Gangliosides have been considered as potential targets for immunotherapy because they are overexpressed on the surface of melanoma cells. However, immunization with purified gangliosides results in a very poor immune response, usually mediated by IgM antibodies. To overcome this limitation, we immunized mice with R24, a monoclonal antibody (mAb) that recognizes the most tumor-restricted ganglioside (GD3); our goal was to obtain anti-idiotype (Id) antibodies bearing the internal image of GD3. Animals produced anti-Id and anti-anti-Id antibodies. Both anti-Id and anti-anti-Id antibodies were able to inhibit mAb R24 binding to GD3. In addition, the anti-anti-Id antibodies were shown to recognize GD3 directly. Anti-Id and anti-anti-Id mAb were then selected from two fusion experiments for evaluation. The most interesting finding emerged from the characterization of the anti-anti-Id mAb 5.G8. It was shown to recognize two different GD3-expressing human melanoma cell lines in vitro and to mediate tumor cell cytotoxicity by complement activation and antibody-dependent cellular cytotoxicity. The biological activity of the anti-anti-Id mAb was also tested in a mouse tumor model, in which it was shown to be a powerful growth inhibitor of melanoma cells. Thus, activity of the anti-anti-Id mAb 5.G8 matched that of the prototypic anti-GD3 mAb R24 both in vitro and in vivo. Altogether, our results indicate that the idiotype approach might produce high affinity, specific and very efficient antitumor immune responses. (Cancer Sci 2011; 102: 64–70)
Melanomas and other tumors of neuroectodermal origin have a distinct profile of cell-surface ganglioside expression.(1) The relevance of these carbohydrate antigens as immune targets in cancer cells can be inferred from earlier studies that described the ability of monoclonal antibodies (mAb) raised against gangliosides to induce complement-dependent cytotoxicity (CDC) in melanoma, neuroblastoma, sarcoma and astrocytoma cell lines.(2) Generally, immunization with whole tumor cells or cell lysates induces low titer IgM antibodies to carbohydrate antigens, but the use of conjugated vaccines can produce high titers of IgM and IgG antibodies.(3) The best vaccine design involves the conjugation of the antigen to keyhole limpet hemocyanin (KLH) and the use of saponins QS-21 and GPI-0100 as adjuvants.(4)
In contrast to normal cells, transformed melanocytes abundantly express disialoganglioside 3 (GD3). R24 is a mouse mAb that specifically recognizes GD3 and mediates in vitro effector functions such as CDC and antibody-dependent cellular cytotoxicity (ADCC).(5) It also stimulates proliferation of GD3-expressing T cells derived from human peripheral blood(6) and was shown to enhance lymphocyte RNA expression of IL-4, IL-10 and IFN-γ.(7) Normal melanocytes are not lysed by the R24-directed immune response due to their low GD3 expression. In contrast to the potent in vitro activity of R24, its effect in nu/nu mice bearing human melanoma grafts is much more modest; tumor inhibition was observed only when R24 treatment started within 3 days of tumor cell inoculation and no effect was shown on established tumors.(8) As a single agent, R24 was shown to induce clinical responses in patients with metastatic melanoma, including complete remissions.(9,10) Nevertheless, the dose-dependent toxicity of R24 can be substantial and constitutes a serious limitation for its clinical use.(11–14)
Anti-idiotype (anti-Id) antibodies that mimic a defined antigenic epitope are a relatively unexplored albeit potentially useful therapeutic tool. We hypothesized that an anti-Id antibody, being a protein, should be more immunogenic than GD3, a thymus-independent antigen. In support of this assumption, promising results have been reported with anti-Id antibodies in the context of several experimental models based on pathogen-derived(15–17) and tumor-associated(18–27)antigens. We have also previously described an anti-Id monoclonal antibody that was able to mimic glycoprotein carcinoembryonic antigen (CEA) and elicit an anti-anti-Id mAb that recognized the antigen in vitro and in vivo.(28–30) In the present study, we immunized BALB/c mice with the mAb R24 with the expectation of obtaining anti-Id antibodies bearing the internal image of GD3. Animals were shown to produce anti-Id and anti-anti-Id antibodies of IgM and IgG classes. One anti-Id antibody clone was selected for detailed evaluation. However, the most interesting finding emerged from the characterization of the anti-anti-Id antibody. The well-defined prototypic anti-GD3 monoclonal antibody R24 was used as a control in the assays of binding, cytotoxic activity in vitro and for the ability to protect against tumor challenge. We found that this anti-anti-Id mAb is a powerful growth inhibitor of melanoma cells in vitro and in vivo.
Materials and Methods
Animals. Eight-week-old female BALB/c and C57BL/6 mice were obtained from the National Institute of Pharmacology of the Federal University of São Paulo (UNIFESP) and from the São Paulo University animal facilities. All animal experiments were carried out in compliance with the NIH-Guidelines for the care and use of laboratory animals, and approved by the Animal Ethics Committee of UNIFESP.
Cell lines. Culture conditions followed standard protocols. Additional information is available in the Supporting Information.
Immunization. BALB/c mice were immunized with 100 μg of purified antibody (R24, 5.E3 or irrelevant mAb), coupled to KLH (Sigma, St Louis, MO, USA), as previously described.(29)
Production of anti-Id and anti-anti-Id mAb. Two days before cell fusion, mice received a final booster of 100 μg of KLH-coupled mAb R24 administered intravenously. Hybridoma cells were produced as previously described.(31,32) Double screening, made by both competitive inhibition and indirect enzyme immunoassays (EIA), allowed the detection of anti-Id and anti-anti-Id colonies. After cloning by limiting dilution and expansion of positive clones, large amounts of antibodies were obtained by the production of ascites in BALB/c mice. The mAb were purified by affinity chromatograghy in a Protein G-Sepharose column (Amersham Biosciences, Uppsala, Sweden). Immunoglobulin isotyping was performed with a Mouse Typer Isotyping Panel kit (Bio-Rad Laboratories, Hercules, CA, USA).
In vivo protection experiment. Immune deficient C57BL/6 mice were generated by a single 6-Gy dose of whole body γ-irradiation from a 137Cs source as previously described;(23–35) animals were maintained in a semi-sterile environment and fed with autoclaved meal and water containing 2 mg/mL gentamycin. 1 × 106 SKMel-28 cells in 0.1 mL sterile PBS were inoculated subcutaneously in two sites at the femur base of irradiated 8-week-old C57BL/6 mice on day 1. Four groups of five animals received an intraperitoneal injection of 50 μg of either one of the mAb (5.G8, R24 or the irrelevant 5.D11) or 100 μL of PBS. Injections were administered daily for 7 days from day 1. The animals were evaluated daily for tumor size by measuring the two largest diameters as described elsewhere.(36)
Statistical analysis. Statistical analysis was performed using the GraphPad Prism V.3 statistical software (La Jolla, CA, USA). Data were analyzed by t-tests when two groups were compared and by one-way analysis of variance (anova), followed by Dunnett’s post-test, in the cases of multiple comparisons. A P value <0.05 was considered significant. Means ± SD are shown.
The EIA, CDC, ADCC and antibody binding assays were performed according to standard protocols. A detailed description is presented in the Supporting Information.
Evaluation of the antibody response to mAb R24 immunization. Serum samples from BALB/c mice immunized with the mAb R24 conjugated to KLH were analyzed by competitive inhibition EIA. Figure 1A shows that the pool of sera from mice that have been immunized twice was able to inhibit 42.20 ± 10.78% of the binding of the mAb R24 to GD3 adsorbed to a solid phase. In addition, it was possible to detect the presence of anti-anti-Id antibodies in the sera by indirect EIA. Figure 1B shows the titration of GD3-binding anti-anti-Id antibodies in the sera of mice that have been immunized three times with the mAb R24.
Isolation of monoclonal antibodies. Given the high titers of anti-Id and anti-anti-Id antibodies present in the serum of mAb R24-immunized mice, we decided to use the spleens of these animals as the source of clonal cell populations for the production of both types of antibodies. Two fusion experiments raised eight anti-Id and five anti-anti-Id mAb. Isotyping of the anti-Id antibodies revealed one IgM, one IgG1 and six IgG2a isotypes. Anti-anti-Id antibodies were also checked and two IgM, two IgG1 and one IgG3 isotypes were identified. All of the obtained mAb had kappa light chains (Table 1). Two of these antibodies, 5.E3 and 5.G8, were characterized in detail.
Table 1. Antibody isotyping
All antibodies had kappa light chains. Bold refers to the antibodies selected for further analysis. Id, idiotype.
Figure 2A shows the inhibition curve of the binding of the mAb R24 to GD3 using increasing amounts of the anti-Id mAb 5.E3. A 50% binding inhibition was achieved with a 10 μg/mL concentration. This inhibition curve is similar to that obtained with other high-affinity anti-Id antibodies described in the literature.(30,37,38)
Next, we examined the reactivity of the anti-anti-Id mAb 5.G8 against crude extracts of the GD3-positive SKMel-28 cells. Figure 2B shows the binding activity to the polar lipid fraction by indirect EIA. The mAb 5.G8 and the positive control antibody R24 efficiently recognized the fraction containing glycosphingolipids that includes gangliosides, as opposed to the irrelevant antibody control (P < 0.001). Altogether, our data allowed us to envisage a model in which the mAb R24, 5.E3 and 5.G8 belong to an interactive idiotype-anti-idiotype network. In order to test this model, we biotinylated the anti-Id mAb 5.E3 and measured its binding ability to the mAb R24 and to the anti-anti-Id mAb 5.G8. The anti-Id antibody recognized equally well the mAb R24 and the mAb 5.G8 as illustrated in Figure 2C.
Anti-GD3 antibodies generated by anti-Id mAb 5.E3 immunization. Groups of five mice were immunized with the anti-Id mAb 5.E3 or an IgG2a isotype control mAb, both conjugated to KLH. Figure 3 shows the presence of GD3-binding anti-anti-Id antibodies in the sera of mice immunized twice with the mAb 5.E3. The reactivity of the sera derived from 5.E3-immunized animals was significantly higher than that observed with samples derived from isotype-vaccinated control mice (P < 0.0001). In addition, no GD3-binding antibody was detected in the pre-immune sera.
The anti-anti-Id antibody recognizes intact human melanoma cells. In order to verify whether the anti-anti-Id mAb 5.G8 could recognize the GD3 antigen in its native status in the context of intact cells, we tested its reactivity against intact SKMel-28 melanoma cells. Figure 4 shows that the mAb 5.G8 and the control mAb R24 were capable of recognizing 87.73 ± 2.75% and 100% of the cells, respectively. Similar data were obtained with another human melanoma cell line, MeWo, that was 65.55 ± 1.42% positive for binding to the mAb 5.G8 and 72.68 ± 1.72% positive for binding to the mAb R24 (Fig. 4). In contrast, both antibodies have shown a much lower reactivity to the human keratinocyte cell line Hacat: 16.78 ± 0.47% for the mAb 5.G8 and 18.96 ± 0.36% for the mAb R24. They have also reacted very poorly against another epithelial tumor, the human colon adenocarcinoma CO112 (3.78 ± 0.29% for the mAb 5.G8 and 1.87 ± 0.45% for the mAb R24), and against the murine embryonic fibroblast cell line NIH3T3 (7.21 ± 3.35% for the mAb 5.G8 and 9.43 ± 2.40% for the mAb R24).
Cytotoxic activity of the anti-anti Id mAb 5.G8 in vitro. The mAb 5.G8 was able to promote antibody-mediated complement-dependent lysis of SKMel-28 cells (Fig. 5A); these cells were also targets for antibody-dependent cellular cytotoxicity (Fig. 5B). The specific lysis achieved in the complement fixation assay was 65.47 ± 13.23% for the mAb 5.G8 and 97.95 ± 0.82% for the mAb R24; the specific lysis values in the ADCC test were 61.41 ± 1.57% and 71.00 ± 4.49%, respectively. The cytotoxicity levels induced by these antibodies in both tests were substantially higher than the background levels produced by the isotype control (P < 0.001). In contrast to the mAb 5.G8, the mAb R24 could lyse target cells in the absence of complement or effector cells.
In vivo antitumor activity of the anti-anti-Id mAb 5.G8. C57BL/6 mice were rendered immune deficient by irradiation preconditioning and were subsequently inoculated with GD3-expressing human melanoma SKMel-28 cells. The animals were divided into four groups that received one of the following treatments: (i) the mAb 5.G8; (ii) the mAb R24; (iii) an irrelevant mAb; or (iv) PBS, as described in the “Materials and Methods”. All mice (5/5) in the two negative control groups developed tumors by day 28 compared with 4/5 and 1/5 of the animals injected with the mAb 5.G8 or the mAb R24, respectively. Although there was no major difference between the mAb 5.G8 group and the negative control groups with regards to the number of tumor-free animals, there were substantial differences in tumor size and growth profile (Fig. 6). Tumor growth was observed in the two negative control groups as early as day 17 as opposed to the mice injected with the 5.G8 and anti-GD3 control antibodies in which the tumors took longer to be measurable. Figure 6 shows the tumor growth curve between days 20 and 28 from one representative experiment out of three. The lesions were barely visible on day 22 in the groups injected with the mAb 5.G8 or the mAb R24 and the average volume was smaller than 1 mm3 on day 28 (0.071 ± 0.073 cm3 for 5.G8 and 0.04 ± 0.1 cm3 for R24). Nevertheless, tumor progression was much more evident in the negative control groups, on day 28 reaching an average volume of 0.480 ± 0.249 cm3 in the animals that received the irrelevant antibody and 0.513 ± 0.284 cm3 in those that were injected with PBS. The evaluation on day 28 of tumor growth by anova followed by Dunnett’s post-test revealed a significant difference between the PBS control group and the mAb 5.G8 group (P < 0.001) and between the PBS and the mAb R24 groups (P < 0.001). Similar differences were found when the irrelevant antibody group was compared with the mAb 5.G8 group (P < 0.05) and with the mAb R24 group (P < 0.05). In conclusion, the anti-melanoma effect in vivo demonstrated with the anti-anti-Id mAb 5.G8 was virtually identical to that observed with the mAb R24 (Fig. 6).
Immunological tolerance is a major issue in cancer immunotherapy. To break the inability of the host’s immune system to recognize and react properly against malignant cells is particularly challenging given that tumor-associated antigens are self, poorly immunogenic molecules. Normal melanocytes express the mono-sialo GM3 as their major ganglioside, whereas the growth and metastatic potential of malignant melanoma correlate with excessive synthesis of GD3.(39) Although little is known about the precise composition of many gangliosides that are expressed in tumors, shedding of GD3 into the circulation has been observed in patients bearing certain malignancies.(39) It has been shown by several groups that this ganglioside might promote tumor growth and cell–cell adhesion.(40,41) Although GD3 is present in melanoma cells, its expression in normal tissues is limited and not related to altered cellular behavior. Altogether, these reports support the idea that GD3 might be a good target for the antitumor immune response and suggest the potential usefulness of an anti-GD3 antibody-based strategy in cancer therapy.(9,14,42,43) However, as the main antigenic epitopes in GD3 are carbohydrates, only a very weak and T-cell-independent immune response can be stimulated when it is used as an immunogen. Several attempts have been made to address the therapeutic potential of the anti-GD3-specific immune response.(14,42–50) Nevertheless, the anti-GD3 immune response was shown to be short-lived and mainly constituted by IgM antibodies.(1)
The anti-idiotypic antibody approach represents an alternative to circumvent the poor immunogenicity of GD3 and the toxicity of the mAb R24(11–14) in cancer therapy. The remarkable ability of anti-Id antibodies to mimic the original antigen has been shown in several models.(15–28) Anti-Id antibodies might induce a specific immune response against tumor-associated antigens, such as the carcinoembryonic antigen(29,51) and the human high molecular weight melanoma-associated antigen(52) among others.(22–27) In addition, two anti-idiotype R24 antibodies (BEC2 and BEC3) have been previously isolated.(53) Although BEC3 was incapable of eliciting anti-GD3 responses, BEC2 was shown to induce IgM but no IgG anti-GD3 antibodies in a fraction (22%) of immunized melanoma patients.(54–56) The application of these antigen surrogates in cancer immunotherapy is particularly appealing when the antigen is not a protein, as in the case of gangliosides. However, the fine specificity profile of anti-GD3 antibodies and of their anti-idiotypic counterparts might vary considerably with significant differences in binding, affinity and immunogenicity.(44–50,53) Thus, the search for novel anti-Id and anti-anti-Id antibodies, such as the ones described in this paper, might lead to the identification of molecules with the ability to induce high affinity IgG responses with increased antitumor activity in a larger fraction of immunized individuals.
In the present report, we have used the mAb R24 coupled to KLH to immunize BALB/c mice. An anti-Id immune response was identified after the first immunization as the serum inhibited binding of the mAb R24 to GD3. Anti-anti-Id antibodies capable of binding GD3 could also be detected in the third immunization. In view of these results, both anti-Id and anti-anti-Id mAb were produced in mice in order to evaluate the efficiency of the GD3-associated idiotypic network. We found that the anti-Id and anti-anti-Id antibody secreting clones belonged to different Ig classes and subclasses. The observed Ig isotype diversity suggests an effective T cell cooperation in the development of the anti-Id and anti-anti-Id immune responses; it is also likely to contribute to a higher efficiency of the immune response because it might augment the cytotoxic effect triggered and/or mediated by antibodies.
The monoclonal anti-Id antibody selected for detailed analysis in the present study, clone 5.E3, was able to inhibit the binding of the mAb R24 to GD3 in a dose-dependent fashion. When used as an immunogen, the mAb 5.E3 also elicited a GD3-binding anti-anti-Id humoral immune response in mice. The participation of the mAb 5.E3 in a cascade of idiotype–anti-idiotype interactions is supported by its ability to bind the mAb R24 as well as the anti-anti-Id mAb 5.G8.
The mAb 5.G8 and the mAb R24 were capable of recognizing the GD3 ganglioside in the polar lipid fraction of SKMel-28 cell extracts. Reactivity of the anti-anti-Id antibody was also tested in intact cells; five cell lines, including two human GD3-expressing melanoma cell lines, were tested for antibody binding. The mAb 5.G8 was shown to recognize both melanoma cell lines at comparable levels to those obtained with the original anti-GD3 mAb R24. It is noteworthy that both antibodies recognized SKMel-28 cells somewhat better than MeWo cells (25% higher reactivity); this finding is in agreement with the reported higher expression of GD3 in SKMel-28 cells.(8)
Antibody-dependent cellular cytotoxicity and CDC are important immune effector mechanisms and the ability to promote target lysis might be a valuable asset of antibodies considered for therapeutic use. Our analysis of the mAb 5.G8 revealed that it was highly effective in mediating lysis of target cells either by ADCC or by fixing complement. The ADCC activity was observed when activated peritoneal macrophages were used as effector cells, but not when splenocytes were the source of the effector cells (data not shown). At first sight, the mAb 5.G8 seemed to be less efficient than the mAb R24 in ADCC and CDC. However, the mAb R24 exhibits direct cytotoxic activity as evidenced by the aggregation of melanoma cells in culture, which usually leads to detachment and subsequent death.(8) In addition, the mAb R24 is an IgG3 that might form noncovalent molecular aggregates at a high concentration, thereby increasing the lysis of target cells in vitro.(57–60) If one accounts for this direct cytotoxicity, the mAb R24 and the mAb 5.G8 have fairly similar activities in ADCC and CDC. Indeed, our results indicate that the mAb 5.G8 was even slightly more efficient than the mAb R24 in its ability to fix complement (59.83 ± 12.29% for the mAb 5.G8 and 52.66 ± 0.82% for the mAb R24), as well as to promote ADCC (58.34 ± 1.57% and 49.20 ± 2.78%, respectively) if one discounts in each case the lysis obtained by the antibody alone.
In our in vivo tumor model, animals were preconditioned by sublethal irradiation and inoculated with the human melanoma cell line SKMel-28. All animals in the negative control groups developed tumors that were approximately 0.5 cm3 on day 28. This was in sharp contrast with groups that were treated with the anti-anti-Id or GD3-specific antibodies in which there were tumor-free animals at the end of the observation period and the average volume of the lesions when present was under 1 mm3. Thus, our data indicate that passive transfer of the anti-anti-Id mAb 5.G8 confers protection in the experimental model described here by delaying tumor growth.
We present evidence that immunization with the anti-GD3 mAb R24 induces a cascade of idiotype–anti-idiotype interactions. While anti-Id interactions are likely to occur in the context of natural immune responses, a definitive picture of their immune-regulatory importance remains to be established. However, our results demonstrate that the anti-Id approach works even in a T-cell-independent system and suggests a potential role to play in cancer immunotherapy. The isotype diversity also constitutes an advantage of the anti-Id strategy; this is particularly relevant if we consider that such diversity occurred in a system where the original antigen is a glycolipid that is unlikely to stimulate an efficient immune response on its own. To our knowledge, this is the first report on a melanoma-specific anti-anti-Id mAb that shows direct antitumor activity in vivo in a T-cell-independent way; this anti-anti-Id mAb might represent an alternative for cancer treatment if further studies identify a favorable profile in terms of toxicity, inflammatory reaction and tumor killing.
This work was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP). A.S.R. and C.B.P. were recipients of scholarships from FAPESP and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), respectively. The authors are indebted to Dr Roger Chammas for his critical comments and suggestions, Ivan Hariton Cordeiro for technical assistance and Mauro Cardoso Pereira for animal care support.