T cells, the warriors of the immune system, are able to recognize and destroy tumorous and virus infected cells. T cell response is dictated by both antigen-specific signals from the T cell receptor (TCR) and antigen-independent signals from co-signaling receptors. Two sets of co-signaling receptors are expressed on the T cell surface: costimulatory receptors , which deliver positive signals that are essential for full activation of T cells, and co-inhibitory receptors which attenuate T cell signaling, and act as ‘molecular brakes’ or ‘checkpoints’ against uncontrolled T cell activities. Cancer and viruses are able to hijack T cell co-inhibitory receptors to escape immune attack. Antibodies that block the co-inhibitory receptors PD-1 or CTLA-4 have proven durable clinical benefit in a variety of cancer indications.
Despite the huge clinical success of extracellular antibodies in cancer immunotherapy, the fundamental intracellular mechanisms of coreceptor signaling remain poorly understood, due to the lack of viable in vitro and ex vivo models. We recently developed novel in vitro and ex vivo models to push the field towards an in-depth, quantitative understanding of immune checkpoints. These include three systems:
Fig. 1. Cancer immunotherapy via PD-L1/PD-1 blockade. Ideally, tumor cells are perceived as foreign cells by T cells, and killed, because tumor cells display proteins that can stimulate T cells. However, many tumor cells have evolved to produce membrane proteins (e.g., PD-L1, PVR, and Gal-9) that can blind the T cells by triggering T cell inhibitory coreceptors (e.g., PD-1, TIGIT and TIM-3), causing T cell hypofunction and tumor survival. Antibodies that block PD-L1/PD-1 binding can reactivate T cells to kill some but, not all, tumors. ‘+’: activating, ‘-’: inhibitory.
1. Cell free reconstitution
Receptor-ligand binding on the cell surface delivers a signal to the cell interior via protein-protein interactions, enzymatic reactions, and spatial reorganization events at the plasma membrane. However, biochemistry of T cell signaling reactions is traditionally done in solution, missing the membrane geometry which we found to dramatically alter the reaction kinetics. To this end, I established a cell-free reconstitution system in which purified receptors, enzymes and adaptor proteins are assembled onto a model membrane that mimics the plasma membrane of T cells. First, I reconstituted the enzyme network that governs the T cell receptor (TCR) activation onto liposomes, developed a kinetic, fluorescence readout of TCR phosphorylation, and revealed a switch-like behavior of the TCR proximal signaling network (Hui & Vale, Nature Struct. Mol. Biol., 2014). Next, we dissected the mechanism of inhibitory signaling through PD-1. By titrating PD-1 signal in a biochemical reconstitution system, we discovered that the central costimulatory receptor CD28 rather than the widely postulated TCR, is highly susceptible to PD-1 inhibition (Hui et al., Science, 2017).
Fig. 2. Cell-free reconstitution. Purified receptors, kinases, protein tyrosine phosphatases (PTPase), and adaptor proteins are reconstituted onto an artificial lipid bilayer (liposome or planar supported lipid bilayer). To mimic the cellular geometry, membrane proteins are attached to the lipid bilayer, cytosolic proteins are presented in the solution. Yellow oval: membrane-associated kinase. Grey oval: a cytosolic kinase.
2. Cell-bilayer hybrid.
Coordinated reorganization of signaling components is critical for T cell responses. Previous microscopy studies have shown that several TCR components form microclusters. Yet much less is known about the spatiotemporal dynamics of co-inhibitory receptors. We leveraged a T cell–bilayer hybrid system and total internal reflection fluorescence (TIRF) microscopy to visualize PD-1 signaling in cytotoxic T cells, and discovered that PD-1 co-migrates with CD28, but shows much less overlap with TCR. Such spatial organization might function to partition PD-1 signaling away from TCR components.
Fig. 3. Cell-bilayer hybrid for visualization of immune checkpoints in live cytotoxic T cells. (A) Coverslip supported lipid bilayers were reconstituted with ligands, to activate receptors in T cells. (B) Image of PD-1, CD28 and TCR microclusters in a T cell that landed onto the bilayer. Scale bar: 5 μm.
3. Cell–cell stimulation.
Receptor phosphorylation in T cells often occurs rapidly and transiently, posing challenges for quantitative analysis. Non-physiological stimuli (e.g., agonist antibodies, phosphatase inhibitors) are often used to trigger robust signals. Moreover, the expression of immune checkpoints is often transient, making it difficult to study their cell biology. To this end, we developed a clean, robust cell line system that enables precise control of both the strength and timing of PD-1 stimulation. By titrating the PD-1 signal in this system, we verified that the costimulatory receptor CD28 is a primary target of PD-1 mediated inhibition (Hui et al., Science, 2017), indicating CD28 as a much needed predictive biomarker for PD-1 based cancer immunotherapy.
Building on the unique combination of assays and expertise, we are currently dissecting the molecular mechanisms of not just PD-1, but the intriguing and significant other T cell checkpoints that include CTLA-4, BTLA, TIGIT and VISTA. Why do T cells express so many checkpoints? Does each deliver a unique inhibitory signal? Do these immune checkpoints allow some cancers to ‘turn off’ T cells even if the PD-1 pathway is blocked?