The anti-cancer potential of T cell receptor-engineered T cells

Abstract

Adoptively transferred T cell receptor (TCR)-transgenic T cells (TCR-T cells) are not restricted by cell surface expression of their targets and are therefore poised to become a main pillar of cellular cancer immunotherapies. Addressing clinical and laboratory data, we discuss emerging features for the efficient deployment of novel TCR-T therapies such as selection of ideal TCRs targeting validated epitopes with well-characterized cancer cell expression and processing, enhancing TCR-T effector function, trafficking, expansion, persistence and memory formation by strategic selection of substrate cells, and gene-engineering with synthetic co-stimulatory circuits. Overall, a better understanding of the relevant mechanisms of action and resistance will help prioritize the vast array of potential TCR-T optimizations for future clinical products.

TCR-T cells represent a versatile platform to target cancer- associated antigens

Immunotherapy is currently revolutionizing the treatment of cancer, demonstrating the power of immune cells in recognizing and killing cancer cells. Direct evidence on the curative potential of T cell immunotherapy comes from experience with ex vivo expanded tumor-infiltrating T cells/lymphocytes (TILs) from cancer patients that were reinfused to the patient and produced reproducible and sometimes durable tumor responses.[1] Building upon advances in T cell receptor (TCR) isolation and gene-engineering technologies, TCRs recognizing a wide range of specific peptide/HLA combinations can be now expressed in a patient’s T cells to generate T cell receptor -transgenic T (TCR-T) cells, redirecting those cells to recognize tumor-associated antigens and kill tumor cells.

In parallel, immune checkpoint blockade (ICB) therapy has been effective in some tumor settings, reinvigorating exhausted tumor-infiltrating T cells/lymphocytes (TILs) and/or recruiting tumor-reactive T cells from the naive T cell compartment, but insufficient T cell priming or immune escape occurs in a substantial proportion of patients.[2] Because TCR gene transfer confers a novel specificity to the treated patient’s T cells, TCR-T cells do not rely on a patient’s preexisting endogenous T cell repertoire and can help overcome resistance to ICB. TCR-T cells are also not restrained by the limited availability of cancer-specific cell surface proteins that are required for successful targeting by chimeric antigen receptor (CAR)-engineered T cells. However, challenges remain in establishing a robust population of TCR-T cells that recognize reliably presented targets, have sufficient avidity (see Glossary) and functional capacity to eliminate existing tumor, and persist long term to prevent recurrences.[3] Here, we review novel developments in optimizing TCR-T therapies, focusing on rational target selection and enhancing engineered T cell function. The rules of optimal target selection are currently being refined based on a better understanding of processing, presentation, dynamic changes on target expression during immune-editing as well as an increasing number of validated TCR-T targets beyond the initial model antigens. The prospect of enhancing T cell function recently entered the TCR-T field because of technological advances in gene engineering as well as by progress made in the CAR-T field. We believe that integrating emerging principles discussed herein will make TCR-T cells more effective and will help establish the importance of this modality in the immunotherapy armamentarium (Figure 1, Key Figure).

Glossary box.

Adoptive T cell therapy

A type of immunotherapy in which living T cells are administered to treat diseases such as infections or cancer. This approach typically involves in vitro selection and expansion, and might include genetic manipulation, of T cells.

Affinity

The binding strength of a single TCR molecule to its peptide-MHC target -a physical property inherent to a given TCR.

Antigen

We refer here to cellular proteins that can potentially yield targets for TCR-T as antigens. If processed and presented, an antigen might yield different epitopes of various HLA-restrictions.

Avidity

Binding avidity describes the binding of a cell with particular TCRs to a target peptide-MHC reagent, e.g. an MHC tetramer. Functional avidity, an important measure of a T cell’s utility, refers to the sensitivity of a T cell with a given TCR to peptide -MHC on a stimulator cell, expressed as amount/concentration of target required to elicit a given response. Avidity is determined by several factors including expression level of the TCR, co-receptor binding and the functional readout used.

Epitope

The precise stretch of the antigenic peptide loaded onto MHC molecules and recognized by a TCR.

Neoantigen

An antigen with altered amino acid sequence compared to its normal counterpart, typically as a consequence of a genetic mutation. When presented as neoepitopes, TCRs can recognize such peptides as foreign.

Epitope

The precise stretch of the antigenic peptide loaded onto MHC molecules and recognized by a TCR.

Epitope spreading

Triggering of an endogenous immune response by immunotherapy against epitope targets that are not directly targeted by the administered immunotherapy. In case of TCR-T, epitope spreading refers to T cell recognition of targets different from targets of the TCR-T cells.

HLA ligandome/immunopeptidome

The totality of all peptides presented on the cell surface by HLA molecules and therefore potentially able to elicit T cell responses.

Immune synapse

The interface between a T cell and its target cell, comprising a region of the cell membrane with distinct spatial architecture containing microdomains with distinct clusters of receptor-ligand composition. Formation of the immune synapse is thought to occur on the T cell side around TCRs engaged by peptide-MHC, involves co-receptors, costimulatory-receptors and other transmembrane proteins, resulting in T cell activation, polarization and ultimately effector functions.

Switch receptors

A synthetic transmembrane receptor which binds inhibitory signals. In addition to acting as a decoy, switch receptors convert the inhibitory signal into a stimulatory signal by virtue of their intracellular signaling domains, typically derived from co-stimulatory receptors.

Figure 1. (Key figure).

Optimizing TCR-T therapies requires integration of emerging knowledge on tumor antigen expression, antigen processing, and T cell biology.

One size does not fit all: optimizing TCR-T to target various cancer-associated antigens

T cells recognize processed peptides bound to/presented by human leucocyte antigen (HLA) by way of their T cell receptors (TCRs), which have evolved to bind to a broad array of intracellular and membrane antigens with exceptional sensitivity, detecting even very low levels of peptide-bound HLA and differentiating between normal and over-expressed levels of self-antigens as well as abnormal, non-self-antigens.[4] Peptides displayed on HLA Class-I molecules generally represent sampling of the cell’s protein output, rendering the cell’s whole peptidome a target of circulating CD8+T cell scrutiny. Peptide presentation by HLA Class-II molecules has more specialized functions in the context of inducing CD4+ helper cell responses by antigen presenting cells and eliciting a coordinated, polyclonal adaptive immune response. Since most cancer cells exclusively express HLA Class-I molecules and CD8+ cytotoxic T cells are the major effectors of anti-tumor immune responses, TCR-T therapies are primarily intended to redirect CD8+ T cells to HLA Class-I restricted targets, although both engaging CD4+/CD8+ T cells towards HLA Class-II targets[5] and redirecting CD4+ T Cells to HLA-Class-I epitopes[6] is feasible and might improve the efficacy of TCR-T therapies.

Conceptually, TCRs can target any intracellular protein if peptide epitopes are presented on the HLA. Selecting the specific protein and epitope to target using TCRs has critical downstream consequences for TCR-T immunotherapy (Figure 2a) as it directly affects the expected frequency of TCRs in the repertoire, the affinity of TCRs, diseases to be treated and potential escape mechanisms. We will highlight below these aspects for the three major classes of TCR-T targets: viral antigens, over-expressed self -antigens and neoantigens.

Figure 2. Main avenues for optimizing TCR-T therapies.

a) Choice of a homogenously expressed antigen and robustly processed epitope will maximize TCR-T therapeutic effect and limit immune-escape.

b) Interrogation of a large number of target-specific TCRs in high-throughput assays will help identify the ideal TCR for clinical translation.

c) Synthetic biology and gene editing will further enhance the function of well-defined substrate cells.

Targeting virus-driven cancers

Ideally, the TCR-T target epitope is derived from a protein that is expressed at high and homogenous levels exclusively in cancer cells, where it is essential for survival and proliferation, such that elimination of cells expressing the target are unfavorable for cell survival. Although relatively rare, oncogenic viruses directly contribute to the development of several cancers, including cervical cancer and head-and-neck cancer (human papilloma virus, HPV), Hodgkin lymphoma and nasopharyngeal lymphoma (Epstein-Barr virus, EBV), Merkel cell carcinoma (Merkel cell polyoma virus, MCPyV) and can produce peptide epitopes with most of these characteristics.[7] A fraction of virus-related cancers can be successfully treated with ICB[8] and adoptive transfer of autologous polyclonal virus-specific T cells[9], indicating that patients harbor T cells against these viral epitopes and their targeting is indeed beneficial. Reinvigorating or expanding autologous virus-specific T cells is however rarely curative,[10] probably as a consequence of immune-evasion mechanisms including clonal deletion of the T cells with the highest affinity.[11,12] Conversely, isolation of such anti-viral TCRs with high-affinity is possible from healthy donors as neither negative thymic selection nor clonal deletion by the tumor has occurred.[13] Successful transfer of viral-specific T cells from the stem -cell donor or from third-party cell banks in the context of allogeneic hematopoietic stem cell transplantation in fact demonstrated the feasibility of adoptive immunotherapy for viral infection.[14] TCR transfer to autologous T cells overcomes both cumbersome expansion of polyclonal autologous T cells as well as HLA-incompatibility of third party viral specific cells. It remains to be seen to what extent viral immune-escape mechanisms such as impaired HLA expression and antigen presentation, [15] that commonly enable virus-driven cancers in the first place and limit the efficacy of ICB, will also affect TCR-T therapies. Ongoing TCR-T clinical trials in HPV-(e.g. NCT03912831) , EBV- (NCT03925896) or MCPyV- driven (NCT03747484) cancers will establish the role of TCR-T cell therapies as a rational strategy to reduce the burden of these diseases.

Targeting mutated proteins

Point mutations, chromosomal translocations and other cancer-causing genetic alterations can create abnormal neoantigens that are not represented in the thymus or any other healthy tissue and can therefore be targeted with high-affinity TCR-T cells without the risk of on-target/off-tumor toxicity.[16] Frequent genomic instability and genotoxic exposure produce a vast array of neoantigens, including antigens derived from cancer driver mutations that are unlikely to be downregulated, but also antigens comprising passenger mutations that can be successfully T cell targeted, as with ICB and/or TIL therapy.[17,18] Especially since TCRs can discriminate between polymorphic peptides differing only in a single amino acid,[19] mutated targets are appealing.

Isolation of clinically relevant TCRs remains challenging as identifying and characterizing which mutations generate recognizable, HLA-presented peptides largely remains unclear.[16] Growing clinical datasets on cancer mutations, HLA -genotype and response to immunotherapy, as well as refined bioinformatic algorithms and large cooperative efforts (e.g. the Tumor Neoantigen Selection Alliance -TESLA[20]) provide increasingly accurate predictions about cancer-specific neoantigens. Complementary efforts are underway to directly characterize cancer-specific HLA-bound peptides (the HLA ligandome/immunopeptidome)[16]. Although TCRs have been identified that recognize peptides derived from common oncogenic hotspot mutations, such as KRAS[19] or TP53[21] (public neoepitopes), the majority of cancer mutations are relatively patient-specific, rendering neoantigen-specific TCR-Ts highly individualized products. Developing TCR-T cell therapies against private neoepitopes poses unique technical, regulatory and commercialization challenges: The pipeline encompassing neoantigen detection typically by next-generation sequencing, epitope prediction, TCR identification and clinical product generation must be completed within weeks to be useful for patients with often rapidly deteriorating, therapy-resistant disease. At the same time, preclinical TCR-T product characterization including potency, dose finding, safety and identity, that are traditionally required for regulatory approval of a first-in-man clinical trial, are not feasible due to time limitations and the lack of appropriate disease models. The individualized nature of the TCR-T product makes interpretation of clinical data and correlative, translational analyses also challenging as both disease characteristics and the drug product differ between patients. Finally, the complexity of the individualized approach might render it less attractive for commercial development.

A strategy to overcome these obstacles is a hybrid personalized/TCR library approach. Whole exome sequencing of cancer samples and simplified gene engineering and TCR-T manufacturing increasingly enable the generation of personalized TCR-T therapies for the first stage of this strategy. Using a patient’s own TCRs requires minimal safety testing, and thus can bypass some regulatory hurdles. Feasibility and efficacy of this approach are currently being tested in the clinic (NCT03970382, NCT03412877). At a later stage, TCRs targeting public or private/personalized neoepitopes could be used in a library strategy with which patients are screened for the presence of mutations and the targeted HLA allele, and treated with the most appropriate TCR-T product selected from a set of fully characterized TCRs (NCT04102436).

Targeting over-expressed self-antigens

Over-expressed self-antigens, such as cancer-testis antigens (CTAs; NY-ESO-1, MAGE family, PRAME) expressed in embryonic tissues or in HLA-negative germ cells[22], or lineage-specific antigens (MELAN-A/MART-1, WT1, Mesothelin, Alpha Fetoprotein -AFP)[23], are abundant across human cancers. Most TCR-T self-antigen targets were identified through screening for self-proteins recognized by TILs,[24] providing a strong biological rationale for clinical translation.

Because self-antigens are expressed in the thymus, potent, high-affinity self-reactive T cells are naturally deleted during thymic development,[25] consequently limiting the circulating pool of high-affinity TCRs from which to identify clinical reagents. To bypass this limitation, high-affinity TCRs were obtained by affinity enhancement through random mutagenesis of a naturally obtained TCR[26] or immunization of HLA-transgenic mice.[27] Unfortunately, two trials led to fatal toxicity by on-target off-tumor reactivity or off-target recognition of unrelated proteins.[28,29] Since then, only TCRs that have undergone stringent pre-clinical testing are being utilized.[30]

Another methodology takes advantage of human diversity and the T cells’ ability to discriminate between antigen levels[23] to identify TCRs that are of high-enough affinity to recognize and kill cells overexpressing self-antigens, but not to react with cells expressing normal levels of antigen. Because such high-affinity TCRs are necessarily infrequent, this strategy requires massive high-throughput screening of the naïve T cell repertoires from matched healthy donors (Figure 2b).[31] Ultimately, the safety of these approaches will have to be evaluated in human trials, due to paucity of fully predictive mouse models.

Protein-degradation pathways that influence epitope production must be considered to avoid immune evasion. These pathways include the ubiquitin-proteasome system that yields short peptides, in some cases further trimmed by ERAP1/2 and other aminopeptidases, and finally loaded onto HLA for T cell recognition.[32] Components of the proteasome complex and other proteins of the antigen processing pathway that can modulate epitope presentation[33–35] are mostly regulated epigenetically and cell lineage-restricted, and are influenced by extrinsic factors, such as interferon-gamma (IFN-γ) signaling.[36] For example, cells of mesodermal origin tend to express the standard proteasome while hematopoietic cells and other cells present in inflamed environments tend to express the immunoproteasome, and these two primary proteasome types produce largely non-overlapping sets of peptides.[34,37] Thus, to prevent TCR-T therapy resistance by a modulation of degradation pathways, target epitopes must be rationally selected as those that are preserved across tumor types and throughout disease progression, based on comprehensive characterizations of their processing. Improved HLA-peptide elution methods,[38] analysis of endogenous, productive T cell responses of TILs,[18] as well as TCR generation using endogenously-processed protein fragments instead of exogenously loaded peptides,[39] will all contribute to resolving this challenge.

Enhancing engineered T cell function

Although individual patients have benefitted from TCR-T cell therapy,[40] efforts to improve in vivo T cell expansion/persistence and avoid rapid loss of effector function are needed to increase success rates in the face of large tumor burdens and obviate the need for producing up to 10¹⁰ TCR-T cells administered in current clinical trials (Figure 2c). The transgenic TCR is the critical protein for tumor target recognition[41] and beyond target selection, TCRs with critical physical characteristics including affinity, or more generally with desired functionality, can be identified by high-throughput TCR isolation, cloning and screening methods.[42] Accumulating evidence suggests however that insertion of a TCR alone may not be sufficient to mediate a robust anti-tumor response. Therefore, co-stimulatory cellular receptors and other features of the TCR “substrate” cell present opportunities to surmount these hurdles and improve clinical efficacy.

Selecting substrate cells

Utilizing less differentiated naïve, central memory and stem memory CD8+ T cell subsets has mediated increased persistence and associated anti-tumor effectiveness of adoptive T cell therapy in preclinical models.[43] This has been further supported in a pilot clinical trial using WT-1-specific TCR-transduced donor Epstein Barr virus (EBV)-specific CD8+ T cells to treat patients with acute myeloid leukemia. Transferred TCR-T cells maintained co-stimulatory receptor expression while inhibitory receptor expression remained low, possibly mediating the robust in vivo TCR-T frequency and persistence observed.[44] However, not all patients who received EBV-specific substrate cells experienced clinical benefit, especially in the face of large tumor burdens and presumably activation-induced T cell death, suggesting alternative approaches are necessary.

Engaging CD4 T cells

To date, TCR-based immunotherapies have primarily exploited CD8+ T cells, which recognize tumor antigens presented by Class I HLA .[3] However, co-transferred CD4+ T cells can enhance anti-tumor effects, including by promoting expansion and survival of tumoricidal CD8+ T cells, as initially shown in a murine leukemia model[45] and extensively shown in CD19-directed CAR therapy.[46] Tumor-specific Class II-restricted CD4+ T cells promote Class I-restricted CD8+ T cell proliferation, survival and effector functions, in part by producing interleukin-2 and facilitating dendritic cell-mediated activation to broaden the range of immune responses (epitope spreading).[47] CD4+ T cells expressing Class II HLA-restricted TCRs exhibited direct cytolytic activity against metastatic melanoma and had anti-tumor activity against human cholangiocarcinoma.[48,49] However, Class II expression is rare in solid tumors[50] and the identification of both Class I and II-restricted TCRs that recognize the same tumor antigen is logistically and financially challenging. Virus or neoantigen-specific Class I TCRs that are of sufficiently high affinity to engage CD4+ as well as CD8+ T cells can be identified in human donors,[6] but thymic selection makes such TCRs that recognize overexpressed self-antigens exceedingly rare. Identification of such CD8-independent HLA Class-I restricted TCRs require large screening campaigns and is not always successful for every target. Co-expression of CD8 alpha and beta chains with a TCR has now been utilized as an alternative strategy to engage both CD4+ and CD8+ T cells, potentially applicable to any Class I-restricted TCR.[51] In some pre-clinical models, tumor-specific CD4+ T cells support CD8+ T cell proliferation and function when large quantities of antigen are present.[52] However, CD4+ T cell engagement might still be insufficient, especially in solid tumors where maximum T cell fitness is necessary to overcome inhibitory interactions in complex tumor microenvironments.

Adding Signal 2 and beyond

The incorporation of a co-stimulatory domain (most notably CD28 and 41BB but also ICOS and OX40) in the intracellular CAR-T signaling domain has been shown to be necessary for CAR-T cell function and led to the spectacular anti-tumor efficacy of CD19 CAR-T cells observed against hematological malignancies.[53] Expressing a TCR with or without the CD8αβ co-receptor, upon peptide-HLA (pHLA) binding, results in phosphorylation of the immunoreceptor tyrosine-based activation motifs of CD3 and ZAP70 by LCK, which leads to activation of essential T cell signaling pathways (signal 1), but not necessarily a co-stimulatory, enhancing signal 2. Indeed, full T cell activation requires independent triggering of co-stimulatory receptors by the matching tumoral ligands,[41] while simultaneously overcoming inhibitory receptor signaling via ligands that are abundantly expressed in the tumor microenvironment (TME).[54]

Preclinical data suggests that enhanced co-stimulatory signaling promotes TCR-T cell proliferation, cytokine production and cytotoxicity.[55] Constructs tailored for optimal TCR-T cell activation have yet to be deployed in the clinic as many factors need to be considered. First, the selection of the co-stimulatory domain can greatly influence mitochondrial biogenesis and overall cellular metabolism, leading to differences in effector vs central memory T cell differentiation and contributing to variations in T cell kinetics.[55,56] Second, the localization of the co-stimulatory domain in the immune synapse has consequences on design. So-called “switch receptors,” such as CD200R/CD28 or PD1/CD28, have demonstrated increased efficacy in vitro and in murine models, by converting an inhibitory signal into a stimulatory signal.[57,58] However, this approach is dependent on additional ligand interactions in the TME other than the TCR:pHLA interaction, and inserting these constructs alongside a TCR requires two separate vectors or a very large vector, which leads to limited transduction efficacy. Third, the abundance of target and TCR affinity are critical, as increased TCR:pHLA interactions amplify TCR signaling and can either lead to deleterious cytokine storms or T cell tolerance, with obvious implications for safety and anti-tumor efficacy.[6,59]

Furthermore, CAR-T cells (with intrinsic co-stimulation) have shown somewhat limited efficacy in the solid tumor setting,[60] indicating that TCR-T therapies may require engineering beyond increased co-stimulatory signaling. Promising possibilities include methods to remove signals that dampen TCR triggering,[55] increase localization and penetrance into tumors,[61] and/or alter T cell metabolism.[62] The number of pre-clinical constructs that have been designed to enhance T cell function downstream of TCR triggering highlights the need for unbiased systematic library screening to find potential synergies between antigen target, transgenic TCR and increased T cell fitness via protein engineering.

Overcoming tumor heterogeneity

The extent of antigen heterogeneity and its impact on the efficacy of TCR-T cell therapies remains unclear. Many cancer-associated antigens are variably expressed throughout the tumor, such that specific antigen-targeting can allow some tumor cells to escape, which can lead to clonal outgrowth and therapy resistance in mouse models,[63] in patients treated with ICB,[64] and with certain adoptive T cell therapies.[15,65] Although limited to T cell products created by lentiviral transduction, due to cost and feasibility, one concept that is gaining steam is the design of multi-TCR products specific to several antigens expressed by a patient’s tumor. Indeed, gene transfer using CRISPR-Cas9 can be accomplished with DNA plasmid- or polymerase chain reaction-derived templates, eliminating the need for viral vector manufacturing.[66] With optimization, this general approach has the potential to target specific gene knock-outs/knock-ins , co-expressing an array of TCRs and co-stimulatory proteins, while simultaneously eliminating inhibitory signals. The characteristics of super-enhanced TCR-T products thus designed to simultaneously overcome limited T cell expansion, prevent T cell dysfunction, overcome tumor escape and control toxicity have yet to be elucidated with solid preclinical and clinical data.

Concluding Remarks

TCR-T cells are emerging as a widely applicable, powerful cancer immunotherapy modality. The complexity of the TCR-T approach makes its preclinical optimization and clinical translation challenging, but, as highlighted in this review, also offers several key themes that can be improved to exploit the full potential of TCR-T therapies. Building upon experience with the first individual TCR-T programs, the field is currently systematically establishing the rules of TCR-T target antigen selection, optimal TCR-T identification and product manufacturing. At the same time, technological and conceptual advances open up completely novel avenues to enhance the function of TCR-T cells by parallel gene-engineering eliciting an optimal anti-tumor immune response. Given this complexity, optimizing TCR-T cell therapies is a truly interdisciplinary endeavor integrating cutting-edge knowledge and technologies not only from immune-oncology, but also from cancer biology and gene-engineering fields.

Translating the next generation of TCR-T cell therapies into the clinic has its own challenges. Complex gene-engineering including multiple gene knock-outs as well as individualized treatment paradigms stretch the traditional regulatory framework for safety and efficacy testing. We hope that the positive safety experience in the first trials with TCR-T, CAR-T and other advanced gene-engineered cell products will pave the way for a more straightforward regulatory process.

We believe that addressing these individual challenges and integrating them into a next-generation clinical TCR-T product will advance TCR-T cells as an essential component of anti-cancer therapy (see also Outstanding Questions).

Outstanding questions box.

• -Which preclinical in vitro assays and animal models will be predictive of clinical success with TCR-T therapies?
• -How can we prioritize and test the almost infinite number of T cell enhancing strategies?
• -To what extent can complex gene-engineering approaches be translated into the clinic?
• -How can we integrate TCRs targeting different epitopes into a single TCR-T product?

Highlights/trends box.

• -Early trials with TCR-T cells have shown direct tumor killing in select patients but limited broad efficacy highlighting room for improvement in first generation products.
• -TCR-T cells offer the flexibility to target a wide variety of cancer-associated proteins, not limited to cell surface expression.
• -Advanced synthetic T cell engineering approaches enable modulation of activation, inhibition and TCR downstream signaling.
• -Future TCR-T therapies are poised to integrate our enhanced understanding of T cell fitness as well as target antigen presentation.