In the treatment of B cell leukemia, relapse due to antigen to loss or downregulation remains a major challenge. Tumor antigen escape may be mitigated by multi-targeting CAR T cells redirected to CD19 and CD22, and possessing a superior sensitivity to low-density antigens. Using lentiviral transduction of primary T cells, flow cytometry, cell-based assays, and xenograft mouse models, we systematically optimized 22-19 CAR architecture and co-stimulatory domains for best functionality.

Fully-human tandem 22-19 CARs with co-stimulatory domains derived from 4-1BB, CD28, ICOS, OX40 or CD27, and hinge and transmembrane domains derived from CD8, CD28, or OX40 were evaluated. The tandem targeting ScFv domain orientation 22-19 was selected based on greater expression and cytotoxicity vs 19-22. All CARs achieved high T cell expression (mean 50-90%), and efficient dose-dependent killing of RajiCD19+CD22+, 293TCD19+, 293TCD22+, but not 293TCD19-CD22- target cells, and elaborated IL-2, IFN-γ, and TNF-a in antigen-dependent manner. CARs' potency in vitro varied by co-stimulatory domain: 4-1BB< OX40, ICOS, CD27<CD28, 4-1BB-CD28. The 22-19 tandem CAR with hinge and transmembrane domains (H/TM) derived from CD8 (CD8/CD8) performed as well as the CAR with H/TM CD28/CD28 configuration, whereas configuration OX40/OX40 was superior to CD8/OX40.

In RajiCD19+CD22+ mouse xenografts, the potency of tumor rejection by the 2nd generation tandem 22-19 CARs was also dependent on co-stimulatory domain, ranking 4-1 BB<CD27, ICOS<OX40<CD28. Tandem 3rd generation CAR 22-19, combining CD28 and 4-1BB co-stimulation, achieved the greatest anti-tumor effect. In antigen-heterogeneous Raji xenografts of a mixture of CD19-CD22+, CD19+CD22-, and CD19+CD22+ clones, the 22-19 tandem CARs with CD28, OX40 or ICOS co-stimulation mediated rapid and complete tumor rejection, whereas the conventional single-targeting CAR19 or CAR22 enabled tumor progression due to antigen escape.

Low antigen density Raji clones were generated by CRISPR-Cas9-mediated disruption of both CD19 and CD22 expression, followed by lentiviral transduction to express a limited number of antigen molecules on the cell surface. When challenged with Raji CD22 low clone in vitro, 2nd generation tandem CARs with CD28 and ICOS co-stimulation, and the 3rd generation tandem CAR combining CD28 and 4-1BB co-stimulatory domains were more effective than 4-1BB-, CD27-, or OX40-containing tandem CARs. Against RajiCD19 low clone, CARs with CD27 and OX40 domains were more effective than CARs with 4-1BB, and CARs with ICOS or CD28 co-stimulation were the most potent.

In summary, the fully-human tandem 22-19 CARs incorporating ICOS and CD28 co-stimulatory domains mitigate tumor antigen escape, exhibit robust anti-tumor function in pre-clinical models, enable superior lysis of CD22low and CD19low tumor clones, and may help improve clinical outcomes.

Disclosures

Hu:Lentigen, a Miltenyi Biotec Company: Current Employment, Patents & Royalties: CAR-T immunotherapy. Xiong:Lentigen, a Miltenyi Biotec Company: Current Employment. Wu:Lentigen, a Miltenyi Biotec Company: Current Employment. Zhu:Lentigen, a Miltenyi Biotec Company: Current Employment, Patents & Royalties: CAR-T immunotherapy. Dropulic:Lentigen, a Miltenyi Biotec Company: Current Employment, Patents & Royalties: CAR-T immunotherapy. Schneider:Lentigen, a Miltenyi Biotec Company: Current Employment, Patents & Royalties.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution