Enriching and Characterizing T Cell Repertoires from 3′ Barcoded Single-Cell Whole Transcriptome Amplification Products

Tasneem Jivanjee, Samira Ibrahim, Sarah K. Nyquist, G. James Gatter, Joshua D. Bromley, Swati Jaiswal, Bonnie Berger, Samuel M. Behar, J. Christopher Love, Alex K. ShalekMethods 2022

TL;DR

This paper introduces a novel molecular and computational workflow to retroactively recover T-cell receptor (TCR) sequences from samples already processed with standard 3’ single-cell RNA sequencing (scRNA-seq). By enriching specific transcripts from "leftover" whole transcriptome amplification (WTA) products, researchers can now link a T-cell’s specific antigen-binding identity to its full gene expression profile without needing specialized 5’ sequencing kits.

What Problem Does This Paper Address?

In immunology, knowing what a T-cell is doing (its phenotype/gene expression) is only half the story; you also need to know who it is (its TCR sequence, which determines antigen specificity).

High-throughput 3’ scRNA-seq methods (like 10x Genomics 3’, Drop-seq, or Seq-Well) are the workhorses of the field, but they have a major flaw: they typically generate short fragments (~300-500 bp) from the 3’ end of mRNA. The critical "identity" region of a TCR—the CDR3—is located roughly 1 kilobase away from that 3’ end. Consequently, this vital information is usually lost during standard library preparation. While 5’ sequencing methods exist to solve this, they require different reagents and cannot be applied to samples that have already been processed using 3’ methods. This paper provides a "best of both worlds" solution to recover TCR data from 3’ WTA products.

Methods

The authors developed a targeted enrichment strategy that plugs into existing 3’ scRNA-seq workflows (specifically optimized for Seq-Well S^3 but adaptable to others).

  1. Targeted Enrichment: They take a fraction of the 3’ barcoded WTA product and use biotinylated oligonucleotide probes to "fish out" TCR$\alpha$ and TCR$\beta$ constant regions.
  2. Affinity Capture: These probes are captured using magnetic streptavidin beads, separating the TCR transcripts from the rest of the transcriptome.
  3. Primer Extension & Amplification: They use species-specific V-region primers containing a second universal primer site (UPS2) to extend the transcripts. A final PCR step creates a library that contains the CDR3 region, the 3’ cell barcode, and the Unique Molecular Identifier (UMI).
  4. Custom Sequencing: The resulting library is sequenced on Illumina platforms using custom primers designed to read through the constant region into the CDR3.
  5. TCRGO Pipeline: The authors released TCRGO, a Python-based computational pipeline that identifies V/J regions, reconstructs CDR3 sequences, and matches them back to the original scRNA-seq cell barcodes using Hamming distance.

Datasets

The method was validated using:

  • Human and Mouse T-cells: Primary samples used to optimize primer mixes and probe sets.
  • Non-human Primates (NHPs): Specifically, crab-eating macaques (Macaca fascicularis), for which primer sets are currently being optimized.
  • Reference Databases: The pipeline utilizes known V and J region sequences from the IMGT (International ImMunoGeneTics information system).

Key Insights and Ideas

  • Retroactive Recovery: The most significant "aha moment" is that TCR information isn't actually "lost" in 3’ WTA products; it’s just diluted. By using a fraction of the WTA material that is usually discarded or archived, researchers can add TCR clonotype data to existing datasets.
  • Universal Adaptability: While the paper focuses on TCRs, the authors note that this enrichment logic can be adapted to capture any transcript of interest—such as viral RNAs or oncogenes—from a 3’ WTA library.
  • Cost-Effective Scaling: By avoiding the need for specialized 5’ kits or expensive long-read sequencing (like Nanopore), this method lowers the barrier to entry for paired TCR/transcriptome profiling.

Key Results

  • High Mapping Precision: The TCRGO pipeline achieves a mapping percentage of approximately 98%, as most reads successfully map to at least one V and J region.
  • Library Quality: The workflow produces distinct peaks on a Fragment Analyzer (~1.3 kb for TCR$\alpha$ and ~1.1 kb for TCR$\beta$), confirming the recovery of near full-length variable regions.
  • Single-Cell Linkage: The method successfully matches TCR sequences to specific cell clusters in t-SNE/UMAP space, allowing researchers to see exactly which clonotypes are "exhausted," "activated," or "memory" cells.
  • Throughput: The authors suggest a sequencing depth of 8-12 million reads per lane (on a MiSeq), allotting roughly 6,000 T-cells per lane.

Limitations and Future Work

  • Primer Sensitivity: The method relies heavily on the design of V-region primer pools. If a specific V-gene isn't covered by the primer mix, that clonotype won't be recovered.
  • WTA Contamination: The authors note that miscellaneous DNA products from the original WTA can lead to "noise" in concentration measurements (Fragment Analyzer vs. qPCR), though these non-target products generally do not interfere with sequencing.
  • Species Expansion: While human and mouse are well-supported, expanding to other model organisms (like the macaque work mentioned) requires significant manual primer optimization and testing.
  • Barcode Swapping: To minimize barcode swapping during PCR, the authors recommend splitting samples into multiple reactions, which adds a small amount of labor to the protocol.

Summary updated

Abstract

Antigen-specific T cells play an essential role in immunoregulation and diseases such as cancer. Characterizing the T cell receptor (TCR) sequences that encode T cell specificity is critical for elucidating the antigenic determinants of immunological diseases and designing therapeutic remedies. However, methods of obtaining single-cell TCR sequencing data are labor and cost intensive, requiring cell sorting and full length single-cell RNA-sequencing (scRNA-seq). New high-throughput 3' cell-barcoding scRNA-seq methods can simplify and scale this process; however, they do not routinely capture TCR sequences during library preparation and sequencing. While 5' cell-barcoding scRNA-seq methods can be used to examine TCR repertoire at single-cell resolution, it requires specialized reagents which cannot be applied to samples previously processed using 3' cell-barcoding methods. Here, we outline a method for sequencing TCR$α$ and TCR$β$ transcripts from samples already processed using 3' cell-barcoding scRNA-seq platforms, ensuring TCR recovery at a single-cell resolution. In short, a fraction of the 3' barcoded whole transcriptome amplification (WTA) product typically used to generate a massively parallel 3' scRNA-seq library is enriched for TCR transcripts using biotinylated probes, and further amplified using the same universal primer sequence from WTA. Primer extension using TCR V-region primers and targeted PCR amplification results in a 3' barcoded single-cell CDR3-enriched library that can be sequenced with custom sequencing primers. Coupled with 3' scRNA-seq of the same WTA, this method enables simultaneous analysis of single-cell transcriptomes and TCR sequences which can help interpret inherent heterogeneity among antigen-specific T cells and salient disease biology. This method can be adapted to enrich other transcripts of interest from 3' and 5' barcoded WTA libraries.
BibTeX
@inproceedings{jivanjee2022enriching,
  author = {Tasneem Jivanjee and Samira Ibrahim and Sarah K. Nyquist and G. James Gatter and Joshua D. Bromley and Swati Jaiswal and Bonnie Berger and Samuel M. Behar and J. Christopher Love and Alex K. Shalek},
  title = {{Enriching and Characterizing T Cell Repertoires from 3′ Barcoded Single-Cell Whole Transcriptome Amplification Products}},
  booktitle = {Methods},
  year = {2022},
}
Figure