Decoding the Immune System

T cells spend their lives talking the language of disease, and we should listen to what they have to say.

Almost every cell in the body chews up a small fraction of its proteins and presents them as antigens on its surface to enable immune system monitoring. T cells surveil these antigens and represent the critical first arm of the adaptive immune system. This component of the overall immune system kills diseased cells, coordinates antibody production by B cells, and provides the essential memory of past disease. For every antigen that a T cell probes with its highly variable T cell receptor (TCR), the cell must integrate the resulting signaling with a variety of its past and current experiences. These include the core functional training the T cell received in the thymus, memories of previous antigen exposure, all previous and concurrent immune interactions, and the microenvironment and functional state of the target cell it is probing. In choosing among many potential paths, each T cell, and the T cell compartment as a whole, must find a balance between killing cancer and aberrantly attacking healthy tissue, thereby causing autoimmune disease. T cells must walk that fine line every day, and in doing so they collectively demonstrate a fundamental and severely underappreciated ability to continuously monitor every cell in our body.

The enormous potential and limitations of current T cell therapeutics are best exemplified by two of today’s most vaunted avenues of therapeutic research: checkpoint inhibitors that disinhibit T cells to enable antigen-specific tumor killing, and CAR T cell therapies that redirect a patient’s own T cells to known antigen targets on cancer cells. Both therapeutics leverage T cell potency to kill cancer cells, but they do it with almost no knowledge of the native antigen specificity of the cells, often resulting in toxicity—particularly with checkpoint inhibitors, when the activated T cells turn against healthy tissue to cause autoimmunity.

What we can learn from B cell development

The current state of T cell antigen discovery can be understood by considering its sibling B cells. With B cells, revolutions in antibody screening and protein engineering have allowed antibodies and their derivatives to transform medicine and dominate the biologics market. As with T cell therapeutics today, the first B cell therapeutics were poorly defined mixtures of B cell–produced antibodies targeted to specific antigens. The result was antiserum therapeutics that were initially used as antibacterials (and predated antibiotics by over 40 years) and are still used to great effect today as antivenoms. The turning point came in the 1970s with technologies that allowed high-throughput antibody interrogation. Hybridoma technology and, later, surface display technologies provided high-throughput screening methods to identify individual B cell antibody clones that bind a specific target antigen. This ability to produce, screen, and optimize antibodies resulted in five of the top six biologics on the market today, whose mechanisms include the core targeting moieties of both the checkpoint inhibitors and CAR T therapeutics.

T cells should be our teachers

T cells should follow a similar path to antigen discovery and therapeutic value, but there are critical differences in T cell biology that will increase both the difficulty of antigen discovery and the potential value of the antigens and therapeutic modalities that result. T cells function through direct cell-cell interactions in which they use their highly diverse T cell receptors (TCRs) to inspect antigens loaded on a target cell’s surface. Upon interaction with their target antigen, activated T cells often stay at the target site, rapidly expand, and kill the target cell. This makes T cells a potent weapon against infections and diseases such as cancer. But it also makes them ideal immune teachers, because their presence can be used not only to isolate the TCRs that bind to diseased cells but also to identify antigens that protect against infectious disease, cancer, and autoimmune disease. Perhaps most important, these antigens are derived from proteins in all cellular compartments, including the cytoplasm and organelles, providing a much larger range of disease-specific targets than are possible for B cell antibodies, which target only surface and secreted molecules.

What we need is a method to decipher the functional state of T cells, the TCRs they carry, and the immunogenic antigens that they see.

T cells see color, not black and white

Many of the antigens that T cells see in these diseases are not the ones that immunologists typically look for. Antigens generated by pathogens and cancer mutations have never been produced by the body or encountered by the body’s T cell repertoire, so they are typically described as foreign, or non-self. Cells that display these non-self antigens are recognized by T cells and killed. Then again, a core tenet of T cell biology is that all T cells that react to self are killed or converted to regulatory T cells; self-reactive cells that escape this clearance and central and peripheral tolerance cause autoimmunity. And yet time and again self-reactive T cells are found in tumors. Early examples were identified from testes and other immune-privileged sites where cells and their presented antigens are shielded from T cell surveillance. Proteins from immune-privileged sites that are aberrantly expressed in cancer cells can be immunogenic, and these cancer testes antigens, such as NY-ESO-1 and MAGE-A1, are the foundation of many of today’s targeted T cell therapy trials.

Yet this vision of immune-privileged proteins as sources of cancer antigens may vastly underestimate the ability of T cells to recognize disease. The entire human genome can theoretically produce immunogenic antigens if the antigens are transcribed, translated, and presented on the surface of the cell for T cell recognition. Around 1.5 percent of the genome codes for conventional proteins that may be subject to T cell deletion and tolerance. But some fraction of those genes, such as many endogenous retrovirus genes, are transcriptionally repressed, and they are never expressed in normal tissue or thymus, leading to immune recognition when they are activated in disease. No doubt a large fraction of the remaining 98.5 percent of the genome is non-coding and does not produce protein. But a significant portion does in fact encode short or frameshift proteins that can be highly immunogenic when aberrantly expressed in cancer.

Recent examples of these germline cancer antigens are almost certainly the first of many potential therapeutic targets that could be far more broadly useful and cost effective than current personalized neoantigen mutations that are rarely shared between patients.

Clearance of cancer and infection may be the least interesting thing T cells do

Even this broadened view of T cell surveillance dramatically underestimates their role in the body. The body selects a significant subset of T cells, known as regulatory T cells (Tregs), to specifically recognize self and actively protect cells and tissues expressing these antigens from immune attack. Thus, Tregs sit at the interface of self and non-self to convert the conventionally stark black-and-white image of self and non-self into a spectrum of grays. They are most often studied in the context of barrier homeostasis in the intestine and lung, but their emerging role in adipose and cardiovascular tissue strongly suggests that they play essential homeostatic roles in all tissue. Importantly, the antigens that these cells see in tissue are almost entirely unknown.

The size of the puzzle

Why don’t we already know what T cells see? T cells can produce at least 10^13 different TCRs. They can recognize a theoretical diversity of over 10^11 antigens loaded on surface receptors known as major histocompatibility complex (MHC) proteins. These proteins are the most polymorphic in the human population, with thousands of variants. The problem becomes one of massive potential diversity on both sides. When targeted to any one person, the practical diversities of the TCRs and antigens are restricted to the person’s defined set of MHCs and the roughly 10-100 million T cell clones that exist in any person at any one time—but the numbers remain nonetheless daunting. Tetramer technology, developed in seminal work over 20 year ago, enabled visualization and isolation of T cells that recognize a single antigen. But inherent limitations in fluorescence and isotope-based isolation have severely restricted the number of antigens that can be effectively screened. Cell-based screening methods enable high-throughput antigen discovery, but they are limited to the interrogation of relatively small numbers of T cell clones. On the other hand, recent advances in single-cell-sequencing technology now allow high-throughput characterization of T cells and the TCRs they encode. Single-cell sequencing has already begun to revolutionize all areas of immunology, but it will be essential to couple high-throughput antigen detection with single-cell sequencing to break the T cell field wide open.

T cell therapeutics

A remarkable range of insights and drugs might result from the new T cell technologies. Every nucleated cell in our body presents MHC-bound antigens on its surface, and those antigens provide a distinct signature of the cell’s identity and functional state. Novel cancer-specific antigens will dramatically improve autologous T cell therapy, in which a patient’s own T cells are specifically activated and expanded to increase anti-cancer activity. Furthermore, the TCRs that bind these antigens will provide an entirely new reservoir for recombinant TCRs that can be used to redirect patient cells to cancer, much like CAR T technology today.

Beyond their immediate utility in cancer treatment, TCRs and their antigen targets will almost certainly be useful in fighting all immune diseases, particularly autoimmunity and transplantation, in which T cell activity must also be redirected to ameliorate disease. Therapeutic modalities that provide antigen-specific immune tolerance are in early development, and they are in desperate need of better antigen targets. But most broadly and perhaps most impactfully, these antigens and TCRs can target any cell in the body, either healthy or diseased, delivering therapeutic cargo wherever it is needed.

Finally, high-throughput interrogation of antigen–T cell interactions would provide an enormous data set that could be used to predict the antigen specificity of potentially every TCR. This would dramatically improve the accuracy and safety of TCR therapeutics. It would also provide an easy and ubiquitous method to diagnose disease at potentially a very early stage. Periodic blood draws and T cell sequencing would provide the functional state and TCR sequence of circulating T cells. For example, activated T cells with specificity to pancreatic cancer antigens could catch early-stage pancreatic cancer that is treatable but largely asymptomatic. A drop in vaccine-induced memory T cells to measles antigens could be followed by a simple booster vaccine. T cells with reactivity to specific cardiomyocyte antigens could predict underlying heart disease or subclinical atherosclerosis. All of these benefits may be achieved with a blood draw, standard sequencing, and an AI platform trained on a massive antigen-TCR interaction map.

If we ask the right questions with the right technology, the answers will come quickly. And if we listen to T cells carefully, they will show us where to look and what we will find. At Flagship Pioneering, we have built a foundational platform within Repertoire Immune Medicines to decode T cell–antigen interactions at unprecedented scale, and we have begun to apply it across cancer, infection, and autoimmune disease. Our ears are attuned.