Profiling the T-cell receptor beta-chain repertoire by massively parallel sequencing

Abstract

T-cell receptor (TCR) genomic loci undergo somatic V(D)J recombination, plus the addition/subtraction of nontemplated bases at recombination junctions, in order to generate the repertoire of structurally diverse T cells necessary for antigen recognition. TCR beta subunits can be unambiguously identified by their hypervariable CDR3 (Complement Determining Region 3) sequence. This is the site of V(D)J recombination encoding the principal site of antigen contact. The complexity and dynamics of the T-cell repertoire remain unknown because the potential repertoire size has made conventional sequence analysis intractable. Here, we use 5'-RACE, Illumina sequencing, and a novel short read assembly strategy to sample CDR3(beta) diversity in human T lymphocytes from peripheral blood. Assembly of 40.5 million short reads identified 33,664 distinct TCR(beta) clonotypes and provides precise measurements of CDR3(beta) length diversity, usage of nontemplated bases, sequence convergence, and preferences for TRBV (T-cell receptor beta variable gene) and TRBJ (T-cell receptor beta joining gene) gene usage and pairing. CDR3 length between conserved residues of TRBV and TRBJ ranged from 21 to 81 nucleotides (nt). TRBV gene usage ranged from 0.01% for TRBV17 to 24.6% for TRBV20-1. TRBJ gene usage ranged from 1.6% for TRBJ2-6 to 17.2% for TRBJ2-1. We identified 1573 examples of convergence where the same amino acid translation was specified by distinct CDR3(beta) nucleotide sequences. Direct sequence-based immunoprofiling will likely prove to be a useful tool for understanding repertoire dynamics in response to immune challenge, without a priori knowledge of antigen.

Figure 1.

(A) Representation of the TCR_β locus at human chromosome 7q34. The TCR_β locus spans 620 kb and includes over 50 TRBV genes (green) belonging to 30 subgroups. There are two TRBC genes (light blue) each downstream from a TRBD (dark blue) and six or seven TRBJs (yellow). Recombination first occurs between TRBJ and TRBD genes, followed by recombination to a TRBV gene. (Red lines) Addition of nontemplated bases. After transcription, intervening sequences are spliced out so that a TRBC is adjacent to the recombined V-D-J sequence. Gene width and distances are not to scale. (Red asterisks under TRBC) Location of primers used for 5′-RACE and PCR reactions. Refer to Supplemental Figure 1A for primer locations. A detailed locus map can be obtained from IMGT (www.imgt.org/textes/IMGTrepertoire). (B) Flowchart illustrating 5′-RACE and Illumina library construction. For more detail, please refer to Methods and Supplemental Figure 1.

Figure 2.

(A) TCR_β diversity. A total of 33,664 TCR_β clonotypes were identified from complete and in-frame CDR3_β sequences assembled by iSSAKE. Clonotypes with a copy number of one (clonotypes identified by a single iSSAKE contig) account for 65.3% of all clonotypes. Clonotypes identified by two to 19 iSSAKE contigs account for 32.6% of all clonotypes, and high-abundance clonotypes (contig depth ≥20) account for 2.1% of the total. (B) Saturation analysis. In duplicate experiments we chose independent sets of 5, 10, 15, 20, 30, and 35 M reads at random from the pool of 40,582,229 total sequence reads in our data set. These subsets of reads were assembled and clonotypes counted as the set of complete, in-frame, nonredundant CDR3_β sequences. The number of clonotypes (mean ± SD for the duplicate experiments) is plotted as a function of the number of reads. Error bars are contained within the symbols.

Figure 3.

TRBV and TRBJ usage. (A) Relative frequency of usage of TRBV segments was for the subset of clonotypes with an unambiguous TRBV gene segment assignment (n = 22,704). (B) Relative frequency of usage of TRBJ segments for the set of all clonotypes (n = 33,664).

Figure 4.

Frequencies of V-J pairing calculated from the subset of clonotypes with an unambiguous TRBV gene segment assignment (n = 22,704). (Blue to purple rectangular bands) TRBJ segments, (red to cyan rectangular bands) TRBV segments. The width of the bands is proportional to the number of times the TRBV and TRBJ genes connected by the band co-occur in CDR3_β sequences. TRBV and TRBJ segments are arranged from left to right and right to left, respectively, and ordered by total pairing links they share. (This illustrates the data contained in Supplemental Table 1.) This figure was generated using the Circos software package (Krzywinski et al. 2009).

Figure 5.

CDR3_β nucleotide length distribution and sequence composition of the most abundant CDR3_β length. To explore CDR3_β length variation we used a precise length criterion, defined as all bases between the last cysteine of TRBV and the phenylalanine in the TRBJ segment motif FGXG. Of the 33,664 total clonotypes, 30,366 could be classified in this manner, and the length distribution of this subset was plotted (A). The most frequently observed length was 45 nt. For the subset of clonotypes with 45 nt CDR3_β sequences, we created logos for the nucleotide (B) and inferred amino acid (C) composition, using WebLogo (Crooks et al. 2004).