Immunotherapy Two antigens are better than one

Nature Reviews Cancer | Research Highlight

Immunotherapy
Two antigens are better than one

Sarah Seton-Rogers

Nature Reviews Cancer 16,128–129(2016)doi:10.1038/nrc.2016.17

Published online
19 February 2016


T cells carrying chimeric antigen receptors (CARs) or engineered T cell receptors (TCRs) have shown remarkable efficacy against some tumour types, primarily B cell malignancies. However, the use of these engineered T cells is limited to tumours that express highly specific antigens; in most cases tumour antigens are also expressed in normal 'bystander' tissues, and T cells that target them can cause lethal side effects.

Jennie Vallis/NPG

One solution to increase the specificity of engineered T cells might be to design a new receptor that is independent of CAR and TCR pathways, but that can reliably activate a CAR for a second antigen.
Morsut, Roybal et al. first designed a modular receptor based on Notch (SynNotch receptor) that binds to a target antigen and then triggers receptor cleavage to release a transcriptional activation domain that can activate a target gene(s) of choice.
Roybal et al. then used this platform to design a system in which one tumour antigen activates SynNotch in T cells, thereby activating transcription of a CAR that recognizes a second tumour antigen, ultimately leading to T cell activation.
The authors designed a proof-of-concept experiment in Jurkat T cells using a SynNotch receptor that recognized CD19 and contained a tetracycline-transactivator (tTa) domain; the tTa domain then activated transcription of a tetracycline response element-driven CAR against mesothelin.
In culture, these Jurkat cells were only activated by tumour cells that expressed both CD19 and mesothelin, and not those expressing only one of these antigens.

They then designed a similar system in primary human CD4+ or CD8+ T cells. In these cells, they found that a SynNotch receptor containing the Gal4-VP64 transcriptional activation domain had low basal transcriptional activity, which is required to prevent the induction of CAR expression in the absence of the first antigen.

These cells expressed a SynNotch–Gal4-VP64 that bound green fluorescent protein (GFP) and a CAR against CD19. K562 leukaemia cells that expressed both GFP and CD19 activated these primary T cells effectively in vitro, but K562 cells lacking either antigen did not; only K562 cells expressing both antigens were killed by these engineered T cells. Similar results were observed with two different SynNotch–CAR systems in primary T cells that responded to tumour cell CD19 and mesothelin, or to GFP and mesothelin.

Primary human T cells with the GFP SynNotch and CD19 CAR also exhibited localized activation in tumours in vivo. Immunocompromised mice were injected with CD19+ Daudi B cell lymphoblastoid cells; cells injected in one flank were also GFP+. Following tumour development, engineered T cells were injected; the CD19 CAR was expressed only in T cells within GFP+ tumours. Furthermore, in a similar bilateral tumour model, the engineered T cells promoted clearance of established K562 xenografts expressing both antigens but not of bystander cells that lacked GFP. In addition, when bilateral models were established with K562 cells each expressing only CD19 or GFP, engineered SynNotch T cells were not activated, indicating that SynNotch activation by one cell and then CAR activation by another following T cell migration (which could lead to off-target side effects) does not occur.

“in principle it is possible to use this flexible, modular system to design more effective and safer T cell therapies”

Although much additional work is required before these types of T cell could be tested clinically, these results demonstrate that in principle it is possible to use this flexible, modular system to design more effective and safer T cell therapies for a wider range of tumour types.