Common T-Cell-Receptor Motifs and Features in Patients with Cytomegalovirus (CMV)-Seronegative End-Stage Renal Disease Receiving a Peptide Vaccination against CMV

Abstract

After solid-organ transplantation, reactivation of the cytomegalovirus (CMV) is often observed in seronegative patients and associated with a high risk of disease and mortality. CMV-specific T cells can prevent CMV reactivation. In a phase 1 trial, CMV-seronegative patients with end-stage renal disease listed for kidney transplantation were subjected to CMV phosphoprotein 65 (CMVpp65) peptide vaccination and further investigated for T-cell responses. To this end, CMV-specific CD8⁺ T cells were characterized by bulk T-cell-receptor (TCR) repertoire sequencing and combined single-cell RNA and TCR sequencing. In patients mounting an immune response to the vaccine, a common SYE(N)E TCR motif known to bind CMVpp65 was detected. CMV-peptide-vaccination-responder patients had TCR features distinct from those of non-responders. In a non-responder patient, a monoclonal inflammatory T-cell response was detected upon CMV reactivation. The identification of vaccine-induced CMV-reactive TCRs motifs might facilitate the development of cellular therapies for patients wait-listed for kidney transplantation.

Keywords: CMV; TCR motif; end-stage renal disease; peptide vaccination; single-cell sequencing.

Figure 1

Deep TCRA and TCRB sequencing for patients’ PBMCs in CMV-seronegative end-stage renal disease. (A) Top 100 TCRA and TCRB sequences from responder (#009) and non-responder (#002) patients visualized at baseline (T0) and three post-vaccine timepoints (T2, T3 and T4). Two exemplary patients are shown. Gray, respective clonotypes were not found at time points illustrated. (B) Proportion of the top-ten clonotypes from responder (#003, #007) and non-responder (#002) patients visualized at baseline (T0) and two to four different post-vaccine timepoints (T1–4). Samples were sequenced in technical duplicates.

Figure 2

Unsupervised deep-learning-based clustering of patient repertoires using DeepTCR. Ten CMV-seronegative end-stage renal disease patients waiting for kidney transplantation were vaccinated four times biweekly. #001 and #007 represent patients with existing strong immune responses. Patients classified as responders (#003, #005, #006 and #009; green labels) and non-responders (#002, #004, #008 and #010; red labels) by CMV-tetramer binding formed distinct clades, with a number of CDR3 features (*) best correlating with response.

Figure 3

A “SYE(N)E” motif (left) harboring polar, acidic and neutral amino acids known to bind HLA-A*02-presented CMVpp65 was found only in responders but not in non-responder patients (right).

Figure 4

Single-cell TCR sequencing upon CMV reactivation in non-responder patient #002 five months after transplantation. Combined scTCR-seq and scRNA-seq for HLA-A2*-tetramer-sorted CMV-reactive circulating peripheral T cells after CMV reactivation. (A) Tetramer sorting gating strategy for CMV-reactive circulating peripheral T cells. (B) TCRA and TCRB gene usage heatmap (left) and clonotype percentages (right) of CMV-reactive T-cell clonotypes. Most-abundant TCR beta chain is illustrated as amino-acid sequence.

Figure 5

Combined scRNA-seq and scTCR-seq of patient #002-dominating CD8⁺ CMV-reactive T-cell clonotype. (A) Six differential T-cell clusters of patient #002-dominating CD8⁺ CMV-reactive T-cell clonotypes visualized by UMAP. Expression levels of CD3E and CD8A are shown. Co-visualization of the top TCR clonotype and its distribution within T-cell clusters visualized by UMAP. (B) Heat map of differentially expressed genes in clusters.

Figure 6

Combined scRNA-seq and scTCR-seq of CMV-reactive T cells from vaccine-responding patient #003. (A) Tetramer sorting gate for CMV-reactive circulating peripheral T cells. (B) Frequency of top-ten TCR clonotypes within the enriched CMV-reactive T-cell repertoire. Amino-acid sequence of the second-most-abundant clonotype is shown. (C) Clonotypic evolution of the second-most-abundant sc-sequencing-retrieved TCR for longitudinal TCRB deep-sequencing datasets (left) and longitudinal peripheral pp65-specific ELISpot responses (right). n.d., not detectable; n.a., not assessed.