Programming CAR T Cell Tumor Recognition: Tuned Antigen Sensing and Logic Gating

Abstract

The success of chimeric antigen receptor (CAR) T cells targeting B-cell malignancies propelled the field of synthetic immunology and raised hopes to treat solid tumors in a similar fashion. Antigen escape and the paucity of tumor-restricted CAR targets are recognized challenges to fulfilling this prospect. Recent advances in CAR T cell engineering extend the toolbox of chimeric receptors available to calibrate antigen sensitivity and combine receptors to create adapted tumor-sensing T cells. Emerging engineering strategies to lower the threshold for effective antigen recognition, when needed, and enable composite antigen recognition hold great promise for overcoming tumor heterogeneity and curbing off-tumor toxicities.

Significance: Improving the clinical efficacy of CAR T cell therapies will require engineering T cells that overcome heterogeneous and low-abundance target expression while minimizing reactivity to normal tissues. Recent advances in CAR design and logic gating are poised to extend the success of CAR T cell therapies beyond B-cell malignancies.

Figure 1.

Structural design and sensitivity of CARs and CD3 complex–based receptors. A, CARs encompassing the 4-1BB (left) and CD28 (middle) costimulatory domains exhibit distinct antigen sensitivity in vitro and in vivo, with CD28-based CAR having superior antigen sensitivity (i.e., requires lesser antigen density for cytolysis). T cells expressing an HLA-independent TCR (HIT) receptor (right) that contains the same V_L and V_H domains require 10-fold lower antigen density than CD28ζ. B, CD3 complex–based receptors. HIT, synthetic TCR and antigen receptor (STAR), TCR-like CAR (TCAR), and antibody-TCR (AbTCR) receptors are based on fusing antibody variable domains to TCR constant regions (human Cα, Cβ for HIT and TCAR; mutated mouse Cα, Cβ for STAR, and human Cγ, Cδ for AbTCR). STAR, TCAR, and AbTCR are expressed using lentiviral or γ-retroviral, which results in heterogeneous expression. HIT T cells are engineered by targeting a V_H-Cβ-P2A-V_L-Cα (exon1) transgene into the TRAC locus, which leads to the disruption of the endogenous TCR and expression of the HIT receptor depending on the endogenous promoter and polyA signal. In ε-TRuC T cells, an scFv–CD3ε fusion is overexpressed using lentiviral vectors. This fusion is expected to compete with the endogenous CD3ε to get incorporated into the full TCR–CD3 complex. The TAC receptor is formed between the endogenous TCR–CD3 complex with a fusion composed of two scFvs (in tandem) and a truncated CD4 (lacking the MHC interacting domain), with one scFv specific to the antigen and the other specific to CD3ε, which leads to TCR–CD3 activation. Except for HIT T cells, all these T cells are expected to express residual TCR, which can lead to alloreactivity. Ag, antigen; KI, knockin; KO, knockout.

Figure 2.

"OR" logic-gate CAR T cell designs. A, Dual-CAR, two distinct fully functional CARs targeting different antigens are coexpressed on the surface of the same T cell. Identical or different costimulatory domains can be used. B, Tan-CARs, left, bivalent single CAR chain with two binding domains (scFvs) in tandem targeting two distinct antigens. Middle, bivalent single CAR chain with two distinct binding domains (VHH, single-domain antibodies) in tandem targeting different epitopes of the same antigen. Right, multivalent single CAR chain using designed ankyrin repeat proteins (DARPIn) in tandem targeting more than two antigens. C, Parallel-CAR, one fully functional CAR (second generation) and a chimeric costimulatory receptor (CCR) with identical hinge domains that allow CAR-CCR heterodimerization are coexpressed on the surface of the same T cell. D, Zip-CAR, Zip-CAR expressed on the surface of T cells binds to distinct Zip-Fv (Zip-scFv) targeting different antigens. Zip-CAR uses a leucine zipper adapter expressed on the T cells that bind to administered Zip-Fv specific for antigens A or B. E, CAR-BiTEs, CAR T cell coexpressing bispecific T cell engagers (BiTEs). BiTEs are directed against CD3 and an antigen distinct from the CAR target.

Figure 3.

Principles of logic-gated CAR T cells. A, OR-gate, T cells coexpress fully functional CARs targeting distinct tumor antigens A and B. B, AND-gate, T cells coexpress a CAR specific for antigen A and a CCR specific for antigen B. CAR T cells are fully activated when the CAR and the CCR simultaneously engage with antigens A and B coexpressed on the tumor but not on normal cells. C, NOT-gate, T cells coexpress a fully functional CAR specific for antigen A and an inhibitory CAR (iCAR) specific for antigen B. T cells are fully activated when the CAR engages with antigen A expressed exclusively on tumor cells. iCAR engagement with antigen B expressed on normal cells reversibly inhibits CAR T cells. D, IF-THEN-gate, T cells coexpress synthetic Notch (SynNotch) receptor specific for antigen A. Engagement of SynNotch receptor with antigen A (left) induces transient expression of fully functional CAR specific for antigen B (right) in the tumor environment. The decay of CAR expression (spatiotemporal regulation) in circulating T cells should allow for the protection of normal cells expressing the tumor-associated antigen B. The gray arrow between implies time. E, IF-BETTER-gate, T cells coexpress a fully functional CAR specific for antigen A and a CCR specific for antigen B. CAR T cells are fully activated if the CAR engages with antigen A expressed at high levels on tumor cells. If tumors express antigen A at low levels (small-size antigen A, middle tumor cell), full T cell activation requires CCR engagement with antigen B on the same tumor cells. Antigen A can be expressed alone, not with antigen B, at low levels in normal tissues.

Figure 4.

Site-specific integration of CAR cDNA in T cells. A, In TRAC-CAR T cells, the CAR gene is inserted upstream of the TRAC exon 1, and it is flanked by splicing acceptor (SA) and 2A sequences to the 5′ end and polyadenylation (pA) sequence to the 3′ end. CAR expression is controlled by the endogenous TCRα promoter. This strategy also leads to the disruption of endogenous TCRα expression and consequently to the disruption of the TCR–CD3 complex surface expression. A number of other advantageous features are highlighted to the right. AAV, adeno-associated viruses; dsDNA, double-stranded DNA; HDR, homology-directed repair; KI, knockin; ssDNA, single-stranded DNA. B, In PDCD1-EF1α-CAR T cells, an EF1α-CAR-pA transcription unit is inserted in the exon 1 of the PDCD1 locus in an orientation opposite to PDCD1 transcription directionality. This strategy shares some features with the TRAC-CAR approach, indicated at the right, but there also are clear distinctions.