T cell receptor and antibody specifically recognizing frameshift neoantigen

Abstract

According to the present invention, neoantigen (shared frameshift neoantigen) which is commonly observed in a plurality of patients and can be targeted by cancer immunotherapies was extracted, a peptide that can be presented by HLA was selected, a neoantigen-specific T cell was induced, and the TCR sequence was identified. Furthermore, an scFV that binds to a neoantigen sequence presented by HLA was obtained, and a bispecific antibody that recognizes the presented neoantigen and CD3 was produced. This method makes it possible to provide a T cell receptor and a bispecific antibody that target a shared neoantigen. A T cell receptor that targets the shared frameshift neoantigen or a bispecific antibody that recognizes the shared frameshift neoantigen can be utilized in immunotherapies that are applicable to a wide range of patients.

Description

T cell receptors and antibodies that specifically recognize frameshift neoantigens

[0001] This invention relates to cancer immunotherapy, particularly to a T cell receptor and antibody that specifically recognize frameshift neoantigens.

[0002] Cancer treatment has primarily involved surgery, radiation therapy, and chemotherapy. In recent years, immunotherapy, which utilizes the patient's own immune response to attack and treat cancer cells, has become increasingly common. However, the efficacy rate of immune checkpoint inhibitors such as anti-PD-1 antibodies and anti-CTLA-4 antibodies remains at only 10-40%, highlighting the need for the development of new immunotherapies.

[0003] In addition to checkpoint inhibitors, cancer immunotherapy involves many treatments related to the immune system, including treatment with small molecule compounds such as IDO (indoleamine 2,3-dioxygenase) inhibitors that block immunosuppression by cancer cells, cancer vaccines that make the immune system recognize antigens specific to cancer cells, such as peptide vaccines and RNA / DNA vaccines, and T cell therapy that uses T cells that act specifically on cancer, such as tumor-infiltrating T cells and CAR-T cells. Among these, neoantigens, which are antigens that arise from somatic mutations in cancer, are attracting attention as therapeutic targets because they originate from somatic mutations in cancer cells, are presented on HLA molecules on cancer cells, and are recognized by T cell receptors (TCRs) expressed on the surface of T cells.

[0004] Neoantigens are considered promising targets for immunotherapy because they are thought to have fewer side effects due to their high specificity for cancer cells, and because they can also be targeted as intracellular proteins due to being presented on the cell surface by HLA. However, immunotherapy against neoantigens requires individualization because it is derived from tumor-specific mutations and the peptides presented differ depending on the patient's HLA type. Therefore, there is a need for the development of immunotherapies using neoantigens that can be applied more broadly. Neoantigens that are commonly found in multiple patients (called shared neoantigens) have been pointed out as having the potential for broader clinical application, and methods for identifying neoantigens are also being developed (Patent Document 1).

[0005] The inventors have identified a shared neoantigen derived from the FGFR Y373C mutation (indicating a mutation where Y (tyrosine) at position 373 of the FGFR gene is replaced by C (cysteine)) found in many bladder cancer patients through a comprehensive screening of shared neoantigens (Non-Patent Literature 1). However, since many bladder cancer patients also have mutations other than the FGFR Y373C mutation, and the genes in which mutations are observed vary depending on the type of cancer, accumulating information on shared neoantigens is essential for the treatment of many cancer patients.

[0006] Frameshift mutations caused by insertions and deletions (collectively referred to as indels) have been suggested to potentially generate highly immunogenic neoantigens (called frameshift neoantigens) (Non-patent documents 2 and 3). Indels create new open reading frames (ORFs), resulting in unique protein sequences downstream of the mutation site. As a result, novel peptides not encoded in the normal genome are generated. Even if the indel site is located in a different position, partially identical peptides may be generated within the new ORF, and such frameshift mutation clusters (FSCs) are highly likely to generate shared neoantigens. However, frameshift neoantigens derived from indels have been largely unanalyzed.

[0007] Special Publication No. 2020-532323

[0008] Tate, T. et al.,Cancer, 2023, 15(4):1031. doi: 10.3390 / cancers15041031Turajlic S, etal., Lancet Oncol 2017; 18:1009-21Sena LA, et al.Oncologist 2021; 26:e270-e278.Zhang Z, et al., Nat Commun,2021; 12:1226.Padovan E, et al.,Science 1993; 262:422-4.Skora AD, et al., Proc Natl AcadSci USA 2015; 112:9967-72.Carter P, et al. Proc NatlAcad Sci USA 1992; 89:4285-9.Fellouse FA, etal. JMol Biol 2007; 373:924-40.Douglass J, et al. Sci Immunol 2021; 6(57):eabd5515. doi: 10.1126 / sciimmunol.abd5515.Hsiue EH, et al. Science 2021; 371(6533):eabc8697. doi:10.1126 / science.abc8697

[0009] The application of neoantigens to drug discovery has been attempted for some time. However, a method for efficiently extracting shared neoantigens common to multiple patients and producing shared neoantigen-specific TCR-T cells and bispecific antibodies targeting neoantigens has yet to be developed. This invention relates to a method for extracting shared frameshift neoantigens and producing TCR-T cells and bispecific antibodies. In particular, the objective is to provide TCR-T cells and bispecific antibodies against neoantigens caused by APC (adenomatous polyposis coli) gene mutations, which are frequently detected in colorectal cancer patients.

[0010] This invention relates to a method for extracting the following shared frameshift neoantigens, producing shared frameshift neoantigen-specific TCR-T cells, bispecific antibodies targeting frameshift neoantigens, and TCR-T cells and bispecific antibodies against frameshift neoantigens due to APC gene mutations.

[0011] (1) A T cell receptor that specifically recognizes an APC gene-derived shared neoantigen, or an antibody that specifically recognizes an HLA that presents a shared neoantigen, wherein the shared neoantigen is APC-F2-1472 ＊ , or APC-F3-1512 ＊ A T-cell receptor or antibody characterized by being a neoantigen produced by a cluster. APC-F2-1472 ＊ In 4.9% of colorectal cancer patients, APC-F3-1512 ＊ Since this mutation occurs in 1.5% of cases, T cell receptors and antibodies that specifically recognize these neoantigens could be effective therapeutic agents.

[0012] For example, neoantigen-targeted TCR therapy or bispecific antibody therapy can be performed as follows: (a) A cancer panel or whole genome / exome sequencing can be used to identify APC-F2-1472 ＊ , APC-F3-1512 ＊Screen patients who have the mutation. (b) Determine the patient's HLA type by an existing bead-based method or whole genome / exome sequencing, and confirm that the patient's HLA can present neoantigens. For patients satisfying (a) and (b), in TCR therapy, PBMCs are isolated from the patient's peripheral blood, the TCR sequences shown below are introduced by retrovirus to prepare TCR-T cells, which are expanded in vitro. The expanded patient-derived TCR-T cells are infused back into the patient. In addition, for bispecific antibody (BsABs) therapy, the formulated antibody is administered to patients satisfying (a) and (b). Alternatively, the antibody can be conjugated with a drug or radioisotope before administration. Since it is administration of an antibody preparation, treatment can be started immediately, which is greatly beneficial to the patient. BsAbs therapy may be used in combination with TCR therapy, but can also be used alone. TCR-like CAR-T cells can be prepared based on the obtained sequence information, and cell therapy can also be performed. In particular, APC is often a driver mutation in colorectal cancer, APC-F2-1472 ＊ For patients with a mutation, therapeutic efficacy is expected regardless of the disease stage.

[0013] (2)SETD1B-F2-34 ＊ , RPL22-F2-19 ＊ , NPM1-F3-297 ＊ , XYLT2-F2-606 ＊ , ZBTB20-F2-734 ＊ , JAK1-F2-875 ＊ , RNF43-F2-699 ＊ , ACVR2A-F2-441 ＊ , APC-F2-1506 ＊ , APC-F2-1472 ＊ , SLC3A2-F2-330 ＊ , DOCK3-F2-1896 ＊ , UBR5-F2-2148 ＊ , VHL-F2-158 ＊ , VHL-F3-130 ＊ , GATA3-F3-505 ＊ , CTCF-F3-228 ＊ , SMAD7-F2-323 ＊, MVK-F2-158 ＊ , PLEKHA6-F2-499 ＊ , NFASC-F2-176 ＊ , TP53-F2-344 ＊ , TP53-F2-43 ＊ , TP53-F2-169 ＊ , TP53-F2-246 ＊ , TP53-F2-122 ＊ , TP53-F3-147 ＊ , GATA3-F3-350 ＊ A shared frameshift neoantigen arising from any of the following FSCs. Since the above frameshift neoantigens are present in a relatively large number of patients, they are effective targets for treatment.

[0014] (3) A method for producing a T cell receptor that specifically recognizes a shared neoantigen, comprising: selecting a shared frameshift neoantigen derived from the FSC described in (2) and an HLA type to which the shared frameshift neoantigen can be presented; and using a PBMC having the target HLA type, the shared frameshift neoantigen-responsive CD8 ＋ Induces T cells and shared frameshift neoantigen-responsive CD8 ＋ A method for analyzing the TCR sequence of T cells and constructing a T cell receptor based on the obtained TCR sequence. For each frameshift neoantigen, an HLA type in which presentation of the neoantigen is predicted is selected, and neoantigen-responsive CD8 ＋ By inducing T cells, a sequence of a T cell receptor that recognizes frameshift neoantigens can be obtained. Based on the obtained sequence, it becomes possible to create cells that express a T cell receptor that can be used for therapeutic purposes. This method can produce a shared frameshift neoantigen-responsive T cell receptor regardless of the shared frameshift neoantigen derived from the FSC or any HLA type.

[0015] (4) A method for selecting an scFV (single-chain variable fragment) antibody that specifically recognizes a neoantigen, comprising: selecting a shared frameshift neoantigen generated from the FSC described in (2) and an HLA type on which the shared frameshift neoantigen can be presented; using an scFv-phage display library, causing the target type of HLA to present the shared frameshift neoantigen; and selecting the scFV to which it binds. An antibody that specifically binds to a neoantigen can be a useful therapeutic agent, especially by making it a bispecific antibody with an antibody that binds to CD3. By using this method, it is possible to produce an antibody that binds to any shared frameshift neoantigen generated from the FSC and to any HLA type.

[0016] A schematic diagram illustrating the generation of frameshift and shared frameshift neoantigens by indel. A diagram showing the method of displaying FSCs. FSCs are indicated by the gene name, frame, and the position of the PTC (premature termination codon). HLA-A ＊ Using PBMCs from healthy individuals who tested positive for 24:02, we investigated CD8 reactions to neoantigens generated by TP53 frameshift. ＋ Figure showing the results of analysis by ELISPOT assay to determine whether T cells were induced. PMA / ionomycin was used as a positive control. Figure showing a heatmap of the frequency of FSCs in each cancer type in TCGA (The Cancer Genome Atlas) data. FSCs present in 3% or more cases in at least one cancer type are shown. Cancer type abbreviations are based on the TCGA definition. APC-F2-1472 ＊ , and APC-F3-1512 ＊ Nucleotide and protein sequences around the cluster. * represents PTC. Refer to frame 1 and define as ORF. APC summarizes the frameshift neoantigens resulting from FSCs. Shared: Peptides that may be processed from two clusters identified as frameshift neoantigens, and HLA-A ＊This summarizes the coupling predictions for 24:02. HLA-A ＊ Using PBMCs from healthy individuals who tested positive for 24:02, we investigated CD8 reactions to neoantigens generated by APC frameshift. ＋ Figure showing the results of analysis of T cell induction using the ELISPOT assay. PMA / ionomycin is a positive control. Figure showing the results of PBMC stimulation with peptide 1472LP or 1512LP presented by autologous dendritic cells (DCs), and analysis of reactivity using the ELISPOT assay. APC neoantigen-responsive CD8 ＋ HLA-A cells were proliferated and loaded with peptide 1472SP2 or 1512SP3. ＊ Flow cytometry analysis of interactions with the 24:02 tetramer. Figures showing pie charts of TCRα and TCRβ frequencies and TCR sequences for APC frameshift neoantigen-responsive T cell lines 1-4. Figure showing the results of analyzing concentration-dependent IFN-γ secretion by ELISPOT assay after stimulating TCR-T cells expressing 1472TCR-1 or 1472TCR-3 with C1R-A24 cells loaded with 1472LP peptide. Figure showing the results of analyzing concentration-dependent IFN-γ secretion by ELISPOT assay after stimulating TCR-T cells expressing 1512TCR-1 or 1512TCR-3 with C1R-A24 cells loaded with 1512LP peptide. Figure showing the results of analyzing the HLA-restrictiveness of 1472LP peptide using C1R cells expressing various HLA-A strains. This figure shows the results of analyzing the HLA-restrictiveness of the 1512LP peptide using C1R cells expressing various HLA-A strains. It also shows the HLA-A strains recognized by T cells expressing 1472TCR-1 and 1472TCR-3. ＊ Figure 24:02 showing the results of analysis of the core peptide presented on the molecule. HLA-A recognized by T cells expressing 1512TCR-1 and 1512TCR-3. ＊Figure showing the results of analysis of the core peptide presented on the 24:02 molecule. Figure showing the cytotoxic activity of four T cell lines. Analysis is performed by real-time impedance measurement. 293-A24-APCTMG target cells and each cell type are co-cultured so that the effector-to-target ratio is 10:1. Impedance is measured by taking nCI (normalized cell index) at 15-minute intervals (left), and cytotoxicity is calculated based on the difference in nCI from the control (right). Figure schematically showing a method for isolating scFv that can specifically bind to neoantigen peptides using an scFv-phage display library. Screening of 1472SP2-HLA-A24 complex-specific phage clones by ELISA. 1472SP2-HLA-A24, 1512SP3-HLA-A24 loaded with the unrelated peptide 1512SP3, or an empty HLA-A24 monomer were bound to a plate, incubated with the supernatant of the phage clone, and then detected with anti-M13 antibody. The absorbance of the empty HLA-A24 monomer was set to 1.0. Figure shows the scFv sequence obtained from the phage clone specific to the obtained 1472SP2-HLA-A24 monomer. Figure shows the results of Western blotting analysis of the purified single-stranded variable fragment (scFv). Figure shows the results of ELISA analysis of the binding of 1472SP2scFv to 1472SP2-HLA-A24 monomer. Figure schematically shows the structure of the prepared bispecific antibody (BsAb). Figure 1: Calculation of the binding of 1472SP2 BsAb to 1472SP2-HLA-A24 (left) and CD3 (right) by ELISA. Figure 2: Calculation of the binding of 1472SP2 BsAb to CD8 by ELISA. ＋ Effects on T cell cytotoxic activity. The figure below shows the results of cross-reactivity analysis using the alanine-substituted 1472SP2 mutant peptide (top). The binding affinity of the variant peptide was predicted using netMHCv4.0, and related peptides were searched using BLAST, UniProtKB, and IEDB.

[0017] The following specification primarily describes APC-based shared frameshift neoantigens, neoantigen-specific TCR-T cells, and bispecific antibodies. However, it is not limited to APCs; similar extraction and production of neoantigen-specific TCR-T cells and bispecific antibodies can be achieved for any molecular shared frameshift neoantigen.

[0018] Furthermore, here we will discuss HLA-A, which is a major HLA type among Japanese people. ＊ While this study targets shared frameshift neoantigens presented at 24:02, it goes without saying that shared frameshift neoantigens presented by other HLA types can also be selected in the same way to produce neoantigen-specific TCR-T cells, bispecific antibodies, and the like.

[0019] The following describes methods for extracting shared frameshift neoantigens, producing neoantigen-specific TCR-T cells, and creating bispecific antibodies, using data as a guide. However, the methods presented here are just examples, and other methods commonly used in this field can be substituted.

[0020] [FSC Extraction] To screen for frameshift neoantigens shared among multiple cancer patients, somatic mutation data (MAF files) analyzed using Mutect2 for TCGA (The Cancer Genome Atlas) from 10,182 samples across 33 cancer types were downloaded from the NCI's Genomic Data Commons (GDC) (Non-Patent Literature 4). For indels generating 121,042 frameshifts in the coding region, downstream frameshift sequences were obtained based on Ensembl transcription data (Release 79) and translated into peptide sequences up to the position of immature stop codons (PTCs). As schematically shown in Figure 1A, common frameshift neoantigens may be generated even indels at different positions. Therefore, FSC analysis has the potential to extract shared frameshift neoantigens common to multiple patients, making it possible to select neoantigens that can be targeted for treatment.

[0021] Based on the common peptide sequences of the extracted frameshift indels, a total of 80,698 FSCs were classified. Further, the selection was narrowed down to 729 FSCs found in recurrent cancers with a frequency of 1.0% or more in specific cancer types and found in at least three patients.

[0022] Overall, TP53-F2-344 ＊ This was the most common, appearing in 20 different cancer types and accounting for approximately 1.0% (0.20%–3.5%) of all cases. Here, FSC neoantigens are labeled and displayed based on their gene (protein) name (TP53), shifted frame (F2), and PTC location (344) (Figure 1B). Indels that share an ORF and cause frameshifts terminating at the same PTC location are grouped as FSCs. Frame 1 is defined as the reference ORF, and frames 2 and 3 are defined as frames with 3n+2bp insertions or 3n+1bp deletions, and frames with 3n+1bp insertions or 3n+2bp deletions, respectively.

[0023] TP53-F2-344 was the most common type of cancer. ＊ Regarding this, we analyzed whether neoantigens could be antigen-presented and serve as therapeutic targets. TP53-F2-344 ＊ The peptide produced as a neoantigen is HLA-A, which is a common HLA type among Japanese people. ＊ Since binding was predicted at 24:02, this peptide is HLA-A ＊ It was demonstrated that it would be presented at 24:02.

[0024] HLA-A ＊ Using peripheral blood mononuclear cells isolated from healthy donors carrying the 24:02 gene (Cellular Technology Ltd.), shared frameshift neoantigen-responsive CD8 ＋ T cell induction was performed as follows: Using the Dynabeads CD8 Positive Isolation Kit (ThermoFisher Scientific), CD8 cells were induced. ＋ T cells and CD8 －T cells were isolated. Monocyte-derived dendritic cells were treated with a plastic bonding method to isolate CD8 cells. － T cells were isolated. Immature dendritic cells were matured by adding OK-432 (Chugai Pharmaceutical) at a final concentration of 0.1 KE / mL. Mature DCs were harvested on day 7, and 20 μg / mL of TP53 neoantigen long peptide (TP53-F2-344LP, SEQ ID NO: 74) was added and left to stand overnight. For T cell activation, CD8 cells were incubated in AIM-V medium supplemented with 5% human type AB serum, 10 ng / mL IL-7 (R&D Systems), and 24 IU / mL IL-2 (R&D Systems) for 7 days. ＋ T cells were co-cultured with dendritic cells (DCs) loaded with peptides.

[0025] TP53-F2-344 ＊ PBMCs were stimulated with the cluster-derived peptide TP53-F2-344LP, and the presence of IFN-γ-secreting cells was analyzed using the ELISPOT assay (Human IFN-γ ELISPOT PRO kit (Mabtech)) (Figure 1C). In the figure, the CMV peptide is HLA-A ＊ This is a known cytomegalovirus peptide presented at 24:02 (SEQ ID NO: 75). Compared to a control without the peptide, cells with the peptide added showed a greater number of cells secreting IFN-γ, and TP53-F2-344 ＊ Shared frameshift neoantigens were suggested to be a potential treatment option.

[0026] TP53-F2-344 ＊ Following the cluster, the cluster that showed a high frequency across all cases was RPL22-F2-19. ＊ 0.91% of RNF43-F2-699 ＊ 0.87% of SETD1B-F2-34 ＊ The percentage was 0.82%. In the analysis of the most frequent FSCs in each cancer type, SETD1B-F2-34 ＊ This was particularly pronounced in endometrial cancer (UCEC), accounting for 9.1% of patients, and was also observed in 6.8% of patients with colorectal adenocarcinoma (COAD). (NPM1-F3-297) ＊ Cluster and WT1-F3-383 ＊Clusters were detected in 8.4% and 3.5% of patients with acute myeloid leukemia (LAML), respectively, as ZBTB20-F2-734 ＊ cluster was detected in 7.1% of patients with gastric adenocarcinoma (STAD). In clear cell renal cell carcinoma (KIRC), VHL-F2-158 ＊ cluster and VHL-F3-130 ＊ clusters were detected in 5.1% and 4.2% of patients, respectively.

[0027] FSCs shown in Figure 1D and SETD1B-F2-34, which are considered important as therapeutic targets ＊ , RPL22-F2-19 ＊ , NPM1-F3-297 ＊ , XYLT2-F2-606 ＊ , ZBTB20-F2-734 ＊ , JAK1-F2-875 ＊ , RNF43-F2-699 ＊ , ACVR2A-F2-441 ＊ , APC-F2-1506 ＊ , APC-F2-1472 ＊ , SLC3A2-F2-330 ＊ , DOCK3-F2-1896 ＊ , UBR5-F2-2148 ＊ , VHL-F2-158 ＊ , VHL-F3-130 ＊ , GATA3-F3-505 ＊ , CTCF-F3-228 ＊ , SMAD7-F2-323 ＊ , MVK-F2-158 ＊ , PLEKHA6-F2-499 ＊ , NFASC-F2-176 ＊ , TP53-F2-344 ＊ , TP53-F2-43 ＊ , TP53-F2-169 ＊ , TP53-F2-246 ＊ , TP53-F2-122 ＊ , TP53-F3-147 ＊ , GATA3-F3-350 ＊Shared frameshift neoantigens, which are frequently found in cancer, are useful as targets for cancer treatment. Therefore, T cell receptors and bispecific antibodies produced targeting shared frameshift neoantigens are useful as therapeutic agents.

[0028] [FSCs in the APC gene] FSCs were frequently observed within the APC gene, which is consistent with the high frequency of APC mutations in colorectal cancer. There are eight APC FSCs found in colorectal cancer at a frequency of more than 1% (APC-F2-860). ＊ , APC-F2-1320 ＊ , APC-F2-1414 ＊ , APC-F2-1472 ＊ , APC-F2-1506 ＊ , APC-F3-1512 ＊ , APC-F3-1557 ＊ , APC-F2-1564 ＊ ).

[0029] For the FSCs identified in these APCs, we predicted whether neoantigens would be presented as antigens for nine HLA-A types common in Japanese and Cabians. Table 1 shows the predicted binding with the nine HLA-A types, and Table 2 shows the sequences of the neoantigens. Of the eight FSCs, APC-F2-860 ＊ , APC-F2-1320 ＊ and APC-F3-1557 ＊ HLA-A in the five FSCs, excluding the one mentioned above. ＊ 02:01, HLA-A ＊ 02:06, HLA-A ＊ 03:01, HLA-A ＊ 11:01, HLA-A ＊ 24:02, HLA-A ＊ 26:02, HLA-A ＊ 31:01 and HLA-A ＊ It was predicted that there would be neoantigens that bind to 33:03. Among these neoantigens, APC-F2-1472 ＊ In 4.9% of colorectal cancer patients, APC-F3-1512 ＊Since it is detected in 1.5% of cases, HLA-A is a common HLA type among Japanese people. ＊ 24:02 presents a highly promising shared neoantigen that is likely to be useful in treatment. Figure 2A shows the amino acid sequence around the FSC that occurs, and Figure 2B shows the sequence that occurs as a neoantigen and the processed HLA-A ＊ This shows the peptides that may be presented at 24:02, along with their binding predictions.

[0030] a: The abbreviations for cancer types are based on the definition of TCGA. b: Y is IC ５０ <This shows an HLA type that is predicted to bind to neoantigens at 500 nM.

[0031]

[0032] [Evaluation of Neoantigen Immunogenicity] To evaluate the immunogenicity of APC neoantigens predicted from their sequences, two HLA-A ＊ PBMCs were obtained from a 24:02 positive healthy donor, and T cell induction experiments were performed. HLA-A ＊ Using peripheral blood mononuclear cells isolated from healthy donors carrying the 24:02 gene (Cellular Technology Ltd.), shared frameshift neoantigen-responsive CD8 ＋ T cell induction was performed as follows: Using the Dynabeads CD8 Positive Isolation Kit (ThermoFisher Scientific), CD8 cells were induced. ＋ T cells and CD8 － T cells were isolated. Monocyte-derived dendritic cells were treated with a plastic bonding method to isolate CD8 cells. － T cells were isolated. Immature dendritic cells were matured by adding OK-432 (Chugai Pharmaceutical) at a final concentration of 0.1 KE / mL. Mature DCs were harvested on day 7, and 20 μg / mL of APC neoantigen long peptide (1472LP or 1512LP, see Figure 2B) was added and allowed to stand overnight. For T cell activation, CD8 cells were incubated in AIM-V medium supplemented with 5% human type AB serum, 10 ng / mL IL-7 (R&D Systems), and 24 IU / mL IL-2 (R&D Systems) for 7 days. ＋T cells were co-cultured with dendritic cells (DCs) loaded with peptides.

[0033] APC-F2-1472 ＊ PBMCs were stimulated with the cluster-derived peptide 1472LP, and the ELISPOT assay (Human IFN-γ ELISPOT PRO kit (Mabtech)) was performed to analyze the presence of IFN-γ-secreting cells. Compared to a control group without peptide (No peptide), two different CD8 cells secreting IFN-γ were identified from donor A. ＋ T cell lines (T cell lines 1 and 2) were identified. However, no cells responsive to the 1472LP peptide were detected in donor B (Figures 3A, 3B). Although the data is not shown here, APC-F2-1472 ＊ When using short peptides corresponding to the cluster, T cell line 1 showed reactivity to peptide 1472SP2 but not to 1472SP1.

[0034] APC-F3-1512 ＊ When PBMCs are stimulated with the long peptide 1512LP derived from the cluster, one CD8 molecule is released from donor A. ＋ The T cell line (T cell line 3) also received one CD8 cell from donor B. ＋ T cells (T cell line 4) were obtained as cells that reacted to 1512LP (Figures 3A, 3B). Although data is not shown here, T cell line 3 was APC-F3-1512 ＊ Among the eight peptides corresponding to the cluster, HLA-A ＊ At 24:02, it recognizes the short peptide 1512SP3, which showed the strongest binding affinity.

[0035] [Identification of neoantigen-specific TCRs] These four T cell lines were grown in vitro and loaded with 1472SP2 or 1512SP3 HLA-A ＊T cells that recognize peptides were selected by flow cytometry using a 24:02 tetramer (Figure 3C), and the TCR sequence was determined (Figure 3D). The sequences of TRAV24, TRAJ22, etc., described in the V and J regions are known sequences and are registered in databases. Furthermore, within the limits where peptide recognition does not change, a few amino acid substitutions, deletions, or additions may be made to the sequences disclosed in the specification. For regions other than the CDR, homology of 90% or more, more preferably 95% or more, is sufficient. The frequencies of TCRα and TCRβ in each cell line are shown in the pie chart (Figure 3D, top).

[0036] CD8 selected from four T cell lines ＋ In T cell lines 2 and 3, a pair of TCRα and TCRβ sequences, thought to correspond to antigen-specific TCRα / TCRβ, was detected. On the other hand, two TCRα chronotypes and one TCRβ chronotype were detected in T cell lines 1 and 4. In T cell line 1, the two TCRα chronotypes were detected at frequencies of 54.8% and 40.1%, respectively, and in T cell line 4, at frequencies of 49.3% and 37.8%. Since it has been reported that two different TCRα sequences can be expressed in a single T cell (Non-Patent Literature 5), it is possible that T cell lines 1 and 4 also express two different TCRα sequences. Therefore, using Jurkat76-NFAT-EGFP reporter cells, APC-F2-1472 ＊ and APC-F3-1512 ＊ The function of TCRα that recognizes frameshift neoantigens was investigated. In Jurkat76 cells, which lack endogenous TCR expression, APC-F2-1472 was introduced. ＊ , or APC-F3-1512 ＊ Two sets of TCRα / TCRβ potentially responsive to were introduced to generate T cells. Using these generated T cells, peptide 1472SP2 was loaded into HLA-A cells. ＊ The reactivity to the 24:02 tetramer was analyzed. The results are not shown here, but HLA-A ＊The 24:02 tetramer recognized TCR1472-1 (T1472-1) transdermated cells, but did not recognize 1472TCR-2 (T1472-2) transdermated cells. Therefore, TCRα1-1 and TCRβ1 were identified as APC-F2-1472 ＊ It was shown to be a functional pair that recognizes frameshift neoantigens. Similarly, analysis of HLA-A loaded with peptide 1512SP3 showed ＊ The 24:02 tetramer detected 1512TCR-3 (T1512-3), but not 1512TCR-2 (T1512-2). Therefore, the TCRα4-2 / TCRβ4 heterodimer is APC-F3-1152. ＊ It was shown to function as a TCR against neoantigens.

[0037] [Immunological Activity of Shared Neoantigen-Specific TCR-T Cells] To confirm the immunoreactivity of the four TCRs, TCR-modified T cells were generated using PBMCs isolated from healthy donors, and their immunological activity was analyzed by IFN-γ secretion using the ELISPOT assay. TCR-T cells expressing 1472TCR-1 or 1472TCR-3 were co-cultured with C1R-A24 cells loaded with varying concentrations of 1472LP. 10 －７ Co-culturing with C1R-A24 cells pulsed at peptide concentration M resulted in IFN-γ secretion being observed in all TCR-modified T cells. However, IFN-γ secretion was observed with the unrelated peptide 1512LP (Unrelated). －５ No response was observed under condition M (Figure 4A). Note that C1R cells are B lymphoblast cells lacking endogenous HLA-A and HLA-B expression. Other TCR-T cell lines were analyzed in a similar manner. TCR-T cells expressing 1512TCR-1 or 1512TCR-3 responded in a dose-dependent manner to C1R-A24 cells loaded with 1512LP (Figure 4B).

[0038] Next, C1R cells expressing various HLAs, C1R-A0201 (HLA-A ＊This indicates cells expressing 02:01. The same notation applies below.) Using C1R-A0206, C1R-A1101, C1R-A2402, C1R-A3101, and C1R-A3303, the constraint of these TCRs on HLA was investigated (Figures 4C, 4D). In all cell lines, IFN-γ production was significantly detected when co-cultured with C1R-A2402 cells, indicating that HLA-A ＊ This suggested a T cell response constrained at 24:02.

[0039] To analyze the core peptide sequences of long peptides that can be presented on the HLA-A24:02 molecule, an IFN-γ ELISPOT assay was performed using 1472 or 1512 consecutive short peptides (1472SP or 1512SP, respectively; see Figure 2B). CD8 induced by the previously described peptides... ＋ Consistent with the results using T cells, TCR-T cells expressing 1472TCR-1 and 1472TCR-3 responded to the 10-mer 1472SP2 but not to the 9-mer 1472SP1 (Figure 4E). On the other hand, TCR-T cells expressing 1512TCR-1 or 1512TCR-3 showed different recognition patterns for the 1512SP peptide. TCR-T cells expressing 1512TCR-3 responded to all eight short peptides (1512SP1 to 1512SP8), while TCR-T cells expressing 1512TCR-1 showed responsiveness to six short peptides (1512SP3 to 1512SP8) and produced IFN-γ, but did not respond to 1512SP1 or 1512SP2 (Figure 4F).

[0040] The cytotoxic response of these TCR-T cells to target cells was evaluated by impedance assay. Cytotoxicity was analyzed using the xCelligence RTCA S16 Real-Time Cell Analyzer (Agilent Technologies), and real-time cell volume monitoring was performed using impedance measurements (Cell Index (CI)). The analysis used 293-A24-APCTMG cells. 293-A24-APCTMG cells were obtained by knocking out HLA using the CRISPR-Cas9 system to create 293HLA-KO cells, and then adding HLA-A ＊ 24:02 Transfect with a pCAGGS expression vector encoding cDNA to stably produce HLA-A ＊ We obtained 293-A24 cells expressing 24:02, and then four APC FSCs (APC-F2-1414) ＊ , APC-F2-1472 ＊ , APC-F2-1506 ＊ , APC-F3-1512 ＊ Cells with high expression of HLA-A*24:02 and EGFP were obtained by transfecting a pcDNA3.1 expression vector, which was created by cloning a gene that linked the tandem mini gene of ) and the reporter EGFP cDNA, and then selecting cells with high expression of HLA-A*24:02 and EGFP using a cell sorter. In other words, 293-A24-APCTMG cells are HLA-A ＊ These are cells that express 24:02 and four APC FSCs.

[0041] 293-A24-APCTMG cells were seeded at a density of 25,000 cells / well, and TCR-T cells were added in an effector-to-target ratio of 10:1 when stable impedance indicating target cell adhesion was measured (within approximately 20-25 hours). Impedance was recorded at 15-minute intervals, and CI was normalized (nCI) to the value recorded immediately before T cell addition. The value measured in cell culture medium alone was used as the background impedance. The real-time cytotoxicity rate was calculated according to the following formula: Cytotoxicity (%) = [Control nCI - (Experimental nCI - T cell nCI) / Control nCI] × 100 Here, control nCI refers to the nCI of 293-A24-APCTMG cells without T cells as baseline. Experimental nCI represents the nCI of 293-A24-APCTMG cells co-cultured with T cells, and T cell nCI represents the impedance contribution by T cells alone.

[0042] The cytotoxic activity of these four TCR-T cells was analyzed using 293-A24-APCTMG cells as a target. Although the cytotoxic activity differed, all four TCR-T cells were shown to have significant cytotoxic activity against the target cells (Figure 5). 1512TCR-3 and 1472TCR-3 showed the strongest cytotoxicity against target cells expressing the corresponding neoantigen, killing all target cells within 15-20 hours. On the other hand, untransformed T cells (Mock) showed almost no cytotoxic activity against 293-A24-APCTMG cells. This indicates that the cytotoxic activity is specific to cells that present neoantigens.

[0043] [Production of Bispecific Antibodies] Next, we will explain the production of bispecific antibodies that target neoantigens. Here, we will use APC-F2-1472 ＊This section describes the production of an antibody that recognizes 1472SP2 (SEQ ID NO: 25), a neoantigen derived from [unspecified substance]. As mentioned above, 1472SP2 is a frameshift neoantigen of FSC found in 3.0% of colorectal cancer patients. Bispecific antibodies can also be produced against neoantigens of other APCs, or neoantigens of molecules other than APCs, using the same method as described below.

[0044] We selected phages that display specific single-stranded variable fragments (scFv) for the 1472SP2 peptide-HLA-A24 monomer. Based on the humanized 4D5 framework, we attempted to isolate scFv that can specifically bind to the 1472SP2 neoantigen peptide on the HLA-A24 molecule using two types of scFv-phage display libraries, in which mutations were introduced into four CDR3 (L3, H1, H2, H3) or five CDR3 (L2, L3, H1, H2, H3) respectively (Non-patent Literature 6-8, Figure 6A). Using three different amplification stages—enrichment stage (round 1), competition stage (rounds 2-4), and final selection stage (rounds 5 and 6)—we performed more than six panning cycles to gradually enrich the neoantigen-HLA-A24 specific phages.

[0045] After six panning cycles, a total of 63 phage clones were selected, and specific binding to the 1472SP2-HLA-A24 monomer was evaluated by ELISA. For ELISA, 50 ng of biotinylated peptide-HLA-A24 (as a peptide, 1472SP2 or a peptide unrelated to APC) or an empty HLA-A24 monomer was added to 50 μL of blocking buffer consisting of PBS containing 0.5% BSA, 2 mM EDTA, and 0.1% sodium azide in a streptavidin-coated 96-well plate (ThermoFisher Scientific), and the plates were left standing overnight at 4°C. After six washes with 1×TBST (TBS, 0.05% Tween-20), the bound phages were detected with anti-M13 antibody (Sino Biological). Detection was performed by measuring the absorbance at 450 nm using a Victor Multilabel Plate Reader (Perkin Elmer).

[0046] Of the 63 clones examined, three phage clones, clones 20, 44, and 48, bound only to 1472SP2-HLA-A24 and did not bind to empty HLA-A24 or 1512SP3-HLA-A24 monomer (Unrelated HLA-A24) (Figure 6B). Next, the phage DNA was PCR-amplified using primers adjacent to the CDR, and the base sequence was determined to identify two scFv sequences: scFv#1 from clone 20 (VL: SEQ ID NO: 63, VH: SEQ ID NO: 64) and scFv#2 from clones 44 and 48 (VL: SEQ ID NO: 65, VH: SEQ ID NO: 66) (Figure 6C). With respect to the CDR region, the scFv sequences may have a few amino acid substitutions, deletions, or additions to the sequences disclosed in the specification, as long as the binding affinity does not change. For regions other than the CDR, a homology of 90% or more, more preferably 95% or more, is sufficient. In Figure 6C, the numbers in the lower row of the amino acid sequence of each CDR indicate the SEQ ID NO.

[0047] To produce scFv that is not bound to the phage protein M13pIII, ssDNA was purified from phages, the scFv region was amplified by PCR, and then cloned into a pET-44a(+) vector (Novagen). Western blot analysis of the purified scFv revealed a protein band of approximately 80 kDa corresponding to the scFv (Figure 7A). ELISA analysis showed that all scFvs were reactive to 1472SP2-HLA-A24, but did not bind to the 1512SP3-HLA-A24 monomer carrying a peptide unrelated to APC at concentrations up to 200 nM (Figure 7B). A bispecific antibody was constructed using scFv#2, which showed a stronger signal at low concentrations.

[0048] For the construction of the bispecific antibody, we decided to use a single-chain diversity (scDb) here (Non-Patent Documents 8-10). However, it goes without saying that known forms of antibodies such as BiTE, DART, IgG-scFv, and CrossMab can also be constructed. The anti-1472SP2 scDb consists of two different single-chain variable fragment (scFv) fragments. One is configured to target the 1472SP2 neoantigen / HLA-A24 complex, and the other is configured to recruit and activate T cells by binding to the CD3 protein. A short glycerol-rich linker was used to link the heavy chain (VH) and light chain (VL) domains of the variable region of the scFv (Figure 8A). The specific composition is a 1472SP2 bispecific antibody (BsAb) to which anti-1472SP2-HLA-A24 VL, GGGGS (SEQ ID NO: 67) short linker, VH (SEQ ID NO: 69) of anti-CD3 (clone UCTH1, VL: SEQ ID NO: 68, VH: SEQ ID NO: 69, Non-Patent Literature 9), a long linker (GGGGS) 3 in which three GGGGS are linked in tandem, anti-CD3 VL (SEQ ID NO: 68), a GGGGS short linker, and anti-1472SP2-HLA-A24 VH is bound. This fragment was cloned into a pcDNA3.1 expression vector (ThermoFisher Scientific) containing a 6×His tag to produce the anti-1472SP2 bispecific antibody (BsAb). In this study, the CD3 binding region is constructed using the sequence of cloned UCTH1, but it is also possible to substitute it with the binding region of a known CD3 antibody. Furthermore, since the CDR3 sequence of the L chain of the region that specifically binds to CD3ε / δ in UCTH1 is GNTLPW (SEQ ID NO: 70), the CDR1 sequence of the H chain is SGYSFTGY (SEQ ID NO: 71), the CDR2 sequence is LEWMGLINPYKGVSTYN (SEQ ID NO: 72), and the CDR3 sequence is GYYGDSDWYFD (SEQ ID NO: 73), any antibody having at least these CDR sequences can be used.

[0049] When the binding ability of 1472SP2 BsAb to the 1472SP2-HLA-A24 monomer was evaluated by ELISA, 1472SP2 BsAb showed a concentration-dependent, specific interaction with 1472SP2-HLA-A24 up to a concentration of 100 nM, but no interaction with an empty HLA-A24 monomer was observed (Figure 8B left). Furthermore, it was confirmed that 1472SP2 BsAb binds to the CD3ε / δ complex in a concentration-dependent manner (Figure 8B right). These results indicate that 1472SP2 BsAb functions as a bispecific antibody that recognizes both 1472SP2 and CD3 presented on HLA-A24.

[0050] [Immunological function of the bispecific antibody 1472SP2 BsAB] We analyzed whether 1472SP2 BsAb specifically recognizes the 1472SP2 peptide / HLA-A24 complex on the cell surface. To confirm whether the expected neoantigen sequence is processed in the cell, we added 1472SP2 long peptide (1472LP) to C1R-A24 cells modified to express HLA-A24, and cells without the addition, and measured CD8 in the presence of 1472SP2 BsAb or isotype control BsAb. ＋ T cells were co-cultured, and T cell activation was evaluated by the ELISPOT assay (Figure 9A). 1472SP2 BsAb selectively induced IFN-γ secretion in a concentration-dependent manner in C1R-A24 cells pulsed with the peptide. However, the isotype control BsAb did not induce IFN-γ secretion even at a high concentration of 5 nM. Therefore, it was shown that 1472SP2 BsAb recognizes 1472SP2-HLA-A24 cells in which endogenous antigens are processed and 1472SP2 is presented on the cell surface.

[0051] To confirm the cytotoxicity of BsAb against target cells, C1R-A24 cells and CD8 cells were treated with or without the 1472SP2 long peptide. ＋T cells were co-cultured with varying concentrations of 1472SP2 BsAb or isotype control antibody, and cytotoxicity assays were performed (Figure 9B). Similar to the results of the ELISPOT assay, cytotoxic activity was observed even when 1472SP2 BsAb was added at a low concentration of 1 nM, but no cytotoxic activity was observed when using the isotype control antibody, even when added at a concentration of 10 nM.

[0052] [Potential Cross-Reactivity with HLA-A24 Binding Peptides] To evaluate potential cross-reactivity, mutant peptides (1472SP2-1A to 10A; SEQ ID NOs. 53-62, Figure 10 top) were created by substituting alanine amino acids at each position of the 1472SP2 peptide, and an alanine scan was performed. When the binding of these peptides to HLA-A24:02 was predicted using NetMHC v4.0, binding was not predicted except for 1472SP2-2A (predicted IC). ５０ (<500 nM). The immunoreactivity of 1472SP2 BsAb against these mutant peptides was tested using the ELISPOT assay with C1R-A24 cells (Figure 10, bottom). As a result, two peptides in which the 1st or 4th position of 1472SP2 BsAb was substituted with alanine (1472SP2-1A and -4A) showed IFN-γ secretion similar to that of the original 1472SP2 peptide. To further investigate the possibility of these two peptides existing in nature, sequences were searched in the BLAST, UniProtKB human protein database, and the Immunoepitope Database (IEDB) HLA-peptide database. However, despite performing both exact match searches and 90% BLAST searches, neither of these two peptides were found to exist in the human proteome. Therefore, it was suggested that they are not naturally presented by HLA-A24 and that cross-reactivity is unlikely (Figure 10, top).

[0053] In summary, using APC as an example, we predict shared frameshift neoantigens by extracting FSCs, select target HLAs, and perform neoantigen-reactive CD. ＋We have demonstrated a method for obtaining neoantigen-specific TCRs and antibodies by selecting and analyzing eight cells. As mentioned above, this method can be applied not only to APCs but also to neoantigens generated by frameshift in other genes.

Claims

1. A T cell receptor that specifically recognizes a shared neoantigen derived from the APC gene, or an antibody that specifically recognizes an HLA that presents a shared neoantigen, wherein the shared neoantigen is APC-F2-1472 ＊ , or APC-F3-1512 ＊ A T-cell receptor or antibody characterized by being a neoantigen produced by a cluster.

2. The T cell receptor or antibody according to claim 1, wherein the shared neoantigen is produced by an indel mutation.

3. The shared neoantigen is HLA-A ＊ The T cell receptor or antibody according to claim 1, as presented by 24:

02.

4. The shared neoantigen is APC-F2-1472 ＊ The T cell receptor according to claim 3, wherein the CDR3 sequence of the α chain of the T cell receptor is CACNSGSARQLTF (SEQ ID NO: 35) and the CDR3 sequence of the β chain is CASSFSFYNQPQHF (SEQ ID NO: 36), or the CDR3 sequence of the α chain of the T cell receptor is CAATNFGNEKLTF (SEQ ID NO: 38) and the CDR3 sequence of the β chain is CASSQDQVLEGFNSPLHF (SEQ ID NO: 39).

5. The shared neoantigen is APC-F3-1512 ＊ The T cell receptor according to claim 3, wherein the CDR3 sequence of the α chain of the T cell receptor is CASEFQGAQKLVF (SEQ ID NO: 40) and the CDR3 sequence of the β chain is CASSYSLQGPSYEQYF (SEQ ID NO: 41), or the CDR3 sequence of the α chain of the T cell receptor is CAMREGPDTDKLIF (SEQ ID NO: 44) and the CDR3 sequence of the β chain is CASSSAYRVTSGRAVGGEFF (SEQ ID NO: 43).

6. Shared Neoantigen APC-F2-1472 ＊ The antibody according to claim 3, wherein the CDR3 sequence of the L chain is QQYSSSPLT (SEQ ID NO: 45), the CDR1 sequence of the H chain is GFNIYAYKM (SEQ ID NO: 46), the CDR2 sequence is SVYPDDYNTY (SEQ ID NO: 47), and the CDR3 sequence is RDWYWGYAFDY (SEQ ID NO: 48), or the CDR3 sequence of the L chain is QQYFYQPV (SEQ ID NO: 49), the CDR1 sequence of the H chain is GFNVTYQGM (SEQ ID NO: 50), the CDR2 sequence is QIFDYHDYTY (SEQ ID NO: 51), and the CDR3 sequence is RAWSYYYFDY (SEQ ID NO: 52).

7. The antibody according to claim 6, wherein the antibody is a bispecific antibody and further specifically binds to CD3ε / δ.

8. The antibody according to claim 7, wherein the CDR3 sequence of the L chain in the region that specifically binds to the CD3ε / δ is GNTLPW (SEQ ID NO: 70), the CDR1 sequence of the H chain is SGYSFTGY (SEQ ID NO: 71), the CDR2 sequence is LEWMGLINPYKGVSTYN (SEQ ID NO: 72), and the CDR3 sequence is GYYGDSDWYFD (SEQ ID NO: 73).

9. SETD1B-F2-34 ＊ , RPL22-F2-19 ＊ , NPM1-F3-297 ＊ , XYL T2-F2-606 ＊ , ZBTB20-F2-734 ＊ , JAK1-F2-875 ＊ , RNF43-F2-699 ＊ , ACVR2A-F2-441 ＊ , APC-F2-1506 ＊ , APC-F2-1472 ＊ , SLC3A2-F2-330 ＊ , DOCK3-F2-1896 ＊ , UBR5-F2-2148 ＊ , VHL-F2-158 ＊ , VHL-F3-130 ＊ , GATA3-F3-505 ＊ , CTCF-F3-228 ＊ , SMAD7-F2-323 ＊ , MVK-F2-158 ＊ , PLEKHA6-F2-499 ＊ , NFASC-F2-176 ＊ , TP53-F2-344 ＊ , TP53-F2-43 ＊ , TP53-F2-169 ＊ , TP53-F2-246 ＊ , TP53-F2-122 ＊ , TP53-F3-147 ＊ , GATA3-F3-350 ＊ A shared frameshift neoantigen produced from any of the above FPCs.

10. A method for producing a T cell receptor that specifically recognizes a shared neoantigen, comprising: selecting a shared frameshift neoantigen generated from an FSC as described in claim 9; an HLA type to which the shared frameshift neoantigen can be presented; and using a PBMC having the target HLA type, the shared frameshift neoantigen-responsive CD8 ＋ Induces T cells and shared frameshift neoantigen-responsive CD8 ＋ A method for analyzing the TCR sequence of T cells and constructing a T cell receptor based on the obtained TCR sequence.

11. A method for selecting an scFV antibody that specifically recognizes a neoantigen, comprising: selecting a shared frameshift neoantigen generated from an FSC as described in claim 9; selecting an HLA type on which the shared frameshift neoantigen can be presented; using an scFv-phage display library, causing the target type of HLA to present the shared frameshift neoantigen; and selecting the scFV to which it binds.

12. The method according to claim 9, further comprising producing an antibody with bispecificity by linking it with an scFV that binds to CD3.

Images