Pancreatic ductal adenocarcinoma-related hla-i epitope peptide, polyepitope nucleic acid vaccine and application thereof

Abstract

The present application relates to the field of immunopeptidomics, and discloses pancreatic ductal adenocarcinoma related HLA-I epitope peptides, including pancreatic ductal adenocarcinoma related HLA-I epitopes as shown in SEQ ID NO: 1-16, wherein SEQ ID NO: 1-4 are derived from pure new transcripts, SEQ ID NO: 5-8 are derived from known transcripts, and SEQ ID NO: 9-16 are derived from non-classical sources. The epitopes are identified by immunopeptidomics mass spectrometry to be capable of being combined with HLA-A24:02 and presented by antigen presenting cells, and verified by IFN-gamma ELISpot to have immunogenicity; a polyepitope polypeptide containing the epitopes, a coding nucleic acid, an mRNA-LNP delivery system and a pharmaceutical composition are also provided. A representative polyepitope mRNA-LNP of the present application induces an immune response mainly in the form of effector memory CD8 positive T cells in transgenic mice, and produces an anti-tumor effect in a pancreatic cancer tumor-bearing model; the present application also provides a preparation method of the above products and application thereof in the preparation of a drug for preventing or treating pancreatic cancer.

Description

Technical Field

[0001] This invention relates to the field of immunopeptidomics technology, specifically to a pancreatic ductal adenocarcinoma-associated HLA-I epitope peptide, a multi-epitope nucleic acid vaccine, and their applications. Background Technology

[0002] The working principle of tumor vaccines is as follows: by providing tumor-associated antigens or neoantigens to the immune system, antigen-specific T cells are induced. The T cells then hunt down and kill the tumor cells. Finding a good target (epitope) is a key step in vaccine development.

[0003] Pancreatic ductal adenocarcinoma (PDAC) is the most common and aggressive type of pancreatic cancer, characterized by atypical early symptoms, frequent diagnosis at an advanced stage, low resectable rate, and high risk of recurrence and metastasis. Current treatments include surgery, chemotherapy, radiotherapy, targeted therapy, and immunotherapy. However, because the tumor microenvironment suppresses immune cells, and the tumor itself exhibits antigenic heterogeneity, current technologies lack a sufficient number and quality of targets for the immune system to precisely attack the tumor.

[0004] The traditional approach to finding targets involves performing whole-exome sequencing and short-read RNA sequencing on tumor and normal tissues to identify unique gene mutations (SNVs, i.e., single-base changes) in tumor cells. These mutations are then translated into a standard reference proteome database to see which mutations produce different peptides and to predict whether these mutant peptides can bind to MHC molecules. If they can, they are considered as candidate epitopes.

[0005] However, PDAC typically has a low mutation burden, with far fewer gene mutations than melanoma. The number of candidate epitopes obtained by finding mutations is very limited. Relying solely on mutations to design vaccines will either result in not finding enough targets or not covering enough patients.

[0006] Besides gene mutations, tumor cells can also produce peptides not included in the standard reference proteome due to aberrant splicing, aberrant transcription initiation or termination, intron retention, translation of non-coding transcripts, pseudogene expression, and non-canonical open reading frames. Theoretically, these peptides can also be cleaved by proteasomes, transported to the cell surface by MHC molecules, and recognized by T cells, thus becoming vaccine targets.

[0007] However, in most cases, researchers use standard reference proteome databases (such as the known proteins listed in UniProt). If a peptide originates from an "abnormal splicing" or "non-classical ORF" as mentioned earlier, and this sequence is not found in the standard database, then even if mass spectrometry detects it, it will not match the sequence, and the functional effect of this peptide cannot be proven.

[0008] It is evident that, given the low mutation burden in pancreatic ductal adenocarcinoma (PDAC) leading to a severe shortage of epitopes from traditional SNV sources, and the inability of standard reference proteomics databases to cover non-classical peptides, resulting in the inability to identify the true presenting peptides detected by mass spectrometry, there is an urgent need for a method that can systematically mine and validate non-classical HLA-I epitopes and provide new peptides. Summary of the Invention

[0009] The purpose of this invention is to provide an HLA-I epitope peptide, a multi-epitope nucleic acid vaccine, and their applications related to pancreatic ductal adenocarcinoma, in order to solve the technical problems in the prior art where the low mutation burden of pancreatic ductal adenocarcinoma leads to a severe shortage of epitopes from traditional SNV sources, and the standard reference proteome database cannot cover peptides from non-classical sources, resulting in the inability to identify the true presenting peptides detected by mass spectrometry.

[0010] To solve the above-mentioned technical problems, the present invention specifically provides the following technical solution: A pancreatic ductal adenocarcinoma-associated HLA-I epitope peptide, wherein the amino acid sequence of the epitope peptide is shown in any one of SEQ ID NO:1-16; Alternatively, it may be a variant based on any of the sequences in SEQ ID NO:1-16, with conserved amino acid substitutions, while retaining HLA-A*24:02 binding activity and CD8-positive T cell recognition function.

[0011] Furthermore, the epitope peptide is derived from pancreatic ductal adenocarcinoma cells, and its sources include: novel transcripts, known transcripts, and non-classical sources; Among them, SEQ ID NO:1-4 are derived from novel transcripts, SEQ ID NO:5-8 are derived from known transcripts, and SEQ ID NO:9-16 are derived from non-classical sources; The non-classical sources include aberrant splicing, non-classical open reading frames, pseudogenes, non-coding RNA translation regions, frameshifts, upstream or overlapping open reading frames.

[0012] The present invention further provides a multi-epitope polypeptide comprising at least two epitope peptide modules selected from SEQ ID NO:1-16, with adjacent epitope peptide modules connected by flexible linker peptides.

[0013] Furthermore, the multi-epitope polypeptide comprises at least one of the epitope peptides and at least one known HLA-A*24:02 tumor-associated epitope module selected from SEQ ID NO:17-24, wherein the epitope peptide and the HLA-A*24:02 tumor-associated epitope module can form a peptide-MHC complex.

[0014] Furthermore, the amino acid sequence of the multi-epitope polypeptide is shown in any one of SEQ ID NO:25-28; Alternatively, a variant that has at least 85% sequence identity with any of SEQ ID NO:25-28 and retains the HLA binding activity and immunogenicity of each epitope module.

[0015] Furthermore, the flexible linker peptide is a linker peptide rich in glycine and serine; The multi-epitope polypeptide also includes a signal peptide at the N-terminus and a transmembrane anchoring domain associated with MHC-I class molecule transport at the C-terminus.

[0016] The present invention further provides an isolated nucleic acid molecule. The nucleic acid molecule encodes the epitope peptide described in any of the preceding claims or the multi-epitope polypeptide described in any of the preceding claims.

[0017] Further, the nucleic acid molecule is a messenger RNA, which sequentially comprises a 5′ cap structure, a 5′ untranslated region, an open reading frame, a 3′ untranslated region, and a poly(A) tail; the messenger RNA contains N1-methylpseudouridine and / or 5-methylcytidine modification, and optionally contains a CleanCapAG cap structure.

[0018] The present invention further provides a delivery system comprising the above-mentioned nucleic acid molecules and lipid nanoparticles; The lipid nanoparticles contain ionizable lipids, cofactor phospholipids, cholesterol, and PEGylated lipids; The lipid nanoparticles have an average particle size of 125-140 nm and an encapsulation efficiency of >90%.

[0019] The present invention further provides a pharmaceutical composition comprising the above-described epitope peptide, the above-described multi-epitope polypeptide, the above-described nucleic acid molecule or the above-described delivery system, and a pharmaceutically acceptable carrier.

[0020] The present invention further provides the use of the above-mentioned epitope peptide, the above-mentioned multi-epitope polypeptide, the above-mentioned nucleic acid molecule, the above-mentioned delivery system or the above-mentioned pharmaceutical composition in the preparation of a medicament for the prevention and / or treatment of pancreatic ductal adenocarcinoma, said medicament being used to induce or amplify antigen-specific CD8 positive T cells, effector memory T cells and / or tumor-infiltrating T cells.

[0021] Furthermore, it also includes the combined use of the drug as an immune checkpoint inhibitor, chemotherapy, radiotherapy, cell therapy, or TCR engineered cells.

[0022] Compared with the prior art, the present invention has the following advantages: This invention combines long-read transcriptome sequencing with a customized mass spectrometry database to establish a systematically identifiable model for peptides from non-classical sources. In the PDAC cell line, this method established 22,784 high-confidence transcript isoforms, of which 46.1% were novel isoforms. Using this as the search space, MHC-I immunopeptidomics identified 45,892 unique binding peptides, of which 18.4% were encoded by novel isoforms. The enrichment ratio in the HLA-A*24:02 high-affinity peptide was 3.47 (P<0.001). IFN-γ ELISpot functional validation showed that 16 out of 64 candidate peptides reached the high-confidence response threshold, with a positive rate of 25%. This indicates that this method can efficiently obtain candidate epitopes with real presentation evidence and immunogenicity from tumors with low mutation burden, effectively compensating for the target deficiency problem of traditional SNV-driven strategies. The multi-epitope mRNA-LNP constructed based on the above-mentioned epitopes induced an immune response dominated by effector memory CD8-positive T cells in HLA-A*24:02 transgenic mice. The sq3 construct containing only 8 non-classical epitopes, with 40% of the epitopes in sq4, achieved a CD8-positive T cell ratio of 52.0% and additionally induced the expansion of CD4-positive T cells, suggesting that non-classical epitopes have higher immunogenicity density and a broader response spectrum. Furthermore, in the S2-VP10 tumor-bearing adoptive metastasis model, the vaccine-immunized cells extended the median survival from 21 days to 52 days with a hazard ratio of 0.35 (P=0.003), and significantly increased CD8-positive T cell infiltration in tumor tissue. This suggests that the mRNA-LNP vaccine of this invention can induce effector memory T cells with tumor homing and killing capabilities and produce a clear anti-tumor effect in animal models. Attached Figure Description

[0023] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0024] Figure 1 To verify the sequencing process shown in Example 1 for constructing a full-length PDAC isoform map by combining long-read and short-read transcriptomes.

[0025] Figure 2 To verify the isoform classification (b) and BUSCO integrity (c) of the PDAC full-length isoform map constructed by combining long-read and short-read transcriptomes as shown in Example 1.

[0026] Figure 3To verify the cross-cohort detection results (d) and related pathway enrichment (e) of the PDAC full-length isoform map constructed by combining long-read and short-read transcriptomes as shown in Example 1.

[0027] Figure 4 To verify the flowchart of long read length enhanced reference database (LRDB) construction and epitope identification shown in Example 2.

[0028] Figure 5 To verify the secondary mass spectrum of the HLA-A*24:02 peptide shown in Example 3.

[0029] Figure 6 To verify the IFN-γELISpot screening heatmap of 64 candidate peptides by three HLA-A*24:02 matched healthy donors shown in Example 4.

[0030] Figure 7 To verify the design schematic diagram of the sq1-sq4 mRNA-LNP construct (including 5' cap, 5' / 3' UTR, signal peptide, epitope module, (GGGGS)3 linker peptide, MITD anchor, and poly(A)) based on the epitopes of this invention and known epitopes shown in Example 5.

[0031] Figure 8 To verify the particle morphology (b), particle size (c), PDI (c), surface potential (d), encapsulation efficiency (d), in vitro translated mRNA electrophoresis (e), Western blot (f), dendritic cell CD86 upregulation and DC-T cell IFN-γ response (g) of sq1-sq4 mRNA-LNP shown in Examples 6-8.

[0032] Figure 9 To verify the timeline of the immunization regimen of HLA-A*24:02 transgenic mice on days 0, 7, and 14 (a), the representative flow cytometry diagram of CD8 positive T cell expansion on day 21 (bc), the CD44 positive / CD62L negative effect memory phenotype, and the IFN-γ ELISpot well diagram of spleen cells (d) shown in Example 9.

[0033] Figure 10 To verify the timeline (a), relative tumor volume (b), endpoint tumor weight (c), survival curve (d), and tumor CD8 immunohistochemical results (e) of the S2-VP10 tumor-bearing mice induced by the vaccine for adoptive transfer, as shown in Example 10.

[0034] Figure 11 A flowchart for the discovery, validation, and application of HLA-I epitopes. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] As shown in Table 1, the present invention first provides a group of pancreatic ductal adenocarcinoma-related HLA-I epitope peptides, the amino acid sequences of which are shown in SEQ ID NO:1-16, or are variants based on the sequences shown in SEQ ID NO:1-16 with conserved amino acid substitutions, and retaining HLA-A*24:02 binding activity and CD8 positive T cell recognition function.

[0037] The acquisition of this group of epitope peptides did not rely on somatic nonsynonymous mutations identified by conventional whole-exome sequencing and short-read RNA sequencing, but was based on a complete long-read and short-read transcriptome combined analysis process.

[0038] Specifically, this invention selected five PDAC cell lines—AsPC-1, KLM-1, MIA PaCa-2, PK-8, and S2-VP10—for PacBioIso-Seq long-read sequencing, and integrated publicly available PANC-1 data and corresponding short-read RNA-seq data. CCS sequences underwent quality control (single-base accuracy ≥99%), Iso-Seq3 was used to identify full-length non-chimeric reads, Minimap2 was used for alignment to GRCh38, redundant isoforms were folded at approximately 95% sequence consistency, and SQANTI3 and other tools were used in conjunction with splice ligation support, non-standard splice sites, read count, consistency with reference annotations, and short-read support for structural classification and filtering.

[0039] Through the above process, the main transcriptome contained a total of 22,784 high-confidence isoforms, of which 12,287 were fully spliced-matched isoforms and 10,497 were novel isoforms (including approximately 4,143 ISMs, 4,558 NICs, and 1,726 NNCs). Novel isoforms accounted for approximately 46.1% (if only the standard reference protein database is used for searching, nearly half of the transcript sources would be systematically missed).

[0040] Based on this, the present invention performs open reading frame (ORF) prediction for high-confidence isomers by combining CPAT, ORF length, start codon, Kozak environment, hexamer usage preference, homology with known proteins, and protein domain information.

[0041] The study found that the average CPAT coding probabilities of ISM, NIC, and NNC were approximately 91.7%, 97.3%, and 95.4%, respectively; approximately 55.1% and 56.8% of the predicted translation products in NIC and NNC contained sequences that were not fully annotated by the standard reference protein.

[0042] This data directly demonstrates the enormous potential of novel isoforms to encode non-classical peptides. Subsequently, this invention merges UniProt human reference protein, non-classical ORF databases such as nuORFdb, long-read isoform prediction proteins, and contaminating proteins and reverse sequences used to control mismatches to form the Long Read Enhanced Reference Database (LRDB). The database retains information for each protein sequence, including its source transcript ID, gene, structural class, genomic coordinates, ORF coordinates, splice connections, presence in the reference protein, cell line expression level, and normal tissue expression.

[0043] It is worth noting that conventional immunopeptidomics analysis typically uses only a standard reference proteome (such as UniProt) as the search database. This means that peptides derived from anomalous isoforms or non-classical open reading frames (ORFs) cannot be correctly identified even if detected by mass spectrometry. This invention fundamentally solves this technical obstacle by constructing an LRDB. The mass spectrometry search database is no longer limited to known proteins but extends to include all possible sequences encoded by novel isoforms and non-classical ORFs.

[0044] After the database was built, this invention collected approximately 5 × 10⁻⁶ samples. 7 PDAC cells were used to immunoaffinity enrich the HLA-I heavy chain-β2 microglobulin-peptide complex using the W6 / 32 antibody. HLA-I binding peptides were released by acid elution, desalted, and concentrated before LC-MS / MS analysis. Raw mass spectrometry data were searched using MaxQuant or equivalent software in the LRDB, with peptide and peptide profile matching FDR controlled at 1%.

[0045] The preferred quality control scheme includes: peptide length of 8-14 amino acids, MaxQuantScore of not less than 80, PEP of not more than 0.147, and precursor quality error mainly within ±10ppm.

[0046] A total of 45,892 unique MHC-I binding peptides were obtained after quality filtration, of which 8,437 (approximately 18.4%) were encoded by novel isomers, with 9 peptides being the predominant length type.

[0047] The statistical significance is as follows: the enrichment ratio of novel isoform-derived proteins was approximately 3.12 (95% CI 2.12–3.44); the enrichment ratio of HLA-A*24:02 high-affinity peptides was approximately 3.47 (95% CI 3.28–3.67, P < 0.0001). This indicates that peptides from novel isoforms not only actually exist in the MHC-I presentation system, but also exhibit significant enrichment relative to their genomic proportions. Therefore, peptides from non-classical sources are not only other available peptides, but may also lead to preferential attention from the immune system.

[0048] Furthermore, this invention uses NetMHCpan 4.1 or an equivalent model to calculate IC50 and percentile rank, and DeepImmuno or an equivalent model to evaluate immunogenicity. Peptides with IC50 < 500 nM and DeepImmuno score > 0.45 are retained. Multi-indicator screening is performed, combining peptide profile matching quality, retention time consistency, source transcript expression, tumor / normal differences, cross-cohort reproducibility, and normal protein exclusion. UniProt human proteome and IEDB data are used for complete matching or high similarity exclusion to reduce the influence of self-antigen cross-reactivity and known common epitopes. After multi-indicator screening, 147 preferred candidates were obtained, from which 64 were synthesized for human PBMC experiments.

[0049] HLA-A*24:02 is one of the dominant alleles in the PDAC cohort and has a high frequency in East Asian populations, so it was selected as a key target for validation.

[0050] Recruit three healthy HLA-A*24:02 matched donors, isolate PBMCs, and add approximately 3 × 10⁻⁶ M cells per well. 5 Cells were stimulated with a single synthetic peptide at a final concentration of 20 μg / mL for approximately 20 hours, and peptide-specific responses were evaluated using IFN-γ ELISpot. A positive criterion was generally >50 SFU / 102 after background subtraction. 6 Cells; not less than twice that of the negative control; statistical P < 0.05. The high-confidence response threshold was set to >200 SFU / 102 6 Cellular. Of the 64 candidate peptides, 16 reached the high-confidence response threshold (approximately 25%).

[0051] The above identification and verification process is a systematic process. It is not a simple prediction, but a process that uses mass spectrometry to confirm that the selected peptides do exist on the MHC-I molecules on the cell surface. Then, it uses ELISpot to confirm that the peptides can be recognized by T cells and trigger a functional response. In the final identified SEQ ID NO:1-16, SEQ ID NO:1-4 are from pure new transcripts, SEQ ID NO:5-8 are from known transcripts, and SEQ ID NO:9-16 are from non-canonical sources (including aberrant splicing, non-canonical open reading frames, pseudogenes, non-coding RNA translation regions, frameshifts, upstream or overlapping open reading frames).

[0052] By combining prediction, screening, and validation, this invention enables the discovery of immunogenic epitopes in pancreatic ductal adenocarcinoma (PDAC) tumors from dimensions inaccessible by traditional methods. This provides new polypeptide fragments for existing technologies and offers more options for the development of pancreatic ductal adenocarcinoma tumor vaccines.

[0053] The core determinants of HLA-I epitope binding to MHC molecules lie in the chemical properties of specific anchoring residues (usually position 2 and the C-terminus) and the spatial conformation of the epitope as a whole at the TCR interface. Provided these key characteristics are preserved, substitutions of the same amino acid at non-anchoring sites (such as substitutions between hydrophobic residues or between polar residues) generally do not significantly affect HLA binding affinity and T cell recognition specificity.

[0054] Therefore, the present invention also includes variants based on any one of the sequences in SEQ ID NO:1-16 with conserved amino acid substitutions, while retaining HLA-A*24:02 binding activity and CD8-positive T cell recognition function.

[0055] Based on a single epitope peptide, the present invention further provides a multi-epitope polypeptide comprising at least two epitope peptide modules selected from SEQ ID NO:1-16, with adjacent epitope peptide modules connected by flexible linker peptides.

[0056] First, adjacent epitope modules are separated by flexible linker peptides rich in glycine and serine. Glycine has extremely high main chain flexibility because its side chain consists of only hydrogen atoms, while serine can enhance the water solubility of the linker peptide by forming hydrogen bonds with water molecules through its hydroxyl groups. The periodic arrangement of the two (such as (GGGGS)3) can provide sufficient independent folding space for each epitope module, avoiding steric hindrance between adjacent epitopes from interfering with their binding to MHC molecules or TCRs.

[0057] Secondly, the multi-epitope polypeptide also includes an N-terminal signal peptide and a C-terminal MHC-I trafficking domain (MITD) transmembrane anchoring domain. The signal peptide (tPA signal peptide in this invention, sequence MDAMKRGLCCVLLLCGAVFVSP) guides the nascent polypeptide chain into the endoplasmic reticulum lumen, which is the entry point for the MHC-I antigen processing and presentation pathway. The MITD domain (sequence VGIVAGLAVLAVVVIGAVVATVMCRRKSSGGKGGSYSQAASSDSAQGSDVSLTA) anchors the polypeptide to the endoplasmic reticulum membrane or cell membrane through its transmembrane region, thereby prolonging the polypeptide's residence time in antigen-presenting cells and improving the efficiency of proteasome cleavage and loading onto MHC-I molecules. The synergistic effect of the signal peptide and MITD enables the multi-epitope polypeptide encoded by nucleic acid molecules to enter the MHC-I presentation pathway efficiently, rather than being rapidly degraded in the cytoplasm or remaining in non-functional locations.

[0058] The present invention may further comprise at least one known HLA-A*24:02 tumor-associated epitope module selected from SEQ ID NO:17-24 and at least one of the epitope peptides forming a peptide-MHC complex.

[0059] The epitopes AFYPGYLCSL, KVYLRVRPLL, TCQPTCRSL, VYAETKHFL, VYGFVRACL, RWPSCQKKF, AYACNTSTL, and QYSWFVNGTF shown in SEQ ID NO:17-24 are known HLA-A*24:02 tumor-associated epitopes and can be used as components or controls of multi-epitope combinations.

[0060] Based on the above design principles, specifically, this invention provides four multi-epitope constructs: 1. The amino acid sequence of sq1 is SEQ ID NO:25 (164aa), containing 4 novel transcript-derived epitopes SEQ ID NO:1-4; 2. sq2 is SEQ ID NO:26 (240aa), containing 8 epitopes SEQ ID NO:1-8; 3. sq3 is SEQ ID NO:27 (238aa), containing 8 non-classical source epitopes SEQ ID NO:9-16; 4. sq4 is SEQ ID NO:28 (471aa), which contains 12 epitopes of the present invention SEQ ID NO:1-4 and 9-16, and 8 known epitopes SEQ ID NO:17-24, for a total of 20 epitope modules.

[0061] SEQ ID NO:25-28 do not contain His or Flag tags, thus avoiding the additional immunogenicity or interference with epitope presentation that may be introduced by exogenous tag sequences, making the vaccine-induced immune response more focused on the target epitope itself.

[0062] The present invention also includes variants that have at least 85% sequence identity with SEQ ID NO:25-28 and retain the HLA binding activity and immunogenicity of each epitope module.

[0063] In this context, fine-tuning of the linker peptide region length and substitution of a few amino acids in the signal peptide or MITD typically do not alter the core function of the multi-epitope polypeptide. As long as the sequences of each epitope module remain unchanged or only the aforementioned conserved substitutions occur, and the linker peptide maintains its flexible characteristics, the function of the entire multi-epitope polypeptide can be preserved. The 85% sequence identity threshold provides room for potential sequence optimization while ensuring the technical equivalence between the variant and the original construct through functional constraints.

[0064] The present invention further provides an isolated nucleic acid molecule encoding the epitope peptide or multi-epitope polypeptide provided by the present invention.

[0065] This nucleic acid molecule is a messenger RNA (mRNA), which consists of a 5′ cap, a 5′ untranslated region (UTR), an open reading frame (ORF), a 3′ untranslated region, and a poly(A) tail.

[0066] The messenger RNA contains N1-methylpseudouridine and / or 5-methylcytidine modifications, and may also contain a CleanCap AG cap structure.

[0067] The selection of mRNA as a vaccine molecule has a clear molecular biological basis. Compared with DNA vaccines, mRNA does not need to enter the cell nucleus and can be directly translated into antigen proteins in the cytoplasm, avoiding the risk of genome integration. Compared with peptide vaccines, mRNA can continuously express antigens in antigen-presenting cells, realizing the processing and presentation of endogenous antigens and more effectively activating CD8-positive T cell responses.

[0068] Each structural element of mRNA performs an irreplaceable function. The 5′ cap (Cap1, such as CleanCapAG) protects mRNA from degradation by 5′→3′ exonucleases and binds to the eukaryotic translation initiation factor eIF4E to initiate translation. The 5′UTR and 3′UTR regulate the translation efficiency and stability of mRNA; their sequences can be optimized to avoid undesirable secondary structures and regulatory elements. The length of the poly(A) tail directly affects the half-life of mRNA and translation initiation efficiency.

[0069] The present invention further employs N1-methylpseudouridine and / or 5-methylcytidine for nucleoside modification. The unmodified exogenous mRNA can be recognized by innate immune sensors such as Toll-like receptors (TLR3, TLR7, TLR8) and RIG-I, triggering an excessive inflammatory response and inhibiting translation.

[0070] N1-methylpseudouridine and / or 5-methylcytidine modification of nucleosides can prevent overactivation by pattern recognition receptors, significantly reduce the immunogenicity of mRNA, and the modified mRNA has higher ribosome affinity while improving translation efficiency. CleanCapAG co-transcriptional capping technology ensures that the mRNA molecule acquires the native Cap1 structure during in vitro transcription, eliminating the need for subsequent enzymatic capping steps.

[0071] During the preparation process, mammalian preferred codons were optimized based on the multi-epitope polypeptide sequence to avoid generating unfavorable repetitive sequences, aberrant splicing sites, excessively strong secondary structures, or potential immunosuppressive motifs. The optimized CDS was cloned into a plasmid containing the T7 promoter, 5′UTR, 3′UTR, and poly(A) template, linearized, and then transcribed in vitro.

[0072] The transcription system contained ATP, GTP, N1-methylpseudouridine triphosphate, and 5-methylcytidine triphosphate, and was co-transcribed and capped using CleanCap AG. Post-transcriptional treatment with DNase followed by purification via chromatography, membrane filtration, or magnetic beads to remove short fragments, free nucleotides, and double-stranded RNA, with an optimal RIN ≥ 8.5 and a full-length RNA content > 95%. Quality control measures ensured the integrity and purity of the mRNA product, laying the foundation for subsequent LNP encapsulation and in vivo expression titers.

[0073] The present invention also provides a delivery system comprising the aforementioned nucleic acid molecules and lipid nanoparticles (LNPs).

[0074] The lipid nanoparticles contain ionizable lipids, cofactor phospholipids, cholesterol, and PEGylated lipids, with an average particle size of 125-140 nm and an encapsulation efficiency greater than 90%.

[0075] Ionizable lipids (such as SM-102, ALC-0315, or MC3) carry a positive charge in acidic pH conditions (such as citrate buffer, pH 4.0), enabling them to form tight lipid-nucleic acid complexes with negatively charged mRNA molecules through electrostatic interactions.

[0076] At physiological pH (7.4), ionizable lipids tend to be neutral, reducing the surface positive charge of LNPs and thus decreasing nonspecific adsorption to serum proteins and clearance by the reticuloendothelial system. Helper phospholipids (such as DSPC) provide structural rigidity to the lipid bilayer, cholesterol regulates membrane fluidity and stability, and PEGylated lipids (such as PEG2000-DMG) form a hydration layer on the LNP surface to prolong in vivo circulation time.

[0077] During preparation, mRNA is mixed with a lipid phase containing the aforementioned lipid components and self-assembled to form LNPs via microfluidic mixing. The LNPs are then replaced with a neutral buffer via dialysis, tangential flow filtration, or ultrafiltration. Cryoprotectants such as sucrose and trehalose can be added. Representative batches have an average particle size of approximately 125-140 nm, a PDI of approximately 0.10-0.20, a near-neutral surface potential, and an encapsulation efficiency >90%.

[0078] A particle size range of 125-140 nm offers significant advantages. Particles within this size are sufficient to passively target tumor tissue through the EPR effect (enhanced permeability and retention), while also being effectively taken up by antigen-presenting cells. An encapsulation efficiency >90% ensures efficient encapsulation of the mRNA payload, reducing non-specific degradation of free mRNA in vivo. Formulation stability was evaluated by measuring particle size, PDI, encapsulation efficiency, RNA integrity, and in vitro expression potency after short-term storage at 4°C, cryopreservation, and freeze-thaw cycles.

[0079] At the pharmaceutical composition level, the present invention provides compositions comprising any of the aforementioned product levels (epitope peptides, multi-epitope polypeptides, nucleic acid molecules or delivery systems) and a pharmaceutically acceptable carrier.

[0080] Pharmaceutically acceptable carriers may include buffers (such as PBS, citrate buffer), stabilizers (such as sucrose, trehalose), cryoprotectants, and combinations thereof.

[0081] The selection of the above-mentioned carriers is optimized based on the route of administration (intramuscular injection is preferred) and storage conditions (freezing or refrigeration) of the formulation to ensure the physicochemical stability and biological activity of the composition during preparation, storage and administration.

[0082] To verify the technical effectiveness of the above design, the following specific verification examples are provided: Unless the context otherwise requires, an "HLA-I epitope" or "MHC-I epitope" is a peptide that can bind to an MHC-I molecule and be recognized by CD8-positive T cells. "Antigenic peptide" includes free peptides, salts, permitted terminally modified peptides, vector-coupled peptides, peptides as part of longer polypeptides, and peptides produced by post-expression processing of nucleic acids within cells. "Encoding" includes nucleic acids directly encoding the target peptide, as well as precursor polypeptides encoding the target peptide, said precursors being processed by intracellular proteasomes or other mechanisms to produce the target peptide.

[0083] "Purely novel transcript-derived epitopes" are epitopes supported only by transcripts newly added relative to the adopted reference annotation; "Known transcript-derived epitopes" are epitopes generated from transcripts already present in the reference annotation but identified and sorted by the process of this invention; "Non-classical epitopes" are epitopes derived from aberrant splicing, atypical transcripts, non-classical open reading frames, pseudogenes, non-coding RNA translation regions, frameshifts, upstream or overlapping open reading frames.

[0084] "Known epitopes" refer to peptides whose HLA-restricted, T-cell immunogenicity, or tumor-related properties have been reported in published literature prior to this application. SEQ ID NO:17-24 are known epitopes and serve as constituent modules, benchmarks, or reinforcing components of specific multi-epitope combinations in this invention.

[0085] "Approximately" indicates a reasonable range of experimental error in the art; unless otherwise stated, it may mean ±10% of the stated value.

[0086] Validation Example 1: Combined analysis of PDAC cell lines and long / short read transcriptomes Transcriptome and isoform identification results are shown in Figure 1-3 As shown.

[0087] Five PDAC cell lines—AsPC-1, KLM-1, MIA PaCa-2, PK-8, and S2-VP10—were selected for PacBioIso-Seq long-read sequencing, and publicly available PANC-1 data and corresponding short-read RNA-seq data were integrated. CCS sequences underwent quality control (single-base accuracy ≥99%). Iso-Seq3 was used to identify full-length non-chimeric reads, Minimap2 was used for alignment to GRCh38, and redundant isoforms were folded at approximately 95% sequence consistency. SQANTI3 and other tools were used in conjunction with splice ligation support, non-standard splice sites, read count, consistency with reference annotations, and short-read support for structural classification and filtering.

[0088] All five PDAC cell lines were confirmed by STR typing and were negative for mycoplasma. They were cultured at 37°C and 5% CO2. S2-VP10, KLM-1, and AsPC-1 cells were cultured in RPMI-1640 medium (Gibco, catalog number 11875-093) supplemented with 10% heat-inactivated fetal bovine serum (Gibco, catalog number 16000-044), penicillin (100 U / mL), and streptomycin (100 μg / mL) (Gibco, catalog number 15140-122). MIA PaCa-2 and PANC-1 cells were cultured in high-glucose DMEM medium (Gibco, catalog number 11965-092). Cells were passaged using trypsin-EDTA (Gibco, catalog number 25200-056) for no more than 20 passages.

[0089] according to Figure 2 and Figure 3 The main transcriptome contained 22,784 high-confidence isoforms, of which 12,287 were fully spliced-matched isoforms and 10,497 were novel isoforms (including approximately 4,143 ISMs, 4,558 NICs, and 1,726 NNCs), with novel isoforms accounting for approximately 46.1%. Completeness was assessed using BUSCO; the detection rate of novel isoforms in the six independent PDAC cohorts ranged from approximately 40% to 80%, supporting cross-cohort reproducibility.

[0090] Alternative splicing events (SE, A3, A5, AF, RI, MX, AL) were generated using SUPPA2 or equivalent software, and PSI was calculated using short-read data from tumor and normal tissues. In 13 PDAC tumors and 6 normal tissues from the GSE196009 Japanese cohort, 48 significantly aberrant splicing events were identified with a threshold of |ΔPSI|≥0.2 and P<0.01. Representative events involved genes such as ACSL5, GSN, JUP, LOXL2, PVR, TKFC, and VNN1. Differentially spliced ​​genes were enriched in pathways related to small GTPase binding, cell adhesion, EMT, inflammation, and metabolism.

[0091] Verification Example 2: Construction of a Reference Database for Open Reading Frame Prediction and Long Read Length Enhancement Open reading frame (ORF) prediction for high-confidence isoforms can be performed by combining CPAT, ORF length, start codon, Kozak environment, hexamer usage preference, homology with known proteins, and protein domain information. In this study, the average CPAT encoding probabilities for ISM, NIC, and NNC were approximately 91.7%, 97.3%, and 95.4%, respectively; approximately 55.1% and 56.8% of the predicted translation products in NIC and NNC contained sequences not fully annotated by the standard reference protein.

[0092] The UniProt human reference protein database, nuORFdb and other non-classical ORF databases, long-read isoform prediction proteins used in this study, and contaminating proteins and reverse sequences used to control mismatches were merged to form the Long Read Enhanced Reference Database (LRDB). The database retains information for each protein sequence, including its source transcript ID, gene, structural class, genomic coordinates, ORF coordinates, splice connections, presence in a reference protein, cell line expression level, and normal tissue expression.

[0093] Figure 4 This paper demonstrates the construction of the Long Read Enhancement Reference Database (LRDB) and the epitope identification process based on this database. In the database construction phase, the classification results of novel isoforms discovered through long read sequencing serve as the core input, integrating the UniProt human reference protein database, the nuORFdb non-classical ORF database, as well as contaminating proteins and reverse sequences (used for FDR control) to form the LRDB. In the epitope identification phase, HLA-I immunoaffinity enrichment, LC-MS / MS mass spectrometry detection, and multi-dimensional computational screening (HLA binding prediction, immunogenicity score, expression level, etc.) are performed sequentially, ultimately outputting a list of candidate epitopes.

[0094] Validation Example 3: HLA-I Immunoaffinity Enrichment and Mass Spectrometry Identification The specific conditions for immunoaffinity enrichment are: collect approximately 5 × 10⁻⁶ samples each. 7 The above-mentioned PDAC cells were lysed on ice for 30 minutes in lysis buffer containing protease inhibitors (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% CHAPS). The cells were centrifuged at 20,000×g and 4°C for 30 minutes to collect the supernatant. The supernatant was pre-cleaned with Protein A / G magnetic beads for 2 hours, and then anti-HLA-I pan-reactive antibody W6 / 32 (5 μg / mg protein) was added and incubated overnight at 4°C by rotation. Protein A / G magnetic beads were then added and incubated for another 4 hours. The magnetic beads were washed 5 times with PBS containing 0.1% CHAPS and 2 times with ice-cold PBS. The HLA-I peptide complex was eluted with 10% acetic acid (pH 2.6-2.8) and immediately neutralized to pH 7.0-8.0 with 1 M Tris. The eluent was collected as free peptides through a 10 kDa molecular weight cutoff ultrafiltration tube, lyophilized, and reconstituted with 0.1% formic acid for analysis.

[0095] In this validation example, the HLA-I binding peptide was desalted on a C18 column and separated on a 15 cm self-made C18 column (75 μm inner diameter, 1.9 μm particle size). Detection was performed using a Bruker Nano-Elute liquid chromatography-tandem time-to-flight Pro2 mass spectrometer. Mobile phase A was water containing 0.1% formic acid, and mobile phase B was acetonitrile containing 0.1% formic acid, with a flow rate of 0.3 μL / min. The chromatographic gradient was 4% B to 20% B from 0-90 min, 20% B to 30% B from 90-105 min, and 30% B to 100% B from 105-110 min, followed by a 100% B wash for 10 min. Mass spectrometry was performed in DDA-PASEF mode (positive ion mode, m / z 100-1700, 1 MS scan + 10 PASEF scans, ion mobility 1 / K0 = 0.7-1.4 V·s / cm). 2 Collision energy 20-59 eV, dynamic exclusion 30s). Raw data were retrieved from LRDB using MaxQuant (v2.5.2), employing a non-specific enzyme digestion mode that allowed for variable modifications such as methionine oxidation and N-terminal acetylation. Peptide lengths ranged from 8 to 14 amino acids, and the peptide match (PSM) level (FDR) was controlled at 1%.

[0096] A total of 45,892 unique MHC-I binding peptides were obtained after quality filtration, of which 8,437 (approximately 18.4%) were encoded by novel isomers, with 9 peptides being the predominant length type. The enrichment ratio of novel isomer-derived proteins was approximately 3.12 (95% CI 2.12–3.44); the enrichment ratio of HLA-A*24:02 high-affinity peptides was approximately 3.47 (95% CI 3.28–3.67, P < 0.0001).

[0097] Figure 5 The image shows the secondary mass spectrum of the HLA-A*24:02 peptide. According to the image, the MS / MS spectra of QYGGDLTNTF, VYGSIFSF, and EYDSSFPNW show fragments consistent with the theoretical b / y ion series.

[0098] IC50 and percentile rank were calculated using NetMHCpan 4.1 or an equivalent model, and immunogenicity was evaluated using DeepImmuno or an equivalent model. Peptides with IC50 < 500 nM and DeepImmuno score > 0.45 were retained, further evaluated by considering peptide profile matching quality, retention time consistency, source transcript expression, tumor / normal differences, cross-cohort reproducibility, and normal protein exclusion. UniProt human proteome and IEDB data were used for complete matching or high similarity exclusion to reduce the influence of self-antigen cross-reactivity and known common epitopes. After multi-indicator screening, 147 priority candidates were obtained, from which 64 were synthesized for human PBMC experiments. HLA-A*24:02 is one of the dominant alleles in the PDAC cohort and has a high frequency in East Asian populations; therefore, it was selected as a key validation target.

[0099] Validation Example 4: Validation of IFN-γ ELISpot in healthy donor PBMCs Recruit three healthy HLA-A*24:02 matched donors, isolate PBMCs, and add approximately 3 × 10⁻⁶ M cells per well. 5 Cells were stimulated with a single synthetic peptide at a final concentration of 20 μg / mL for approximately 20 hours, and peptide-specific responses were evaluated using IFN-γ ELISpot. DMSO was used as a negative control, and a non-specific T-cell stimulant was used as a positive control. Each candidate peptide was prepared in duplicate. A general positive criterion was >50 SFU / 10-1 after background subtraction. 6 Cells; not less than twice that of the negative control; statistical P < 0.05. High confidence response threshold set to > 200 SFU / 102 6 cell.

[0100] IFN-γ ELISpot assay was performed using a human IFN-γ ELISpot kit (Mabtech, catalog number 3420-4HPW-2), following the kit instructions. Spot count was quantified using an ELISpot reader, and results were expressed in increments of 10⁻⁶. 6 Cell spot count (SFU / 10) 6 The data are expressed as mean ± SEM (n = 3 donors). Candidate peptides (9-12 mers) were synthesized by Nanjing Peptide Valley Biotechnology Co., Ltd., and the lyophilized powder was dissolved in DMSO (Sigma-Aldrich, catalog number D2650) and diluted to 2 μg / μL before use. The PBMC collection from healthy donors and the ELISpot human sample procedure involved in this validation case were approved by the Ethics Committee of Beijing Yuhe Integrated Traditional and Western Medicine Rehabilitation Hospital (Approval No.

[2023] Ethics Review 016, Protocol No. LP202210 v2.0), and all donors signed informed consent forms.

[0101] Figure 6A heatmap (including SFU values ​​and statistics) of IFN-γ ELISpot screening for 64 candidate peptides in three HLA-A*24:02 matched healthy donors. Figure 6 It can be seen that 16 out of 64 candidate peptides reached the high confidence response threshold (approximately 25%).

[0102] Validation Example 5: Multi-epitope peptide design To compare the effects of epitope number, source type, and their combination with known epitopes on immune responses, four multi-epitope constructs were designed. Each construct contains an N-terminal tPA signal peptide, epitope modules separated by glycine / serine-rich linker peptides, and a C-terminal MITD transmembrane anchoring region. The amino acid sequence of construct sq1 is SEQ ID NO:25 (164 aa), containing 4 epitopes of pure novel transcript origin SEQ ID NO:1-4; sq2 is SEQ ID NO:26 (240 aa), containing 8 epitopes SEQ ID NO:1-8; sq3 is SEQ ID NO:27 (238 aa), containing 8 epitopes of non-classical origin SEQ ID NO:9-16; and sq4 is SEQ ID NO:28 (471 aa), containing 12 epitopes of this invention SEQ ID NO:1-4 and 9-16, and 8 known epitopes SEQ ID NO:17-24, for a total of 20 epitope modules. None of SEQ ID NO:25-28 contain the His or Flag tags.

[0103] Figure 7 This is a schematic diagram of the design of the sq1-sq4 mRNA-LNP construct based on the epitopes of this invention and combinations of known epitopes.

[0104] Validation Example 6: mRNA Preparation and Quality Control Mammalian preferred codons were optimized based on multi-epitope polypeptide sequences to avoid unfavorable repetitive sequences, aberrant splicing sites, excessively strong secondary structures, or potential immunosuppressive motifs. The optimized CDS was cloned into a plasmid containing the T7 promoter, 5'UTR, 3'UTR, and poly(A) template, linearized, and then transcribed in vitro. The transcription system contained ATP, GTP, N1-methylpseudouridine triphosphate, and 5-methylcytidine triphosphate, and co-capped transcription was performed using CleanCap AG. Post-transcriptional treatment with DNase followed by purification by chromatography, membrane filtration, or magnetic beads to remove short fragments, free nucleotides, and double-stranded RNA.

[0105] In vitro transcription was performed using the T7 Co-Transcription RNA Synthesis Kit (All In One High-Yield T7 Co-Transcription Kit (Cap1AG / N1ψ), Shanghai Hexincheng Biotechnology Co., Ltd., catalog number C3050). Urate was replaced with N1-methylpseudouridine (N1ψ), and cytidine was replaced with 5-methyl-CTP. CleanCap AG co-transcription was then performed for capping. The integrity of the purified mRNA was detected using the Agilent Bioanalyzer (RNA Nano microarray). All constructs had a RIN ≥ 8.5 and a full-length band percentage > 95%.

[0106] Validation Example 7: mRNA-LNP Preparation mRNA-LNPs were prepared using a NanoAssemblr Ignite microfluidic system. The lipid formulation consisted of ionizable lipid SM-102, helper phospholipid DSPC, cholesterol, and PEGylated lipid DMG-PEG2000 in a molar ratio of 50:10:38.5:1.5, with an N / P ratio of 6:1 (mRNA concentration was quantified using the RiboGreen method). Particle size and zeta potential were determined using a Malvern Zetasizer Nano ZS90. LNP morphology was observed using transmission electron microscopy (TEM). Encapsulation efficiency was determined using a modified Quant-iTRiboGreen method (with / without 1% Triton X-100), calculated as: Encapsulation efficiency (%) = (1 - free mRNA / total mRNA) × 100%. Formulation stability was assessed by incubation at 4°C for 14 days and three freeze-thaw cycles at -80°C. The deviations of particle size and encapsulation efficiency from baseline for each construct were ≤5%.

[0107] Formulation stability was evaluated by measuring particle size, PDI, encapsulation efficiency, RNA integrity, and in vitro expression potency after short-term storage at 4°C, cryopreservation, and freeze-thaw cycles.

[0108] Figure 8 To verify the particle morphology (b), particle size (c), PDI (c), surface potential (d), encapsulation efficiency (d), in vitro translated mRNA electrophoresis (e), and Western blot (f) of sq1-sq4 mRNA-LNP shown in Example 7, as follows: Figure 8 As shown in (c) and (d), the representative batches have an average particle size of about 125-140 nm, a PDI of about 0.10-0.20, a surface potential close to neutral, and an encapsulation efficiency of >90%.

[0109] Validation Example 8: Expression, Dendritic Cell Activation, and In Vitro Function Different doses of sq1-sq4 mRNA were added to suitable cell or antigen-presenting cell culture systems, and the target multi-epitope products were detected by Western blot (major expression signals consistent with the expected molecular weight were detected in each construct, and the expression increased with dose). After incubating the mRNA with dendritic cells, maturation / co-stimulatory molecules such as CD83 and CD86 were detected. Specifically, bone marrow-derived dendritic cells (BMDCs) were transfected with different doses of sq1-sq4 mRNA-LNP, with 2 μg mRNA per construct. After 24 hours, the expression of maturation / co-stimulatory molecules such as CD83 and CD86 was detected by flow cytometry.

[0110] Figure 8 To verify the upregulation of sq1-sq4 mRNA in dendritic cells and the IFN-γ response in DC-T cells (g) shown in Example 8. Based on... Figure 8 (g) It can be seen that the CD86 positivity rate of blank cells is approximately 2.64%, while that of cells treated with sq1, sq2, sq3, and sq4 is approximately 12.2%, 23.8%, 33.2%, and 26.8%, respectively. When IFN-γ ELISA was performed using the treated antigen-presenting cells, it was observed that sq1-sq4 cells could induce a higher response than the blank cells.

[0111] Validation Example 9: Prophylactic Immunization of HLA-A*24:02 Transgenic Mice This validation example used HLA-A*24:02 transgenic C57BL / 6 mice [Beijing Biocytogen Biotechnology Co., Ltd., B-HLA-A24.02], randomly divided into 7 groups: PBS, empty LNP, sq1, sq2, sq3, sq4, and a normal C57BL / 6J control, with 6 mice in each group. On days 0, 7, and 14, 50 μL (mRNA dose 30 μg) was injected intramuscularly into the quadriceps femoris muscle. On day 21, the mice were sacrificed, and spleen and peripheral blood were collected. Flow cytometry was performed using a BD LSRFortessa X-20 flow cytometer, and analysis was performed using FlowJo v10.8 software. Antibodies used for surface staining included anti-CD45, CD3, CD4, CD8α, CD44, CD62L, CD69, CD86, and CD83 (BioLegend). Animal experiments were approved by the Experimental Animal Management and Welfare Ethics Committee of Beijing Huayuan Times Biotechnology Co., Ltd. (Approval No. HYSD2025-08).

[0112] Figure 9 The timeline of the immunization regimens for HLA-A*24:02 transgenic mice on days 0, 7, and 14 (a), the representative flow cytometry diagram of CD8-positive T cell expansion on day 21 (bc), the CD44-positive / CD62L-negative effect memory phenotype, and the IFN-γ ELISpot diagram of spleen cells (d).

[0113] Flow cytometry showed that CD8-positive cells accounted for approximately 3.9% and 3.6% of CD3-positive T cells in the PBS and empty LNP groups, respectively; while in the sq1, sq2, sq3, and sq4 groups, the proportions were approximately 33.8%, 32.0%, 52.0%, and 58.5%, respectively. Approximately 91%-96% of the expanded CD8-positive cells exhibited a CD44-positive, CD62L-negative effector memory phenotype. An increase in CD4-positive T cells was also observed in the sq3 group (approximately 18.4% ± 3.2%, compared to approximately 5.1% ± 1.8% in the PBS group). After immunization, spleen cells were pooled or restimulated with a single epitope for 48 hours before IFN-γ ELISApot treatment; all groups from sq1 to sq4 showed specific responses to the corresponding peptides of the constructs.

[0114] Validation Example 10: PDAC tumor-bearing model and adoptive immune cell transfer A subcutaneous S2-VP10 xenograft model was established using NOD / SCID mice [Beijing Huayuan Times Biotechnology Co., Ltd.]. On day 1, approximately 2.5 × 10⁻⁶ mice were subcutaneously inoculated. 6 One S2-VP10 cell. HLA-A*24:02 transgenic mice were separately immunized with sq1-sq4 or a control, and spleen cells were isolated. Approximately 2 × 10⁶ S2-VP10 cells were intravenously infused into recipient mice on days 5, 8, and 11 post-tumor implantation. 7 One corresponding donor spleen cell. Measure the tumor's long and short diameters every 2-3 days, and calculate the volume as (long diameter × short diameter). Calculations were performed; tumor volume and weight were analyzed as endpoints on day 25, while another batch of animals was used for survival observation until day 50. The tumor growth analysis group had n=5 per group, and the survival analysis group had n=8 per group. Tumor volume and endpoint tumor weight data are expressed as mean ± SD.

[0115] Figure 10 Timeline (a), relative tumor volume (b), endpoint tumor weight (c), survival curve (d), and tumor CD8 immunohistochemical results (e) for adoptive transfer of spleen cells induced by vaccine in S2-VP10 tumor-bearing mice.

[0116] according to Figure 10 The results showed that, compared with PBS and the unstimulated control, tumor growth was slower in the sq1-sq4 immune cell metastasis groups, and the tumor weight at day 25 in the sq4 group was approximately 50% lower than that in the PBS group. In independent survival batches, median survival was extended from approximately 21 days to 52 days, with a hazard ratio of approximately 0.35 (P=0.003). Anti-CD8α staining of tumor tissue showed increased infiltration of CD8-positive cells in the vaccine groups, particularly in the sq3 and sq4 groups. This model directly demonstrates the anti-tumor effect of vaccine-induced immune cells.

[0117] The above epitope identification process is repeatable. Specifically, the epitopes of this invention can be obtained in the following order: sample quality control → long read library construction and CCS → full-length non-chimeric read identification → genome alignment and transcript folding → SQANTI3 classification and filtering → short read quantification and splicing verification → ORF prediction → LRDB establishment → HLA-I immunoaffinity enrichment → LC-MS / MS → 1% FDR retrieval → HLA attribution → binding and immunogenicity ranking → normal tissue exclusion → peptide synthesis → PBMC / animal function verification.

[0118] To reduce false discoveries caused by database expansion, the number of new sequences added at each step should be recorded, the target-decoy method should be used to control FDR, and proteins supported by only a single peptide should be manually verified by spectrograms. For peptides derived from new isoforms, it is preferable that the peptide crosses a new splice link or that its complete sequence does not exist in the reference proteome.

[0119] Mass spectrometry validation of candidate peptides includes at least two of the following: precursor mass error, b / y ion coverage, retention time consistency, and co-elution with synthetic standards. Parallel reaction monitoring or targeted mass spectrometry confirmation using relabeled peptides is preferred. Functional validation of candidate peptides includes at least one of the following: healthy donor PBMCs, patient PBMCs, HLA-matched transgenic mice, peptide-MHC multimers, intracellular cytokines, cytotoxic cells, or TCR reporter cells. Two complementary methods are preferred. Epitopes obtained through the above process can be used to construct shared vaccines, personalized vaccines, peptide-MHC detection reagents, TCR screening reagents, antigen-presenting cells, and T cell products. The overall process of this invention is as follows: Figure 11 As shown, Figure 11 This is a flowchart illustrating the discovery, validation, and application of HLA-I epitopes, sequentially including sample quality control and long-read sequencing, new transcript identification and splicing validation, ORF prediction and LRDB establishment, HLA-I enrichment and LC-MS / MS detection, 1% FDR search, candidate epitope screening and synthesis, immune function validation, and downstream applications such as vaccines, diagnostic reagents, and cell products. The diagram below indicates database control requirements (target-bait search, manual spectral verification) and the dual mass spectrometry and functional validation requirements for selecting novel isoform-specific peptides.

[0120] Statistical Analysis: The experimental data of each validation case in this application were statistically analyzed using R (v4.2.0) and GraphPad Prism (v9.3.1) software. Comparisons between two groups were performed using the two-tailed Student's t-test or the Mann-Whitney U test; comparisons among multiple groups were performed using one-way ANOVA (Tukey's HSD post-hoc test) or the Kruskal-Wallis test; survival curve comparisons were performed using the log-rank test; multiple test corrections were performed using the Benjamini-Hochberg method; MHC-I immune peptide enrichment analysis was performed using the two-tailed Fisher's exact test (odds ratio and 95% confidence interval were calculated). Data Presentation: ELISpot and in vitro functional experimental data are expressed as mean ± SEM; animal tumor growth and survival data are expressed as mean ± SD. Significance Level: ns indicates P ≥ 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

[0121] In summary, the present invention has the following technical effects: 1. Compared to strategies that rely solely on reference protein databases, the identification process employed in this invention incorporates aberrant splicing and non-classical open reading frames, reducing systematic missed detections of tumor-specific peptides. A total of 22,784 high-confidence isoforms were established in the PDAC system, of which 10,497 were novel isoforms (approximately 46.1%), achieving detection support of approximately 40%-80% in six independent cohorts. Compared to using only a standard reference protein database, the identification process of this invention additionally covers approximately 46.1% of high-confidence transcript isoforms, expanding the source space of candidate epitopes.

[0122] 2. In the PBMCELISpot experiment with healthy HLA-A*24:02 matched donors, 16 out of 64 preferred candidate peptides reached the high-confidence immune response threshold (positive rate 25.0%). The inclusion of new isoforms and non-classical sources brought a patient coverage gain of approximately 16.7%-44.6% compared to using only known transcripts. It is these "extra" epitopes that may become the only immunotherapy targets for some patients.

[0123] 3. The multi-epitope mRNA-LNP provided by this invention can induce an immune response dominated by effector memory CD8-positive T cells. In HLA-A*24:02 transgenic mice, CD8-positive cells accounted for approximately 58.5% of CD3-positive T cells in the sq4 group; in the PDAC tumor-bearing adoptive metastasis model, the median survival was extended from approximately 21 days in the PBS group to approximately 52 days (hazard ratio approximately 0.35, P=0.003), and tumor CD8-positive T cell infiltration increased. These data directly support the anti-tumor effect of vaccine-induced immune cells.

[0124] Table 1

[0125] Complete IVT plasmid sequence: The complete sq1 IVT plasmid sequence is shown in SEQ ID NO:29, with a length of 3538 bp and 4 novel transcript-derived epitopes (SEQ ID NO:1-4). The complete sq2 IVT plasmid sequence is shown in SEQ ID NO:30, with a length of 3766 bp and 8 epitopes (SEQ ID NO:1-8, including novel and known transcript sources): The complete sq3 IVT plasmid sequence is shown in SEQ ID NO:31, with a length of 3760 bp and 8 non-classical epitopes (SEQ ID NO:9-16): The complete sq4 IVT plasmid sequence is shown in SEQ ID NO:32, with a length of 4459 bp and 20 epitopes (SEQ ID NO:1-4+9-16+17-24, which contains a combination of the epitopes of this invention and the known HLA-A*24:02 tumor-related epitopes).

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. All such modifications or substitutions should be covered within the protection scope of this application, and should not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A pancreatic ductal adenocarcinoma-associated HLA-I epitope peptide, characterized in that, The amino acid sequence of the epitope peptide is shown in any one of SEQ ID NO:1-16; Alternatively, it may be a variant based on any of the sequences in SEQ ID NO:1-16, with conserved amino acid substitutions, while retaining HLA-A*24:02 binding activity and CD8-positive T cell recognition function.

2. The epitope peptide according to claim 1, characterized in that, The epitope peptide is derived from pancreatic ductal adenocarcinoma cells, and its sources include: novel transcripts, known transcripts, and non-classical sources; Among them, SEQ ID NO:1-4 are derived from novel transcripts, SEQ ID NO:5-8 are derived from known transcripts, and SEQ ID NO:9-16 are derived from non-classical sources; The non-classical sources include aberrant splicing, non-classical open reading frames, pseudogenes, non-coding RNA translation regions, frameshifts, upstream or overlapping open reading frames.

3. A multi-epitope polypeptide, characterized in that, The multi-epitope polypeptide comprises at least two epitope peptide modules selected from SEQ ID NO:1-16, and adjacent epitope peptide modules are connected by flexible linker peptides.

4. The multi-epitope polypeptide according to claim 3, characterized in that, The multi-epitope polypeptide comprises at least one of the epitope peptides, and at least one known HLA-A*24:02 tumor-associated epitope module selected from SEQ ID NO:17-24; The epitope peptide and the HLA-A*24:02 tumor-associated epitope module can form a peptide-MHC complex.

5. The multi-epitope polypeptide according to claim 3 or 4, characterized in that, The amino acid sequence of the multi-epitope polypeptide is shown in any one of SEQ ID NO:25-28; Alternatively, a variant that has at least 85% sequence identity with any of SEQ ID NO:25-28 and retains the HLA binding activity and immunogenicity of each epitope module.

6. The multi-epitope polypeptide according to any one of claims 3-5, characterized in that, The flexible linker peptide is a linker peptide rich in glycine and serine. The multi-epitope polypeptide also includes a signal peptide at the N-terminus and a transmembrane anchoring domain associated with MHC-I class molecule transport at the C-terminus.

7. An isolated nucleic acid molecule, characterized in that, The nucleic acid molecule encodes the epitope peptide according to any one of claims 1-2 or the multi-epitope polypeptide according to any one of claims 3-6.

8. The nucleic acid molecule according to claim 7, characterized in that, The nucleic acid molecule is a messenger RNA, which sequentially comprises a 5′ cap structure, a 5′ untranslated region, an open reading frame, a 3′ untranslated region, and a poly(A) tail; the messenger RNA is modified with N1-methylpseudouridine and / or 5-methylcytidine. The 5′ cap structure is a CleanCap AG cap structure.

9. A delivery system, characterized in that, It comprises the nucleic acid molecule and lipid nanoparticles as described in claim 7 or 8; The lipid nanoparticles contain ionizable lipids, cofactor phospholipids, cholesterol, and PEGylated lipids; The lipid nanoparticles have an average particle size of 125-140 nm and an encapsulation efficiency of >90%.

10. A pharmaceutical composition, characterized in that, It comprises the epitope peptide of any one of claims 1-2, the multi-epitope polypeptide of any one of claims 3-6, the nucleic acid molecule of any one of claims 7-8, or the delivery system of claim 9, and a pharmaceutically acceptable carrier.

11. The use of the epitope peptide according to any one of claims 1-2, the multi-epitope polypeptide according to any one of claims 3-6, the nucleic acid molecule according to any one of claims 7-8, the delivery system according to claim 9, or the pharmaceutical composition according to claim 10 in the preparation of a medicament for the prevention and / or treatment of pancreatic ductal adenocarcinoma, characterized in that, The drug is used to induce or amplify antigen-specific CD8-positive T cells, effector memory T cells, and / or tumor-infiltrating T cells.

12. The application according to claim 11, characterized in that, It also includes the combined use of the aforementioned drug as an immune checkpoint inhibitor, chemotherapy, radiotherapy, cell therapy, or TCR engineered cells.

Images