Identification of shared TCR sequences from T cells in human breast cancer using emulsion RT-PCR

Abstract

Infiltration of T cells in breast tumors correlates with improved survival of patients with breast cancer, despite relatively few mutations in these tumors. To determine if T-cell specificity can be harnessed to augment immunotherapies of breast cancer, we sought to identify the alpha-beta paired T-cell receptors (TCRs) of tumor-infiltrating lymphocytes shared between multiple patients. Because TCRs function as heterodimeric proteins, we used an emulsion-based RT-PCR assay to link and amplify TCR pairs. Using this assay on engineered T-cell hybridomas, we observed ∼85% accurate pairing fidelity, although TCR recovery frequency varied. When we applied this technique to patient samples, we found that for any given TCR pair, the dominant alpha- or beta-binding partner comprised ∼90% of the total binding partners. Analysis of TCR sequences from primary tumors showed about fourfold more overlap in tumor-involved relative to tumor-free sentinel lymph nodes. Additionally, comparison of sequences from both tumors of a patient with bilateral breast cancer showed 10% overlap. Finally, we identified a panel of unique TCRs shared between patients' tumors and peripheral blood that were not found in the peripheral blood of controls. These TCRs encoded a range of V, J, and complementarity determining region 3 (CDR3) sequences on the alpha-chain, and displayed restricted V-beta use. The nucleotides encoding these shared TCR CDR3s varied, suggesting immune selection of this response. Harnessing these T cells may provide practical strategies to improve the shared antigen-specific response to breast cancer.

Keywords: T-cell receptors; T-cell repertoire profiling; breast cancer; emulsion RT-PCR; high-throughput sequencing.

Fig. 1.

Recovery of TCR pairs by emulsion RT-PCR from 5KC hybridomas. (A) Ten 5KC T-cell hybridomas were mixed in equal numbers, and emulsion RT-PCR was performed either with or without emulsion phase. Sequences were analyzed for pairing with the correct alpha-TCR (black) or beta-TCR (gray). Percentages represent the number of reads that paired with the input alpha or beta out of the total number of reads for the analyzed alpha or beta (n = 5 replicate experiments). (B) Frequency that a given TCR pair was recovered after being added in equivalent numbers is shown as a percentage of the total reads recovered (n = 5 replicate experiments). (C) Real-time quantitative PCR for the transcript abundance in the various 5KC hybridomas before addition to the emulsion RT-PCR normalized to the amount in the 1234 hybridomas (n = 3 replicate experiments). Error bars represent SD from the mean. The average read count was 670,000 across five experiments.

Fig. 2.

Pairing efficiency of patient-derived TCRs exhibits predominant pairing with a single partner. (A) Sequence read counts, cell counts, and unique TCR pairs are shown for each tissue repertoire analyzed (n = 21 TIL, n = 16 LN, n = 25 PBL). (B) Frequency of pairing of the dominant TCR pair out of all recovered pairs for either the dominant alpha- or beta-TCR.

Fig. 3.

Comparison of TIL repertoires with the corresponding sentinel LN repertoires. (A) Number of TCRs out of the total TIL repertoire that were also present in the tumor-infiltrated (LN⁺, n = 5) or non–tumor-infiltrated (LN⁻, n = 5) sentinel LN repertoires is shown as a percentage (*P = 0.01). Patient 20 presented with tumors in each breast. The tumors were resected, along with the sentinel LN from each axilla. (B) Repertoires of the TILs and LNs [either left (L) or right (R)] were compared with each other and percentage of overlap was determined as in A. (C) Twenty-nine TCRs shared between the tumor repertoires of patient 20 are shown, with TCRs found in all five (blue), four (red), three (green), or TIL-only (dark gray) repertoire. TCRs not found in a specific repertoire are represented in light gray. n, total number of unique TCR sequences in the TIL.

Fig. 4.

Sharing of TCR pairs across breast cancer tumors reveals a shared response among HLA-A2⁺ patients. TCR pairs shared between seven or more patient tumors are listed (Right) with the number of patients (Left) where the specific TCR pairs were identified. A color value was assigned corresponding to the presence of the TCR in a repertoire in the tumor (blue), in the blood (red), not found (white), or not done (gray). Amino acids corresponding to the germ-line V and J sequences are underlined. Patients 1–16 are HLA-A2:01⁺ (HLA-A genotypes of other samples are shown in Table 1).

Fig. S1.

TCR repertoire overlap among patients with breast cancer. (A) Number of unique TCRs shared between six or fewer patient tumor repertoires is shown. The TCRs shared between greater than six patient tumor repertoires is shown in Fig. 4. (B) Percentage of TCRs from a given patient repertoire unique to that repertoire (“private” TCRs) is shown. NPBL, control PBL. Error bars represent the SEM (n = 21 tumor, n = 21 PBL, n = 11 LN, and n = 6 NPBL).