Reversion analysis reveals the in vivo immunogenicity of a poorly MHC I-binding cancer neoepitope

Abstract

High-affinity MHC I-peptide interactions are considered essential for immunogenicity. However, some neo-epitopes with low affinity for MHC I have been reported to elicit CD8 T cell dependent tumor rejection in immunization-challenge studies. Here we show in a mouse model that a neo-epitope that poorly binds to MHC I is able to enhance the immunogenicity of a tumor in the absence of immunization. Fibrosarcoma cells with a naturally occurring mutation are edited to their wild type counterpart; the mutation is then re-introduced in order to obtain a cell line that is genetically identical to the wild type except for the neo-epitope-encoding mutation. Upon transplantation into syngeneic mice, all three cell lines form tumors that are infiltrated with activated T cells. However, lymphocytes from the two tumors that harbor the mutation show significantly stronger transcriptional signatures of cytotoxicity and TCR engagement, and induce greater breadth of TCR reactivity than those of the wild type tumors. Structural modeling of the neo-epitope peptide/MHC I pairs suggests increased hydrophobicity of the neo-epitope surface, consistent with higher TCR reactivity. These results confirm the in vivo immunogenicity of low affinity or 'non-binding' epitopes that do not follow the canonical concept of MHC I-peptide recognition.

Fig. 1. Definition of the precise neoepitope of Ccdc85c^MUT that mediates tumor rejection.

a The sequences of the 18-mer wild type and mutant peptides derived from Ccdc85c gene as well as their corresponding allelic fractions (the number of mutant/normal reads divided by the total number of reads (coverage) at a specific genomic position) are shown. b The top panel shows a schematic diagram of immunization and tumor challenge in BALB/cJ mice. The bottom panel (left) shows tumor growth in BALB/cJ mice immunized with Ccdc85c^MUT or Ccdc85c^WT and challenged with Meth A as described in Methods. Each line represents tumor growth in a single mouse (n = 5 mice per group). AUC for each group is plotted in the panel on the right. Data are presented as mean ± SD. P values were calculated using 1-way ANOVA test adjusted for multiple comparisons (Tukey’s multiple comparison test). c Several truncated versions of the 18-mer Ccdc85c^MUT peptide were tested in tumor rejection assay. BALB/cJ mice were immunized and tumor challenged. Each line represents tumor growth in a single mouse. Although mice were immunized with individual peptides, the data for multiple peptides are grouped into one with the composition of the peptides shown on the right. The tumor rejection data for individual peptides are shown in Supplementary Fig. 1. Tumor rejection score (TRS) for each neoepitope is shown in the yellow box, where five represents a complete tumor protection and zero means no tumor rejection. d On the left panel, total Area Under the Curve (AUC) scores for each group in B are plotted. Each bar shows the average total AUC score for the indicated group (TRS = 5; n = 35, TRS = 4–4.5; n = 40, TRS = 3–3.3; n = 25, TRS = 2; n = 30, TRS = 0.1–1.5; n = 60). Error bars represent standard deviation (SD). The P values corresponding to the comparison of TRS = 0 with TRSs 5.0, 4.0–4.5 and 3.0–3.3 were respectively <0.0001, <0.0001 and 0.0002. P values were calculated using 1-way ANOVA test adjusted for multiple comparisons. On the right, peptides with the highest and the lowest TRS are shown. e Targeted MS-based detection of TYIRPFETKVK and YIRPFETKVK among MHC I peptides eluted from BMDCs pulsed with the 18-mer Ccdc85c^MUT. Heavy labeled synthetic peptides were spiked into the peptide samples; the labeled amino acid is marked with a bold character and the mutation is in red. Matched peak lists for the “heavy” and “light” ions were extracted and monitored, while only single charge y ions were plotted. See Methods for details. f Predicted (by NetMHC4.0) and measured IC50 values of the binding of candidate precise neoepitopes of Ccdc85c^MUT to K^d are shown. The candidate neoepitopes include those defined by tumor rejection and MS as in panels c and e. The other three candidate neoepitopes were predicted by NetMHC4.0 alone and were not active in tumor rejection. The affinities of the MS/TRS predicted neoepitopes were also measured for D^d and L^d; measured affinities were below the level of detection.

Fig. 2. Immune response in mice immunized with Ccdc85c^MUT and control neoepitope Alms1.1^MUT.

a Tumors from mice immunized with Ccdc85c^MUT or Alms1.1^MUT as well as two groups of control mice immunized with BMDCs were harvested 10 days after tumor challenge. Tumor infiltrating CD45 + cells were sorted and analyzed by scRNA seq as described in Methods and Supplementary Fig 9b. Combined scRNA seq data from four libraries were analyzed. The t-SNE plots of all intra-tumoral immune cells as well as myeloid and lymphoid compartments are shown in the top left panel. Myeloid and lymphoid composition for each library is shown on the right panel. The myeloid and lymphoid sub-clusters are shown in the bottom panels. See text for definition of each sub-cluster. b Heat maps of the indicated genes in the myeloid and lymphoid compartments are shown. The heat map genes were selected from the list of differentially expressed genes, genes with a high average TF-IDF score, and typical cell type markers. The average gene expression for each cluster was normalized by dividing it from the maximum value of each row (gene). c Summary heat map of selected genes associated with cytotoxicity for pooled activated T cells is shown. The heat map illustrates scaled (Z-scored) average gene expression by library.

Fig. 3. Intratumoral CD8 T cell response in mice challenged with the original and CRISPR edited Meth A cell lines.

Tumors from mice challenged with MUT1, REV, and MUT2 Meth A cell lines were harvested ten days after the tumor challenge. Tumor infiltrating CD45 + cells were sorted (Supplementary Fig. 9b) and analyzed by scRNA sequencing, as described in Methods. Combined scRNA sequencing data from the three resulting libraries were analyzed. a Top panel shows a simple hierarchical clustering of MUT1, REV and MUT2 libraries, based on the normalized average expression vector of top ~1,500 informative genes (selected by the highest average TF-IDF score) where, the Y value reflects the distance between clusters. The bottom panel represents the violin heat map plots of the top average TF-IDF scoring gene expression for the three libraries (expression of 26 genes is shown). Significant difference in distance of the REV versus MUT1 and MUT2 in the hierarchy is indicated by asterisk (please refer to Sup Fig. 6 for more details). b Top panel depicts the expression percentage of the genes involved in CD8 T cell activation of CD8 T cells that are derived from MUT1, MUT2, and REV libraries. Bottom panel shows the violin plots for the expression of genes involved in cytotoxicity, early response and other effector functions of CD8 T cells that are derived from the three libraries. c Top panel illustrates clone networks resulting from applying GLIPH-algorithm to the mixed pool of T cells TCR sequencing data, using igraph R package. Each node, represented by a circle, is a TCR clone. The diameter of a node is representative of the number of cells with the same TCR. Existence of a link between two nodes indicates global or local similarity between the two nodes as defined by the GLIPH algorithm. Further, a dense cluster in the network, characterized by high number of connections within a cluster and a low number of connections to neighboring clusters, suggest higher similarity and hence higher specificity within the cluster. A large number of dense clusters might suggest higher diversity in the network. Bottom panel represents the violin plots for the expression of genes involved in cytotoxicity, early response and other effector functions of the top 10 clonally expanded CD8 T cells that are derived from each library.

Fig. 4. Models of peptide/MHC I complexes indicate structural and physical correlates with immunogenicity.

a For the tumor rejecting TYIRPFETKVK neoepitope, the leucine-to-phenylalanine substitution at position 6 is predicted to increase hydrophobic solvent accessibility by 17 Ų, with the aromatic phenylalanine ring partially exposed for interactions with T cell receptors. An overlay of the neoepitope and its wild type counterpart demonstrates the substantial differences between the wild type peptide and neoepitope. b For the tumor rejecting YIRPFETKVK neoepitope, with the leucine-to-phenylalanine substitution now at position 5, the modeling predicts structural alterations in exposed side chains in response to the mutation, as well as a reduction in exposed hydrophobic solvent accessible surface area of 23 Ų.