An Optimized GD2-Targeting Retroviral Cassette for More Potent and Safer Cellular Therapy of Neuroblastoma and Other Cancers

Abstract

Neuroblastoma is the commonest extra cranial solid cancer of childhood. Despite escalation of treatment regimens, a significant minority of patients die of their disease. Disialoganglioside (GD2) is consistently expressed at high-levels in neuroblastoma tumors, which have been targeted with some success using therapeutic monoclonal antibodies. GD2 is also expressed in a range of other cancer but with the exception of some peripheral nerves is largely absent from non-transformed tissues. Chimeric Antigen Receptors (CARs) are artificial type I proteins which graft the specificity of a monoclonal antibody onto a T-cell. Clinical data with early CAR designs directed against GD2 have shown some promise in Neuroblastoma. Here, we describe a GD2-targeting CAR retroviral cassette, which has been optimized for CAR T-cell persistence, efficacy and safety.

Conflict of interest statement

Competing Interests: MP has received contract research funding from Cellectis. He owns stock and receives salary contribution from Autolus Ltd. He is an inventor on patents filed by UCLB and has received and may receive royalties therefrom. He has received honoraria from Roche and Amgen for speaking. JA and ST own stock from Autolos Ltd. They are inventors on patents filed by UCLB and may receive royalties therefrom. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.

Figures

Fig 1. Comparison between huK666 and muK666 based CARs.

(a) Comparison of amino acid sequences of huK666 and muK666 scFvs are compared. Complementarity determining regions are shown in red. The linker sequence is shown in green. On right, architectures of the CARs generated for initial comparison of huK666 with muK666. Both differ only in the scFvs used being either the original murine muK666 sequences, or the humanized (CDR grafted) huK666. Otherwise the CARs comprise of a human IgG1 hinge-CH2-CH3 spacer, a TM domain derived from CD28 and a compound endodomain comprising of fusions between endodomains from CD28, OX40 and CD3-Zeta. (b) Stability of muK666 vs huK666 CARs. T-cells from 5 donors were transduced with equal titers of retroviral supernatant coding for muK666 or huK666 CARs. CAR expression was detected by anti-human-Fc polyclonal antibodies. MFI was identical in both constructs. A representative example is shown with NT (green), muK666 (blue) and huK666 (red) histograms overlaid. These CAR T-cells were challenged with LAN-1 (a Neuroblastoma cell line expressing GD2) and A204 (a rhabdomyosarcoma cell line which is GD2 negative). Cytotoxicity is shown in (c), proliferation at day 7 is shown in (d), IFN-γ in (e) and IL-2 release in (f). Data shown as means +/-SEM from 6 independent experiments with different donors.

Fig 2. Effects of spacer on CAR function.

Several huK666 CARs were cloned into an identical format whereby the scFv was tagged with an HA-tag, and the CAR was co-expressed with a truncated FMD-2A sequence with truncated CD34. The expression cassette is shown in (a) and cartoons of these formats are shown in (b). Co-expression of CAR (HA) and CD34 marker gene from the different cassettes. The four hinge regions were compared in terms of cytotoxicity (c) (Lan1 = GD2 positive target and A204 = GD2 negative target), IFN-γ secretion (d), IL-2 (e) and target-specific proliferation (f). Data shown as means +/-SEM from 4 independent experiments with different donors.

Fig 3. Removal of FcR binding motifs in IgG1 spacer.

(a) amino acid sequence of the wild type (top) and mutated (PVAA) CH2 regions responsible for FcγR binding. (b) flow staining with anti-Fc antibody to show comparable level of expression of receptor with or without the PVAA mutation. Identical CARs with and without PVAA were compared side by side in terms of cytotoxicity against GD2 engineered SupT1 and GD2 negative wild type SipT1 cells (c), cytotoxicity against GD2 positive Lan1 cells and FcRγ positive THP-1 cells (d), and IL-1β release on culture with THP-1 cells (e). Data shown as means +/-SEM from 4 independent experiments with different donors.

Fig 4. Optimization of expression cassette for co-expression with iCasp9.

(a) Nine different expression cassettes were compared: Three different retroviral vectors were generated–wild-type SFG, SFG with scaffold-attachment region (SAR) inserted into the 3’UTR and SFG with CHS4 inserted into the 3’LTR U3 region. (retroviral vectors with both SAR and CHS4 were generated but produced very low vector titers and were not compared further); Into these retroviral vectors wild-type iCasp9-2A-huK666 CAR constructs were inserted and iCasp9 were inserted in 3 forms; codon-optimized, wild-type, and with the catalytic domain mutated. (b) Histogrammes at day 3 after transduction (blue lines). Overlaid T-cells cultured in 20nM CID are shown in red. Bargraphs of MFI of cells in the absence of CID at day 3 (c) and (d) day 7. Data shown as means +/-SEM from 5 independent experiments with different donors.

Fig 5. Consequence and function of iCasp9.

The SAR codon-optimized cassette was taken further and compared with the original cassette without iCasp9. (a) Expression of the CAR was unchanged. Depletion is shown by facs after addition of CID. The function of cassettes with and without iCasp9 were assessed by (b) Killing (c) IFN-γ release and (d) IL-2 release. Data shown as means +/-SEM from 4 independent experiments with different donors.

Fig 6. In vivo testing of GD2 CAR.

Balb-C mice were innocultaed with 1x106 CT26 or CT26-GD2 cell mixed in matrigel. 10 days post tumour inoculation, mice were sub-lethally irradiated (200 rads) and two weeks post tumour injection, mice were intravenously transplanted with a total of 1.5x106 transduced splenocytes or non transduced controls by tail vein injection. Wild type CT26 were used as antigen negative controls (a,c) whereas CT26 –GD2 tumors treated with untransduced T cells (b) did not regress whilst CT26-GD2 tumors challenged with CAR T cells regressed. Each line represents one mouse.