Overcoming a Critical Obstacle Towards Effective and Safe CAR T-Cell Therapeutics

Chimeric Antigen Receptor (CAR) T-cells have shown great promise in achieving stable remission of B cell acute lymphoblastic leukemia and are being utilized to treat a variety of other hematologic and non-hematologic malignancies. However, this therapy is associated with significant adverse effects (e.g., cytokine release syndrome and neurotoxicity), largely related to the high cell doses requisite for achieving therapeutic response(s). E-selectin, an endothelial molecule expressed on microvessels of bone marrow and tumors, is critical for extravasation of blood-borne cells. E-selectin binds to sialyl Lewis X (sLe^X), a sialofucosylated tetrasaccharide motif decorating specialized glycoproteins and glycolipids on circulating cells. Given that better site-specific delivery of administered CAR T-cells would yield more efficacious immunotherapy while limiting off-target effects, we sought to determine whether these cells express E-selectin ligands.

CAR T-cells targeting glioblastoma were created by lentiviral transduction of human T-cells with anti-EGFR-BBz-mCherry CAR construct. Transduced cells (T) and untransduced cells (UT) were subjected to stimulation with anti-CD3/CD28 microbeads followed by culture expansion in presence of interleukin-2 (IL2) for 10 days ("10"). In some cases, cells were expanded for 7 additional days by co-culturing with irradiated U87 glioblastoma cell line ("17"). These cells were compared to resting T-cells (non-expanded; NE). Expression of sLe^X on each T-cell population (NE, T10, T17, UT10 or UT17) was measured by flow cytometry.

We observed that NE T-cells display moderate expression of sLe^X as previously reported in literature (Silva, Fung et al., J. Immunol., 2017). T10 CAR T-cells display significantly lower sLe^X compared to NE T-cells, while T17 CAR T-cells completely lack sLe^X. Similar pattern of sLe^X expression was observed on UT10 and UT17 populations, indicating that culture expansion progressively depletes sLe^X expression on T-cells, regardless of transduction with a CAR construct.

To analyze the capacity of CAR T-cells to bind E-selectin under hemodynamic shear conditions, we utilized a microfluidic flow chamber wherein T-cells are perfused over human umbilical vein endothelial cell (HUVEC) monolayers treated with TNFα to induce E-selectin expression. Consistent with the observed cell surface sLe^X levels, NE T-cells display robust rolling interactions on TNFα-stimulated HUVECs, whereas T10 and T17 CAR T-cells exhibit essentially no endothelial interactions. To overcome this shortfall, we assessed whether sLe^X expression can be enforced on the CAR T-cell surface via glycoengineering using glycosyltransferase programmed stereosubstitution (GPS), i.e., fucosyltransferase-mediated α(1,3)-fucose installation that converts terminal sialylated type 2 lactosamines to sLe^X. GPS engenders markedly increased CAR T-cell sLe^X expression, yielding full (100%) sLe^X -expression of the CAR T-cell populations, while increasing sLe^X expression from a baseline of 20-30% to maximally 60-70% of the NE T-cell populations. The glycoengineered CAR T-cells display markedly enhanced (>20 fold) adhesive interactions on TNFα-stimulated endothelial cells compared to that of unfucosylated CAR T-cells. Collectively, our data indicate that culture expansion induces down-regulation of endogenous α(1,3)-fucosylation, and, importantly, the missing fucose can be restored via GPS. Finally, to assess whether the enforced E-selectin binding has a meaningful biologic effect, we examined homing of GPS-modified cells to an E-selectin bearing bed (marrow) within 24 hours of intravenous administration. We consistently observed >3-fold increase in marrow infiltrates of culture-expanded GPS-modified T-cells compared to their unmodified counterparts.

Overall, these results draw attention to an inherent inability of CAR T-cells to migrate to sites where they are needed. Our data reveal that culture expanded T-cells lack E-selectin binding capacity, however, sLe^X display can be enforced by cell surface glycoengineering. Improved E-selectin binding yields superior tissue colonization. As such, glycoengineering CAR T-cells may translate into lowering of cell dosing, thereby improving both the efficacy and safety, and reducing both cell expansion and clinical care costs, of this promising therapeutic approach.