Abstract
Introduction
Antigen density has emerged as an important factor regulating CAR-T cell activity with low antigen density being associated with reduced potency in pre-clinical models across several targets. In clinical trials of CD22 CAR-T cells, the low and heterogeneous CD22 expression in B-ALL enabled CD22lo blasts to escape CAR-T cell detection leading to antigen dim relapses (Nat Med. 2018, 24:20-28).
We have previously reported on the development of novel CD21 directed CAR-T cells (CAR21) for the treatment of T-cell acute lymphoblastic leukemia (T-ALL) (Blood. 2021, 138:902). Donor derived CAR21 T cells targeting membrane proximal epitopes and utilizing a Fab-CAR architecture were capable of specific cytotoxicity and cytokine secretion against CD21 positive target cells in vitro and were active in vivo. However, in co-culture assays, CAR21 cells secreted less IL-2 against cell lines and primary tumors expressing CD21 at very low antigen density (<1000 copies/cell).
We have shown that CD21 is expressed in ~60% of primary T-ALL samples (n=58) although at low antigen density (mean 1262 copies/cell) compared to more established targets such as CD19. We wanted to explore the impact of this low antigen density on CD21 CAR-T function and whether it was possible to upregulate CD21 density by pharmacological means.
Results
We have previously shown that CD21 is a direct transcriptional target of the NOTCH-MYC pathway and that PI3K/mTOR inhibition of T-ALL results in activation of this pathway and consequent CD21 up-regulation (Leukaemia. 2013, 27:650).
We therefore incubated six T-ALL cell lines with four inhibitors of the PI3K/mTOR/AKT pathway (AZD5363, GDC0941, BEZ235 and rapamycin) for 48 and 72hrs and showed surface CD21 up-regulation by flow cytometry of > two-fold at both time points for all inhibitors, with BEZ235 showing the greatest fold increases across all cell lines. Incubation with the gamma-secretase inhibitor L685 resulted in significantly down regulated CD21 expression across all cell lines in keeping with the known NOTCH regulation of CD21 expression.
We were next interested in the persistence of this enhanced CD21 expression after drug withdrawal. In two T-ALL cell lines, PF382 and RPMI8402, the impact of BEZ235, rapamycin and L685 was sustained at 48hrs after washing off.
PI3K/ mTOR mediated CD21 manipulation did not occur in a CD21 negative T-ALL cell line, normal donor B and T cells or B-ALL patient samples suggesting the effect was specific to T-ALL.
When clinically approved PI3K inhibitors (Idelalisib, Duvelisib, Copanlisib, Everolimus) were used, an increase in CD21 expression was again seen across 5 T-ALL cell lines tested. Greatest fold increases (> 2-fold) in CD21 expression were seen with the dual PI3Kα and PI3Kδ inhibitor copanlisib and the mTOR inhibitor rapamycin. No consistent up regulation of CD7 or CCR9 by flow was seen after 48-hour copanlisib incubation suggesting that the effect was specific to CD21.
PI3K inhibition was additionally able to mediate 2-fold increases in CD21 expression in a CD21 low patient-derived xenograft T-ALL sample. Further, in a MOLT-4 murine xenograft model of T-ALL, 2 doses of copanlisib were able to produce a 2-fold increase in CD21 expression.
These PI3K-mediated increases in CD21 antigen density translated into improved in vitro CAR21 functionality. Across 3 T-ALL cell lines - MOLT-4, CUTLL-1 and RPMI-8402, 48hr pre-incubation with both rapamycin and copanlisib prior to co-culture with CAR21 or CAR19 cells, resulted in improved CD21 specific cytotoxicity, IFN-G and IL-2 release.
However, in the same co-culture assay, when copanlisib was added directly to the co-culture rather than via pre-incubation, there was a significant attenuation of IFN-G and IL-2 production by CAR21 suggesting a direct inhibitory effect on CAR-T effector functions in keeping with the known immunosuppressive effects of PI3K inhibitors.
Conclusion
In summary, we have shown that CD21 expression in T-ALL can be increased through PI3K pathway inhibition and that enhanced CD21 antigen density through pharmacological priming results in improved in vitro CAR21 efficacy. In vivo models analyzing the impact of copanlisib tumor priming on CAR-T efficacy and persistence are ongoing. Target antigen modulation is a promising strategy for enhancing CAR efficacy and reducing risk of antigen negative relapse in low antigen density targets.
Disclosures
Maciocia:Autolus Ltd: Current equity holder in publicly-traded company. Maciocia:Autolus Ltd: Current equity holder in publicly-traded company. Pule:Autolus Ltd: Current Employment, Current equity holder in publicly-traded company.
Author notes
Asterisk with author names denotes non-ASH members.