Antigen discovery tools for adaptive immune receptor repertoire research

Abstract

The adaptive immune system has evolved to recognize with incredible precision a large diversity of molecules. Innovations in high-throughput sequencing and bioinformatics have accelerated large-scale immune repertoire analyses and given us important insights into the behavior of the adaptive immune system. However, establishing a connection between receptor sequence and its antigen-specificity remains a challenge despite its central role in determining T and B cell fate. We discuss recent large-scale antigen discovery technologies which can be combined with adaptive immune receptor repertoire (AIRR) studies. We highlight important discoveries made using repertoire analyses in the field of host-microbe interactions.

Introduction

The adaptive immune system has evolved to recognize a large array of molecules via the antigen receptors present on T and B cells: the T-cell receptor (TCR) and the B-cell receptor (BCR), respectively. Both BCRs and TCRs possess variable complementarity determining regions, important for interaction with antigen, and a constant region. The diversity of the variable regions is determined in the course of their assembly by somatic gene rearrangement of diverse genetic elements, a process known as V(D)J recombination. Allelic exclusion ensures that each T and B cell usually expresses a single antigen receptor on its surface. The ensemble of antigen receptors present on lymphocytes is termed the adaptive immune receptor repertoire (AIRR). By providing a global picture of the adaptive immune system, repertoire analyses can be used for diagnostic purposes or as a source of therapeutic agents, for example to build new chimeric antigen receptor T cells (CAR-T) or identify broadly neutralizing antibodies. However, the massive diversity of antigen receptors poses an analytical challenge. In recent years, great strides have been made in the development of sophisticated methodologies that can analyze the complex data sets generated by AIRR analyses (reviewed elsewhere^1,2). While single-cell sequence analyses can link receptor sequence to phenotype, they do not provide information on antigen specificity³. Here we discuss novel methodologies that can connect repertoire analyses with antigen specificity in high-throughput fashion. Finally, we highlight important discoveries made using repertoire analyses in the field of host-microbe interactions.

T cell antigen discovery

While T cell receptor specificity can dictate T cell fate, large scale approaches for antigen discovery remain limited. Repertoire analyses provide a global picture of the set of TCRs present in a given sample, but these alone cannot establish antigen specificity. Historically, T cell specificity was assessed by in vitro experiments in which T cells are co-cultured with antigen-pulsed antigen presenting cells (APCs). Upon recognition of a cognate peptide loaded onto major histocompatibility complex (pMHC) molecules, a signaling cascade is triggered that leads to T cell activation, as assessed by upregulation of surface markers (e.g. CD69), T cell proliferation, production of cytokine or cytolytic activity^4–8. However, these techniques are constrained by the required high number of T cells. Furthermore, such assays are low throughput and expensive, making them difficult to extend to the vast numbers of cells that must be interrogated for adequate sampling.

One common strategy to identify antigen-specific T cells relies on soluble multimers of pMHCs, typically referred as MHC tetramers. Because pMHC multimers can bind TCRs with sufficient stability, they can be used to identify antigen-specific T cells among a heterogenous population of T cells, typically using flow cytometry⁹. The identification and characterization of, for example, microbe-specific T cells in the intestine often relies on such methods^10–14. Sensitivity of detection can be improved by increasing the avidity of the pMHC reagents, for example using nanoparticles harboring >10⁴ tetramers per particles¹⁵. However, this technique is difficult to scale up because (i) antigen-specific T cells are found at very low frequency, (ii) a panel of antibodies is required to characterize phenotypic properties of T cells based on surface markers, and (iii) multi-color tetramers are typically needed to confidently measure specificity. The use of mass cytometry with heavy metal ions as tags (CyTOF) on pMHC multimers is a possible alternative to address this limitation. Unfortunately, the destructive nature of CyTOF does not allow it to be combined with TCR sequencing^16–19. The advances described above are limited by the number of fluorescent and metal ion tags available and typically can only screen 10–100 different epitopes simultaneously^20–22.

The use of pMHC multimers can be combined with DNA barcodes to screen >1000 T cell specificities²³. When combined with deep sequencing, this technology permits the simultaneous sequence analysis of the TCRs and identification of antigen specificity at the single-cell level²⁴ (Fig. 1a). Assuming that adequate numbers of cells are available for analysis, the theoretical diversity of the DNA barcodes used in these studies is up to 10¹⁰, making single-cell sorting a critically limiting step of this strategy²⁵. While various pMHC technologies have been essential in the exploration of T cell specificities, it remains true that working with pMHC multimers can be technically challenging, especially for Class II MHC products. pMHCs can be difficult to produce, may suffer from a lack of stability and require knowledge of the antigen(s) to be tested. Not only is cross-reactivity a common feature of TCRs, the process of selection implies some measure of reactivity with ‘self’ pMHC in the course of T cell development²⁶.

Figure 1: New technologies for T cell antigen discovery.

(a) Schematic of TetTCR-seq, a high-throughput method to pair TCR sequences with antigen-specificity. DNA-barcoded peptide MHC (pMHC) tetramers are generated by in vitro transcription and translation. The library of tetramers is then incubated with a population of polyclonal T cells. T cells specific for a given tetramer are sorted out and single-cell sequenced, allowing to link a TCR sequence to antigen-specificity. (b) Cell-based approaches to identify antigen-specificity of a T cell clone. SABRs (signaling and antigen-presenting bifunctional receptors) and novel pMHC–TCR (MCR-TCRs) are chimeric receptors expressed on reporter cell line capable to present peptides to T cells (or carrier cells expressing TCRs). Upon recognition of a cognate TCR by pMHC carrying cells, a reporter gene is expressed. Activated cells are then sorted by flow cytometry and the expressed antigen sequenced, thereby allowing to connect TCR sequences to antigen-specificity.

More recently, the development of microfluidic devices has opened the door to new strategies for the study of TCR-pMHC interactions at higher throughputs^15,27–29. Such strategies typically require smaller number of cells, an advantage when dealing with clinical samples and limited numbers of cells²⁸. These innovative technologies are still in their infancy but hold promise in overcoming the technical limitations of linking antigen specificity to TCR sequences in high-throughput at the single-cell level.

The techniques described above rely on knowing the identity of the antigens to be tested. Alternatively, TCRs of known sequence can be tested against a library of different antigens. In the most common strategy, a library of peptides or whole genome libraries of a known pathogen or commensal can be tested in vitro against a T cell clone^10,30. Antigen libraries can also be displayed on the surface of baculoviruses, yeast or mammalian cells^31–40. To identify cognate antigens, soluble fluorescently-tagged TCRs are incubated with a library of baculovirus-infected cells or yeast cells that display suitably engineered covalent pMHC complexes. Carrier cells that display the cognate antigen and are recognized by a given TCR can be sorted. The antigen is then identified by sequencing. This strategy allowed Birnbaum and colleagues to study the extent of TCR cross-reactivity³⁴. This approach permits an examination of up to 10⁶–10⁸ different peptides but requires soluble TCR tetramers, which are non-trivial to produce. These also might not represent physiological TCR-pMHC interactions⁴¹. More recently, a number of cell-based strategies have been developed that circumvent some of the mentioned limitations (Fig. 1b). One key feature of these newer strategies is that the readout exploits signaling induced by TCR-pMHC interactions, rather than relying on mere physical interaction. In one approach, chimeric receptors named “signaling and antigen-presenting bifunctional receptors” (SABRs) contain an extracellular pMHC moiety and an intracellular TCR signaling co-receptor domain. These SABR-expressing cells are then used as a platform for T cell antigen discovery. Upon recognition of a cognate pMHC by TCRs expressed on accessory cells, signaling downstream of the chimeric receptor is triggered. Activated SABR-expressing cells can then be sorted out using a fluorescent reporter gene³⁶. In a second strategy, chimeric receptors named MCR-TCR made of a class II MHC molecule fused to intracellular domains of the TCR are used in similar fashion³⁷. In lieu of TCR signaling, other studies have relied on trogocytosis, a phenomenon in which transfer of membranous materials between T cells and antigen-presenting cell occurs only when the cognate antigen is presented to T cells^39,42. In yet another strategy, Kula et al. and Sharma et al., exploited the release of granzyme B by T cells upon antigen recognition^38,40.These promising new technologies expand the toolkit available to identify antigen specificity for a given TCR.

B cell antigen discovery

Antibodies and their derivatives are the largest class of therapeutic biologicals and are indispensable tools for basic science⁴³. As a result, BCR repertoire analyses could serve as a rich source of potential therapeutic antibodies, provided the difficulties of linking antigen specificity to BCR sequences can be overcome. Until recently, most methods to identify B cell specificities have been fairly low throughput, relying on single cell sorting with labeled versions of the antigen as baits^44,45 or in vitro culture with primary or immortalized B cells^46–49. Alternatively, antibody sequences of interest can be recombinantly produced and tested individually⁵⁰. Increased throughput was enabled by expression of antibody fragments in yeast-display systems, similar to the examples described above used to identify T cell specificities^51–53. While these technologies have identified high-affinity neutralizing antibodies, the number of antigens that can be tested using such approaches is still fairly limited. New technologies address these limitations. Building on traditional hybridoma technology, an array named PETAL of 62,208 monoclonal antibodies was built by sequentially immunizing and raising antibodies against 15,199 peptides from diverse proteomes⁵⁴ (Fig. 2a). This painstaking process generated a powerful tool to identify novel antibodies against antigens of choice. In recent years, the use of microfluidic technologies has emerged as a highly robust and precise method to interrogate antigen specificity at the single-cell level. This year, Gerard and colleagues developed a highly versatile droplet-based microfluidics device named CelliGO that enable high-throughput single-cell screening of primary B cells⁵⁵ (Fig. 2b). Using this technology, the authors were able to sort cells that produce antibodies of the appropriate specificity for the antigen of interest at the single-cell level. This strategy addresses a key limitation over previous approaches: antigen-specific B cells that secrete antibodies but do not express BCRs on their surface (i.e. plasma cells) can still be identified, sorted and sequenced. In addition, the antibodies produced can then be screened for a range of functional properties: cross-reactivity, binding to cell-surface of target cell, inhibition of target activity, or internalization by the target cell⁵⁵. In a different approach, Setliff et al. developed an elegant strategy inspired by a technology developed to identify T cell specificities^24,56 (Fig. 2c). In LIBRA-seq (linking B cell receptor to antigen specificity through sequencing), B cells are mixed with DNA-barcoded antigens. Antigen-bound B cells are sorted and then encapsulated with bead-delivered oligos using microfluidics, indexed and sequenced, thus allowing linkage of antigen specificity to BCR sequence.

Figure 2: New technologies for B cell antigen discovery.

(a) Schematic of Proteome Epitope Tag Antibody Library (PETAL), a library of 62,208 monoclonal mouse antibodies printed on a microarray. PETAL can be used to screen for antibodies specific for a given antigen or proteome. (b) Schematic of Celli GO. Polyclonal B cells are sorted based on phenotypic properties (here binding to antigen), encapsulated and their B cell receptor is sequenced using next generation sequencing. (c) Schematic of linking B cell receptor to antigen-specificity through sequencing (LIBRA-seq). Fluorescently-labelled, DNA-barcoded antigens are mixed with a polyclonal population of T cells. Single B cells specific for given antigen can be sorted out, encapsulated using droplet microfluidics and sequenced.

Application of repertoire analyses to study host-microbe interactions

The complexity of both the immune system and the microbiome make scaled experimental tools especially important for the study of host-microbe interplay. Analysis of TCR and BCR repertoires at mucosal surfaces has yielded important insights into the developmental pathways of lymphocytes in response to commensal microbes. These studies try to address how antigen specificity is linked to lymphocyte function and development. A few examples illustrate the applicability of repertoire analyses to a biological problem.

TCR repertoire analyses of colonic lamina propria T cells show that the majority of colonic Tregs are induced locally in response to microbiota-derived antigens⁵⁷. Repertoire analysis can also inform on the development of a particular T cell subset. Overlapping repertoires suggest developmental relationships between different lymphocyte subsets. Colonic Th17 and RORγt⁺Foxp3⁺ Tregs share a subset of TCRs, suggesting that RORγt⁺Foxp3⁺ Tregs could be the precursors of some Th17 cells or share a common developmental origin. This hypothesis was subsequently validated using a commensal-specific TCR transgenic line^58,59. Similarly, lymphocytes present in the epithelium of the small intestine share TCRs with Tregs, which they can be derived from^60–62. Interestingly, these intraepithelial lymphocytes (IELs) harbor a restricted set of TCRs, suggesting that such IELs may recognize a limited number of antigens⁶⁰. This observation led to the discovery that IELs recognize antigens derived from abundant microbes present in the intestine^12,61. By sorting Vβ14⁺ T cells, Littman and colleagues found that 60% of Th17 had a unique CDR3 TCR sequence³⁰. Remarkably, close to 70% of murine Th17 TCRs recognize the commensal segmented filamentous bacteria. In addition to T cell responses, commensals can also affect the BCR repertoire. Using a Rag reporter model and BCR repertoire analyses, Wesemann and colleagues showed that V (D)J recombination could occur in the lamina propria of the intestine in response to microbes⁶³. By comparing germ-free and conventional mice, they showed that the microbiome influences the preimmune B cell repertoire by enriching for anti-bacterial specificities early in life⁶⁴. Furthermore, B cells present in germinal centers in Peyer’s patches can harbor public antibody sequences that can be selected by commensals⁶⁵. These studies illustrate how commensals shape both the T and B cell compartments in the intestine.

Conclusions

Repertoire analyses have been instrumental to our understanding of the behavior of the adaptive immune system and led to the development of novel therapeutic, diagnostic and research approaches. New high-throughput technologies not only allow a molecular description of the set of antigen receptors present, but also make it possible to link these structures to their antigen specificity. Such studies have allowed the rapid identification of neoantigens, broadly neutralizing antibodies and TCRs that recognize neoantigens but not their corresponding wild-type versions. These technologies will continue to advance both basic research and clinical applications.

Highlights.

• AIRR analyses provide valuable insights into the behavior of the adaptive immune system.

• Antigen-specificity is difficult to derive from receptor sequences.

• Technologies able to link antigen-specificity to receptor sequences are emerging.