Selection of anti-cancer antibodies from combinatorial libraries by whole-cell panning and stringent subtraction with human blood cells

Traditional strategies for the identification of cell-surface cancer targets often fall short of their objective. For example, whole-cell panning of antibody libraries to isolate a diverse panel of antibodies directed against targets on cancer cells often identifies all immunogenic and/or abundant cell-surface antigens, not simply tumor-specific or tumor-associated antigens. Here we describe the use of stringent negative selection in combination with positive panning to increase tumor specificity and clinical relevance of selected antibodies. Sera from cancer cell-immunized mice showed strong binding to immunizing cancer cell lines but also cross-reacted strongly with human blood cells. Antisera blood cell binding was considerably decreased after stringent subtraction with human red blood cells (RBCs) and white blood cells (WBCs), yet cancer cell specificity was retained. In order to select for a higher percentage of clinically relevant antibodies for potential therapeutic use, stringent negative selection by RBC subtraction was employed in whole-cell panning of a disease-specific phage displayed antibody library on the prostate cancer cell line, PC-3. Isolated antibodies were found to bind to target antigens implicated in tumorigenicity and cancer cell migration and/or invasion, and included CD26, CDCP1, and the integrin complexes alpha 2/beta 1, alpha 3/beta 1, alpha 5/beta 1, and alpha 6/beta 4. Compared with traditional cell panning, this method considerably increased the selectivity of antibodies to tumor-associated antigens. (C) 2007 Elsevier B.V. All rights reserved.