Background

What are T-cell receptors?

In humans and closely related species, cellular immunity is mediated by T cells (or T lymphocytes), which participate directly in the detection and neutralization of pathogenic threats. Essential to T-cell function are highly specialized extracellular receptors (T-cell receptors or TCRs) that selectively bind specific antigens displayed by major histocompatibility complex (MHC) molecules on the surface of antigen-presenting cells (APCs) (Figure 1, Panel A). Antigen recognition by TCRs activates T cells, causing them to proliferate rapidly and mount immune responses through the release of cytokines.

Given the relative specificity of TCR-antigen interactions, a tremendous diversity of TCRs are required to recognize the wide assortment of pathogenic agents one might encounter. To this end, the adaptive immune system has evolved a system for somatic diversification of TCRs that is unrivaled in all of biology. The vast majority of TCRs are heterodimers composed of two distinct subunit chains (α- and β-), which both contain variable domains and, in humans, are encoded by single-copy genes. The term "clonotype" is typically used to refer either to a particular TCR variant (TCR-α or TCR-β subunit) or to a particular pairing of TCR subunit variants (TCR-α + TCR-β) shared among a clonal population of T cells. TCR diversity is generated during the early stages of T-cell development. T-cell progenitors are derived from hematopoietic stem cells (HSCs) in the thymus, and as these cells divide, extensive recombination occurs between the V- and J-segments, and the V-, D-, and J-segments, in the TCR-α and TCR-β genes, respectively, via a mechanism that also incorporates and deletes additional nucleotides (Figure 1, Panel B). Ultimately, this process—commonly referred to as "V(D)J recombination"—yields a population of T cells with sufficient TCR diversity to collectively recognize any peptide imaginable. The region of TCR-β that spans the V-D and D-J junctions, known as "complementarity determining region 3" (CDR3), is unique to each TCR-β variant and is frequently used to quantify TCR diversity in high-throughput profiling experiments. Following somatic diversification, T cells that lack sufficient affinity for MHC molecules and those that recognize self-antigens are eliminated (positive and negative selection, respectively), yielding a functional T-cell repertoire.

The seemingly endless number of potential TCR clonotypes—estimates range from 10⁶–10⁷ (Six et al., 2013) to 10¹⁵–10²⁰ unique clonotypes (Murphy et al. 2012; Laydon, Bangham, and Asquith 2015)—poses significant challenges for researchers seeking to characterize T-cell repertoires in the context of human development and disease, as extensive amounts of data must be obtained. While low-throughput approaches incorporating conventional cloning and Sanger sequencing and protein-based methods for identifying antigen-specific TCRs (e.g., tetramer assays) have yielded many important insights, the development of next-generation sequencing (NGS) technologies has dramatically expanded the prospects for this field of research.

Why do TCR profiling?

High-throughput TCR profiling experiments have already yielded fundamental insights regarding T-cell development and TCR repertoire diversity (Calis and Rosenberg 2014; Woodsworth, Castellarin, and Holt 2013). For example, these approaches have demonstrated that TCR variation does not determine T-cell fate (Wang et al. 2010) and that there is considerable overlap in the population at large for so-called "public TCRs" or "public clones", which occur much more frequently than would be expected by chance (Robins et al. 2010). A sampling of different populations has revealed that TCR repertoire diversity declines linearly with age and is significantly reduced in patients suffering from autoimmune diseases or cancer, relative to healthy individuals (Britanova et al. 2014; Sherwood et al. 2013; Klarenbeek et al. 2012).

In the clinic, TCR profiling has been used to analyze the recovery of the immune system in patients who have undergone hematopoietic stem cell transplants (HSCT), and to compare the efficacy of approaches aimed at accelerating this process (van Heijst et al. 2013). Looking to the future, high-throughput TCR profiling holds tremendous promise as both a diagnostic tool, and as a means for developing new therapeutics and treatment modalities (Calis and Rosenberg 2014; Woodsworth, Castellarin, and Holt 2013). For example, TCR repertoire analysis could be used to evaluate a candidate vaccine's capacity to trigger a protective immune response.

Sequencing approaches for TCR repertoire analysis

The vast majority of TCR-profiling experiments performed thus far have focused on capturing genomic DNA or mRNA sequences that correspond to the CDR3 region of the TCR-β subunit chain (Calis and Rosenberg 2014; Woodsworth, Castellarin, and Holt 2013). Given that the CDR3 region is thought to be unique to each TCR-β variant, sequence variation in this region has served as a useful proxy for overall T-cell repertoire diversity.

While sequencing genomic DNA may be preferable for certain TCR-profiling applications—including those that involve quantifying various T-cell subpopulations—this approach is not without its limitations, and methods that involve analyzing mRNA sequences carry several important advantages. TCR mRNA templates are likely to be more highly represented than DNA templates in any one T cell, such that mRNA sequencing approaches will afford greater sensitivity and allow for more comprehensive identification of unique TCR variants, including those that are present in a very small proportion of T cells. Another important benefit of sequencing mRNA rather than genomic DNA is that it specifically allows for the identification of expressed TCR sequences that have undergone splicing and post-transcriptional processing and are likely to yield functional proteins. DNA-based approaches, by contrast, do not identify TCR sequences in their translated forms, and will unavoidably yield many nonproductive sequences that are functionally irrelevant. For this reason, mRNA sequencing is the preferred option for researchers interested in exploring functional aspects of specific TCR variants. TCR profiling approaches that involve sequencing genomic DNA are also subject to significant technical limitations. Due to the lack of splicing, DNA-derived templates are considerably longer than their RNA counterparts, such that amplification of genomic DNA corresponding to TCR variable regions (including CDR3) requires multiplex PCR and is potentially susceptible to biases imposed by the various primer pairs. As demonstrated below, the relatively shorter length of TCR mRNA templates allows for simpler amplification schemes in which TCR-α and TCR-β variable regions are captured with single primer pairs, minimizing the potential for amplification biases and allowing for analysis of both subunit chains in the same experiment.

Experimental workflow

First-strand cDNA synthesis and template switching

This approach utilizes leukocyte RNA extracted from human peripheral blood or intact human T cells as starting material. First-strand cDNA synthesis is dT-primed (TCR dT Primer) and performed by the MMLV-derived SMARTScribe Reverse Transcriptase (RT), which adds nontemplated nucleotides upon reaching the 5′ end of each mRNA template (Figure 2, Panel A). The SMART-Seq v4 Oligonucleotide—enhanced with Locked Nucleic Acid (LNA) technology for increased sensitivity and specificity—then anneals to the nontemplated nucleotides, and serves as a template for the incorporation of an additional sequence of nucleotides to the first-strand cDNA by the RT (this is the template-switching step). This additional sequence—referred to as the "SMART sequence"—serves as a primer-annealing site for subsequent rounds of PCR, ensuring that only sequences from full-length cDNAs undergo amplification.

cDNA amplification and incorporation of Illumina adapters by semi-nested PCR

Following reverse transcription and extension, two rounds of PCR are performed in succession to amplify cDNA sequences corresponding to variable regions of TCR-α and/or TCR-β transcripts. The first PCR uses the first-strand cDNA as a template and includes a forward primer with complementarity to the SMART sequence (SMART Primer 1), and a reverse primer that is complementary to the constant (i.e. nonvariable) region of either TCR-α or TCR-β (TCRa/b Human Primer 1); both reverse primers may be included in a single reaction if analysis of both TCR subunit chains is desired. By priming from the SMART sequence and constant region, the first PCR specifically amplifies the entire variable region and a considerable portion of the constant region of TCR-α and/or TCR-β cDNA (Figure 2, Panel B).

The second PCR takes the product from the first PCR as a template, and uses semi-nested primers (TCR Primer 2 and TCRa/b Human Primer 2) to amplify the entire variable region and a portion of the constant region of TCR-α and/or TCR-β cDNA (once again, either or both TCR subunit chains may be amplified in a single reaction). Included in the forward and reverse primers are adapter and index sequences which are compatible with the Illumina sequencing platform (read 2 + i7 + P7 and read 1 + i5 + P5, respectively). Following post-PCR purification, size selection, and quality analysis, the library is ready for Illumina sequencing.

Library quality control and Illumina sequencing

Prior to sequencing, libraries are purified and size selected using Solid Phase Reversible Immobilization (SPRI) beads. To confirm the success of library amplification and purification, samples are run on a Fragment Analyzer or Bioanalyzer (Figure 3). The position and shape of electropherogram peaks vary depending on whether TCR-α and/or TCR-β sequence fragments are included in the library, the nature of the sample input, and the analysis method. Once the quality and size of each purified library have been confirmed, samples are sequenced on the Illumina platform using 300 bp paired-end reads, which fully capture the TCR sequence included in each cDNA molecule.

Cross-sample clonotype comparisons

The SMARTer Human TCR a/b Profiling Kit was used to process RNA extracted from replicate peripheral blood mononuclear cell (PBMC) samples obtained from eight blood cancer patients. Sequencing outputs for the 16 resulting cDNA libraries were analyzed using a computational immunology platform as reported in the Methods section below. One form of analysis involved determining the number of common TCR-α and TCR-β clonotypes shared between samples, based on sequences that mapped to the CDR3 regions of TCR-α or TCR-β. The results are depicted using a heat map, in which different tile colors correspond to Log₁₀-transformed counts for the number of clonotypes shared between the indicated pairings (Figure 4). Not surprisingly, the greatest level of overlap is observed between replicate PBMC samples from the same patient. In contrast, the level of overlap is dramatically lower for PBMC samples from different patients. These results speak to both the reproducibility and sensitivity of the SMARTer approach, in that it generates similar TCR profile data for replicate samples, and markedly divergent TCR profile data for samples obtained from different patients.

Sequencing reads on target

To evaluate the performance of the kit for a range of input amounts, the SMARTer workflow was performed on three different amounts of peripheral blood RNA (10 ng, 100 ng, and 1,000 ng) and the resulting cDNA libraries were sequenced as above. Sequencing outputs were downsampled to either ~260,000 or ~275,000 reads, depending on the analysis method, and processed using an application provided by Illumina via the BaseSpace website, as reported in the Methods section, below. For each RNA sample amount analyzed, ≥70% of sequencing reads mapped to a CDR3 region in either TCR-α or TCR-β, with the 10-ng sample amount yielding the highest percentage of on-target reads (Figure 5, Panel A). These results demonstrate that the SMARTer approach can capture and amplify TCR sequences from total RNA with considerable specificity across a wide range of sample input amounts.

Correlation of clonotype count data for varying sample input amounts

Another form of analysis involved plotting counts of the 100 most highly represented clonotypes for varying sample input amounts. Comparison of clonotype count data for the 100 ng and 1,000 ng sample amounts yielded a Pearson correlation coefficient of 0.80 and a Spearman coefficient of 0.80 (Figure 5, Panel B), a result that attests to the robustness of the SMARTer approach for input sample amounts that vary by at least one order of magnitude.

Visual representation of TCR-β clonotype distributions

The distribution of TCR clonotypes identified in the sequencing data can also be depicted visually using chord diagrams (Figure 6). Included here are chord diagrams representing the observed distributions of the indicated TCR-β Variable-Joining (V-J) segment combinations for each RNA input amount. Each arc (on the periphery of the diagram) represents a V or J segment and is scaled lengthwise in proportion to that segment's distribution in the dataset. Each chord (connecting the arcs) represents a set of clonotypes which include the indicated V-J combination and is weighted according to that set's distribution in the dataset. Comparison of the three diagrams suggests that the indicated clonotypes are identified at similar proportions for each RNA input amount.

Assessing the sensitivity and reproducibility of the SMARTer approach

In order to assess the sensitivity and reproducibility of the SMARTer approach, the protocol was performed in replicate on PBMC RNA samples spiked at varying concentrations (10%, 1%, 0.1%, 0.01%, and 0.001%) with RNA obtained from a homogenous population of Leukemic Jurkat T cells (TRAV8-4-TRAJ3, TRBV12-3-TRBJ1-2 clonotype). The number of TRBV12-3-TRBJ1-2-specific sequence reads obtained for each spiked sample was normalized by subtracting the number of corresponding reads obtained for negative control samples consisting of unspiked PBMC RNA. Following Log₁₀ transformation, plotting of the data (spike-in dilution vs. normalized read count) and linear regression analysis revealed a statistically significant correlation (p = 3.93 x 10^-10, R² = 0.99) between the amount of spiked-in Jurkat RNA and the number of TRBV12-3-TRBJ1-2-specific sequence reads (Figure 7, Panel A). This result demonstrates that differences in the relative abundance of transcripts for a particular TCR clonotype are faithfully and reproducibly represented in sequencing libraries generated using the SMARTer approach. Comparison of the number of TRBV12-3-TRBJ1-2-specific sequence reads obtained for the control vs. spike-in samples suggests that added Jurkat RNA at a concentration of 0.1% is detectable above background in the sequencing output (p <0.005) at a depth of ~275,000 reads, evidence of the sensitivity afforded by the SMARTer approach (Figure 7, Panel B).

Our SMARTer Human TCR a/b Profiling Kit provides a powerful solution for those seeking to elucidate the diversity of TCR-α and/or TCR-β subunits present in samples consisting of human leukocyte RNA or intact T cells. In contrast with profiling methods that involve the amplification of genomic DNA, our 5' RACE-based approach uses total RNA as input material. Starting from RNA allows for the capture of complete TCR V(D)J variable regions, the avoidance of multiplex PCR amplification strategies, and the detection of low-abundance TCR clonotypes. Sequencing data derived from TCR RNA may also yield worthwhile insights regarding the function of corresponding subunit proteins. The results presented here indicate that the SMARTer Human TCR a/b Profiling Kit generates data that is highly reproducible over a range of sample input amounts and consists largely of on-target reads. They also demonstrate that our method allows for the detection of TCR clonotypes present at proportions of 0.1% of total RNA at a relatively shallow read depth. On the basis of its reliability and elegance, our SMARTer approach to TCR profiling will help forward basic and applied research objectives involving the analysis of human immune repertoires.

All sequencing libraries were generated using protocols and reagents included in the SMARTer Human TCR a/b Profiling Kit (Takara Bio, Cat. Nos. 635014, 635015, 635016).

Input material

Libraries included in the quality analysis (Figure 3) were generated from 10 ng of commercially available input RNA. Both Human Blood, Peripheral Leukocytes Total RNA (Takara Bio, Cat. No. 636592) and Total RNA - Human Tumor Cell Line: Jurkat (BioChain, Cat. No. R1255815-50) were used. For the cross-sample clonotype comparisons (Figure 4), libraries were generated from 5 ng–2 µg of RNA obtained from Conversant Biologics, Inc., which sourced the PBMC samples and extracted the RNA. Libraries included in the analyses of reads on target (Figure 5) and TCR-β clonotypes identified (Figure 6) were generated from 10 ng, 100 ng, and 1,000 ng of Human Blood, Peripheral Leukocytes Total RNA, respectively. Libraries included in the sensitivity and reproducibility assays (Figure 7) were generated from 10 ng of Human Blood, Peripheral Leukocytes Total RNA spiked with serial-diluted Total RNA - Human Tumor Cell Line: Jurkat at the indicated concentrations.

PCR parameters

As indicated in the SMARTer Human TCR a/b Profiling Kit user manual, all libraries were subject to two rounds of semi-nested PCR amplification. Libraries included in the quality analysis (Figure 3) and in the sensitivity and reproducibility assays (Figure 7) were amplified using 21 cycles for PCR 1, and 20 cycles for PCR 2. Libraries included in the cross-sample clonotype comparisons (Figure 4) were amplified using 20 cycles for PCR 1 and 15–20 cycles for PCR 2 (depending on the RNA input amount). For the analyses of reads on target (Figure 5) and TCR-β clonotypes identified (Figure 6), the following numbers of PCR cycles were used for each input amount:

• 10 ng: 21 cycles for PCR 1, 20 cycles for PCR 2
• 100 ng: 21 cycles for PCR 1, 15 cycles for PCR 2
• 1,000 ng: 21 cycles for PCR 1, 12 cycles for PCR 2

For all experiments performed, mixtures of TCR-α- and TCR-β-specific primers were included in both PCR 1 and PCR 2 at a 2:1 ratio (TCR-α:TCR-β).

Library purification

Amplified libraries were purified using the Agencourt AMPure XP PCR purification kit (5-ml size: Beckman Coulter Item No. A63880; 60-ml size: Beckman Coulter Item No. A63881). Libraries validated on a Fragment Analyzer were subject to one double size selection, while libraries validated on a Bioanalyzer were subject to two double size selections. Beads were pelleted using a SMARTer-Seq Magnetic Separator - PCR Strip (Takara Bio, Cat. No. 635011).

Library validation

Library validation was performed using an Advanced Analytical Fragment Analyzer and the High Sensitivity NGS Fragment Analysis Kit (Advanced Analytical, Cat. No. DNF-474), or an Agilent 2100 Bioanalyzer and the DNA 1000 Kit (Agilent, Cat. No. 5067-1504). For validation on the Fragment Analyzer, purified libraries were diluted 1:5, and 1 µl of the diluted library was analyzed. For validation on the Bioanalyzer, 1 µl of purified, undiluted library was analyzed.

Illumina sequencing

Libraries were pooled to a final pool concentration of 4 nM. Pooled libraries were then diluted to a final concentration of 13.5 pM, including a 10% PhiX Control v3 (Illumina, Cat. No. FC-110-3001) spike-in. Libraries were sequenced on an Illumina MiSeq® sequencer using the 600-cycle MiSeq Reagent Kit v3 (Illumina, Cat. No. MS-102-3003) with paired-end, 2 x 300 base pair reads.

Data analysis

For the cross-sample clonotype comparisons (Figure 4), sequencing data were analyzed using the AbGenesis computational immunology platform, provided by Distributed Bio, Inc. For other analyses, sequencing data were downsampled to either ~260,000 reads (Figure 5 and Figure 6) or ~275,000 reads (Figure 7) and analyzed using the MiXCR software package (Bolotin et al. 2015), hosted on BaseSpace, Illumina's cloud computing environment for next-generation sequencing. Statistical analyses were performed using Excel (Microsoft).

Bolotin, D. A. et al. MiXCR: software for comprehensive adaptive immunity profiling. Nat. Methods 12, 380–381 (2015).

Britanova, O. V et al. Age-related decrease in TCR repertoire diversity measured with deep and normalized sequence profiling. J. Immunol. 192, 2689–98 (2014).

Calis, J. J. A. & Rosenberg, B. R. Characterizing immune repertoires by high throughput sequencing: Strategies and applications. Trends Immunol. 35, 581–590 (2014).

van Heijst, J. W. J. et al. Quantitative assessment of T cell repertoire recovery after hematopoietic stem cell transplantation. Nat. Med. 19, 372-7 (2013).

Kenneth Murphy, Paul Travers, Mark Walport, Michael Ehrenstein, Claudia Mauri, Allan Mowat, A. S., Hsu, D. C., Murphy, K. M., Travers, P. & Walport, M. Janeway's Immunobiology. Garl. Sci. 7, (2008).

Klarenbeek, P. L. et al. Inflamed target tissue provides a specific niche for highly expanded T-cell clones in early human autoimmune disease. Ann. Rheum. Dis. 71, 1088–1093 (2012).

Laydon, D. J., Bangham, C. R. M. & Asquith, B. Estimating T-cell repertoire diversity: limitations of classical estimators and a new approach. Phil Trans R Soc B 370, 20140291- (2015).

Robins, H. S. et al. Overlap and effective size of the human CD8+ T cell receptor repertoire. Sci. Transl. Med. 2, 47ra64 (2010).

Sherwood, A. M. et al. Tumor-infiltrating lymphocytes in colorectal tumors display a diversity of T cell receptor sequences that differ from the T cells in adjacent mucosal tissue. Cancer Immunol. Immunother. 62, 1453–1461 (2013).

Six, A. et al. The past, present, and future of immune repertoire biology - the rise of next-generation repertoire analysis. Front. Immunol. 4, (2013).

Wang, C. et al. High throughput sequencing reveals a complex pattern of dynamic interrelationships among human T cell subsets. Proc. Natl. Acad. Sci. U. S. A. 107, 1518–23 (2010).

Woodsworth, D. J., Castellarin, M. & Holt, R. A. Sequence analysis of T-cell repertoires in health and disease. Genome Med 5, 98 (2013).