Quantitative Characterization of the T Cell Receptor Repertoire of Naïve and Memory Subsets Using an Integrated Experimental and Computational Pipeline Which Is Robust, Economical, and Versatile

Abstract

The T cell receptor (TCR) repertoire can provide a personalized biomarker for infectious and non-infectious diseases. We describe a protocol for amplifying, sequencing, and analyzing TCRs which is robust, sensitive, and versatile. The key experimental step is ligation of a single-stranded oligonucleotide to the 3′ end of the TCR cDNA. This allows amplification of all possible rearrangements using a single set of primers per locus. It also introduces a unique molecular identifier to label each starting cDNA molecule. This molecular identifier is used to correct for sequence errors and for effects of differential PCR amplification efficiency, thus producing more accurate measures of the true TCR frequency within the sample. This integrated experimental and computational pipeline is applied to the analysis of human memory and naive subpopulations, and results in consistent measures of diversity and inequality. After error correction, the distribution of TCR sequence abundance in all subpopulations followed a power law over a wide range of values. The power law exponent differed between naïve and memory populations, but was consistent between individuals. The integrated experimental and analysis pipeline we describe is appropriate to studies of T cell responses in a broad range of physiological and pathological contexts.

T‐cell receptor and B‐cell receptor repertoire profiling in adaptive immunity

“…RNA‐based methods employ 5′ rapid amplification of cDNA ends (RACE)‐based template switching or single‐stranded RNA ligation to add a universal oligo at the 5′ end of the first strand cDNA, and thus minimize quantitative bias via the use of a single primer pair for amplification. RNA‐based methods can utilize unique molecular identifiers (UMIs), which comprise a stretch of 8–12 random nucleotides that is used to mark each initial cDNA with a unique ‘molecular barcode’.…”

Some deterministic and stochastic mathematical models of naïve T‐cell homeostasis

“…Data available to date suggest that apart, perhaps, from a few large clonotypes, the repertoire of naïve T cells in a mouse or human consists of many clonotypes with small clonal sizes. Thus, given current sample sizes, the majority of TCRs found in typical samples will be found on only one cell per sample . We must conclude that most TCR clonotypes are unobserved.…”

“…Thus, given current sample sizes, the majority of TCRs found in typical samples will be found on only one cell per sample. 95 We must conclude that most TCR clonotypes are unobserved. These observations, no doubt a source of frustration to the experimentalist, are the basis of convenient approximations for mathematicians.…”

“… A power‐law distribution of clonal sizes . In the Figure, we use the exponent b = 2.5, which gives a mean clonal family size of cells. A geometric distribution of clonal sizes.…”

“…Similarly, if asequence is present as two copies, is it possible to establish whether there were two cells in the pool of cells, or two RNA copies of the same TCR chain in one cell? Even at the large end of the spectrum, the distribution of family sizes is difficult to evaluate reliably because of the possibility of inhomogeneities in the PCR processes used 95. Single-cell measurements, where PCR and sequencing are performed on one cell at a time, avoid these problems and biases, and can identify paired α and β chains 96,97.…”