Sensitive and frequent identification of high avidity neo-epitope specific CD8 + T cells in immunotherapy-naive ovarian cancer

Abstract

Immunotherapy directed against private tumor neo-antigens derived from non-synonymous somatic mutations is a promising strategy of personalized cancer immunotherapy. However, feasibility in low mutational load tumor types remains unknown. Comprehensive and deep analysis of circulating and tumor-infiltrating lymphocytes (TILs) for neo-epitope specific CD8⁺ T cells has allowed prompt identification of oligoclonal and polyfunctional such cells from most immunotherapy-naive patients with advanced epithelial ovarian cancer studied. Neo-epitope recognition is discordant between circulating T cells and TILs, and is more likely to be found among TILs, which display higher functional avidity and unique TCRs with higher predicted affinity than their blood counterparts. Our results imply that identification of neo-epitope specific CD8⁺ T cells is achievable even in tumors with relatively low number of somatic mutations, and neo-epitope validation in TILs extends opportunities for mutanome-based personalized immunotherapies to such tumors.

Fig. 1

Identification of neo-epitope specific PBLs. a, b Representative example of peripheral blood CD8⁺ T lymphocytes (PBLs) response against the HHAT_L75F neo-epitope in patient CTE-0013. a T-cell reactivity measured by IFNγ ELISpot against the neo-epitope (mut) from HHAT_L75F gene but not against the wild-type (wt) peptide (PHA: phytohemagglutinin; No Ag: no peptide). Shown are the average of triplicate + SD. b Representative example of validation of neo-epitope specific CD8⁺ T cells using peptide-MHC multimers. c T-cell reactivity measured by IFNγ ELISpot against the neo-epitope from KIR2DS4_I7S gene but not against the wild-type peptide in patient CTE-0012 (PHA: phytohemagglutinin; No Ag: no peptide). Shown are the average of triplicate + SD. d Representative example of neo-epitope validation by polychromatic intracellular cytokine staining (ICS) in patient CTE-0012 without (No Ag) or after stimulation with a cognate neo-epitope (KIR2DS4_I7S). All other PBL neo-epitopes are described in Supplementary Fig. 1 and Table 2. e Proportion of EOC patients with documented peripheral blood CD8⁺ T-cell (PBL) response against neo-epitopes. f Frequency and g number (mean ± SEM) of neo-epitope specific CD8⁺ PBLs per patient, detected after one round of in vitro stimulation (Supplementary Table 2). h Cumulative analysis showing the cytokine profiles of neo-epitope specific CD8⁺ PBLs. The SPICE analysis represents the functional composition of cytokine-producing neo-epitope specific CD8⁺ T cells (mean ± SEM). i Heatmap showing expression of all the genes from Reactome’s antigen processing and cross-presentation pathway^, . The number of neo-epitopes recognized by PBL of each patient is shown on top of the heatmap. Genes from this pathway are ordered based on the differential expression fold change between patients with or without PBL neo-epitope recognition. j Gene set enrichment analysis (GSEA) curves for various Reactome pathways significantly different between patients with or without PBL neo-epitope recognition. Vertical bars indicate the ranked positions of each gene from the respective gene set. Statistical significance of these enrichments is reported by the FDR of the GSEA. Corresponding heatmaps for IFNγ signaling, PD-1 signaling and collagen formation are reported in Supplementary Fig. 2

Fig. 2

Identification of neo-epitope specific TILs. a–c Representative example of neo-epitope (ZCCHC11_P1265H)-specific TILs in patient CTE-0015. a IFNγ ELISpot showing the average of triplicate + SD (PHA: phytohemagglutinin; No Ag: no peptide) and peptide-MHC multimer staining. All other neo-epitopes are described in Supplementary Fig. 3 and Table 2. b Intracellular cytokine staining showing the frequency of viable IFNγ, TNFα, and IL-2 cytokine-producing TILs. c Proportion of patients with neo-epitope reactive TILs. d Cumulative analysis (SPICE) showing the functional composition (cytokine profiles) of neo-epitope specific TILs (mean ± SD). e Repertoire of neo-epitopes in PBLs and TILs in 19 patients. Each colored square represents one neo-epitope for which T-cell reactivity was validated either exclusively in PBLs (blue), exclusively in TILs (red), in both compartments (orange) or in none of them (white). For 5 out of 19 patients no tumor samples were available for TIL generation (black squares). f Neo-epitope-specific TILs from patient CTE-0011 were FACS sorted with HLA-A1101-SEPT9_R289H multimer. Sorted cells (left) underwent TCR sequencing, as well as autologous bulk PBLs, bulk tumor and bulk TIL cultures. The Manhattan plot reports the V/J recombination of the T-cell receptor β (TCRβ) of SEPT9_R289H-specific T cells: V and J segments are represented according to chromosomal location on the x and y-axis, respectively. On the right, the distribution of the TCR frequencies is shown for PBL, tumor and TILs. Dominant SEPT9_R289H-specific TCRβ are identified by red bars (and arrows) in tumor and TILS, whereas no SEPT9_R289H-specific TCRs was identified in PBLs. gTRIM26_G497W-specific circulating T cells from patient CTE-0010 were isolated based on CD137 upregulation and immediately processed for TCR-sequencing analysis (center). TCR repertoire analysis was performed in parallel on matched bulk PBLs, bulk tumor, and bulk TIL cultures (right). The most frequent shared neo-epitope specific TCR was ranked according to its relative frequency in PBL and tumor (red bars and arrows). The same TCR was undetectable in IL-2-expanded TIL cultures. Elements of box plots represent median (line), 25% and 75% confidence limit (box limits) and 10% and 90% confidence limit (whiskers)

Fig. 3

TIL cultures can be enriched for neo-epitope specific T cells. a Schematic representation of the alternative procedures for TIL expansion. Single cell tumor suspensions were plated in the presence of high-dose IL-2 alone (conventional TILs) or alternatively, in the presence of high-dose IL-2 supplemented with pools of predicted mutated peptides (primed TILs). b Representative examples of conventional and primed TILs, interrogated for the presence of neo-epitope specific TILs by multimer staining. c Cumulative analysis of the frequencies of neo-epitope specific CD8⁺ T lymphocytes detected in conventional (x-axis) and primed (y-axis) TIL cultures. d Proportions of patients with documented neo-epitope specific T-cell responses in the different compartments and within the distinct TIL culture conditions. e Landscape of neo-epitope specific T-cell responses in circulating, conventional TILs and primed TILs in nine patients for which primed TILs were available. T-cell responses against mutated epitopes identified exclusively in PBLs (blue), in TILs (red), in both the compartments (orange) or in none of them (white) are indicated

Fig. 4

Functional analyses of neo-epitope specific PBLs and TILs. a Higher functional avidity of HHAT_L75F-specific TIL and PBL clones. Shown are the relative frequencies of IFNγ-producing CD8⁺ T cells (mean ± SD). b Higher functional avidity (mean ± SEM) of HHAT_L75F-specific TILs compared with PBL clones based on TNFα and IL-2 release (Supplementary Fig. 5 shows raw data). c Higher cytolytic capacity of HHAT_L75F-specific TILs compared with PBLs. Data show the percentage of lysis of T2 cells pulsed with different concentrations of the cognate neo-epitope. d Mass cytometry analysis showing the relative phosphorylation of ERK1/2 and S6 proteins involved in TCR-signaling pathways following CD3 stimulation of PBL and TIL clones. Graphs represent mean ± SEM. Supplementary Fig. 6 shows raw data. e Mass cytometry analysis showing the relative expression of different markers on neo-epitope specific PBLs and TILs. Naive and memory bulk peripheral blood CD8⁺ T cells are shown for comparison. f Analysis of the repertoire of the T-cell receptor β (TCRβ) V-J segment recombination of HHAT_L75F-specific CD8⁺ T cells isolated from either PBLs (top) or TILs (bottom) after FACS sorting using multimers. V and J segments are represented according to chromosomal location on the x and y-axis, respectively. Arrows identify dominant TCRβ sequences that were associated to TCRα sequences shown in Supplementary Fig. 8. g Calculated TCR/pMHC complexes for PBL and TIL-related TCRs: HHAT_L75F-PBL-hTRAV05 + hTRAJ34/hTRBV11-2 + hTRBJ02-7 on the top, and HHAT_L75F-TIL-hTRAV38-2 + hTRAJ33/hTRBV12-3 + hTRBJ02-3 at the bottom. The peptide is shown in ball and stick, colored according to the atom types. MHC ribbon is colored in brown, with residues displayed in thick lines and colored according to the atom types, with carbon colored in brown. TCRα ribbon is colored in light blue, with residues displayed in thick lines and colored according to the atom types, with carbon colored in light blue. TCRβ ribbon is colored in pink, with residues displayed in thick lines and colored according to the atom types, with carbon colored in pink. Hydrogen bonds and ionic interactions are shown as thin blue lines. Dotted blue lines indicate hydrogen bonds accessible through thermal fluctuations. Residues are labeled in brown, black, blue, and pink, for MHC, peptide, TCRα, and TCRβ, respectively

Fig. 5

Functional analyses of neo-epitope specific PBLs and TILs. a Representative analyses of the functional avidity of neo-epitope specific PBLs and TILs as assessed by IFNγ ELISpot. HS6ST1_S405I-specific PBLs from patient CTE-0005 and ZCCHC11_P1265H-specific TILs from patient CTE-0015 were stimulated with serial dilutions of cognate neo-epitopes or native (wt) peptides. Data show mean ± SEM of technical replicates. b Cumulative analysis (individual values and median) of the functional avidity of neo-epitope specific CD8⁺ PBLs and TILs (raw data are shown in Supplementary Fig. 10). c Analysis of ODZ3_A2490V-specific (top, high avidity) and SEPT9_R289H-specific (bottom, low-avidity) TCRs (red bars and arrows) among total TCRs from the autologous tumor samples of patients CTE-0015 and CTE-0011, respectively. Elements of box plots represent median (line), 25% and 75% confidence limit (box limits) and 10% and 90% confidence limit (whiskers)