Cellular immunotherapy of pancreatic ductal adenocarcinoma: Discovery and evaluation of novel target candidates

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a devastating disease with a poor prognosis and limited treatment options. In recent years, chimeric antigen receptor (CAR) T cells have achieved remarkable success in the treatment of liquid malignancies but efficacy in solid malignancies is limited to date. A major hurdle is the lack of tumor specific targets. We aimed to address this issue in the case of PDAC by analyzing the expression of almost 400 cell-surface antigens on tumor cells of different patient derived xenograft (PDX) mouse models. Identified candidates were prioritized with respect to their off-tumor expression bioinformatically by using RNA and protein expression databases. The specificity of these target candidates was further validated on primary PDAC specimens using flow cytometry and a newly developed cyclic immunofluorescence (cyclic IF) microscopy technology. We identified Cutaneous Lymphocyte Antigen (CLA), CEACAM6 (CD66c), CDCP1 (CD318) and Tetraspanin-8 (TSPAN8) as having high specificity to tumor cells and restricted off-tumor expression. In addition, we investigated the expression of these targets using flow cytometry and cyclic IF microscopy on healthy tissue arrays. Target expression on healthy tissues was very restricted, which further underlined the quality of the novel target candidates. We generated CAR constructs differing in spacer length (XS, S, M, L) and scFv orientation (Vh-Vl, Vl-Vh) for each target candidate and thoroughly validated these CARs in co-culture assays that utilized cytotoxicity, activation marker expression and cytokine release as readouts. All CLA CARs were dysfunctional due to self-antigenicity on activated T cells. The most promising constructs for the other target candidates were evaluated in two pre-clinical PDAC mouse models. This revealed CD318 Vh-Vl XS, CD66c Vl-Vh XS and TSPAN8 Vh-Vl S as the most effective CAR constructs of the remaining targets. During CAR development we observed inferior in vivo functionality of long IgG-based spacers compared to their in vitro performance. It has been shown that these spacers convey unspecific binding in murine models. However, long spacers are regarded as beneficial when targeting membrane proximal epitopes. Thus, we evaluated a novel class of long spacers, derived from the SIGLEC family, with similar structural attributes as the IgG family but lack the natural unspecific binding sites. In non-solid and solid tumor settings, a novel SIGLEC-4 based spacer demonstrated superior functionality over long IgG spacers in vitro and in vivo with favorable activation marker expression and cytokine release for putative subsequent clinical applications. In summary, this project led to the identification of novel target candidates for PDAC and establishment of a set of highly functional targetspecific CARs that may pave the road for later clinical application and hopefully will help PDAC patients by establishing more efficient treatment options in the future.