Chimeric antigen receptor signaling: Functional consequences and design implications

Abstract

Chimeric antigen receptor (CAR) T cell therapy has transformed the care of refractory B cell malignancies and holds tremendous promise for many aggressive tumors. Despite overwhelming scientific, clinical, and public interest in this rapidly expanding field, fundamental inquiries into CAR T cell mechanistic functioning are still in their infancy. Because CAR T cells are manufactured from donor T lymphocytes, and because CARs incorporate well-characterized T cell signaling components, it has largely been assumed that CARs signal analogously to canonical T cell receptors (TCRs). However, recent studies demonstrate that many aspects of CAR signaling are unique, distinct from endogenous TCR signaling, and potentially even distinct among various CAR constructs. Thus, rigorous and comprehensive proteomic investigations are required for rational engineering of improved CARs. Here, we review what is known about proximal CAR signaling in T cells, compare it to conventional TCR signaling, and outline unmet challenges to improving CAR T cell therapy.

Fig. 1. CAR versus TCR structure.

TCRs (left) are a multisubunit antigen recognition complex in which the TCRα and TCRβ chains recognize peptide in the context of major histocompatibility complex (MHC) molecules and associate with signaling molecules CD3δ, CD3ε, CD3γ, and CD3ζ (shown in gold). TCRs also associate with a coreceptor, either CD4 (shown in green) or CD8. Minimally, CARs (center and right) are built around an antigen-binding extracellular domain, either an antibody-derived scFv (center) or a receptor-binding ligand or peptide (right). These antigen recognition domains are linked through a flexible immunoglobulin domain-containing hinge region (for scFvs; center) or a hinge and immunoglobulin-based scaffold (for receptor-binding constructs; right) to a transmembrane domain (green) and then to signaling domains. First-generation CAR constructs (not shown) have only the cytoplasmic tail of CD3ζ, whereas subsequent generations contain one (second generation; center) or more (third generation example at right) costimulatory domains membrane-proximal to a CD3ζ tail (gold). IgG, immunoglobulin G; F_c, fragment crystallizable.

Fig. 2. Canonical TCR-mediated activation requires three distinct signals, whereas CAR signaling is less discrete.

(A) Antigen-dependent ligation of the TCR complex (left), termed signal 1, initiates T cell activation. The TCR:peptide-MHC complex stabilizes coreceptor-MHC interactions (CD8 shown in green), which results in more recruitment and activation of the Src family tyrosine kinase LCK. Signal 2 is mediated by costimulatory molecules such as CD28, 4-1BB, and OX40; in T cells, signal 2 is not completely contemporaneous with signal 1 and may not occur in exactly the same place. In CAR T cells (right), in contrast, signal 1 and signal 2 are mediated by the same physical event of antigen recognition. Signal 3 for both canonical T cells and CARs is provided through cytokine signaling and usually occurs later than both signal 1 and signal 2. (B) The TCR complex (left) comprises 10 immunoreceptor tyrosine-based activation motifs (ITAMs), depicted in red. Each ITAM incorporates two tyrosines (Y), each of which is phosphorylated by LCK and other Src family tyrosine kinases. Phosphorylated ITAMs serve as docking sites for ZAP-70 and other Src homology 2 (SH2) domain–containing proteins, which nucleate signaling cascades, leading to full activation. CAR constructs (right) only comprise six ITAMs if the CAR dimerizes, and three if it does not. Although it is clear that ITAM multiplicity has strong effects on CAR signaling, how to incorporate ITAM number and position into CAR design has not yet been optimized.

Fig. 3. TCRs and CARs form distinct ISs.

Canonical TCR signaling (left) leads to the formation of an IS with clearly demarcated zones: the central supramolecular activation complex (cSMAC), the zone of closest physical proximity between the T cell and the target cell, incorporates TCR:peptide-MHC complexes as well as costimulatory molecules such as CD28 and CD2; the peripheral SMAC (pSMAC) includes larger molecules such as LFA1, in addition to molecules like CD2; and the distal SMAC (dSMAC) includes large, ligandless proteins such as CD45. This stereotypical bull’s eye–like structure evolves over time, and CD45 is eventually recruited back to the cSMAC before degranulation occurs at the secretory domain. CAR synapses (right), on the other hand, are both spatially and temporally disorganized. They form more quickly and are less stable and comprise much less well-demarcated zones. Rather than a cSMAC, there are central CAR microclusters interspersed with and surrounded by adhesion molecules such as LFA-1 and signaling molecules such as CD45. Around this central area, there is an actin-rich pseudoring that is relatively CAR poor. Moreover, the area of cell-cell contact between CAR T cells and their targets appears to be quite convoluted, perhaps due to the varying extracellular sizes of the numerous proteins at the IS. Nonetheless, there is evidence that this disorganized IS leads to more rapid degranulation and disengagement from target cells, which ultimately may result in faster killing.