Generation and characterization of a IgG monoclonal antibody specific for GM3 (NeuGc) ganglioside by immunizing β3Gn-T5 knockout mice

A murine monoclonal antibody (MAb-1) specific for GM3 has been generated by immunizing β3Gn-T5 knockout mice with purified GM3 ganglioside. The binding specificity of MAb-1 (IgG3 subclass) was established by an enzyme-linked immunosorbent assay (ELISA) and FACS and the antibody showed high binding specificity with GM3. Cell viability assay showed that MAb-1 significantly suppressed cell growth. Immunohistochemistry analysis revealed that MAb-1 was strongly expressed in human ovarian cancer tissues, whereas it was hardly expressed in normal tissues. Finally, antibody-dependent cellular cytotoxicity (ADCC) activities were determined by measuring lactate dehydrogenase (LDH) releasing assay and the results showed high ADCC activities in two representative ovarian cancer cell lines (OVHM and ID8). All of these data indicate that MAb-1 may be potentially used as a therapeutic antibody against ovarian cancers in clinical trials.

After drying up in the air, 5% BSA was added for 2 h at room temperature. A series of diluted MAb-1 were added to the plates and incubated for another 2 h, followed by horseradish peroxidase (HRP)-anti-mouse IgG (Amersham Biosciences) as a secondary antibody. Finally, 10 μl of substrate solution [ortho-phenylene diamine (2 mg) (Sigma) and H 2 O 2 (8 μl) (Sigma) in 5 ml of citrate-phosphate buffer] was added to stop the reaction. The optical density was recorded at 450 nm with a scanner. The isotype control of mouse IgG 3 was purchased from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA).
Flow cytometry. Cell surface expression of GM3 was determined by FACSCaliver TM (Becton Dickinson).
Measurements of affinity of MAb-1. The affinity of MAb-1 antibody against GM3 ganglioside was determined using BIAcore 3000 system (BIA core, Piscataway, NJ). Firstly, the MAb-1 antibody was immobilized on the surface of biosensor chips and coupled with N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide/N-hydroxysuccinimide according to the instructions of the manufacturer. Then, 1% BSA was used as a control. The affinity rate constants (association rate constant, K a ; disassociation rate constant, K d ) were determined. Finally, the affinity of MAb-1 (K) were calculated as K = K a /K d .
Immunohistochemistry analysis by MAb-1. Immunohistochemistry analysis was performed as follows. Briefly, 20% sucrose-fixed human ovarian cancer tissues were cut into 10 μm sections. Then, they were incubated with MAb-1 (15 μg/ml) or isotype control (15 μg/ml) at 4 °C overnight. Next, they were incubated with biotin-conjugated secondary antibody for 30 min at room temperature and then incubated with streptavidin-horseradish peroxidase complex for 30 min. Finally, the sections were incubated with 3, 3′-diaminobenzidine for 10 min and counterstained with hematoxylin. The immunostained slides were evaluated by two independent observers under the microscope. Positive staining was detected as a brown color of the cells. Five high-power fields were randomly selected, and the percentage of positive cells in these fields was counted. Tumors without staining or with weak staining (positive cell rate <10%) were classified as negative while tumors with moderate (10%≤ positive cell rate <75%) to intense staining (positive cell rate ≥75%) were classified as positive.

Isolation of peripheral blood mononuclear cells (PBMCs).
PBMCs of mice were isolated as follows: Firstly, 5 ml peripheral blood was collected and diluted by addition of an equal volume of 1× PBS. Then the diluted blood was carefully loaded over an equal volume of lymphoprep reagent (Mouse Lymphoprep Reagent kit, Beijing Solarbio Science & Technology Co., Ltd) in a 50 ml centrifuge tube. After centrifugation at 800 g for 20 min at room temperature, the distant band at the blood/lymphoprep was carefully transferred into a new tube. The harvested fraction was washed twice and finally a pellet of cells was obtained. ADCC assay. OVHM and ID8 cells were used as target cells and fresh PBMCs were used as effector cells.
The ADCC was evaluated using a LDH release assay (Promega, Madison, USA) in 96-well plates, according to the manufacturer's protocol. Briefly, cells were incubated at 37 °C for 20 h. Then, 100 ng/ml MAb-1 was added into each well. Next, the effector cells were added into the well with the E:T ratios of 200:1, 100:1, 50:1, 20:1, 10:1, 5:1 and 2:1, respectively. After 4 h co-incubation, LDH release in the supernatants was determined at 490 nm. The percentage of cellular cytotoxicity was calculated using CytoTox 96 Non-Radioactive Cytotoxicity Assay TM (Promega) according to the manufacturer's instructions. Data were graphed and analyzed using GraphPad Prism5.0.

Statistical analysis.
The results were reported as mean ± SD. Statistical significances were performed by one-way analysis of variance (ANOVA) followed by Tukey post hoc test between groups and multiple comparisons. A p value of less than 0.05 was evaluated as statistically significant.

Results
Generation of an anti-GM3-specific monoclonal antibody. After immunization of 3 mice with GM3 ganglioside embedded in liposome, the titers in each mouse were determined by ELISA and the mouse with the highest titer of anti-GM3 were chose for the following experiment. Spleen cells were fused with NS-1 myeloma cells. After hypoxanthine-aminopterin-thymidine (HAT) selection, a number of Mabs reactive with GM3 ganglioside were generated. Briefly, about 316 clones were available from 921 clones, and 89 definitely positive clones were identified by immuno-fluorescence (IF) assay. Furthermore, these clones were subcloned and checked by IF assay again. Finally, only 13 clones were found to be significantly positive with glycolipids (data unpublished). Among them, MAb-1 was firstly established and tested on tumor immunity against ovarian carcinoma. As shown in Fig. 2, MAb-1 reacted with GM3 in a dose-dependent manner in ELISA.
Specificity of MAb-1 against GM3 at cellular level. Flow cytometry was used to determine the specificity of MAb-1 at cellular level (Fig. 3A). The purple portion indicates the negative control and the green portion indicates the positive expression of GM3 ganglioside. Both CHO and A431 cell lines, which endogenously express  To further analyze the epitope of MAb-1 antibody, the antibody was tested against other a-series (GM2, GM1), b-series gangliosides (GD3, GD2, GD1), globo-series(Gb3) as well as against GM3 (NeuAc). As shown in Fig. 3B, no cross-reaction with other gangliosides was found. Moreover, the association and dissociation rate constants (K a = 6.08 × 10 4 (mol/l s) −1 and K d = 3.17 × 10 −4 s −1 ) were determined, respectively. The affinity of MAb-1 (K = K a /K d ) was calculated as 1.92 × 10 8 (mol/l) −1 . Similarly, the affinity of 14F7 against GM3 ganglioside was 1.79 × 10 8 (mol/l) −1 , indicating that it has similar affinity compared with MAb-1.

Subclass of MAb-1 antibody. Next, the subclass of MAb-1 antibody was established using Sino Biological
Mouse Mab Antibody isotyping kit (Fig. 3C). The results clearly showed that the subclass of MAb-1 was IgG 3 . Immunohistochemistry analysis by MAb-1 against ovarian tissues. Since GM3 ganglioside is known to be strongly expressed in human ovarian cancer cells, immunohistochemistry analysis was performed to detect the immunoreactivity of MAb-1 on human ovarian cancer tissues. In total, 56 human ovarian cancer samples were checked and 43 were positive. This result indicated the immunoreactivity was about 76.79%. Immunostaining by MAb-1 was negative in normal ovarian tissues (Fig. 5A), but showed strong staining in a cell-surface and cytoplasm pattern in ovarian cancer cells, such as serous adenocarcinoma (Fig. 5B), mucinous adenocarcinoma (Fig. 5C) and metastatic adenocarcinoma (Fig. 5D). These results suggested that MAb-1 specifically reacted with GM3 ganglioside in human ovarian cancer tissues. ADCC by MAb-1. ADCC effects induced by MAb-1 were evaluated by determining the activity of cytosolic LDH released by treating OVHM and ID8 cells, respectively. As shown in Fig. 6, (1    and 50:1, the cytotoxicity (%) in OVHM and ID8 cells showed significantly higher levels compared with that of control groups (p < 0.01); (3) The cytotoxicity showed an E:T ratio-dependent manner when treated with 100 ng/ ml MAb-1; (4) The control mouse IgG did not cause any significant cell lysis (less than 10%). These results indicated that MAb-1 could generate specific ADCC effects in mouse ovarian cancer cells.

Discussion
Since 1985, it has been well known that glycosphingolipids are ubiquitous membrane components of various organs 2 . Although they are natural components of the plasma membranes of vertebrates, numerous evidence indicate that gangliosides are attractive targets for immunotherapy due to different expression patterns during oncogenesis and tumor development [26][27][28][29][30] . Some of these gangliosides have been used as tumor markers or tumor-associated antigens in cancer diagnosis and therapy [31][32][33] . Especially, GM3 (NeuGc) is one of the most common types of sialic acid, which scarcely exists in normal human tissue but strongly expresses in glycoconjugates of human tumors 34,35 . These studies indicate that GM3 can be a potentially attractive target for cancer diagnosis and therapy. In view of the above points, a large number of antibodies, such as L612 36 , GMR6 15 , 8G9D8 37 , AbFCM1 38 , have been established against NeuGc-containing gangliosides. At present, it is a little difficult for us to obtain these antibodies. However, we found that the subclass of L612, 8G9D8, GMR6 and AbFCMl was IgM, but not IgG subclass. In the application of reacting cancer cells with mAbs, IgG subclass mAbs are preferable, since they can be easily purified and possess immunological actions such as ADCC. Also, GMR6 exhibited broader specificities with GM4, GM1b, GD1a and GT1b. In addition to ganglioside GM3, 8G9D8 may bind to glycoproteins or another glycolipid of the stratum corneum in normal skin with a shared carbohydrate sequence. All of the above results have shown that MAb-1 monoclonal antibody in this study has valuable advantage in the future. Recently, 14F7, a IgG1 binding NeuGc-GM3 monoclonal antibody, was proved to bind specifically to GM3 (NeuGc) in breast, melanoma, colon and primary lymphoid tumors [39][40][41] . These results suggest that novel IgG antibodies specific to GM3 ganglioside can be reasonable and practicable.
In the present study, we used purified GM3 ganglioside to immunize β3Gn-T5 knockout mice to generate IgG monoclonal antibodies (IgG 3 subclass) against human ovarian cancer tissues. β3Gn-T5 knockout mice, which lack Lc3-synthase, the key enzyme that controls the expression of lacto-/neolacto-series glycolipids, at some point, can enhance the antigen specificity. ELISA assay has shown that MAb-1 reacted with GM3 in a good dose-dependent manner. This result was also obtained by another independent study 16 . Furthermore, at cellular levels, the newly generated MAb-1 can significantly recognize cell lines that highly expressed GM3. These results further suggest that β3Gn-T5 knockout mice can be suitable animals for generating anti-glycolipid antigens with lacto-/neolacto-series structures.
It is generally accepted that NeuAc-gangliosides are expressed in normal tissues. Most studies have shown that NeuGc-gangliosides are widely expressed in human tumors 39,42,43 and cell lines 44 . To our best knowledge, this is the first description of a murine IgG3 mAb specific for GM3 by immunizing β3Gn-T5 knockout mice. MAb-1 reacted with GM3 (NeuGc), but not GM3(NeuAc), suggesting that NeuGc structure affected MAb-1 binding. On the other hand, MAb-1 showed a little reactivity with GM2 ganglioside, indicating that the GalNAcβ1-4 residue significantly affect the binding efficacy of Mb-1. As for 14F7 antibody, similar results were found before 27 . Both of them showed specific reactivity with GM3 (NeuGc) ganglioside, suggesting that GM3(NeuGc) could be a suitable epitope.
It has been reported that most gangliosides (such as GM1, GD1a and GD3) of tumors, especially tumor microenvironment, can positively influence tumor growth 10 . As for GM3 ganglioside, early studies showed that exogenous GM3 inhibited the cell proliferation of several cancer cells 43,45 . However, the growth of the Siat9 (encoding GM3 synthase) and Galgt 1(encoding GM2 synthase)-deficient knockout tumor cells is significantly impeded both in vivo and in vitro 10,46 . The current data seem to suggest that the role of GM3 ganglioside in cell proliferation remains controversial. Here, in 2 ovarian cancer cell lines, MAb-1 treatment significantly inhibited the cell proliferation and showed good dose-and time-dependent manners. The discrepancies are likely due to the different concentrations of GM3 being used, different cell lines and tumor types and/or different administration of ganglioside-specific IgG antibodies.
It has been well established that ADCC is one of the immune effector mechanisms associated with antibodies against tumor-associated gangliosides 47,48 . The IgG Fc domain can interact with human FcγRs on effector cells and IgG 3 is considered one of the principal human isotypes for activating FcγRs 49 . In the present study, the subclass of MAb-1 is IgG 3 subclass and we further assessed the ability of MAb-1 to induce ADCC in ovarian cancer cells. Our results indicate that MAb-1 can induce remarkable ADCC effects on effector cells.
In conclusion, by immunizing β3Gn-T5 knockout mice with purified GM3 ganglioside, we successfully generated MAb-1 antibody. The MAb-1 reacted with GM3 in a dose-dependent manner in the ELISA assay. Furthermore, we detected the specific reaction of the antibody against ovarian cancer cells via immunohistochemistry and ADCC analysis. These results indicate that MAb-1 is a potentially effective IgG monoclonal antibody which may further be used in antibody-dependent diagnose and therapy of ovarian cancers.