Radiolabeled fragments of monoclonal antibodies against carcinoembryonic antigen for localization of human colon carcinoma grafted into nude mice.

Four monoclonal antibodies against carcinoembryonic antigen (CEA) have been selected from 32 hybrids that produce antibodies against this antigen, by the criteria of high affinity for CEA and low cross-reactivity with granulocyte glycoprotein(s). The specificity of tumor localization in vivo of the four MAb, and their F(ab')2 and Fab fragments was compared in nude mice bearing grafts of a serially transplanted, CEA-producing, human colon carcinoma. The distribution of radiolabeled MAb and their fragments after intravenous injection was analyzed by direct measurement of radioactivity in tumor and normal organs, as well as by whole-body scanning and by autoradiography of tumor sections. Paired labeling experiments, in which 131I-labeled antibody or fragments and 125I-labeled control IgG are injected simultaneously, were undertaken to determine the relative tumor uptakes of each labeled protein. The tumor antibody uptake divided by that of control IgG defines the specificity index of localization. Tumor antibody uptakes (as compared with the whole mouse), ranging between 7 and 15, and specificity indices ranging between 3.4 and 6.8, were obtained with the four intact MAb at day 4-5 after injection. With F(ab')2 fragments of the four MAb, at day 3, the tumor antibody uptakes ranged between 12 and 24 and the specificity indices between 5.3 and 8.2. With the Fab fragments prepared from the two most promising MAb, the antibody uptakes reached values of 34 and 82 at day 2-3 and the specificity indices were as high as 12 and 19. The scanning results paralleled those obtained by direct measurement of radioactivity. With intact MAb, tumor grafts of 0.5-1 g gave very contrasted positive scans 3 d after injection. Using MAb fragments, tumors of smaller size were detectable earlier. The best results were obtained with Fab fragments of MAb 35, which gave clear detections of tumors weighing only 0.1 g as early as 48 h after injection. Autoradiographs of tumor sections from mice injected with 125I-labeled MAb demonstrated that the radioactivity was localized in the tumor tissues and not in the stromal connective tissue of mouse origin. The highest radioactivity concentration was localized in areas known to contain CEA such as the pseudolumen of glands and the apical side of carcinoma cells. The penetration of radioactivity in the central part of tumor nodules and the pseudolumen appeared to be increased with the use of MAb fragments.

V beta-specific stimulation of human T cells by staphylococcal toxins.

The staphylococcal toxins are responsible for a number of diseases in man and other animals. Many of them have also long been known to be powerful T cell stimulants. They do not, however, stimulate all T cells. On the contrary, each toxin reacts with human T cells bearing particular V beta sequences as part of their receptors for major histocompatibility complex protein-associated antigen. The specificity of these toxins for V beta s puts them in the recently described class of superantigens and may account for the differential sensitivity of different individuals to the toxic effects of these proteins.

Common acute lymphoblastic leukemia antigen expressed on leukemia and melanoma cell lines has neutral endopeptidase activity.

We have previously reported that the amino acid sequence of the common acute lymphoblastic leukemia antigen (CALLA, CD10) translated from a normal human kidney cDNA clone is identical to that of neutral endopeptidase (NEP, EC 3.4.24.11). In this study, we show that by flow cytometry, a monoclonal antibody (135A3) produced against rabbit NEP reacted selectively with leukemia and melanoma cell lines expressing CALLA on their surface. A glycoprotein of apparent Mr 100,000 was immunoprecipitated from surface labeled NALM-1 leukemia or Mel-1477 melanoma cells with monoclonal antibodies to NEP (135A3) or CALLA (44C10). mRNAs hybridizing to a NEP-specific probe were present in CALLA+ leukemia and melanoma cell lines, but absent from CALLA- lines. NEP enzymatic activity was detected on intact cells from CALLA+ lines, but not CALLA- lines. The activity was blocked by two selective inhibitors of NEP, thiorphan and phosphoramidon. CALLA antigen purified from the NALM-6 leukemic cell line by affinity to 44C10-IgG Sepharose retained a peptidase activity that was completely blocked by thiorphan and phosphoramidon. Thus the CALLA antigen present at the surface of leukemia and melanoma cell lines is an enzymatically active neutral endopeptidase.

A common neoepitope is created when the reactive center of C1-inhibitor is cleaved by plasma kallikrein, activated factor XII fragment, C1 esterase, or neutrophil elastase.

The reactive center of C1-inhibitor, a plasma protease inhibitor that belongs to the serpin superfamily, is located on a peptide loop which is highly susceptible to proteolytic cleavage. With plasma kallikrein, C1s and beta-Factor XIIa, this cleavage occurs at the reactive site residue P1 (Arg444); with neutrophil elastase, it takes place near P1, probably at residue P3 (Val442). After these cleavages, C1-inhibitor is inactivated and its conformation is modified. Moreover, in vivo, cleaved C1-inhibitor is removed from the blood stream more rapidly than the intact serpin, which suggests that proteolysis unmasks sites responsible for cellular recognition and the uptake of the cleaved inhibitor. In the study reported here, we show, using an MAb, that an identical neoepitope is created on C1-inhibitor after the cleavage of its exposed loop by plasma kallikrein, C1s, beta-Factor XIIa, and by neutrophil elastase.

Ablation of human colon carcinoma in nude mice by 131I-labeled monoclonal anti-carcinoembryonic antigen antibody F(ab')2 fragments.

Pooled F(ab')2 fragments of three MAbs against distinct epitopes of carcinoembryonic antigen (CEA) were used for radioimmunotherapy of nude mice bearing a subcutaneous human colon carcinoma xenograft. 9-10 d after transplantation when tumor nodules were in exponential growth, 36 mice were treated by intravenous injection of different amounts of 131I-labeled MAb F(ab')2. All 14 mice injected with a single dose of 2,200 (n = 10) or 2,800 microCi (n = 4) showed complete tumor remission. 8 of the 10 mice treated with 2,200 microCi survived in good health for 1 yr when they were killed and shown to be tumor free. Four of nine other mice treated with four fractionated doses of 400 microCi showed no tumor relapse for more than 9 mo. In contrast, all 15 mice injected with 1,600-3,000 microCi 131I-control IgG F(ab')2 showed tumor growth retardation of only 1-4 wk, and 15 of 16 mice injected with unlabeled anti-CEA MAb F(ab')2 showed unmodified tumor progression as compared with untreated mice. From tissue radioactivity distributions it was calculated that by an injection of 2,200 microCi 131I-MAb F(ab')2 a mean dose of 8,335 rad was selectively delivered to the tumor, while the tissue-absorbed radiation doses for the normal organs were: peripheral blood, 2,093; stomach, 1,668; kidney, 1,289; lung, 1,185; liver, 617; spleen, 501; small intestine, 427; large intestine, 367; bone, 337; and muscle, 198. These treatments were well tolerated since out of 19 mice with complete tumor remission only 4 required bone marrow transplantation and 17 were in good health for 6-12 mo of observation. The results demonstrate the selective destruction of established human colon carcinoma transplants by intravenous injection of either single or fractionated doses of 131I-MAb F(ab')2.

Detection of the common acute lymphoblastic leukemia antigen in the serum of leukemia patients.

The common acute lymphoblastic leukemia antigen (CALLA) has been detected in biological fluids using a radioimmunoassay based on the inhibition of binding of 125I-labeled monoclonal anti-CALLA antibody to glutaraldehyde-fixed NALM-1 cells. With this assay, we showed first that CALLA was released in culture fluids from NALM-1 and Daudi cell lines but was absent from culture fluids from CALLA negative cell lines. Then, we found that the sera of 34 out of 42 patients (81%) with untreated common acute lymphoblastic leukemia (c-ALL) contained higher CALLA levels than any of the 42 serum samples from healthy controls. The specificity of these results was further demonstrated by testing in parallel the sera from 48 patients with CALLA negative leukemias, including 26 acute myeloid leukemia (AML), 12 T-cell acute lymphoblastic leukemia (T-ALL), and 10 acute undifferentiated leukemia (AUL). All of these sera gave negative results, except for one patient with AUL, who had a significantly elevated circulating CALLA level, and one patient with AML, who had a borderline CALLA level, 3 SD over the mean of the normal sera. Preliminary results suggest that circulating CALLA is associated with membrane fragments or vesicles, since the total CALLA antigenic activity was recovered in the pellet of the serum samples centrifuged at 100,000 g. In addition, the CALLA-positive pellets contained an enzyme considered as a membrane marker, 5'-nucleotidase. Evaluation of the clinical importance of repeated serum CALLA determinations for the monitoring of c-ALL patients deserves further investigation.

Detection of new cross-reacting carcinoembryonic antigen(s) on cultured tumor cells by mixed hemadsorption assay.

Antisera highly specific for carcinoembryonic antigen (CEA) from New Zealand White rabbits and a goat reacted strongly in antibody binding tests with cultured tumor cell lines, irrespective of the ability of the cell lines to produce CEA. The most reactive were colon carcinoma and melanoma cell lines, the former known to produce CEA and the latter not associated with CEA production. The reactivity was not diminished by absorption with perchloric acid extracts of normal lung or spleen, whereas absoprtion with purified CEA preparations abolished the reactivity. Quantitative absorption studies indicated that reactivity against CEA-producing cell lines could be totally removed by absorption with other CEA-producing lines but not with melanoma cell lines. Reactivity against melanoma cell lines could be completely removed by colon carcinoma cells as well as by melanoma cells. Antisera raised against purified CEA, after absorption with extracts of normal lung, still contained two populations of antibodies, one that binds a newly described antigen cross-reacting with CEA which is present on melanoma cells.

Swine monoclonal antibodies of high affinity and specificity to carcinoembryonic antigen.

To avoid the exclusive use of rodent monoclonal antibodies (MAbs) in patients for the detection of tumors by immunoscintigraphy and for radioimmunotherapy, swine MAbs were produced that are directed against carcinoembryonic antigen (CEA). Spleen cells from 2 pigs immunized with purified colon carcinoma CEA were fused with a nonsecreting mouse myeloma cell line by conventional methods, except that a particularly long immunization protocol and large amounts of spleen and myeloma cells were used. Of 1,200 growing hybrids tested, 20 were found initially to produce antibodies binding to radiolabeled CEA. Seven stable clones producing anti-CEA MAbs for more than 6 months were derived from these hybrids by repeated subcloning. The pig origin of the seven MAbs was demonstrated in a solid-phase CEA enzyme immunoassay where anti-pig immunoglobin (Ig) antibodies coupled to peroxidase gave a positive reaction while anti-mouse Ig antibodies were entirely negative. All swine MAbs were of the IgG isotype. Three anti-CEA MAbs showed no cross-reactivity with granulocytes, while four others gave various degrees of reactivity with different granulocyte glycoproteins. Competitive binding to CEA performed for two purified swine MAbs showed that they recognized two different epitopes. The affinity constants measured for these two MAbs by Scatchard plot on purified CEA were high (1.2 X 10(9) and 1.2 X 10(10) liter/mol). One of the MAbs was tested in vivo for tumor localization by injection, after radiolabeling, in nude mice bearing human colon carcinoma xenograft. High ratios of tumor to normal tissue were obtained with mean values of 10.5 for intact MAbs and of 26.8 for F(ab')2 fragments of the porcine MAb. The results showed that heterofusion with this particular protocol can be used to produce swine MAbs of high affinity and specificity for a well-defined tumor marker. These reagents may have an important clinical utility, particularly in patients who became sensitized to mouse immunoglobulins.

Heterogeneity of the induction of HLA-DR expression by human immune interferon on glioma cell lines and their clones.

The modulation of HLA-DR and HLA-A, -B, and -C by human recombinant immune interferon (IFN-gamma) was studied on 10 malignant glioma cell lines established in our laboratory, on 8 clones or subclones derived from these lines, and on a fetal astrocyte cell line. Comparative studies were performed with recombinant leukocyte interferon (IFN-alpha). The results not only confirmed the selective activity of IFN-gamma on the modulation of HLA-DR expression, as opposed to that of IFN-alpha, but also demonstrated a marked heterogeneity in the response of glioma cell lines and their clones to the two types of IFN tested. For example, all 3 clones of an inducible cell line could be modulated to express HLA-DR, whereas only 2 of 5 clones derived from a noninducible line were modulated. This heterogeneity did not seem to be due to the absence of the receptor for IFN-gamma on the surface of these cells, since almost all of the cell lines or clones tested (17 of 19) responded to IFN-gamma by the induction or enhancement of the expression for either HLA-DR or HLA-A, -B, and -C (or both). The heterogeneity of induction was also demonstrated between clones derived from a glioma line that did not express HLA-DR after IFN-gamma treatment. The production of HLA-DR by one of the clones was abundant enough to be confirmed by immunoprecipitation and sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis.

Establishment of a cell line (Co-115) from a human colon carcinoma transplanted into nude mice.

A human colon carcinoma cell line, Co-115, has been established in vitro from solid xenografts maintained in nude mice and subcultured for 95 passages. Co-115 cells grow in vitro as tightly packed, epithelial-like colonies, have a doubling time of about 36 hr, have a relatively low plating efficiency in agar, and release significant amounts of carcinoembryonic antigen to the culture medium. Their epithelial nature has been confirmed by ultrastructural examination. The injection of Co-115 cells into nude mice reinduced the formation of solid tumor masses that could be retransplanted and showed a morphology comparable of that of the original xenograft.

Highly selective recognition of human neuroblastoma cells by mouse monoclonal antibody to a cytoplasmic antigen.

Immunization of a BALB/c mouse with cells from the human neuroblastoma line LAN-1 and fusion of the spleen cells with mouse myeloma cells, NS-1, led to the production of a monoclonal antibody (Mab) with a rather unique reactivity for neuroblastoma. This Mab, named 5 A7, detects an antigen of an apparent molecular weight of 65,000-67,000, localized mainly in the cytoplasm and released into culture medium, as revealed by immunoprecipitation and immunoblotting experiments. By immunoperoxidase staining using a biotin-avidin technique, Mab 5 A7 demonstrates a restrictive staining for neuroblastoma cell lines. Following extensive testing on freshly frozen specimens of neuroblastoma and other tumors, Mab 5 A7 shows a highly selective reactivity for neuroblastomas (13 of 14) and some cells of one primitive neuroectodermal tumor (ependymoblastoma). No reactivity could be detected with Mab 5 A7 on the 57 other tumor tissues tested. Among the normal fetal or adult tissue specimens tested, positive staining is found only on adult brain, colonic crypts, some renal tubules, and fetal medulla of the adrenal gland. Among bone marrow specimens tested, only those infiltrated by neuroblastoma cells gave a positive staining. Normal or malignant hematopoietic cells showed no reactivity with Mab 5 A7. Our results with Mab 5 A7 suggest that this reagent not only provides a valuable probe for the immunohistological diagnosis of neuroblastoma on fresh tumor specimens but also allows the detection of bone marrow infiltration by neuroblastoma cells.

Antibody-dependent cell-mediated cytolysis of human colon carcinoma cells induced by specific antisera against carcinoembryonic antigen.

Antibody-dependent lymphocyte cytotoxicity against human colon carcinoma cells grown in vitro was demonstrated with rabbit anti-carcinoembryonic antigen (CEA) antisera and normal human lymphocytes. The same antisera produced no tumor cell lysis in a complement-dependent cytotoxicity test. The specificity of the reaction was demonstrated by the inhibition of antibody-dependent lymphocyte cytotoxicity after the addition of increasing amounts of purified CEA to the antiserum and by the fact that only tumor cell lines expressing CEA on their surface were lysed. Antibody-dependent lymphocyte cytotoxicity was also observed against two colon carcinoma cell lines that expressed Blood Group A antigen, using a human serum containing anti-Blood Group A antibodies of the immunoglobulin G class. This reaction was specifically inhibited by absorption with Blood Group A red cells, whereas the anti-CEA-dependent cytotoxicity was not inhibited by absorption with red cells of different blood groups.

Monoclonal antibodies to carcinoembryonic antigen: ionic strength as a factor in the selection of antibodies for immunoscintigraphy.

As part of an ongoing effort to improve the technique of immunoscintigraphy for the detection of human carcinomas with radiolabeled monoclonal antibodies (MABs) to carcinoembryonic antigen (CEA), we have developed a series of MABs to CEA and have studied the effects of low- and physiological molarity buffers on their CEA binding and affinity, as well as their cross-reactivity with granulocyte glycoprotein(s). These in vitro results in different buffer systems were then correlated with the use of these MABs to CEA in the detection of human colon carcinoma grafts in nude mice. Our results show that the binding of CEA by some MABs is influenced by ionic strength and that this may be an important factor in their successful use for the immunolocalization of carcinomas in vivo.

Common human melanoma-associated antigen(s) detected by monoclonal antibodies.

Hybridoma cells have been derived from a fusion between mouse myeloma cells (P3-NSI/1Ag4) and spleen cells from a mouse immunized with membrane-enriched fractions from the human melanoma cell line Me-43. Of the 26 hybrids obtained, seven secreted antibodies which reacted with the melanoma cell line used for immunoassay. The specificity of the antibodies produced by the seven positive hybrids was further investigated on 16 melanoma cell lines, 15 other tumors, and 14 lymphoblastoid cell lines. The antibodies from four positive hybrids showed a broad reactivity, whereas those from three hybrids reacted exclusively with melanoma cells. The antibodies from two of these three hybrids, alpha-Mel/5 and alpha-Mel/14, seem to be directed against common melanoma antigen(s) since they reacted with all (with one exception) of the 16 melanoma cell lines tested only with five of the 16 melanoma lines. Reciprocal binding inhibition tests using [3H]leeucine-labeled antibodies showed that alpha-Mel/5 and alpha-Mel/14 antibodies were directed against different antigenic determinants.

Human glioma-associated antigens detected by monoclonal antibodies.

Hybridoma cells were derived from a fusion between mouse P3x63/Ag8 myeloma cells and spleen cells from a mouse immunized with whole cells of a human malignant glioma line. Of 345 hybrids obtained, 36 secreted antibodies that reacted with the glioma cell line used for immunization as assayed by an indirect antibody-binding radioimmunoassay. After a first screening for the absence of reactivity on two nongliogenous cell lines, 3 hybrids were selected and cloned by limiting dilution. The specificity of these monoclonal antibodies was then investigated on a panel of 18 cell lines derived from human malignant gliomas, 18 cell lines from nongliogenous neoplasms, as well as normal peripheral blood lymphocytes, normal skin fibroblasts, and normal spermatozoa. The monoclonal antibodies from two positive hybrids, BF7 and GE2, reacted exclusively with glioma cells and appeared to be directed against common malignant glioma antigen(s). BF7 antibodies bound to 13 and GE2 to 17 of 18 glioma cell lines. The third monoclonal antibody, CG12, showed a broad reactivity since it bound to 10 of 18 glioma lines, five of five melanoma lines, and one of one neuroblastoma line. Absorption with normal adult and fetal brain homogenate did not modify the binding capacity of BF7 and GE2 for glioma cells, while the binding of CG12 antibodies was abolished. Reciprocal binding inhibition tests using [3H]leucine-labeled antibodies showed that BF7, GE2, and CG12 antibodies were directed against different antigenic determinants.

Characterization of receptors for alpha-melanocyte-stimulating hormone on human melanoma cells.

Receptors for alpha-melanocyte-stimulating hormone (alpha-MSH) on human malignant melanoma cell lines were investigated with a specific binding assay and characterized with structural analogues of alpha-MSH and adrenocorticotropic hormone and by photoaffinity cross-linking of the hormone-receptor complex. Specific binding of high-performance liquid chromatography-purified, monoiodinated alpha-MSH in the presence of 1 mM 1,10-phenanthroline as protease inhibitor was highest after a 2-h incubation at 37 degrees C. The nonspecific binding was less than 20% and dissociation of the ligand-receptor complex was relatively slow. Ten out of 12 human cell lines showed specific binding sites for alpha-MSH with Kp values ranging from 0.195 to 2.87 nM and the sites/cell being approximately 400 to approximately 1600. Virtually identical results were obtained in an assay where the cells remained attached to the culture dishes during the entire experiment. The study of hormone analogues with the D10 cell line showed that oxidized alpha-MSH had an approximately 40-fold lower affinity than alpha-MSH whereas [Nle4,D-Phe7]-alpha-MSH displayed a threefold and the adrenocorticotropic hormone fragments (1-17) and (1-24) a 20- and 8-fold higher affinity. Cross-linking of the alpha-MSH-receptor complex of three cell lines using monoiodinated [Nle4,D-Phe7,Trp(2-nitro-4-azidophenylsulfenyl)9]-alpha-MSH as photoaffinity label revealed a major Mr 45,000 protein band on sodium dodecyl sulfate-polyacrylamide gels, analogous to the MSH receptor of mouse B16 melanoma cells.

Comparative localization of murine monoclonal antibody Me1-14 F(ab')2 fragment and whole IgG2a in human glioma xenografts.

Monoclonal antibodies (MAbs) targeted to glioma-associated antigens may allow the selective delivery of imaging and therapeutic agents to brain tumors; the use of MAb fragments may be a strategy to further improve tumor uptake of such agents relative to normal tissues. In this study, we have examined the in vivo localization of radioiodinated MAb Me1-14, a murine immunoglobulin G2a (IgG2a) reactive with gliomas, and its F(ab')2 fragment in s.c. and intracranial xenografts of human glioma cell line D-54 MG in athymic mice. The radiolabeled F(ab')2 fragment of Me1-14 was demonstrated to possess in vitro binding affinity and immunoreactivity comparable to that of whole IgG. Direct comparison of IgG and F(ab')2 biodistribution in s.c. xenograft-bearing mice showed higher tumor: normal tissue ratios of the F(ab')2 fragment compared to IgG. In intracranial tumor-bearing mice paired-label analysis using a nonspecific protein control showed earlier specific tumor localization by the F(ab')2 fragment of Me1-14 compared to IgG. Blood-to-tumor transfer constants (K1) derived for Me1-14 F(ab')2 were significantly greater than those for whole Me1-14 (P = 0.01). Estimated radiation dosimetry revealed that 131I-labeled Me1-14 F(ab')2 would deliver higher radiation doses to tumor than to normal tissues. These studies demonstrate that the F(ab')2 fragment of Me1-14 may be a potential agent for immune-directed brain tumor diagnosis and therapy.

Tumor localization in patients by radiolabeled monoclonal antibodies against colon carcinoma.

A radiolabeled monoclonal antibody (MAb) that has been shown to react specifically in vitro and ex vivo to human colorectal carcinoma and to inhibit growth of human carcinomas grafted in nude mice was administered to 52 colorectal carcinoma patients and 15 patients with other types of cancer. Of 63 colorectal carcinoma tumor sites studied, 34 showed significant accumulation of antibody by external photoscanning and tomoscintigraphy, whereas none of the 20 sites of other cancer types gave positive results. One-third of the patients received F(ab')2 fragments of the MAb, which gave a slightly higher percentage (61%) of positive results than did intact MAbs (51%). A few patients scheduled for tumor resection were given injections simultaneously of 131I-labeled MAb and 125I-labeled normal immunoglobulin G. Antibody concentration in resected tumors was 3.6 to 6.3 times higher than the average antibody concentration in adjacent normal tissues (1.5, 3.4, and 9.4 as compared with normal mucosa, serosa, and fat, respectively), and the specificity indices, calculated by differential radioactivity analysis, ranged from 2.1 to 5.1. The results show the potential value and limitations of this particular MAb for tumor detection by immunoscintigraphy.

Radioimmunotherapy of intracerebral human glioma xenografts with 131I-labeled F(ab')2 fragments of monoclonal antibody Mel-14.

The administration of radiolabeled monoclonal antibodies to improve the treatment of malignant gliomas is dependent upon achieving effective tumor radiation dose while sparing normal tissues. We have evaluated the efficacy of 131I-labeled F(ab')2 fragment of monoclonal antibody Mel-14, an IgG2a reactive with the chondroitin sulfate proteoglycan antigen of gliomas, melanomas, and other neoplasms, in prolonging survival of athymic mice transplanted intracerebrally with D-54 MG human glioma xenografts. Studies indicated that in vitro immunoreactivity, affinity, and tumor localization in vivo of radiolabeled Mel-14 F(ab')2 were maintained at specific activities of 10-13 microCi/micrograms. Intravenous injection of 1500 microCi/115 micrograms or 2000 microCi/154 micrograms 131I-labeled Mel-14 F(ab')2 into mice 6-7 days after xenograft implantation resulted in significant survival prolongation over control animals (P = 0.009 using Wilcoxon rank sum analysis). In another experiment, 1500 microCi/126 micrograms 131I-labeled Mel-14 F(ab')2 improved survival significantly over controls (P = 0.006), while 1500 microCi/220 micrograms 131I-labeled nonspecific antibody did not (P = 0.2). Increasing the injected radiation dose to 3000 microCi 131I-labeled Mel-14 F(ab')2 did not significantly increase survival in tumor-bearing mice, because of supervening radiation toxicity. However, giving 3000 microCi 131I-labeled Mel-14 F(ab')2 in two doses of 1500 microCi, 48 h apart, did significantly prolong animal survival over controls (P = 0.001). Estimated radiation dose to tumor was 915 rad after injection of 3000 microCi 131I-labeled Mel-14 F(ab')2 in two doses, a dose higher than that delivered to normal tissues. The results of this study suggest that radiolabeled Mel-14 F(ab')2 be evaluated as an agent for radioimmunotherapy trials.

Melanoma-reactive human cytotoxic T lymphocytes derived from skin biopsies of delayed-type hypersensitivity reactions induced by injection of an autologous melanoma cell line.

The expression by melanomas of multiple antigens that are recognized by specific MHC class I-restricted CTLs has been clearly demonstrated. The goal of many immunotherapy protocols being developed is, therefore, the induction and/or augmentation of CTLs specific for such antigens. One approach has been to immunize using irradiated autologous melanoma cells. Responses to this type of immunization and others are often subsequently measured by delayed-type hypersensitivity (DTH) reactions. The aim of this work was to characterize whether specific CTL responses occur at such DTH sites. Cutaneous DTH reactions were observed following injection of irradiated autologous melanoma cells expressing known tumor antigens. We isolated lymphocytes from biopsies of DTH reaction sites and could measure melanoma-specific CTL activity after 2-3 weeks of culture. The T-cell receptor-Vbeta repertoire of the cultured lymphocytes, assessed by flow cytometry, was highly skewed in both the CD4(+) and CD8(+) T-cell subsets. The repertoires were different among cultures derived from independent biopsies of simultaneous or subsequent DTH reaction sites and very different to that of fresh peripheral blood lymphocytes (PBLs) or PBLs cultured under the same conditions. No particular T-cell expansions dominated several DTH reaction sites, nor could they be detected in PBLs. It appears that T-cell responses to this type of immunization may be limited to the local microenvironment. Establishing the value of DTH reactions in determining levels of systemic antitumor immunity requires further investigation; however, such reactions may indicate a patient's competence to mount an antitumor immune response and enable the isolation of tumor-specific CTLs for use in tumor antigen identification.