Combining the Expression of CD33.CAR and CXCR4 to Increase CAR-CIK Cell Homing to Bone Marrow Niche and Leukemic Stem Cell Eradication in Acute Myeloid Leukemia

Chimeric Antigen Receptor (CAR) T cell therapy is a promising treatment for acute myeloid leukemia (AML), but a limited efficacy was reported from ongoing clinical trials. The capacity of engineered T cells to infiltrate into the bone marrow (BM) niche, where leukemic stem cells (LSC) reside, strongly impacts the success of the treatment. Ex vivo manipulation of CAR T cells affects the expression of several chemokine receptors and may alter the capacity of infused cells to migrate to BM. The chemokine ligand 12 (CXCL12), released by mesenchymal stromal cells (MSCs) within the medullary niche, and its chemokine receptor 4 (CXCR4) regulate leukocytes trafficking to the BM. In AML, CXCL12 binds CXCR4 over-expressed on blasts, promoting their homing in the niche. CXCR4 expression is drastically downregulated during the culture of cytokine induced killer (CIK) cells, an interesting effector T cell population with acquired NK-like cytotoxicity along with minimal alloreactivity. Therefore, combining the expression of CD33.CAR with the over-expression of CXCR4 might facilitate CAR-CIKs homing to the BM and subsequent leukemia eradication. We designed two bicistronic Sleeping Beauty transposon vectors: CXCR4(IRES)CD33.CAR and CD33.CAR(2A)CXCR4. The monocistronic CD33.CAR was used as control. We observed that both CD33.CAR(2A)CXCR4-CIKs (n=22, P<0.0001) and CXCR4(IRES)CD33.CAR-CIKs (n=9, P<0.0001) maintained CXCR4 over-expression during culture, whereas in CD33.CAR-CIKs was drastically downregulated (n=22). However, CD33.CAR expression was lower in CXCR4(IRES)CD33.CAR-CIKs (n=8, P<0.0001) compared to CD33.CAR-CIKs, while CD33.CAR(2A)CXCR4-CIKs (n=11) exhibited a significant co-expression of both proteins against control (P=0.001). CXCR4(IRES)CD33.CAR-CIKs and CD33.CAR(2A)CXCR4-CIKs maintained all CAR-associated in vitro effector functions, eliminating CD33+ KG1 target cell line, releasing cytokines (IL-2 and IFN-γ) and proliferating in an antigen-specific manner. However, CXCR4(IRES)CD33.CAR-CIKs exhibited lower effector responses against control, due to lower CAR expression. Chemotaxis assays toward recombinant CXCL12 confirmed both CXCR4(IRES)CD33.CAR-CIKs (n=7, P=0.01) and CD33.CAR(2A)CXCR4-CIKs (n=8, P=0.0006) displayed a migration advantage over CD33.CAR-CIKs (n=12) with a mean percentage of migration of 58.5% and 67.2% respectively, compared to 40.1%. Interestingly, CD33.CAR(2A)CXCR4-CIKs (n=2) showed an increased specific chemotactic response toward healthy (n=3) and AML-derived MSC (n=2) supernatants, which could be inhibited by the use of the CXCR4 antagonist Plerixafor. Moreover, when infused intravenously into NSG mice, significantly higher proportions of CD33.CAR(2A)CXCR4-CIKs were recovered in the femur BM compared to controls (P=0.0068). In conclusion, CD33.CAR(2A)CXCR4-CIKs, reaching the medullary niche more effectively, have the potential to more efficiently target the residing LSC responsible for the high relapse rates in AML.