Exploiting TCR Repertoire Analysis to Select Therapeutic TCRs for Cancer Immunotherapy

Abstract

Over the past decade, numerous innovative immunotherapy strategies have transformed the treatment of cancer and improved the survival of patients unresponsive to conventional chemotherapy and radiation therapy. Immune checkpoint inhibition approaches aim to block negative regulatory pathways that limit the function of endogenous T cells, while adoptive cell therapy produces therapeutic T cells with high functionality and defined cancer specificity. While CAR engineering successfully targets cancer surface antigens, TCR engineering enables targeting of the entire cancer proteome, including mutated neo-antigens. To date, TCR engineering strategies have focused on the identification of target cancer antigens recognised by well-characterised therapeutic TCRs. In this review, we explore whether antigen-focused approaches could be complemented by TCR-focused approaches, whereby information of the TCR repertoire of individual patients provides the basis for selecting TCRs to engineer autologous T cells for adoptive cell therapy. We discuss how TCR clonality profiles, distribution in T cell subsets, and bioinformatic screening against continuously improving TCR databases can guide the selection of TCRs for therapeutic application. We further outline in vitro approaches to prioritise TCR candidates to confirm cancer reactivity and exclude recognition of healthy autologous cells, which could provide validation for their therapeutic use even when the target antigen remains unknown.

Keywords: T cell receptor; TCR repertoire; TCR sequencing; TCR-T cells; cancer immunotherapy; cell therapy; clonality; meta-clonotype; tumour infiltrating lymphocytes.

Figure 1

T cell engaging modalities in cancer immunotherapy. This graphic represents immune-engaging modalities directly or indirectly converging onto the activation of anti-tumoural T cells. In adoptive T cell therapy, patient-derived T cells are transduced or electroporated to express Chimeric Antigen Receptors (CARs) or T cell receptors (TCRs) specific to tumour proteins, then adoptively transferred to a patient. In tumour-infiltrating lymphocyte (TIL) therapy, TILs collected from surgical resections are expanded ex vivo, then re-infused into the patient. Immune checkpoint inhibition therapies deliver monoclonal antibodies blocking inhibitory receptors including CTLA-4, PD-1/PD-L1, TIM-3, LAG-3, and TIGIT, thereby relieving functional inhibition of endogenous T cells. Similarly, T cell engagers are bispecific molecules, containing a CD3 binding moiety and a cancer target binding moiety, either derived from scFv for traditional bispecifics or from a TCR for ImmTACs, resulting in local recruitment and activation of endogenous T cells. Lastly, therapeutic cancer vaccines in the form of mRNA, peptide, or dendritic cells loaded with cancer antigens all aim to prime or boost endogenous tumour specific T cells.

Figure 2

Schematic of richness and evenness diversity metrics in TCR repertoires. The two illustrated TCR repertoires have the same richness, each containing 5 unique clonotypes, represented by unique colours in sectors of the pie charts. Despite equal richness, the left repertoire (left) has higher evenness, in which all clonotypes make up the same proportion of the repertoire (20%), while the right repertoire (right) has a more uneven repertoire, which is dominated by a single expanded clonotype, and made up of additional clonotypes with uneven frequencies. Measures such as Simpson and Shannon diversity capture the intuitive notion that the more uneven repertoire is less diverse.

Figure 3

Schematic representation of a TCR meta-clonotype. A TCR meta-clonotype is an ensemble of TCRs with high sequence similarity, or sharing sequence and structural motifs, which are inferred to share antigen specificity. A “centroid” TCR is illustrated, and concentric rings of TCR sequences differing from the centroid TCR by various edit distances: (∆ = 1, 2, 3 substitutions) are illustrated here, and colour-coded according to edit distances. The edit distance may be calculated by comparing the TCRα sequence, the TCRβ sequence, or both, and commonly uses the Levenshtein distance, a metric quantifying the difference between string sequences. TCRs with experimentally determined antigen specificity, represented by a grey boundary, may encompass TCRs with varying edit distances, and this does not necessarily correlate with the number of substitutions.

Figure 4

Schematic of unconventional MHC-I and MHC-II-restricted T cell engineering. (a) Schematic representation of providing an MHC-I-restricted TCR, alongside a CD8 co-receptor, in a CD4+ helper T cell endogenously expressing CD4. Provision of the MHC-I-restricted TCRs allows this T cell to recognise peptide–MHC-I complexes, constitutively expressed on nucleated cells, while preserving helper or memory functions. The CD8 co-receptor stabilises the interaction with peptide–MHC-I complexes. (b) Provision of an MHC-II-restricted TCRs, alongside a CD4 co-receptor, in a CD8+ T cell endogenously expressing CD8. Provision of the MHC-II-restricted TCRs allows this T cell to recognise peptide–MHC-II complexes, while simultaneously preserving cytotoxic functions. CD3εδγζ moieties required for TCR signal transduction are illustrated.

Figure 5

Tumour-reactive TCR identification methods. (a) RNAseq-based identification methods. TILs or PBMCs sequenced with dual V(D)J and gene-expression (GEX) profiling can be scored on gene-expression signatures of T cells experimentally confirmed to display tumour reactivity. TCRs belonging to clusters with inferred tumour-reactive phenotypes can then be taken forward for screening. (b) Flow-cytometry identification methods. Primary TIL or PBMC isolated from patients can be stained for surface markers of putative tumour-reactive signatures, such as CD103, CD39, or CD40L for fluorescence-activated cell sorting (FACS) and TCR sequencing, or cultured for further assays. (c) Primary TILs, PBMCs, or Jurkat lines transduced with TCR pools can be initially co-cultured with autologous cancer targets or cell lines loaded with cancer antigens, followed by staining for activation markers including CD69 or 4-1BB, followed by TCR sequencing, or cultured for further assays. In all diagrams, grey TCRs schematically represent bystander TCRs and green TCRs schematically represent tumour-reactive TCRs.

Figure 6

The Goldilocks’ zone of TCR avidity. Schematic illustrating the avidity of TCRs for target pMHC antigen as concentric circles, with high avidity rings closer to the central TCR and lower avidity furthest away. The name Goldilocks’ zone is taken as an analogy to represent the zone of TCR avidity that is neither “too high” nor “ too low”, but well suited to elicit effective T cell activation. Very-high-avidity interactions lead to the gain of undesirable and toxic cross-reactivities and make T cells lose responses to low-antigen density targets. Very-low-avidity interactions are ineffective at signalling altogether, or induce T cells to differentiate into memory rather than effector phenotypes. A narrow avidity window is optimal to induce potent T cell activation and maintain peptide specificity. The functional avidity window is not a TCR-intrinsic feature and depends on additional factors such as T cell activation state and inflammatory cues in the milieu.