Multiplex Identification of Antigen-Specific T Cell Receptors Using a Combination of Immune Assays and Immune Receptor Sequencing

Abstract

Monitoring antigen-specific T cells is critical for the study of immune responses and development of biomarkers and immunotherapeutics. We developed a novel multiplex assay that combines conventional immune monitoring techniques and immune receptor repertoire sequencing to enable identification of T cells specific to large numbers of antigens simultaneously. We multiplexed 30 different antigens and identified 427 antigen-specific clonotypes from 5 individuals with frequencies as low as 1 per million T cells. The clonotypes identified were validated several ways including repeatability, concordance with published clonotypes, and high correlation with ELISPOT. Applying this technology we have shown that the vast majority of shared antigen-specific clonotypes identified in different individuals display the same specificity. We also showed that shared antigen-specific clonotypes are simpler sequences and are present at higher frequencies compared to non-shared clonotypes specific to the same antigen. In conclusion this technology enables sensitive and quantitative monitoring of T cells specific for hundreds or thousands of antigens simultaneously allowing the study of T cell responses with an unprecedented resolution and scale.

Fig 1. Overview of multiplex assay.

A) MIRA assay outline: 1) Divide cells into equal aliquots (A to N), 2) Create immune assay antigen reagent pools by assigning each reagent (for example, dextramers or peptides; from a much larger set of antigens (1 to i) to a unique aliquot subset address, 3) Perform immune assay by combining each cell aliquot with the corresponding pool of antigens, 4) Sort T cells from each aliquot into two populations: antigen-specific and not antigen-specific, 5) Sequence TCR from all sorted populations, 6) Identify antigen-specific TCR clonotypes as those at higher frequency in the sorted antigen-specific population compared to the not antigen-specific population from the aliquot address assigned to each antigen reagent. B) MIRA set-up with dextramers. The PBMC sample is divided into an equal number of aliquots (A to H, indicated in red at top) matching the total number of dextramer, or antigen pools. Each dextramer is assigned to a unique subset, or “address”, of exactly four of eight pools as indicated in the right column. Individual dextramer assignments are indicated with an “X”. The CMV IE1 dextramer, for example, was assigned to pools B, D, F and H but not A, C, E or G.

Fig 2. Number and frequency of antigen-specific clonotypes identified with dextramer-based MIRA.

Plots show number (bars) and sum frequency (red circles) of antigen-specific clonotypes identified from each of four donors. Note donor 1 time point is the ‘month 2’ sample.

Fig 3. Comparison of MIRA and ELISPOT assay results.

Plot of IFN-γ ELISPOT (y-axis; Spot Forming Cells per 10⁶ cells) and MIRA (x-axis; Antigen-specific T cells per 10⁶ cells) results for the set of four antigens assessed from each of four donors. The ‘antigen-specific T cells per 10⁶ cells’ values for each donor were calculated by summing frequencies of all clonotypes specific for a given antigen and dividing by the proportion of T cells in each donor PBMC sample based on sort gate frequencies.

Fig 4. Number and frequency of antigen-specific clonotypes identified with peptide-based MIRA.

Plots show number (bars) and sum frequency (red circles) of antigen-specific clonotypes identified with peptides from each of five donors. The donor 1 sample used in this experiment was from the earlier of the two time points used from this donor (‘month 0’).

Fig 5. Shared antigen-specific clonotype protein sequences are simpler to generate and more abundant than those that are not shared between individuals.

A) Plot of the relative abundance of each shared (identified in more than one individual) and not shared Flu M1- and EBV BMLF1-specific clonotype protein sequence identified by MIRA in a library of 10⁹ simulated sequences generated by VDJ recombination. The relative complexity of each clonotype was measured as the abundance of each sequence in the library of 10⁹ simulated clonotypes with more abundant clonotypes being easier to generate than those that are less abundant. B) Plot of the frequencies of each shared (identified in more than one individual) and not shared Flu M1- and EBV BMLF1-specific clonotype nucleotide sequence identified by MIRA. Blue lines indicate the median observations (A) and frequencies (B) for each group.

Fig 6. Frequency (Log₁₀) dynamics of antigen-specific clonotypes.

A) Plot of all antigen-specific clonotypes identified from donor 4 with dextramers between time points spaced two months apart. B) Plot of all antigen-specific clonotypes identified from donor 1 between time points spaced two months (left) and seven years apart (right). Antigen-specific clonotypes are indicated with colored dots (see legend) and were identified from cells from the first time point of each donor.