The objective is to examine Tn-O-glycopeptide antigens as targets for immunotherapy in a range of human cancers. A large percentage of common human cancers (e.g., colon, breast, lung, prostate, ovary and pancreas) have defects in O-linked glycosylation thereby generating tumor-specific epitopes on surface glycoproteins of cancer cells. The defect is a truncation of the growing sugar tree after linkage of the first sugar N-acetylgalactosamine (GalNAc), to threonine or serine on the peptide chain of proteins by an ?1-O-glycosyl linkage. The linked single sugar is referred to as Tn (T antigen nouvelle) and can be detected by "anti-Tn" IgM antibodies. The cancer-specific truncation can also be detected by high-affinity "anti-Tn-O-glycopeptide" IgG antibodies. They, in contrast, engulf the single sugar moiety while recognizing surrounding specific peptide sequences. Such IgG antibodies react with all cancers that have the required glycosylation defect and provide the same aberrantly glycosylated surface glycoprotein. Three such monoclonal antibodies, 237, 5E5 and 3H4, have been cloned and sequenced to generate (i) chimeric antibody receptors (CARs) for transduction into T cells for adoptive transfer and (ii) bispecific fusion proteins to engage the host's NK and cells in the margins of large solid tumors. The fusion proteins will consist of the binding domain and IL15 linked to IL15R?. Aim 1 is to determine the potential of 237 CAR-transduced T cells targeting a tumor-specific Tn-O-glycopeptide epitope to eradicate solid non-hematopoietic tumors in a syngeneic mouse model. It will be determined whether these T cells eradicate clinical-size solid tumors without toxicity to normal tissues, by what mechanism destruction or escape occurs, whether the sequential events leading to destruction differ from those caused by TCR transduced T cells, and whether T cells transduced with CARs can destroy cancer cells that micro-disseminate from solid primary tumors. Aim 2 is to determine the potential of fusion proteins carrying IL15-IL15R? and having 237 or 5E5 receptors to target tumor-specific Tn-O-glycopeptide epitopes. It will be determined whether fusion proteins carrying the anti-Tn- O-glycopeptide scFv and IL15-IL15R? can cause the appearance of densely granulated NK cells at the tumor margin and rescue tolerant intratumoral T cells to reject large established tumors. Aim 3 is to generate and evaluate CARs recognizing Tn-O-glycopeptide epitopes on the deglycosylated form of MUC1 found in human cancers. It will be tested whether such CARs lack toxicity towards normal tissues expressing the normal, fully glycosylated form of MUC1 in MUC1-transgenic mice while eliminating clinical size tumors. Furthermore, a panel of isogenic cell lines of common human cancers, all lacking Cosmc function (leading to defects in O-linked glycosylation), will be used to induce and select for additional cancer-specific antibodies recognizing Tn-O-glycopeptide epitopes on other surface glycoproteins. With these antibodies, we will make additional CARs (and fusions proteins) to be used in combination with 5E5 to prevent tumor escape.