Science Spotlight

The human version makes it better

from the Riddell lab (Clinical Research Division)
The figure represent the Chimeric Antigen Receptor CAR construct.
Schematic representation of the CAR construct, including an ectodomain with the variable regions (scFV). The CAR also comprises transmembrane and intracellular regions activating the signaling pathways leading to T cell stimulation when the scFv region binds to the antigen. The scFv variable region is responsible for binding to the targeted antigen expressed at the tumor surface.

Chimeric Antigen Receptor (CAR) T cells are a promising immunotherapy to treat cancer patients. In this therapy, a patient’s own T lymphocytes are isolated, expanded, and modified to present a receptor at their cell surface that will specifically recognize an antigen predominantly expressed at the tumor surface. When injected back into the patient, these modified CAR T cells track down and kill the cancer cells.

The CAR is an assemblage of the antigen receptor protein’s functional domains. The synergy of these domains allows CAR T cell activation following antigen binding and eradication of the tumor cells. However, in some cases these domains originate from a different species, such as mouse, as is the case with the variable region of the FMC63 antibody binding to the human CD19 antigen. Expression of this region at the T cell surface can be recognized by the immune system as non-self and trigger an immune response against the CAR T cells, limiting their persistence and consequently their efficacy.

To address these issues, Drs. Daniel Sommermeyer and Stanley Riddell, along with colleagues from the Clinical Research Division and Juno Therapeutics, selected a pre-existing anti-CD19 CAR construct and improved its design with the goal to reduce CAR immunogenic potential. The results of this study were published in the journal Leukemia.

The FMC63 CAR construct was used as the mouse origin version of the CAR control and for having shown efficacy in human patients in clinical trials. This region was replaced with different variable regions isolated from bone marrow and peripheral blood mononuclear cells from 20 human donors, generating a library initially containing approximately a billion sequences. After several rounds of selection to isolate the antibodies binding specifically and with the highest affinity to the CD19 antigen, the authors selected five clones for the study that bind to the same epitope as FMC63 and with a similar affinity.

Except for the human variable regions, the newly generated CAR constructs were highly similar to the one containing the FMC63 murine clone. All CAR constructs were expressed in CD8+ T lymphocytes and evaluated in vitro for their ability to lyse specifically only those cells expressing the CD19 targeted antigen. All constructs efficiently and specifically killed tumor cell lines expressing high levels of the target and produced interferon gamma cytokines, an indicator of lytic activity by effector T lymphocytes. Additionally, T lymphocytes were able to expand, indicating that CAR expression and function in these cells did not lead to T cell exhaustion (loss of the cell activation and function, observed by expression of the PD1 marker). By tagging the CAR constructs with a GFP (Green Fluorescent Protein) protein, the authors showed that the CAR aggregated in immunological synapse only in the presence of CD19+ target cells, further confirming the specificity of their activity.

Interestingly, in spite of binding to the same region of the antigen with the same affinity and being expressed at the same level as the FMC63 clone, the human versions of the CAR were more efficient at killing target tumor cells. In the antigen sequence, different amino acids were associated with antibody binding of human versus mouse variable region. However, it is not clear how this difference could contribute to a change in CAR T cells efficiency.

Finally the CAR constructs were also tested in immunodeficient mouse models allowing growth of human tumor cells. Since the tumor cells express the luciferase gene, tumor growth in this model can be monitored by luciferase quantitation, a straightforward and non-invasive method. Hence, the authors observed that both CD4+/CAR+ and CD8+/CAR+ T cells were significantly potent in vivo, and confirmed that the human constructs were more efficient than the CAR construct of mouse origin.

To further reduce immunogenicity of the CAR constructs, the authors used the prediction algorithm NetMHC to identify the potentially immunogenic regions in their CAR constructs that could be presented by the major histocompatibility (MHC) complex. One region was considered as a concern and was extended by two amino acids to decrease its MHC presentation potential. Similarly to previously generated CAR constructs, this CAR was also evaluated in vitro and demonstrated intact functional killing ability.

Altogether, these data confirm the possibility of further improvements to cell-based immunotherapy strategies by thoroughly understanding their action mechanism then allowing improvement of the construct design. By reducing the immunogenicity potential of these receptors, the new generation of CAR T cells should present increased efficiency and reduced immunogenicity that will favor their persistence in vivo for increased tumor eradication and a reduced risk of relapse. Additionally, reduced immunogenicity should also be associated with a reduced risk of toxicities.



Funding for this study was provided by  Juno Therapeutics and the National Institutes of Health.


Sommermeyer D,Hill T,Shamah SM,Salter AI,Chen Y,Mohler KM,Riddell SR. 2017. Fully human CD19-specific chimeric antigen receptors for T-cell therapy. Leukemia.