Human Antibodies

Recombinant DNA techniques were utilized successfully to join the coding regions for the variable region of a mouse anti-tumor antibody (BA-Br-1) and the human IgG1 constant region for both the light and heavy chains. After insertion into a mouse myeloma host cell line, the chimeric genes were expressed successfully and the resulting antibody (ING-1) was purified. In this study, we describe biochemical, serological, immunohistochemical, and functional properties of the chimeric ING-1 antibody. Analysis of the synthesized antibody revealed that while it was similar in size to the mouse antibody, it had a different pI as determined by isoelectrofocusing. The flow cytometric binding profiles of the new molecule were found to be essentially identical to the parental mouse immunoglobulin.

The specificity of the chimeric ING-1 and mouse BA-Br-1 antibodies were compared by extensive immunohistochemical analysis on human normal and tumor tissues. The chimeric antibody retained the same broad carcinoma binding activity, showing strong reactivity with greater than 90% of epithelial tumor tissues, as was previously observed for the mouse BA-Br-1 antibody. The chimeric and mouse antibodies also recognized the same selected normal tissues: primarily glandular epithelia, gastrointestinal mucosa, bile ducts, and thyroid follicles.

Analysis of the biological function of the chimeric antibody revealed that it possessed ADCC activity against antigen-bearing tumor targets in vitro which was absent from the mouse form of the antibody. Competent effector cells could be either PBMCs from normal healthy donors, PBMCs from cancer patients receiving LAK/IL-2 therapy, or LAK cells prepared from cancer patients. Enhanced cytotoxicity even in the presence of LAK cell killing was noted with effector cells from the latter two sources. This contrasts sharply with the absence of activity in the same systems when the native murine antibody was used. The in vitro activation of cell-dependent cytolysis observed with the chimeric antibodies when effector cells from both normal and tumor-bearing donors were used strongly suggests that comparable activity would be observed in vivo. These results, along with the broad carcinoma binding activity and minimal normal tissue reactivity, suggest that the ING-1 chimeric antibody may be useful in cancer therapy. The application of the ING-1 chimeric antibody for treatment of tumors thus offers a promising avenue for future research.

Human lymph node lymphocytes from cancer patients were fused with either the UC 729-6 or SHFP-1 human fusion partners. Resulting human-human hybridomas were tetraploid, expressed markers from both parent cells, and secreted approximately 1 μgIg/10⁶cells/ml/day. Immunofluorescence analysis of some of the human MAbs with a panel of normal and malignant cell lines revealed a staining pattern of only the nuclear region. One IgM secreting hybridoma, TLN1F4, derived from a teratocarcinoma lymph node, predominantly stained the nuclear regions of adherent tumor cell lines and no hematopoietic cell lines or normal fibroblasts. PLN3C8, an IgG1 secreting hybridoma, derived from a prostate carcinoma lymph node, predominantly stained the nucleolus of LnCap, a carcinoma of the prostate cell line. CLN2E5, an IgM secreting human hybridoma, derived from a carcinoma of the cervix lymph node, predominantly stained both cytoplasmic and nuclear components to tumor cell lines and not normal fibroblasts or hematopoietic cell lines. These data suggest that the immune response occurring within regional draining lymph nodes is capable of recognizing nuclear-associated antigens.

Splenocytes from a patient with chronic, immune-mediated thrombocytopenic purpura (ITP) were transformed with Epstein-Barr virus. A stable lymphoblastoid cell line (LCL) derived from this transformation (2A3) produces IgM antibody reactive with platelet glycoprotein IIb. 2A3 was fused to the 6-thioguanine-resistant, ouabain-resistant, murine-human heteromyeloma cell line, F6. The resultant heterohybridomas were selected by growth in medium containing hypoxanthine/aminopterin/thymidine and ouabain. One hybridoma line, 2E7, produces high levels of IgM antibody (2 to 4 μg IgM/ml/24 hr/10⁵ cells) reactive with glycoprotein IIb. 2E7 has been repeatedly subcloned by limiting dilution and has been maintained in continuous culture for 26 months. 2E7 binds to human platelets but not endothelial cells, as determined by flow cytometry, and does not react with platelets of patients with Glanzmann's thrombasthenia that lack IIb-IIIa. The epitope recognized by 2E7 is likely to be a contiguous peptide sequence since the antibody binds to the IIb heavy chain in immunoblot assays of denatured, reduced platelet protein. Treatment of intact platelets or purified IIb-IIIa with papain or chymotrypsin, but not SV8 protease, destroys the epitope. Thus, the 2E7 epitope may be at or very close to a site on IIb that is cleaved by these proteases. The expression of the 2E7 epitope is significantly affected by the presence of divalent cations. Treatment of intact platelets with EDTA at 37°C results in a three- to four-fold increase in the number of 2E7 molecules bound per platelet and an eight-fold increase in the affinity of the antibody. The binding of 2E7 to normal platelets does not inhibit any of the functions attributed to IIb-IIIa, such as fibrinogen-dependent platelet aggregation or clot retraction. 2E7 represents the first human monoclonal antibody reported to recognize an epitope on platelet glycoprotein IIb. The epitope is unique to IIb and not shared by other integrin alpha subunits.

Presenting a panel of human hybridomas secreting serospecific antibodies which confer a high degree of protection against fatal infection with Pseudomonas aeruginosa, we report an efficient approach for a systematic generation of antigen specific human monoclonal antibodies with biological activity. This approach is based on active immunization and antigen specific panning. Individuals were immunized with polysaccharides isolated from LPS of Pseudomonas aeruginosa conjugated to toxin A. Specific B cells were isolated and enriched by panning of blood samples taken at the time point with the highest frequency of fuseable cells. These cells were then transformed with Epstein-Barr virus. Arising lymphoblastoid cell lines were screened for the secretion of anti-LPS antibodies by enzyme-linked immunosorbent assay and fused to a murine-human heteromyeloma cell line. Hybridomas were selected for high levels of antibody secretion and binding to intact bacteria as determined by an immunofluorescence microscopy assay. The observation that protective capacity of an antibody was associated with its ability to bind to LPS determinants accessible on the bacterial cell surface allowed for an effective screening for therapeutically interesting human monoclonal antibodies. Out of four immunized individuals, 15 lymphoblastoid cell lines with anti-LPS activity could be isolated, and 8 hybridomas, which cover the majority of the common Pseudomonas aeruginosa serotypes, were characterized further. The generation of monoclonal anti-Pseudomonas aeruginosa toxin A, anti-Klebsiella capsular polysaccharides, and anti-Escherichia coli LPS antibodies suggests that the success of this approach is not limited to the generation of human monoclonal antibodies of a particular specificity or to the use of antigens of a particular character.

Lymphocytes from regional lymph nodes of patients with ovarian carcinoma were immortalized by fusing them with a nonsecreting cell line of murine myeloma (Sp2/0-Ag14). By early cloning and recloning a hybrid cell line, named HMD4, was established. It has secreted human IgG for more than 15 months stably. Chromosome analysis corresponded with the characterization of human-mouse hybridoma. Large quantities of ascites were obtained after hybrid cells injection into the primed nude mice. Human IgG of light chain was detected and purified from the ascites. Twenty-six of 43 (60.5%) epithelial ovarian cancers were positively stained with HMD4 by ABC immunoperoxidase methods while nonepithelial ovarian cancers and almost all benign tumors and normal tissues were negative. The molecular weight of the antigen recognized by HMD4 was 55KDa determined by Western blotting.

¹³¹I labeled HMD4 was administered intraperitoneally to nude mice bearing human ovarian epithelial adenocarcinoma; ¹³¹I labeled normal human IgG and normal murine IgG were used as controls. Measurements of T/NT and T/B ratios of ¹³¹I-HMD4 were done. Radioimaging showed HMD4 clearly localized on tumor regions at 48 and 72 hours and the biodistribution and metabolism of the labeled HMD4 corresponded with the images. The above results indicate that HMD4 was specific to ovarian carcinoma, a hopeful clue for clinical applications.

Human lymphocytes from peripheral blood (MNC) were separated on magnetic beads for the presence of different surface markers. Cells from positive and negative fractions were successfully immortalized by electro fusion with the heteromyeloma line CB-Fu2. B cells, which were separated on anti-CD 19 coated beads, could be immortalized at a rate between 10⁻⁵ and 10⁻⁴ even if the fusion was conducted with just a few hundred thousand cells. Comparison of the frequency of 19-positive hybridomas in B cell, T cell, and unseparated MNC fusions indicated that also non-B cells may give rise to HAT resistant hybridoma clones, although the fusion frequency was low.

In this study, limited sampling models for HA-1A human IgM monoclonal antibody were developed to predict the area under the concentration time curve from timed serum concentrations. Patients were administered 15 minute infusions of 25 mg (11 patients), 100 mg (15 patients), and 250 mg (2 patients). A detailed pharmacokinetic analysis (eight time points) was performed to obtain the area under the curve for each of the 28 patients using a one-compartment model with the values normalized to a dose of 100 mg. Various models were then developed to estimate the area under the curve from one and two timed concentrations using regression analysis methodology. Of these models, four were determined to be of interest with coefficients of determinations ranging from 0.92 to 0.97. They performed quite well in predicting the area under the curve values with relative root mean squared predictive error of 7.1 to 10.8% and relative mean predictive error of −4.5 to +6.9%. These models should be extremely useful in larger scale, Phase II/III studies of HA-1A in correlating the estimated area under the curve with various demographic data, toxicity, and efficacy.