, as described (22). Ganglioside content of this vaccine was determined by extracting the cells as described (2) , performing TLC, and quantitating the gangliosides by densitometric 0 scanning (2) . JB-RH cells (5 x 10') contain 60 micro¬ grams of GM2.

Traditional vaccines. Equal volumes of GM2 in saline and commplete Freund's adjuvant (DIFCO Laboratories, 5 Detroit, MI) were emulsified as descirbed (22) . MPLA and BCG cell wall skeletons (BCG C S) (Ribi Immunochem, Hamiliton, MT) were added to intralipid (Cutter Labora¬ tories Inc., Berkely, CA) and were used to suspend the dried ganglioside. 0

Vaccines containing bacteria. J-5 E. coli and Salmo¬ nella minnesota (American Type Culture collection, Rockville, .D) , and Salmonella minnesota mutant R595 (kindly provided by Dr. Jerry McGhee, University of 5

Alabama) were boiled in 1% acetic acid as described (27) , were washed, and were stored frozen before use. The day before vaccination, these bacteria or BCG (Tice strain: University of Illinois Medical Center) were resuspended in distilled water by sonication and were added to tubes containing dried GM2. The suspension was lyophilized and on the day of vaccination was resuspended in normal saline shortly before administra¬ tion.

Liposoβe vaccines. Liposome preparation 1 was prepared by mixing 200 micrograms of GM2 with 9.3 mg of leci¬ thin, 9.3 mg of sphingomyelin, and 18.4 mg of choles¬ terol (Sigma Chemical Co., St. Louis, MO). Liposomes 1 were formed by sonication by using a probe sonicator with a microtip and an energy level of 7 for 5 sec

(Model 185:Bronson Sonic Power Co., Danbury, CT) . Liposome preparations 2a and 2b were prepared with 200 micrograms of GM2, phosphatidyl choline (800 micro¬ grams) , cholesterol (600 micrograms) , and dicetyl-phos- phate (30 micrograms) . Liposomes 2a were prepared with a probe sonicator as described for liposomes 1, whereas liposomes 2b were formed by sonication in water bath sonicator for 5 min (Bransonic 12; Bransonic Cleaning Equipment Company, Shelton, CT) . Adjuvants to be in- corporated into the liposomes were added to the GM2 before sonication. These adjuvants included the lipoidylamine CP-20, 961 (Pfizer, Groton, CT) , the muramyldipeptide (MDP) succinimidester analog CGP 17107 supplied courtesy of Professor D.G. Braun, CIBA-Geilgy, Basel, Switzerland, endotαxin (highly purified lipopolysaccharide) derived from Salmonella abortus equi and provided by Dr. C. Galanos, Freiberg, Germany, and MPLA.

Cyclophosphaaide (Cy) . Cy (Cytoxan: Mead Johnson and company, Evansville, IN) was administered at a dose of 15 mg per kg i.p. 3 days before the first vaccination.

Administration of vaccines. In each experiment, five mice were immunized with a given vaccine. Mice were selected randomly from the same shipment. Vaccines were administered subcutaneously in a total vol. of 0.1 ml per mouse. Two or more vaccinations containing 50 micrograms of GM2 were given at 1 mo intervals, with the exception of vaccines containing complete Freund's adjuvant (which produced marked induration and draining ulcers), no toxicity Or morbidity was detected as a consequence of the administration of any vaccine.

Human Trials

Patients. For the studies involving vaccination with purified GM2, patients with AJCC stage III melano¬ ma (i.e., metastases restricted to regional skin and lymph nodes) were considered eligible if tumors and regional lymph nodes had been resected within 4 months and if they were free of detectable melanoma. The studies with the whole-cell vaccines involved patients with Clark's level IV or V primary melanoma or palpable regional lymph node metastases, who were scheduled for regional lymph node dissection. In these cases, the initial vaccine was administered at least 10 days prior to surgery. None of the patients had received prior chemotherapy or radiation therapy. Patients were exam- ined at 6-week intervals. Chest x-rays, liver function tests and urir.alysis were performed at 3-month inter¬ vals. Blood for serologic tests was obtained at 2-week intervals.

Gangliosides. GM2 was prepared by treating GMl with beta-galactosidase (G.W. Jourdian, Michigan State Uni¬ versity, Ann Arbor, MI) according to published methods (14) . GDla, GDlb and GT1 were purchased from Supelco (Bellaf onte, PA) . GD2 was generously provided by Herbert iegandt (University of Marburg, Federal Repub¬ lic of Germany) . Ganglioside extraction, identifica¬ tion, and quantification were performed as described (2).

Serological Procedures. Typing of cell lines for ex¬ pression of cell-surface gangliosides with mAbs 5-3, 3F8, and R24 was performed as described (2, 3, 5). The enzyme-linked immunosorbent assay (ELISA) (2) was per¬ formed with rabbit anti-human IgM, anti-human IgG, or protein A conjugated to alkaline phosphatase (Zymed

Laboratories, San Francisco) . Antibody titer was de¬ fined as the highest serum dilution yielding an 0D greater than or equal to 0.190. Complement-depen¬ dent cytotoxicity assays (22) were performed with nor- mal human serum (diluted 1:3) as the complement source.

Reagents for ITLC (Gelman) (2) were peroxidase-conju- gated goat anti-human IgM and goat anti-human IgG (Tago, Burlingame, CA) diluted 1:500.

Whole-Cell Vaccines. Procedures used for establishing human astrocytoma cell line S -MG-14 and human melanoma cell line SK-MEL-31 have been described (25) . The JB- RH mouse melanoma cell line was established by J. Berkelhammer et al. (26) and a subclone, JB-RH-16, selected for high GM2 expression was established in our laboratory (2). Methods for culturing and harvesting cells for vaccine production have been described (19) .

Cells were irradiated with 10,000 rads (1 rad = 0.01 gray) from a Co source and frozen viable in dimethyl

sulfoxide. On the day of vaccination, cells from each line were thawed rapidly, washed three times in phos¬ phate buffered saline (PBS) , pooled in roughly equal numbers, and injected. The median total number of cells per vaccine was 2.2 x 10 suspended in 1 ml of PBS, with approximately 40% viability (as judged by trypan blue exclusion) . No bacterial adjuvant was used. Two or three vaccinations were administered at 5-day intervals immediately prior to lymph node dissec¬ tion, two or three additional vaccinations were given at 4-weeks after surgery for a total of five vaccina¬ tions. The vaccine was administered intrader ally on a rotating basis involving all extremities.

Purified GM2 Vaccines. To prepare GM2 vaccines without adjuvants, 100 micrograms of GM2 was dissolved in 1 ml of PBS. For vaccines containing BCG, 10' viable units of BCG (Tice strain. University of Illinois) , or 3 x 10 units in the case of patients showing strong reac¬ tions to BCG, were suspended in distilled water by sonication and added to tubes containing 100 micrograms of dried GM2. The suspension was lyophilized and sus¬ pended in PBS shortly before vaccine administration _[. minnesota mutant R595 (kindly provided by Jerry McGhee, University of Alabama) was boiled in 1% acetic acid for 1 hour as described (27), wahsed, dried, and stored frozen. For vaccine preparation, 0.5 mg of R595 was suspended in PBS by sonication and added to GM2 in the same manner as described for BCG. Patients immu¬ nized with GM2 vaccines received four vaccinations intradermally at 2-week intervals on a rotating basis to uninvolved extremities. In addition, some patients immunized with GM2 vaccines containing BCG or R595 received a booster imunization between 3 and 5 months after the fourth vaccination.

Results

Animal Trjals

Sera were obtained from mice before and at regular intervals after vaccination, and were tested by the IA and PA assays on JB-RH target cells. No sera were reactive before vaccination. The reactivity of sera obtained after two vaccinations in eight separate ex¬ periments are shown in Table I. Ml reactions were detected by the IA assay (detecting IgM) ; no IgG reac¬ tions were induced. Three broad categories of vaccines were tested: traditional vaccines such as irradiated whole cells, GM2 alone, or GM2 incorporated into com¬ plete Freund's adjuvant, GM2 attached to or mixed with bacteria, and GM2 in liposomes.

Traditional vaccines. GM2 administered in saline, in complete Freund's adjuvant, and in intralipid mixed with MPLA and BCG CWS, either with or without Cy, re- suited in serologic responses in only nine of 65 mice

(median titer 1/20, see Table I) . Immunization with 5 x 10' irradiated JB-RH cells (known to contain 60 mi¬ crograms GM2) plus MPLA in Cy-pretreated mice induced a serologic response in four of 20 mice (median titer 1/20) .

Bacterial adjuvants. J. coli was found to be an inef¬ fective adjuvant (resulting in only two of 10 respons¬ es) . Salmonella minnesota and BCG were moderately effective adjuvants (10 of 25 and eight of 15 respons¬ es, respectively) and the R595 mutant of Salmonella minnesota was a highly effective adjuvant (27 of 30 responses, rredian titer 1/64 including nine mice with titers greater than 1/128) for inducing a serologic

-16-

Table 1

Serologtcal response tn IA tests on JB-RH after two GM2 υacclnations

GM2 Do*-. P reilment No. Mice Median

Vacctne Group* No. Mice Responding Titer 4J •tth Cy 13 tι_~ _ Vaccinated filter * 1/30) (reciprocal)

Unvaccinated 0 — 10 0

JB-RH 15 x 10') -f MPLA 60 + 20 5 20

Saline 50 - 10 2 24

Saline 50 25 3 30

Complete Freund ^' s adjuvant 50 +• 10 3 40

MPLA t- HCG CWS 50 — 10 0

MPLA + BCG CWS 50 + 10 1 20

E. colt (O.S mg] 50 + 10 2 32

E. coll (5 mg] 50 + 10 0

S. mtnn. (0.5 mgj 50 — 5 3 27

S. mtnn. (5) + MPLA (50) 50 +• 5 0

S. mtnn. (5 mgj 50 - 5 0

S. mtnn. (0.5) 50 + 20 7 27

S. mtnn. (5) 50 20 5 80

S. mtnn. R595 (0.05) 50 + 5 2 40

S. mtnn. R595 (0.5) 50 + 30 27* 64

S. mtnn. R595 (0.5) 50 — 15 7 80

S. mfnn. R595 (5) 50 + 15 2 120

BCG 10 ^s 50 X 5 2 40

BCG 10 ^β 50 15 8 40

BCG 10' 50 X 5 1 40

BCG 10* 50 - 5 2 160

Liposomes 1 + MPLA (50) 50 - 5 0

Liposomes 1 50 + 5 0

Liposomes 1 + MPLA (50) 50 + 15 3 15

Liposomes 1 + CP20.961 50 + 5 1 64

Liposomes 2a 50 + 5 0

Liposomes 2a + MPLA (50) 50 + 25 16 40

Liposomes 2a ♦ MOP 50 + 5 1 64

Lipoaomes 2a + ^■ MPLA (SO) 50 — 10 3 80

Lipoaomes 2b 50 -> 5 1 40

Lipoaomes 2a + endotαxiπ ( 10) 50 5 0 40

Lipoaomes 2b + MPLA (SO) 50 -r 15 5 15

Lipoaomes 2b + MPLA (5) 50 + 5 I 40

Lipoaomes 2b + CP20.961 50 + 5 1 20

^• Sutlstlcal significance by the Fisher exact test compared with JB-RH +• MPLA + Cy. p » 0 001 : compared with R595 (0.5) - Cy. p * 0.003: compared with BCG 10* + Cy. p ~ 0.009: compared with liposomes 2a + MPLA x Cy. p » 0.026.

response to GM2 (see Table I). The p values for these compared with GM2 plus saline were 0.03, 0.01, and less than 0.001, respectively. The optimal dose of R595 was 0.5 mg, and Cy pretreatment significantly augmented vaccine immunogenicity.

To determine whether binding of GM2 to adjuvant was required for optimal effect, the binding capacity of R595 and BCG for GM2 was determined. At ^" a dose of 50 micrograms G 2 and 0.3 mg R595, trace amounts of GM2

10 were detected in the supernatant; at higher doses of R595 all of the ganglioside was in the pellet and at lower doses of R595 most of the ganglioside was in the supernatant. The R595 dose found most immunogenic here (0.5 mg) coincides therefore with the least amount

^" 15 of R595 capable of binding the full 50 micrograms of

GM2 to its surface. BCG differed from R595 in that approximately 5% of the bacteria were viable and that at the highest dose tolerated by the mice (10') about 60% of GM2 was not attached to the BCG cell pellet

²⁰ (15,600 x G, 30 min) , but was found in the supernatant.

This persisted over a range of BCG:GM2 ratios. To determine whether G 2 in the supernatant was free or attached to subcellular BCG components, the mixtures were sujected to ultracentrif ugation (300,000 x G, 180 5 min) , and TLC plates were prepared from both pellet and the supernatant. At a ratio of 50 micrograms GM2:10 ⁶ viable BCG U, 90% of the GM2 was in the pellet. At higher BCG doses, all BCG was in the pellet and at lower doses most was in the supernatant. Immuno-

30 staining of these plates with anti-GM2 monoclonal anti¬ body 5-3 confirmed that the bands observed were indeed GM2. These results show that in the most immunogenic BCG and R595 vaccines, most GM2 is bound to the bacte¬ ria or bacterial products, the bacteria are maximally 5

saturated with GM2, and the vaccines contain little free GM2.

Liposoae vaccines. Liposomes 1 and 2b were not very immunogenic (three of 15 and five of 15 responses, respectively; see Table I) , but liposomes 2a were mod¬ erately immunogenic (16 of 25 responses, p = 0.001 compared with GM2 plus saline) . The difference between liposome preparations 1 and 2 was their lipid content, whereas the difference between 2a and 2b was the soni¬ cation method; liposomes 2a, which were sonicated more vigorously, are assumed to be smaller. Our results indicate that it is both the composition of the liposomes and the way in which they are formed that determines their immunogenicity. MPLA was required in the liposomes and could not be substituted for by the lipoidylamiπe CP20,961 or MDP. Endotoxin at a dose of 10 micrograms (the highest dose tolerated by the mice) was not an effective substitute, but MPLA at this dose was also not effective. Higher titler IgG antibody titers such as these were not detected again in the subsequent four experiments. Cy appeared to enhance the immunogenicity of the liposomes 2a-MPLA vaccine (16 of 25 vs three of 10 responses) , but the difference was not statisticlly significant (p = 0.13 by the Fisher exact test) .

Specificity of observed responses. Initially, selected sera with high titer reactivity on JB-RH-16 were tested by ELISA assays against a panel of purified ganglio¬ sides to determine specificity. This analysis was complicated, however, by natural antibodies against GM1 detected before and after vaccination in some mice and "sticky sera" from some mice that resulted in positive reactions in wells containing various gangliosides or

no ganglioside at all. Because we were not able to block this later artifact with bovine serum albumin or normal mouse serum, we analyzed the specificity of five high titer sera against JB-RH cells by inhibition tests (see Table II) . We found that the IA reactivity with JB-RH cells was completely inhibited only by GM2; GDla, GT1, GD3 , and GM3 did not inhibit this reactivi¬ ty. GM1, however, partially inhibited reactivity of two sera from BCG GM2-treated mice.

Sera reactive by IA 2 so mediate co__.pl eaent-de pendent cytotoxicity (CDCX) . Initially, sera reactive in IA tests were also tested for CDCX. Titers of reactivity were similar. In one such experiment, sera from five mice immunized with GM2 alone were nonreactive in both assays, whereas those from four of the five GM2 plus

R595 mice were strongly reactive in both assays; median titer by CDCX for 50% kill was 1/40 and by IA, 1/64. We have detected no significant discrepancies between titers obtained by CDCX and IA, but have relied more heavily on IA because it is more rapid and requires less serum.

Attempts to obtain an IgG response and delayed-type hypersensitivity (DTH) . Although sera induced by vari¬ ous vaccines in these experiments were able to mediate potent CDCX with rabbit complement on JB-RH, no consis¬ tent spontaneous conversion to an IgG response was observed. In an attempt to obtain such a conversion, in one series of experiments 10 mice were revaccinated subcutaneously with R595-GM2 2 mo after the last vacci¬ nation. The median IA titer (IgM) on JB-RH cells in this group increased from 1/80 to 1/160, but no PA or anti-IgG assay reactivity (IgG) was observed. Three additional groups of four mice were given "booster"

Table 2

Inhibition by purified gangliosides of lA reactivity in sera of mice immunized with CM2 vaccines

Vaccines

R595 GM2plu» GM2ph» GM2pfua

InhibtUiLg Alone RS9S BCG Upoaome2a Gangllαudes β I 2 3 4 3 β

Reciprocal antibody titer by !A on JB-RH

None 0 160 320 160 80 80

GM2 10 — _ _ _ _

5 — 20 20 — —

2 20 40 20 — 40

1 20 80 40 — 40 GM1 10 80 160 20 20 40

5 40 160 40 20 40

2 40 160 80

1 80 160 160 40 80 GOla 10 80 320 160 40 80

5 80 160 160 40 80

2 160 ' 80

1 160 160 160 80 80 GT1 10 40 160 80 80 80

5 80 160 80 80 80

1 GM3 10 80 160 160 40 80

5 160 160 160 40 80

1 160 320 160 40 80 GD3 10 80 160 160 40 80

5 160 160 160 40 80

1 160 320 160 40 80

vaccinations with 50 micrograms of GM2 administered i.v. either alone or with R595 or BCG 2 mo ^" after the last vaccination. Once again, the median IA titer increased slightly (from 1/40 to 1/80) but no PA or anti-IgG reactivity was detected. These same groups of mice were also tested for DTH reactivity by footpad injection of 1, 5, and 25 micrograms of GM2. No reac¬ tivity was detected.

Human Trials

Vaccine characteristics and Seriological Response of Vaccinated Patients. Table 3 summarizes the charac¬ teristics of the whole-cell vaccine constructed from three cell lines: a mouse melanoma cell line and mela- noma and astrocytoma cell lines of human origin, These cell lines were selected for high surface expression of GM2, as indicated by reactivity with a mouse mAb de¬ tecting GM2 (2). Five vaccines containing purified GM2 were tested, one with GM2 alone and four with BCG or R595 as adjuvants.. In two of these trials patients were pretreated with low-dose c clophosphamide (Cy)

(200 mg/m ² ) 3 days prior to the initial vaccination.

The whole-cell vaccine, though containing no bacterial adjuvant, resulted in induration and erythema (greater than 8cm in diameter at 48 hr) in 5 of 6 patients and low-grade fever (less than 39°C) in 4 patients. Four patients experienced tenderness and swelling in the draining lymph nodes and 5 of the 6 patients showed prominent hyperplas ia in the resected lymph nodes. These reactions in skin and lymph nodes, which in¬ creased with each vaccination, were not seen in our previous trials with human melanoma cell vaccines. It seems likely, therefore, that this heightened inflamma¬ tory reaction is attributable to an antimouse response.

Table 3

GM2, GD2 AND GD3 COMPOSITION OF WHOLE CELL VACCINE

Cell Surface Expression of Gangliosides

Cell Lines Monoclonal JB-RH-16 SK-MG-14 SK-MEL-3

Antibody Antibody Titers (reciprocal)

5-3 (αGM2) 65,000 65,000 22,000

3F8 (αGD2) 0 25,000 2.5 x 10 ⁶

R24 (αGD3) 0 0 i,000

Ganglioside content of cell lines (yg/10 ⁷ cells)

GM2 12.4 4.5 6.9 GD2 0 5.4 2.0 GD3 0 2. 1 1.0

Ganglioside content of combined vaccine (yg/2.2 x 10 ⁸ cells)

GM2 173.8

GD2 54.3

GD3 22.7

Vaccination with GM2 alone or GM2/R595 was well toler¬ ated; no side effects were detected. GM2/BCG vaccines resulted in low-grade fever (less than 39°C) and marked local ulceration in 5 of 11 patients, requiring a de¬ crease in the BCG dose (3 x 10 organisms) or use of R595 in place of BCG for the booster vaccination. No neurologic or other detectable abnormalities were asso¬ ciated with GM2 vaccination.

Figure 1 and Table 4 show the results of ELISAs for GM2 antibody in serum from normal individuals .and from nonvaccinated and vaccinated melanoma patients. The frequency and titer of GM2 antibody in normal individu¬ als and nonvaccinated melanoma patients were simi¬ lar: 80% were negative and only one normal individual had a titer above 1:40. The whole-cell vaccine induced

GM2 antibody in high titer (1:80 or greater) in 5 of 6 vaccinated patients. No GM2 antibody was induced in patients immunized with GM2 alone. Addition of BCG to the purified GM2 vaccine resulted in GM2 antibody pro- duction, particularly in patients pretreated with Cy or given a booster immunization 12-16 weeks after the last vaccine injection. The effect of Cy was also evident in the case of GM2 vaccines with R595 as the adjuvant; 2 of 6 patients pretreated with Cy produced GM2 anti- body, whereas no GM2 antibody was detected in patients not treated with Cy. R595, in contrast to BCG, was not effective as an adjuvant in booster immunizations. No increase in GM2 titers was found in 4 Cy-treated pa¬ tients given the GM2/R595 vaccine and booster immuniza- tions with GM2/R595 (14 weeks after the last vaccine injection) . Because of increasing local inflammatory and systemic reactions induced by BCG in some patients, 5 of the 11 patients initially vaccinated with GM2/BCG

Table 4

CM2 ANTIBODY TITERS (ELISA) OF NORMAL INDIVIDUALS, UNTREATED MELANOMA PATIENTS, AND MELANOMA PATIENTS AFTER IMMUNIZATION WITH A WHOLE CELL VACCINE OR PURIFIED GM2 VACCINES

No. Patients With a Given Titer Statistical Total No. (Reciprocal) Significance****

Patients 0 20 40 80 160 _ 320- __________

[TREATED -

Normal Individuals 44 37 4 2 1 0 0

Stage HI Melanoma Patients * 50 37 8 3 0 0 0

VACCINATED WITH WHOLE CELLS **

SK-MEL-31, SK-MG-14 and JB-RH-16 3 - 2 <.001

VACCINATED WITH PURIFIED GM2

GM2 Alone 5 4 1 0 0 0 0 1.000

GM2/RS95 5 3 2 0 0 0 0 1.000

Cy + GM2/R595*** 6 0 1 3 0 2 0 .009

GM2/BCG* 5 0 1 1 1 1 1 <.001

Cy + GM2/BCG **★

6 0 1 0 2 2 1 < .001

Serum from two patients excluded from evaluation because of non-specific reactivity (refs. 11,24).

Peak titer observed after vaccination.

***

Including booster vaccine.

**** Fishersexact test of number of vaccinated patients with titers 21/80 compared with 92 untreated controls.

vaccine received booster injections of a GM2/R595 vac¬ cine. Four of the 6 patients given the GM2/BCG booster immunization showed a strong rise in GM2 titer: no increase in GM2 titers was found in the 5 patients given the GM2/R595 booster immunization.

Specificity Analysis of Sera from Vaccinated Patinets.

Sera from the 19 patients with anti-GM2 titers of 1:40 or greater were tested for reactivity with GDla, GM1, GD2, GD3, and GM3 by EL ISA and ITLC. Reactivity was restricted to GM2, with the exception of serum from 1 patient in the whole cell vaccine trial that recognized GD2 at a titer of 1:40. All sera with a titer of 1:80 or higher were also analyzed by ITLC. Reactivity was restricted to GM2. Figure 2 shows ITLC tests with four sera having an anti-GM2 titers ^" of 1:80 or higher by ELISA. Sera with lower anti-GM2 titers could not gen¬ erally be analyzed by ITLC. GM2 antibodies in vacci¬ nated patients belonged to the IgM class; tests with an IgG indicator system revealed no IgG anti-GM2.

Complement- Dependent Cytotoxicity of GM2 Antibodies.

Sera from patients developing high titers of GM2 anti¬ body after vaccination were found to be cytotoxic for GM2 positive target cells in the presence of normal human serum as complement source. GM2-negative target cells were not lysed. Table 5 shows the relation be¬ tween anti-GM2 titers detected by ELISAs and cytotoxicity tests (with SK-MG-6 astrocytoma cells, ref 2) . A positive correlation was seen between antibody titers in both assays.

Table 5

GM2 ANTIBODIES IN SERA OF VACCINATED MELANOMA PATIENTS: COMPARISON OF TITERS DETERMINED BY ELISA AND CYTOTOXIC TESTS WITH HUMAN COMPLEMENT

Median Cytotoxicity Antibody Titer •Median Titer (Range)

By E ISA Number of Optical Density (Range) 20S Lysis 50X Lysis (Reciprocal) Patients Ac 1/40 Serum Dilution End Point End Point

0 8 .04 (.01-.06) N.D. N.D.

20 6 .12 (.08-.23) N .D . N .D.

40 4 .25 (.21-.35) 20 (5-80) 0 (0-5)

80 6 .40 (.38-.84) 80 (20-80) 5 (0-20)

160 7 .66 (.41-1.45) 160 (80-320) 20 (10-80)

320 2 .83 (1.40,1.45) 320 (320) 30 (20,40)

Discussion

Animal Trials

The purpose of these studies was to identify methods for consistently immunizing mice against GM2. Immuni¬ zation with irradiated whole cells expressing GM2 and mixed with a suitable adjuvant would have, been predict¬ ed to be the method of choice. Although no other stud¬ ies have compared the relative immunogenicity of a ganglioside on tumor cells and in vaccines containing purified ganglioside, such studies have been done with protein or glycoprotein antigens by using DTH or tumor rejection as end points (39-41). In these studies, antigens expressed on irradiated whole cells have con- sistently been more immunogenic than the same antigens purified. Our results show, however, that with regard to antibody production against GM2, whole cell vaccines are not very immunogenic, inducing an immune response in only occasional mice. Our results also show that GM2 alone or in complete Freund's adjuvant is not immu¬ nogenic, but that GM2 incorporated into certain liposome preparations or mixed with bacteria such as BCG or especially acid-hydrolyzed R595 is much more immunogenic. In fact, GM2 presented in this way is significantly more immunogenic than the same amount of GM2 expressed on whole tumor cells. A consistent cytotoxic antibody response against tumor cells ex¬ pressing GM2 was induced, a finding we have used to produce a new series of monoclonal antibodies against GM2. In comparing these results with those of investi¬ gators working with DTH or tumor rejection and protein or glycoprotein antigens, the conclusiions only appear to be contraditory. Vaccines containing GM2 alone or with complete Freund's adjuvant are not more immunogen-

ic than those expressing GM2 on irradiated tumor cells either. The difference is in the way the antigen is presented, and this has never been thoroughly explored by using protein or glycoprotein tumor antigens.

Use of acid-hydrolyzed mutant Salmonella minnesota strain R595 as a vehicle for increasing the immuno¬ genicity of glycolipids was initially described by Galanos et al. (27). They found that acid-hydrolyzed R595 or complete Freund's adjuvant in terms of increas- ing the immunogenicity of the lipid A component of bacterial lipopolysaccharides. This approach has since been used by Hakomori and Young et al. to augment the immunogenicity of asialo GM2 (42) and other glycolipid blood group antigens (43) and to produce a series of monoclonal antibodies against blood group and tumor glycolipids (44, 45) . We show here that R595 can also be applied to inducing antibodies against gangliosides such as GM2. Since the studies of Galanos et al. (27) with lipid A, little work has been done on identif iying alternative methods for argumenting the immunogenicity of glycolipids. In the studies we have expanded the variety of approaches tested for inducing antibodies against glycolipids by identifying BCG- and MPLA-con- taining liposomes as two other vehicles worthy of ad- ditional study. The • results of Galanos et al. (27), together with our findings that acid-treated R595 was superior to wild-type Salmonella minnesota or E. coli and that vaccines containing the highest GM2 to R595 ratio were also the most immunogenic, suggest the role that acid-treated R595 plays. Lipopolysaccarides are made up of two distinct regions, the hydrophylic polysaccharide portion consisting of C-specific chains and basal core, and the hydrophobic lipid portion lipid A. Our previous studies with whole cell vaccines (20,

22) and those described here with liposomes have iden¬ tified lipid A as the single most potent adjuvant test¬ ed. The R (rough) mutant 595 contains lipi A but no O-specific chains, and the core polysaccharide consists only of 2-keto-3-dioxyoctonate, which is removed by acid- hydrolysis, as originally described by Galanos

(27). The cell surface of acid-treated R595 is there¬ fore highly hydrophobic and ideal for binding and con¬ centrating added glycolipids such as GM2. GM2 is con¬ centrated on the surface and oriented in such a way that cerramide is imbedded and polysaccharide (the antigen) exposed and in close proximity to bacterial lipid A.

Liposomes are another method of concentrating glycolipids on a membrane with an orientation that enhances immunogenicity and of bringing them into close proximity to selected adjuvants. Our results show the importance of both liposome composition and size. They also show that one adjuvant, MPLA, was most effective in liposome preparations. As opposed to our studies with GM2-coated R595 and BCG in which the variables were limited and quickly addressed, our studies with liposomes have only served to suggest additional areas for study. We are currently investigation the effect of lipid composition, liposome size and charge, unilamellar vs multilamellar structure, incorporation of the antigen and lymphokines on to the liposome sur¬ face as opposed to within the liposome or both, and the use of various single adjuvants or combinations of adjuvants.

Many of the mice immunized in these studies continued to produce antibody for more than 6 o without evidence of neurologic or other toxicity. This is reassuring,

because GM2 is present on a small subpopulation of human and (presumably) B-β astrocytes (1) . Although experimental autoimmune encephalomyelitis (EAE) is generally induced by immunization with the basic pro¬ tein of myelin in a suitable adjuvant such as complete Freund's adjuvant, Nagai et al. (46) have reported that immunization of rabbits with GDIa or GM1 in complete Freund's adjuvant resulted in a syndrome similar to that described for EAE: hind leg paralysis and some¬ times death. Furthermore, Rapport et al. (47) have described grand mal seizures in rabbits after intracerebral injections of anti-GMl antibodies, but not injecitons of antibodies against various other unrelated antigens. Although EAE is generally short lived, requiring continuing immunizations for progres¬ sion of the disease, and although there have been no reports of autoimmune phenomena associated with anti- GM2 antibody activity, we observed immunized mice closely for as long as a full year to confirm that there was no association between anti-GM2 antibody and autoimmune disease.

Human Trials

The identification of melanoma cell-surface antigens that are immunogenic in the host of origin has been the object of our analysis of sera (21) , cytotoxic T cells (28,29), and mAbs (15, 30) derived from melanoma pa¬ tients. Three general categories of melanoma cell- surface antigens that are immunogenic or potentially immunogenic in the autologous host have been defined; these range from highly restricted antigens that are detected only on autologous melanoma cells [class 1 (unique) antigens] , to antigens present on a subset of melanomas as well as a limited range of other cell

types (class 2 antigens), to antigens that are widely distributed on melanomas and other cell types (class 3 antigens) (21) . Biochemical characterization of these melanoma antigens is limited, but class 1 (unique) antigens appear to be glycoproteins (31) and one of the best-studied class 2 melanoma antigens is the ganglioside GD2 (13, 14). As reactivity against these antigens is found in only a small percentage of melano¬ ma patients, we have attempted to induce antibodies to class 1 or class 2 antigens using vaccines of irradiat- ed cells expressing these antigens (16-20) . These human trials have not been successful and prompted us to define conditions required for a consistent humoral immune response to tumor antigens in the mouse, includ¬ ing the class 1 antigens of Meth A sarcoma (22, 23) and the ganglioside GM2 (20, 24) . Adjuvants and pretreat- ment with low doses of Cy were important factors in the mouse studies, and results of the present human trials indicate their important in melanoma patients.

Irie, Tai, Morton and colleagues have also identified the immunogenicity of GM2 in their studies of melanoma patients (11, 12) . They isolated stable cultures of Epstein-Barr virus-transformed B cells from a melanoma patient that produced a mAb to GM2 (12) . In addition, Tai et al. have immunized melanoma patients with mela¬ noma cell vaccines containing a mixture of gangliosides and found that reactivity against GM2 was induced in 10/26 patients (11) . As in the present study, reactiv¬ ity against GD2 was only rarely detected (2/26 pa- tients) in their series and no antibody agaist GD3 or

GM3 was found.

We and others have detected low levels of GM2 antibody in some normal individuals (10) and nonvaccinated stage

III melanoma patients. Natural growth of melanoma in the skin and regional lymph nodes does not appear to be a potent stimulus for generating GM2 antibody, since antibody levels in stage III melanoma patients are no higher than in normal individuals. The fact that most melanoma patients can be induced to develop high levels of GM2 antibody after vaccination is surprising in view of the presence of these gangliosides in brain and other tissues of neuroectodermal origin. The other finding of GM2 vaccines is that high titers of GM2 antibody had no demonstrable ill effect on these pa¬ tients.

There is a suggestion from the present study that mela¬ noma recurrence is delayed in patients developing GM2 antibody. Evaluation of 31 vaccinated patients in this study observed for 15 months showed that 5 of 14 pa¬ tients with GM2 titers less than or equal to 1:20 are disease free, as compared to 14 of 17 patients with GM2 titers greater than or equal to 1:40.

References

1. Cochran, F.B. , Jr., Yu, R. K. & Ledeen, R. ^' W. (1982) J. Neurochem, 39, 773-779.

2. Natoli, E. J. , Jr., Livingston, P.O., Cordon-Cardo,

C, Pukel, C.S., Lloyd, K.O. , Wiegandt, H. , Szalay, J., Oettgen, H.F. & Old, L.J. (1986) Cancer Res. 46, 4116-4120.

3. Cheung, N.K.V., Saarinen, ϋ.M. , Neely, J.E. ,

Landmeier, B., Donovan, D. _ Coccia, P.F. (1985) Cancer Res. 45, 2642-2649.

4. Cheresh, D.A., Harper, J.R. , Schulz, G & Reisfeld, R.A. (1984) Proc. Natl. Acad. Sci. OSA 81, 5767-

5771.

5. Dippold, W.G., Lloyd, K.O. , Li, L.T.C. , Ikeda, H. , Oettgen, H.F. & Old, L.J. (1980), Proc. Natl Acad. sci. USA 77, 6114-6118.

6. Pukel, C.S., Lloyd, K.O. , Travassor, L.R. , Dippold, W.G., Oettgen, H.F. & Old L.J. (1982) J.Exp Med. 155, 1133-1147.

7. Nudelman, E. , Hakomori, S. , Kannagi, R. , Levery, S., Yeh, M.H., Hellstrom, K.E. _ Hellstrom, I. (1982) J. Biol. Che . 257, 12752-12756.

8. Cheresh, D.A. _ Klier, F.G. (1986); J. Cell Biol.

102, 1887-1897.

9. Arnon, P., Crisp, E. , Kelley, R. , Ellison, G.W. , Myers, L.W. & Tourtellotte, W.W. (1980) J. Neurol. Sci. 46, 179-186.

10. Richards, R.L. & Alving, CR. (1980) ACS Sy p. Ser., 461-473.

11. Tai, T. , Cahan, L.D. , Tsuchida, T. , Saxton, R. E. , Irie, R. F. & Morton, D.L. (1985) Int. J. Cancer 35, 607-612.

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

13. 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.

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

15. Yamaguchi, H. , Furukawa, K. , Fortunato, S.R. ,

Livingston, ^' P.O., Lloyd, K.O. , Oettgen, H.F. & Old, L.J. (1987) Proc. Natl. Acad. Sci. USA, in press.

16. Livingston, P.O., Watanabe, T. , Shiku, H. , Houghton, A. N. , Albino, A., Takahashi, T. ,

Resnick, A., Michitsch, R. , Pinsky, CM., Oettgen, H.F. & Old, L.J. (1982) Int. J. Cancer 30, 413-422.

17. Livingston, P.O., Takeyama, H., Pollack, M.S., Houghton, A.N. , Albino, A., Pinsky, CM., Oettgen,

H.F. & Old, L.J. (1983) Int. J. Cancer 31, 567-575.

18. Livingston, P.O., Albino, A.P., Chung, T.J.C, Real, F.X., Houghton, A.N. , Oettgen, H.F. & Old, L. j. (1985) Cancer 55, 713-720.

19. Livingston, ?.0., Kaelin, E. , Pinsky, CM., Oettgen, H.F. & Old, L.J. (1985) Cancer -56, 2194- 2200.

20. Livingston, P.O. (1985) UCLA Symp. Mol. Cell. Biol. New Ser. 27, 537-548.

21. Old, L.J. (1981) Cancer Res. 41, 361-3-68.-

22. Livingston, P.O., DeLeo, A.B., Jones, M., Oettgen,

H.F. & Old, L.J. (1983) J. Immunol. 131, 2601.

23. Livingston, P.O., Jones, M. , DeLeo, A.B., Oettgen,

H.F. & Old, L.J. (1985) J. Immunol. 135, 1505-1509.

24. Livingston, P.O., Jones Calves, M. & Natoli, E. J. , Jr. (1987) J. Immunol., in press.

25. Watanabe, T. & Livingston, P.O. (1981) in AACHT Tissue Typing Manual, eds. Zachary, A. A. & Braun,

W.E. (AACHT, New York), pp. 1-21-1-1-21-4.

26. Berkelhammer, J. , Oxenhandler, R.W. , Hook, R.R. , Jr., & Hennessy, J.M. (1982) Cancer Res. 42, 3157- 3163.

27. Galanos, C, Luderitz, 0. _ Westphal, 0. (1971) Eur. J. Biochem. 24, 116-122.

28. Livingston, P.O., Shiku, M. , Bean, M.A. , Pinsky,

CM., Oettgen, H.F. & Old, L.J. (1979) Int. J. Cancer 24, 34-44.

29. Knuth, A., Danowski, B., Oettgen, H.F. _ Old, L.J. (1984) Proc. Natl. Acad. Sci. USA 81, 3511-3515.

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

31. Real, F.X., Jules Mattes, M. , Houghton, A.N. , Oettgen, H.F., Lloyd, K.O. & Old, LΪJ. - (1984) J. Exp. Med. 160, 1219-1233.

32. Svennerholm, L. (1964) J. Lipid. Res. 5, 145-155.

33. Cheresh, D.A. , Honsik, C.J., Staff ileno, L.K. , Jung, G. & Reisfeld, R.A. (1985) Proc. Natl. Acad. Sci. USA 82:5155.

34. Livingston, P.O., Natoli, E. J. , Calves, M.J., Stockert, E. , Oettgen, H.F. & Old, L.J. (1987) Proc. Natl. Acad. Sci. USA. In press.

35. 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

36. Cheung, N-K, Berger, N. , Coccia, P., Kallick, S. , Lazarus, H. , Miraldi, F. , Saarinen, ϋ. & Strandjord, S. (1986) Proc. Am. Assoc. Cancer Res. 27:318.

37. Shiku, H. , Takahashi, T. , Oettgen, H.F. & Old, L.J. (1976) J. Exp. Med. 144:873.

38. ϋeda, R. Shiku, H., Pf reundschuh, M., et al. (1979) J. Exp. Med. 150:564.

39. Saunders, T.L., Kahan, B.D. & Pellis, N.R. (1985) Cancer Immunol. Immunother. 19:22.

40. Ullrich, S.J. , Robinson, E.A. & Appella, E. (1986) Meth A. Mol. Immunol. 23:545.

41. Rogers, M.J., Galetto, G. , Hearing, V.J., Siwarski, D.F. _ Law, L.W. (1984) J. Immunol. 132:3211.

42. Hakomori, S. & Kannagi, R. (1983) J. Clin. Immunol. 71:232.

43. Hakomori, S. & Young, W.W. (1978) Scand. J. Immunol. 7:97.

44. Young, W.W. , Jr., Portoukalian, J. & Hakomori, S- I. (1981) J. Biol. Chem. 256:10967.

45. Hakomori, S. (1985) Cancer Res. 45:2405.

46. Nagai, Y., Momoi, T., Saito, M. , Mitsuzawa, E. & Ohtani, S. (1976) Neurosci. Lett. 2:107.

47. Rapport, M.M, Karpiak, S.E. & Mahadik, S.P. (1979) Fed. Proc. 38:2391.