In the last years, adoptive cellular immunotherapy employing T lymphocytes genetically modified with Chimeric Antigen Receptors (CARs) has demonstrated impressive clinical results, particularly in the treatment of acute/chronic lymphoblastic leukemia and B-cell lymphoma, paving the way towards the possibility to translate this approach also to other hematological malignancies, such as Acute Myeloid Leukemia (AML).

In the AML context, the CD123 antigen (IL-3 receptor alpha subunit) represents a good target antigen, being a poor prognosis over-expressed marker on AML cells and leukemic stem cells (LSCs), a rare population that plays a key role in perpetuating leukemia. However, CD123 is also expressed on the surface of healthy cells such as monocytes and endothelial cells, although at lower levels as compared to leukemic cells. The potential recognition of low antigen positive healthy tissues by CAR-redirected T cells, through the so called "on-target-off-organ" effect, limits a safe clinical employment of this immunotherapeutic approach. CARs are artificial receptors generated by joining the cytoplasmic TCR (T Cell Receptor) signaling modules to the heavy and light chain variable regions of a monoclonal antibody, whose affinity toward a target antigen is a variable capable of influencing the CAR-mediated functional responses.

Therefore, in our study we investigated how the CAR affinity variable in the context of CD123 targeting, together with the CAR and CD123 target antigen density, could impact anti-CD123. CAR-redirected effector cells efficacy against leukemic cells and safety towards the healthy cells.

To this aim, Cytokine-Induced Killer (CIK) effector cells have been genetically modified with four Chimeric Affinity Mutants (CAMs), CAM-1, CAM-2, CAM-3 and CAM-4, identified by means of a computational docking technique.

In vitro cytotoxic assays, cytokine production and proliferation experiments have been performed in order to evaluate both the efficacy and safety profile of the CAR-redirected CIK cells, using un-manipulated CIK cells (NO DNA) and wild-type anti-CD123.CAR condition as controls.

The functional characterization of all the CAMs revealed both the specificity and the effectiveness of CIK-CAR+ cells against the CD123+ THP-1 cell line and primary AML cells. However we observed that, at least in the context of CD123 targeting, a good CAR expression level is necessary for inducing effective later functions, such as proliferation and cytokine production, towards a high CD123+ target.

When introducing leukemic cell lines with different CD123 density on their surface we observed different effector properties minimally influenced by the CAR affinity. In particular, in terms of killing activity, we noticed that a number of ≈1600 CD123 molecules is sufficient to induce a good cytotoxic response of all the CARs tested, with the CAM-2 (2-magnitude log lower in affinity) being less powerful. At the same time, this antigen density is not enough to determine a good proliferative capability which instead occurs with leukemic target cells expressing 5000 or more CD123 molecules. Considering the cytokine production (IL-2 and IFN-gamma), we observed that all CIK-CAR+ cells showed a cytokine release that is directly proportional to the target antigen density, with CAM-2 showing a reduced response towards low-CD123 expressing leukemic targets.

When analyzing the safety profile of the CAMs against low-CD123+ endothelial target cells (≈1600 molecule/cell), we observed a lower functional activity of the CAMs as compared to the leukemic cell lines expressing the same level of CD123 molecules on their surface, with the low-affinity CAM-2 showing a major sparing capability in terms of killing activity (being the only one not statistically different from NO DNA).

In conclusion, exploiting our model of affinity mutants we were able to in vitro characterize the role of the CAR density balanced with the affinity of the anti-CD123.CAR towards AML cells expressing different CD123 levels and CD123-low expressing normal tissues. In particular, the results obtained with CAM-2 suggested a potential threshold of affinity below which, even if the safety profile is preserved, the anti-leukemic efficacy would be impaired.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.