T Cells Engineered With Chimeric Antigen Receptors Targeting NKG2D Ligands Display Lethal Toxicity in Mice.

Ligands for the NKG2D receptor are overexpressed on tumors, making them interesting immunotherapy targets. To assess the tumoricidal properties of T cells directed to attack NKG2D ligands, we engineered murine T cells with two distinct NKG2D-based chimeric antigen receptors (CARs): (i) a fusion between the NKG2D receptor and the CD3ζ chain and (ii) a conventional second-generation CAR, where the extracellular domain of NKG2D was fused to CD28 and CD3ζ. To enhance the CAR surface expression, we also engineered T cells to coexpress DAP10. In vitro functionality and surface expression levels of all three CARs was greater in BALB/c T cells than C57BL/6 T cells, indicating strain-specific differences. Upon adoptive transfer of NKG2D-CAR-T cells into syngeneic animals, we observed significant clinical toxicity resulting in morbidity and mortality. The severity of these toxicities varied between the CAR configurations and paralleled their in vitro NKG2D surface expression. BALB/c mice were more sensitive to these toxicities than C57BL/6 mice, consistent with the higher in vitro functionality of BALB/c T cells. Treatment with cyclophosphamide prior to adoptive transfer exacerbated the toxicity. We conclude that while NKG2D ligands may be useful targets for immunotherapy, the pursuit of NKG2D-based CAR-T cell therapies should be undertaken with caution.

Deciphering epitope specificities within polyserum using affinity selection of random peptides and a novel algorithm based on pattern recognition theory.

While numerous strategies have been developed to map epitope specificities for monoclonal antibodies, few have been designed for elucidating epitope specificity within complex polysera. We have developed a novel algorithm based on pattern recognition theory that can be used to characterize the breadth of epitope specificities within a polyserum based on affinity selection of random peptides. To attribute these random peptides to a specific epitope, the sequences of the affinity-selected peptides were matched against a database of random peptides selected using well-described monoclonal antibodies. To test this novel algorithm, we employed polyserum from patients infected with West Nile virus and isolated 109 unique sequences which were recognized selectively by serum from West Nile virus-infected patients but not uninfected patients. Through application of our algorithm, it was possible to match 20% of the polyserum-selected peptides to the database of peptides isolated by affinity selection using monoclonal antibodies against the virus envelope protein. Statistical analysis demonstrated that the peptides selected with the polyserum could not be attributed to the peptide database by chance. This novel algorithm provides the basis for further development of methods to characterize the breadth of epitope recognition within a complex pool of antibodies.

A Cross-Reactive Small Protein Binding Domain Provides a Model to Study Off-Tumor CAR-T Cell Toxicity.

Tumor-targeted chimeric antigen receptor (CAR)-engineered T lymphocytes (CAR-T cells) have demonstrated striking clinical success, but their use has been associated with a constellation of toxicities. A better understanding of the pathogenesis of these toxicities is required to improve the safety profile of CAR-T cells. Herein, we describe a xenograft model of off-tumor CAR-T cell-associated toxicity. Human CAR-T cells targeted against HER2 using a small-protein binding domain induced acute, dose-dependent toxicities in mice. The inclusion of a CD28 or 4-1BB co-stimulatory domain in the CAR was required to produce toxicity; however, co-stimulation through CD28 was most toxic on a per-cell basis. CAR-T cell activation in the lungs and heart was associated with a systemic cytokine storm. The severity of observed toxicities was dependent upon the peripheral blood mononuclear cell (PBMC) donor used as a T cell source and paralleled the CD4+-to-CD8+ T cell ratio in the adoptive transfer product. CD4+ CAR-T cells were determined to be the primary contributors to CAR-T cell-associated toxicity. However, donor-specific differences persisted after infusion of a purified CD4+ CAR-T cell product, indicating a role for additional variables. This work highlights the contributions of CAR-T cell-intrinsic variables to the pathogenesis of off-tumor toxicity.

A novel computer algorithm improves antibody epitope prediction using affinity-selected mimotopes: a case study using monoclonal antibodies against the West Nile virus E protein.

Understanding antibody function is often enhanced by knowledge of the specific binding epitope. Here, we describe a computer algorithm that permits epitope prediction based on a collection of random peptide epitopes (mimotopes) isolated by antibody affinity purification. We applied this methodology to the prediction of epitopes for five monoclonal antibodies against the West Nile virus (WNV) E protein, two of which exhibit therapeutic activity in vivo. This strategy was validated by comparison of our results with existing F(ab)-E protein crystal structures and mutational analysis by yeast surface display. We demonstrate that by combining the results of the mimotope method with our data from mutational analysis, epitopes could be predicted with greater certainty. The two methods displayed great complementarity as the mutational analysis facilitated epitope prediction when the results with the mimotope method were equivocal and the mimotope method revealed a broader number of residues within the epitope than the mutational analysis. Our results demonstrate that the combination of these two prediction strategies provides a robust platform for epitope characterization.

The chimeric TAC receptor co-opts the T cell receptor yielding robust anti-tumor activity without toxicity.

Engineering T cells with chimeric antigen receptors (CARs) is an effective method for directing T cells to attack tumors, but may cause adverse side effects such as the potentially lethal cytokine release syndrome. Here the authors show that the T cell antigen coupler (TAC), a chimeric receptor that co-opts the endogenous TCR, induces more efficient anti-tumor responses and reduced toxicity when compared with past-generation CARs. TAC-engineered T cells induce robust and antigen-specific cytokine production and cytotoxicity in vitro, and strong anti-tumor activity in a variety of xenograft models including solid and liquid tumors. In a solid tumor model, TAC-T cells outperform CD28-based CAR-T cells with increased anti-tumor efficacy, reduced toxicity, and faster tumor infiltration. Intratumoral TAC-T cells are enriched for Ki-67+ CD8+ T cells, demonstrating local expansion. These results indicate that TAC-T cells may have a superior therapeutic index relative to CAR-T cells.

Designed ankyrin repeat proteins are effective targeting elements for chimeric antigen receptors.

BACKGROUND: Adoptive cell transfer of tumor-specific T lymphocytes (T cells) is proving to be an effective strategy for treating established tumors in cancer patients. One method of generating these cells is accomplished through engineering bulk T cell populations to express chimeric antigen receptors (CARs), which are specific for tumor antigens. Traditionally, these CARs are targeted against tumor antigens using single-chain antibodies (scFv). Here we describe the use of a designed ankyrin repeat protein (DARPin) as the tumor-antigen targeting domain. METHODS: We prepared second generation anti-HER2 CARs that were targeted to the tumor antigen by either a DARPin or scFv. The CARs were engineered into human and murine T cells. We then compared the ability of CARs to trigger cytokine production, degranulation and cytotoxicity. RESULTS: The DARPin CARs displayed reduced surface expression relative to scFv CARs in murine cells but both CARs were expressed equally well on human T cells, suggesting that there may be a processing issue with the murine variants. In both the murine and human systems, the DARPin CARs were found to be highly functional, triggering cytokine and cytotoxic responses that were similar to those triggered by the scFv CARs. CONCLUSIONS: These findings demonstrate the utility of DARPins as CAR-targeting agents and open up an avenue for the generation of CARs with novel antigen binding attributes.

Applying bioinformatics for antibody epitope prediction using affinity-selected mimotopes - relevance for vaccine design.

To properly characterize protective polyclonal antibody responses, it is necessary to examine epitope specificity. Most antibody epitopes are conformational in nature and, thus, cannot be identified using synthetic linear peptides. Cyclic peptides can function as mimetics of conformational epitopes (termed mimotopes), thereby providing targets, which can be selected by immunoaffinity purification. However, the management of large collections of random cyclic peptides is cumbersome. Filamentous bacteriophage provides a useful scaffold for the expression of random peptides (termed phage display) facilitating both the production and manipulation of complex peptide libraries. Immunoaffinity selection of phage displaying random cyclic peptides is an effective strategy for isolating mimotopes with specificity for a given antiserum. Further epitope prediction based on mimotope sequence is not trivial since mimotopes generally display only small homologies with the target protein. Large numbers of unique mimotopes are required to provide sufficient sequence coverage to elucidate the target epitope. We have developed a method based on pattern recognition theory to deal with the complexity of large collections of conformational mimotopes. The analysis consists of two phases: 1) The learning phase where a large collection of epitope-specific mimotopes is analyzed to identify epitope specific "signs" and 2) The identification phase where immunoaffinity-selected mimotopes are interrogated for the presence of the epitope specific "signs" and assigned to specific epitopes. We are currently using computational methods to define epitope "signs" without the need for prior knowledge of specific mimotopes. This technology provides an important tool for characterizing the breadth of antibody specificities within polyclonal antisera.