Claudin-4 (CLDN4) is a tetraspanin transmembrane protein of tight junction structure and is highly expressed in pancreatic and ovarian cancers. In this study, we aimed to generate an anti-Claudin-4 monoclonal antibody (mAb) and evaluate its antitumor efficacy in vitro and in vivo. To isolate specific mAb, we generated CLDN3, 4, 5, 6, and 9, expressing Chinese hamster ovary (CHO) cells, and then used them as positive and negative targets through cell-based screening. As a result, we succeeded in isolating KM3900 (IgG2a), which specifically bound to CLDN4, from BXSB mice immunized with pancreatic cancer cells. Immunoprecipitation and flow cytometry analysis revealed that KM3900 recognized the conformational structure and bound to extracellular loop 2 of CLDN4. Furthermore, binding of KM3900 was detected on CLDN4-expressing pancreatic and ovarian cancer cells, but not on negative cells. Next, we made the mouse-human chimeric IgG1 (KM3934) and evaluated its antitumor efficacy. KM3934 induced dose-dependent antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity in vitro, and significantly inhibited tumor growth in MCAS or CFPAC-1 xenograft SCID mice in vivo (P < 0.05). These results suggest that mAb therapy against CLDN4 is promising for pancreatic and ovarian cancers. (Cancer Sci 2009; 100: 1623-1630)

Human claudin-3 (CLDN3) is a tetraspanin transmembrane protein of tight junction structures and is known to be over-expressed in some malignant tumors. Although a specific monoclonal antibody (MAb) against the extracellular domains of CLDN3 would be a valuable tool, generation of such MAbs has been regarded as difficult using traditional hybridoma techniques, because of the conserved sequence homology of CLDN3s among various species. In addition, high sequence similarity is shared among claudin family members, and potential cross-reactivity of MAb should be evaluated carefully. To overcome these difficulties, we generated CLDN3-expressing Chinese hamster ovary and Sf9 cells to use an immunogens and performed cell-based screening to eliminate cross-reactive antibodies. As a result, we generated MAbs that recognized the extracellular loops of CLDN3 but not those of CLDN4, 5, 6, or 9. Further in vitro studies suggested that the isolated MAbs possessed the desired binding properties for the detection or targeting of CLDN3.