PREDICTING T CELL EPITOPES USEFUL FOR VACCINATION

Abstract

The present invention relates to methods for predicting T cell epitopes useful for vaccination. In particular, the present invention relates to methods for predicting whether modifications in peptides or polypeptides such as tumor-associated neoantigens are immunogenic and, in particular, useful for vaccination, or for predicting which of such modifications are most immunogenic and, in particular, most useful for vaccination. The methods of the invention may be used, in particular, for the provision of vaccines which are specific for a patient's tumor and thus, in the context of personalized cancer vaccines.

Claims

1.-29. (canceled)

30. A method for predicting immunogenicity of an amino acid modification, the method comprising steps of: a) ascertaining a score for binding of a modified peptide comprising a fragment of a modified protein to one or more MHC class II molecules (“its MHC class II binding score”); b) ascertaining a score for expression or abundance (its “abundance score”) of the modified protein; and c) predicting immunogenicity of the amino acid modification based on the score for binding of the modified peptide to the one or more MHC class II molecules and the score for expression or abundance of the modified protein.

31. The method of claim 31, wherein predicting immunogenicity of an amino acid modification comprises predicting whether a peptide comprising an amino acid modification will be immunogenic.

32. A method for selecting and/or ranking immunogenic amino acid modifications, the method comprising steps of: a) ascertaining a score for binding of a modified peptide comprising a fragment of a modified protein to one or more MHC class II molecules (“its MHC class II binding score”); b) ascertaining a score for expression or abundance (its “abundance score”) of the modified protein; c) performing steps a) and b) for two or more different immunogenic amino acid modifications; and d) selecting and/or ranking the two or more different immunogenic amino acid modifications by comparing the MHC class II binding scores of the corresponding modified peptide and the abundance scores for the corresponding modified protein.

33. The method of claim 32, wherein selecting and/or ranking the two or more different immunogenic amino acid modifications comprises a selection and/or ranking of peptides comprising the two or more different immunogenic amino acid modifications for their immunogenicity.

34. The method of claim 32, wherein comparing the MHC class II binding scores and the abundance scores comprises ranking the two or more different amino acid modifications by their MHC class II binding scores and removing amino acid modifications for which the corresponding modified protein has an expression or abundance of less than a given threshold.

35. The method of claim 30, wherein the modified protein comprises a tumor-associated neoantigen of a patient.

36. The method of claim 35, wherein the tumor-associated neoantigen is a peptide or protein comprising a tumor-specific amino acid modification.

37. The method of claim 30, wherein the MHC class II binding score reflects a probability for binding of the modified peptide to the one or more MHC class II molecules.

38. The method claim 30, wherein the abundance score is based on a score for level of expression of a protein to which the one or more amino acid modifications are associated and a score for variant allele frequency for the modified protein.

39. The method of claim 38, wherein the abundance score is determined by multiplying the score for level of expression of the protein to which the one or more amino acid modifications are associated with the score for variant allele frequency of the modified protein.

40. The method of claim 38, wherein the score for variant allele frequency is determined from a sum of detected sequence reads covering mutation site(s) corresponding to the amino acid modification(s) and carrying the amino acid modification(s) divided by a sum of all detected sequence reads covering the mutation site(s).

41. The method of claim 38, wherein the score for variant allele frequency is determined from a sum of mutated nucleotides detected at the mutation site(s) corresponding to the one or more amino acid modifications divided by a sum of all nucleotides detected at the mutation site(s).

42. The method of claim 30, wherein two or more different amino acid modifications are present in the same and/or in different modified proteins.

43. The method of claim 30, further comprising performing step a) on two or more different modified peptides, said two or more different modified peptides comprising the same amino acid modification(s).

44. The method of claim 43, wherein the two or more different modified peptides comprising the same amino acid modification(s) comprise different fragments of a modified protein, said different fragments comprising the same amino acid modification(s) present in the modified protein.

45. The method of claim 44, wherein the two or more different modified peptides comprising the same amino acid modification(s) comprise different potential MHC class II binding fragments of a modified protein, said fragments comprising the same amino acid modification(s) present in the modified protein.

46. The method of claim 43, further comprising selecting the modified peptide(s) from the two or more different modified peptides comprising the same amino acid modification(s) having a probability or having the highest probability for binding to one or more MHC class II molecules.

47. The method of claim 43, wherein the two or more different modified peptides comprising the same amino acid modification(s) differ in length and/or position of the amino acid modification(s).

48. The method of claim 43, wherein a highest score for binding to one or more MHC class II molecules of the two or more different modified peptides comprising the same amino acid modification(s) is assigned to the amino acid modification(s).

49. The method of claim 38, further comprising a step of determining a score for level of expression (an “RNA expression score”) of the RNA encoding the protein to which the one or more amino acid modifications are associated, wherein is determined by measuring the level of expression of RNA encoding the protein.

50. The method of claim 38, further comprising a step of determining a score for variant allele frequency (a “variant allele frequency score”) for the modified protein, wherein the score is determined by measuring the level of expression of RNA encoding the modified protein.

51. The method of claim 30, wherein the modified peptide comprises a fragment of the modified protein, said fragment comprising the amino acid modification(s) present in the modified protein.

52. The method of claim 30, further comprising identifying non-synonymous mutations in one or more protein-coding regions of a transcript corresponding to the modified protein.

53. The method claim 30, wherein amino acid modifications are identified by partially or completely sequencing the genome, exome, or transcriptome of one or more cells.

54. The method of claim 30, wherein mutation(s) corresponding to the amino acid modification(s) are somatic mutation(s).

55. The method of claim 54, wherein said mutation(s) are cancer mutation(s).

56. The method of claim 30, further comprising a step of manufacturing a vaccine that delivers one or more peptides or polypeptides comprising the one or more amino acid modifications predicted as being immunogenic or more immunogenic, and/or one or more modified peptides predicted as being immunogenic or more immunogenic.

57. The method of claim 56, wherein the vaccine is a nucleic acid vaccine comprising a nucleic acid encoding the one or more peptides or polypeptides comprising the one or more amino acid modifications predicted as being immunogenic or more immunogenic, and/or one or more modified peptides predicted as being immunogenic or more immunogenic.

58. The method of claim 56, wherein the vaccine is a peptide or polypeptide vaccine comprising one or more peptides or polypeptides comprising the one or more amino acid modifications predicted as being immunogenic or more immunogenic, and/or one or more modified peptides predicted as being immunogenic or more immunogenic.

59. The method of claim 58, wherein the step of manufacturing comprises in vitro transcription to generate an RNA encoding the one or more peptides or polypeptides comprising the one or more amino acid modifications predicted as being immunogenic or more immunogenic, and/or one or more modified peptides predicted as being immunogenic or more immunogenic.

60. A method for manufacturing a vaccine, the method comprising steps of: a) detecting a mutation corresponding to a non-synonymous amino acid modification by partial or complete sequencing of the genome, exome, or transcriptome of one or more cells from a subject; b) predicting one or more immunogenic amino acid modifications or one or more immunogenic modified peptides by the method of claim 30; and c) manufacturing a vaccine comprising one or more peptides or polypeptides comprising the one or more immunogenic amino acid modifications predicted as being immunogenic or more immunogenic or comprising the one or more immunogenic modified peptides predicted as being immunogenic or more immunogenic, or a nucleic acid encoding the one or more peptides or polypeptides.

61. The method of claim 60, wherein the one or more peptides or polypeptides comprises one or more neo-epitopes or T-cell epitopes.

62. The method of claim 60, wherein the one or more neo-epitopes or T-cell epitopes each comprises one or more tumor-specific amino acid modifications.

63. The method of claim 60, wherein the one or more peptides or polypeptides further comprises one or more epitopes that do not comprise tumor-specific amino acid modifications.

64. The method of claim 62, wherein the one or more neo-epitopes or T-cell epitopes are separated by linkers.

65. The method of claim 60, wherein the vaccine provides WIC class II-presented epitopes that are capable of eliciting a CD4.sup.+ helper T cell response against cells expressing antigens from which the WIC class II-presented epitopes are derived.

66. The method of claim 60, wherein the vaccine provides WIC class I-presented epitopes that are capable of eliciting a CD8.sup.+ T cell response against cells expressing antigens from which the WIC class I-presented epitopes are derived.

67. The method of claim 60, wherein the nucleic acid is ribonucleic acid (RNA).

68. The method of claim 67, wherein the RNA has been modified to increase stability or expression.

69. A vaccine comprising: a peptide or polypeptide comprising one or more amino acid modification(s) predicted as being immunogenic or more immunogenic by the method of claim 30, or a nucleic acid encoding said peptide or polypeptide.

70. The vaccine of claim 69, wherein the vaccine is specific for a subject's tumor.

71. The vaccine of claim 69, wherein the peptide or polypeptide comprises one or more neo-epitopes or T-cell epitopes.

72. The vaccine of claim 71, wherein the one or more neo-epitopes or T-cell epitopes each comprise one or more tumor-specific amino acid modifications.

73. The vaccine of claim 69, wherein the peptide or polypeptide further comprises one or more epitopes that do not comprise tumor-specific amino acid modifications.

74. The vaccine of claim 71, wherein the one or more neo-epitopes or T-cell epitopes are separated by linkers.

75. The vaccine of claim 69, wherein the vaccine provides MHC class II-presented epitopes that are capable of eliciting a CD4.sup.+ helper T cell response against cells expressing antigens from which the MHC class II-presented epitopes are derived.

76. The vaccine of claim 69, wherein the vaccine provides MHC class I-presented epitopes that are capable of eliciting a CD8.sup.+ T cell response against cells expressing antigens from which the MHC class I-presented epitopes are derived.

77. The vaccine of claim 69, wherein the nucleic acid is ribonucleic acid (RNA).

78. The vaccine of claim 77, wherein the RNA has been modified to increase stability or expression.

79. The vaccine of claim 69, wherein the vaccine further comprises a pharmaceutically acceptable carrier.

80. The vaccine of claim 69, wherein the vaccine further comprises one or more adjuvants or stabilizers.

81. A method for analyzing and selecting modified peptides for use in producing a vaccine, the method comprising: employing a computer-based analytical process comprising predicting, ranking, and/or selecting one or more immunogenic amino acid modifications in a protein by the method claim 30.

82. The method of claim 81, wherein the protein comprises a tumor-associated neoantigen of a patient.

83. A method of manufacturing a vaccine, the method comprising: a) performing the method of claim 82; and b) manufacturing a vaccine comprising a peptide or polypeptide comprising the one or more immunogenic amino acid modifications or modified peptides predicted as being immunogenic or more immunogenic, or a nucleic acid encoding the peptide or polypeptide.

84. A method for manufacturing a vaccine, the method comprising steps of: a) detecting a mutation corresponding to a non-synonymous amino acid modification by partial or complete sequencing of the genome, exome, or transcriptome of one or more cells from a subject; b) predicting one or more immunogenic amino acid modifications in a protein comprising a tumor-associated neoantigen by the method of claim 30; and c) manufacturing a vaccine comprising a peptide or polypeptide comprising the one or more immunogenic amino acid modifications or modified peptides predicted as being immunogenic or more immunogenic, or a nucleic acid encoding the peptide or polypeptide.

85. The method of claim 84, wherein the vaccine comprises a synthetic mRNA vaccine encoding one or more neo-epitopes or T-cell epitopes.

Description

FIGURES

[0245] FIG. 1. Non synonymous cancer-associated mutations are frequently immunogenic and pre-dominantly recognized by CD4.sup.+ T cells. a, For immunogenicity testing, mice (n=5 for b and c, n=3 for d) were vaccinated with either synthetic peptides and poly (I:C) as adjuvant (b) or antigen-encoding RNA (c, d) representing the mutated epitopes (two mutations per mouse). Splenocytes were restimulated ex vivo with the mutated peptide or an irrelevant control peptide and tested by IFNγ Elispot (see exemplarily FIG. 2a) and intracellular cytokine and CD4/CD8 surface staining to assess subtype of elicited immune responses. b-c, T cell responses obtained by vaccinating C57BL/6 mice with epitopes mutated in the B16F10 tumor model. Left, prevalence of non-immunogenic, MHC class I or class 11 restricted mutated epitopes. Right, examples for detection and typing of mutation-specific T cells (see Table 1 for data on individual epitopes). d prevalence of non-immunogenic, MHC class I or class II restricted mutated epitopes discovered in the CT26 model. Right, MHC restriction of immunogenic mutated epitopes prioritized based on predicted MHC class I binding and selected based on either good (0.1-2.1) or poor (>3.9) binding scores. See Table 2 for data on individual epitopes. Sequences in a (“Sequence analysis and mutation identification”): TTCAGGACCCA (SEQ ID NO: 93); TTCAGGACCCACACGA (SEQ ID NO: 94); TTCAGGACCCACACGACGGGAAGACAA (SEQ ID NO: 95); TTCAGGACCAACACGACGGGAAGACAAGT (SEQ ID NO: 96); CAGGACCCACACGACGGGTAGACAAGT (SEQ ID NO: 97); ACCCACACGACGGGTAG ACAAGT (SEQ ID NO: 98); ACCCACACGAGCCCTAGACAAGT (SEQ ID NO: 99); GACGGGAAGACAAGT (SEQ ID NO: 100). Sequences in b and c: B16-M27 (SEQ ID NO: 10); B16-M30 (SEQ FD NO: 13).

[0246] FIG. 2. Efficient tumor control and survival benefit in B16F10 melanoma by immunization with an RNA vaccine encoding a single mutated CD4.sup.+ T cell epitope. a, Splenocytes of mice (n=5) vaccinated with B16-M30 RNA were tested by ELISpot for recognition of synthetic peptides. Left, the mutated (B16-M30) versus the corresponding wild type (B16-WT30) sequence. Right, definition of the minimal epitope by testing for recognition of truncated variants of B16-M30 (mean±SEM). b, The mean±SEM tumor growth (left) and survival (right) of C57BL/6 mice (n=10) inoculated subcutaneously with B16F10 tumors cells and left untreated (control) or immunized IV with B16-M30 encoding RNA (B16-M30) with or without administration of CD4 or CD8 depleting antibodies. c, B6 albino mice (n=10) developing lung metastases upon IV injection of luciferase transgenic B16F10 tumor cells (B16F10-Luc) were treated with B16-M30 encoding RNA (B16-M30) or irrelevant control RNA. Median tumor growth was determined by BLI. d, Single cell suspensions of B16F10 tumors of untreated (control, n=x) or B16-M30 RNA immunized mice (n=4) were restimulated with B16-M30 peptide, medium or irrelevant peptide (VSV-NP.sub.52-59) and tested in an IFNγ ELISpot assay (mean±SEM). e, Flow cytometric characterization of tumor infiltrating leucocytes in B16-M30 RNA vaccinated mice. Depicted is the frequency of CD4.sup.+, CD8.sup.+ or FoxP3.sup.+/CD4.sup.+ T cells among CD45.sup.+ cells and Gr-1.sup.+/CD11b.sup.+ cells (MDSCs) of untreated (control) or Mut30 RNA vaccinated C57BL/6 mice (n=3) inoculated subcutaneously with B16F10 tumors cells. Sequences in a: B16-M30 (SEQ ID NO: 13); DWENVSPELNSTDQP (SEQ ID NO: 80); DWE NVSPELNSTDQ (SEQ ID NO: 81); DWENVSPELNSTD (SEQ ID NO: 82); DWENVSPELNS T (SEQ ID NO: 83); DWENVSPELNS (SEQ ID NO: 84); WENVSPELNSTDQP (SEQ ID NO: 85); WENVSPELNSTD (SEQ ID NO: 86); WENVSPELNST (SEQ ID NO: 87); ENVSPELNS TDQP (SEQ ID NO: 88); NVSPELNSTDQP (SEQ ID NO: 89); VSPELNSTDQP (SEQ ID NO: 90).

[0247] FIG. 3. Immunization with RNA pentatopes induces T cell responses against the individual mutated epitopes and confers disease control and significant survival benefit in mouse tumor models. a, Engineering of a poly-neo-epitope RNA vaccine. The RNA pentatope contains five 27mer sequences connected by gly/ser linkers inserted into the pST1-Sp-MITD-2hBgUTR-A120 backbone. (UTR, untranslated region; sp, signal peptide; MITD, WIC class I trafficking domain). b, BALB/c mice (n=5) were vaccinated either with pentatope RNA (35 μg) or the corresponding mixture of five RNA monotopes (7 μg each). T cell responses in peptide stimulated splenocytes of mice were measured ex vivo on day 19 in an IFNγ ELISpot assay (medium control subtracted mean±SEM). c, BALB/c mice (n=10) developing lung metastases upon IV injection of CT26-Luc cells were treated simultaneously with a mixture of two RNA pentatopes or left untreated (control). The median tumor growth by BLI (left), survival data (mid) and lungs from treated animals (right) are shown. d, CD3 stained tissue sections from the lungs of pentatope1+2 treated animals (upper panel). The left side of each panel shows the analyzed sections, the right side the magnifications (scale bar: scan: 1000 μm, upper pictures: 100 μm, lower pictures: 50 μm). CD3.sup.+, CD4.sup.+, FoxP3.sup.+ and CD8.sup.+ (calculated by CD3.sup.+ area-CD4.sup.+ area) areas in consecutive immunohistochemical lung tissue sections of control (n=6) or RNA pentatope (CD3: n=14; CD4, CD8, FoxP3: n=12) treated animals were quantified and proportions of tumor were calculated. The right figure depicts a comparison of tumor area in sections of control (n=18) and Pentatope 1+2 (n=39) treated animals (tumor free animals of pentatope1+2 treatment group were excluded). Depicted are mean±SEM. Sequences in a (“Cloning of template”): GGAAACTTTC (SEQ ID NO: 105).

[0248] FIG. 4. RNA pentatope vaccines with mutations selected for in silico predicted favorable MHC class II binding properties and abundant expression confer potent antitumor control. a, Comparison of MHC II binding scores of immunogenic and non-immunogenic mutations (medians shown). b, Mutations with high expression levels were selected with (‘ME’ mutations) or without (‘E’ mutations) considering MHC class II binding score. See also Table 4. Ten mutations out of each category represented by two pentatopes each were used for vaccination of CT26-Luc lung tumor bearing mice. Tumor growth curves (left), area under the curve (mid) and ink treated lungs (right) are shown. c, Mice (5 per group) were analyzed for T cell responses against the vaccinated pentatopes by restimulation with RNA electroporated syngeneic BMDC in an IFNγ ELISpot assay. Each dot represents the mean spot count of one mouse subtracted by an irrelevant RNA control (mean±SEM). d, Tumor nodules per lung of BALB/c mice (n=10) inoculated IV with CT26 tumor cells and left untreated or injected with irrelevant RNA, pentatope1, pentatope2 or CT26-M19 RNA. e, T cell responses against gp70.sub.423-431 (gp70-AH1) were determined via IFNγ ELISpot assay in blood (pooled from 5 mice, day 20 after tumor inoculation) and spleen (n=5). (Background (no peptide control) subtracted mean±SEM depicted). f, Somatic mutation and RNA-Seq data for individual human cancer samples (black dots) from The Cancer Genome Atlas (TCGA) was employed to identify genomic (upper panel) and expressed (mid panel) non-synonymous single nucleotide variations (nsSNVs). (lower panel) Neo-epitopes predicted to bind to the patients' HLA-DRB1 alleles (percentile rank <10%) are shown (SKCM, skin cutaneous melanoma; COAD, colon adenocarcinoma; BRCA, breast invasive carcinoma).

[0249] FIG. 5: Calculation of variant allele frequency (VAF). The figure shows an idealized gene as a combination of exons on a piece of genomic DNA (upper part) and example read sequences aligned to this locus (lower part, in a higher zoom level). The site of the mutation event (“mutation site”) is shown by a dashed line (upper part) or box (lower part). The mutant nucleotides are colored red, the wild type nucleotides are colored green. Also the sums of those nucleotides in the VAF formula are colored accordingly. Sequences in a: TGCAAGAACGCGT ACTTATTCGCCGCCATGATTATGACCAGTGTTTCCAGTC (SEQ ID NO: 101); CAAGAA CGCGTACTTATTCGCCACCATGATTATGACCAGTGTTTCCAG (SEQ ID NO: 102); AAC GCGTACTTATTCGCCACCATGATTATGACCAGTGTTTCCAGTC (SEQ ID NO: 103); TG CAAGAACGCGTACTTATTCGCCGCCATGATTATGACCAGTGTTT (SEQ ID NO: 104).

[0250] FIG. 6: Influence of the expression of mutated allele on the prediction performance of MHC II-scores. 185 selected mutations from the murine tumor models 4T1, CT26 and B16F10 were tested for their antigenicity. The predictive performance of the calculated MHC II-scores was deduced from the area under the receiver operating characteristic curve (AUC, open circle). This value was subsequently recalculated after applying different thresholds for the total mRNA expression (left panel) and the expression of the mutated allele (right panel, mRNA expression*mutated allele frequency, closed circles). The maximum AUC values are indicated. The expression of the mutated allele contributes more to the improvement of the prediction performance.

[0251] FIG. 7: Comparison of receiver operating characteristic (ROC) curves with and without threshold on the expression. The ROC curves indicate the performance of the antigenicity prediction for all 185 selected mutations from the murine tumor models 4T1, CT26 and B16F10 (dotted curves) and for those mutations, for which the mRNA expression was ≥6 RPKM (left panel, solid curve) or the expression of the mutated allele was ≥4 RPKM (right panel, solid curve). The selected thresholds achieved the maximum AUC values (see FIG. 6).

EXAMPLES

[0252] The techniques and methods used herein are described herein or carried out in a manner known per se and as described, for example, in Sambrook et al., Molecular Cloning: A Laboratory Manual, 2.sup.nd Edition (1989) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. All methods including the use of kits and reagents are carried out according to the manufacturers' information unless specifically indicated.

Example 1: Materials and Methods

[0253] Samples. Female 8-12 week old C57BL/6, BALB/c mice (Janvier Labs) and C57BL/6BrdCrHsd-Tyr.sup.c mice (B6 albino, Harlan) were kept in accordance with federal policies on animal research at the University of Mainz. B16F10 melanoma cell line, CT26 colon carcinoma cell line and 4T1-luc2-tdtomato (4T1-Luc) cells were purchased in 2010, 2011 and 2011 respectively (ATCC CRL-6475 lot #58078645, ATCC CRL-2638 lot #58494154, Caliper 125669 lot #101648) and maintained as suggested by the supplier. Firefly luciferase expressing CT26-Luc and B16F10-Luc cells were lentivirally transduced. Master and working cell banks were generated, of which third and fourth passages were used for tumor experiments.

[0254] Next generation sequencing and data processing was described previously (Castle, J. C., et al., Cancer Res 72, 1081 (2012); Castle, J. C., et al., BMC Genomics 15, 190 (2014)). In brief, exome capture from mouse tumor cells and tail tissue samples of BALB/c or C57BL/6 mice were sequenced in triplicate (4T1-Luc in duplicate). Oligo(dT) based RNA sequencing libraries for gene expression profiling were prepared in triplicate. Libraries were sequenced on an Illumina HiSeq2000 to generate 50 nucleotide single-end (B16F10) or 100 nucleotide paired-end (CT26, 4T1-Luc) reads, respectively. Gene expression values were determined by counting reads overlapping transcript exons and junctions, and normalizing to RPKM expression units (Reads which map per kilobase of transcript length per million mapped reads). Mutation expression was determined by normalization of mutated RNA reads to the total mapped read counts multiplied by 100 million (normalized variant read counts; NVRC).

[0255] Mutation selection, validation and prioritization was described previously (Castle, J. C., et al., Cancer Res 72, 1081 (2012); Castle, J. C., et al., BMC Genomics 15, 190 (2014); Lower, M., et al., PLoS Comput Biol 8, e1002714 (2012)). Mutations to be pursued were selected based on following criteria: (i) present in the respective tumor cell line sequencing triplicates and absent in the corresponding healthy tissue sample triplicates, (ii) occur in a RefSeq transcript, and (iii) cause non-synonymous changes. Further criteria were occurrence in expressed genes of tumor cell lines (median RPKM across replicates). For validation, mutations were amplified from DNA of B16F10, CT26 or 4T1-Luc cells and C57BL/6 or BALB/c tail tissue and subjected to Sanger sequencing. DNA-derived mutations were classified as validated if confirmed by either Sanger sequencing or the RNASeq reads. No confirmation via Sanger sequencing and immunogenicity testing was performed for experiments shown in FIG. 4. For experiments shown in FIG. 1 mutated epitopes were prioritized according to their predicted MHC class I binding based on the consensus method (version 2.5) of the Immune Epitope Database (Vita, R., et al., Nucleic Acids Res 38, D854-D862 (2010)). Mutations targeted in the experiment shown in FIG. 4b-e were selected based on either their expression (NVRC) alone or together with their predicted MHC class II peptide binding capability (IEDB consensus method version 2.5). Retrospective analysis of MHC II binding prediction shown in FIG. 4a was determined with IEDB consensus method version 2.12. For analysis of mutations in human tumors, DNA sequencing data of skin cutaneous melanoma (SKCM, n=308), colon adenocarcinoma (COAD, n=192) or breast invasive carcinoma (BRCA, n=872) retrieved from The Cancer Genome Atlas (TCGA) (august 2014) was filtered to obtain genomic non-synonymous point mutations (nsSNVs). RNA-Seq data (TCGA) of tumor samples with identified genomic mutations was used to define expressed nsSNVs. In order to predict MHC II binding expressed neo-epitopes seq2HLA was employed to identify the patients' 4-digit HLA class II (HLA-DQA1, HLA-DQB1, HLA-DRB1) type. The IEDB consensus binding prediction (version 2.12) was used to predict MHC class II binding from a 27mer peptide and the patients HLA-DRB1 alleles. As recommended from IEDB, neo-eptiopes with a percentile rank below 10% were considered as binders.

[0256] Synthetic RNA and synthetic peptides. Identified non-synonymous mutations were studied in the context of the respective 27mer amino acid epitope with the mutated amino acid in the center (position 14). Either of these mutated peptides were synthesized together with control peptides (vesiculo-stomatitis virus nucleo-protein (VSV-NP.sub.52-59), gp70-AH1 (gp70.sub.423-431) and tyrosinase-related protein 2 (Trp2.sub.180-188) by JPT Peptide Technologies GmbH. Alternatively, sequences encoding mutated 27mer peptides were cloned into the pST1-Sp-MITD-2hBgUTR-A120 backbone (Holtkamp, S., et al., Blood 108, 4009 (2006)) featuring sequence elements for pharmacologically optimized synthetic RNA in terms of translation efficiency and MHC class I/II processing of epitopes either as monotopes or as pentatopes fused to each other by sequences encoding 10 amino acid long glycine-serine linker in between. Linearization of these plasmid constructs, in vitro translation (IVT) of these templates and purification are described in detail elsewhere (Holtkamp, S., et al., Blood 108, 4009 (2006)).

[0257] Mouse Models.

[0258] For experiments investigating the immunogenicity of mutated epitopes age-matched female C57BL/6 or BALB/c mice were vaccinated on day 0, 3, 7 and 14 (immunization with RNA) or day 0 and 7 (immunization with peptide), the read out was performed five to six days after the last immunization. Vaccination was performed either by retro-orbital injection of 200 μl (20 μg per mutation for B16F10, 40 μg per mutation for CT26) RNA complexed with cationic lipids (manuscript in preparation) or subcutaneous injection of 100 μg synthetic peptide and 50 μg poly (I:C) formulated in PBS (200 μL total volume) into the lateral flank. Two mutations per mouse were tested (n=5 for B16F10, n=3 for CT26). For confirmation of immunogenic mutations and subtyping, mice were vaccinated against a single mutation (n=5).

[0259] For therapeutic tumor experiments C57BL/6 mice were inoculated subcutaneously with 1×10.sup.5 B16F10 melanoma cells into the right flank and randomly distributed into treatment groups. Tumor volume was measured unblinded with a caliper and calculated using the formula (AxB.sup.2)/2 (A as the largest and B the smallest diameter of the tumor). In lung metastasis experiments 5×10.sup.5 CT26-Luc or 2×10.sup.5 CT26 cells were injected into the tail vein of BALB/c mice or 1.5×10.sup.5 B16F10-Luc tumor cells into B6 albino mice to obtain lung tumors. Tumor growth of luciferase transgenic cells was traced unblinded by bioluminescence imaging after i.p. injection of an aqueous solution of D-luciferin (250 μl, 1.6 mg, BD Bioscience) on an IVIS Lumina (Caliper Life Sciences). Five minutes after injection emitted photons were quantified. In vivo bioluminescence in regions of interest (ROI) were quantified as total flux (photons/sec) using IVIS Living Image 4.0 software. Mice were randomized based on their total flux values (ANOVA-P method, Daniel's XL Toolbox V6.53). CT26 lung tumor burden was quantified unblinded after tracheal Ink (1:10 diluted in PBS) injection and fixation with Fekete's solution (5 mL 70% EtOH, 0.5 mL formalin, and 0.25 mL glacial acetic acid). In therapeutic experiments mice were administered repeated doses of either monotope (40 μg), pentatope RNA (in total 40 μg) or equimolar amounts of irrelevant RNA. For mechanistic studies repeated doses of CD8 depleting (clone YTS191, BioXcell), CD4 depleting (clone YTS169.1, BioXcell) or CD40L blocking (clone MR1, kind gift of Prof. Stephen Schoenberger) antibodies were administered intraperitoneally as indicated in the figure (200 μg/mouse in 2004, PBS).

[0260] Enzyme-linked immunospot (ELISpot) has been previously described (Kreiter, S., et al., Cancer Res 70, 9031 (2010)). In brief, 5×10.sup.5 splenocytes were cultured over night at 37° C. in anti-INF-y (10 μg/mL, clone AN18, Mabtech) coated Multiscreen 96-well plates (Millipore) and cytokine secretion was detected with an anti-IFN-γ antibody (1 μg/mL, clone R4-6A2, Mabtech). For stimulation either 2 μg/mL peptide was added or spleen cells were coincubated with 5×10.sup.4 syngeneic bone marrow-derived dendritic cells (BMDC) transfected with RNA. For analysis of tumor infiltrating lymphocytes, single cell suspensions of lung metastasis were rested overnight to get rid of living tumor cells via plastic adherence. Viable cells were separated via density gradient centrifugation. All retrieved cells were added to the ELISpot plate. For analysis of T cell responses in peripheral blood, PBMC were isolated via density gradient centrifugation, counted and restimulated by addition of peptide and syngeneic BMDC. Subtyping of T cell responses was performed by addition of a WIC class II blocking antibody (20 μg/ml, clone M5/114, BioXcell). All samples were tested in duplicates or triplicates.

[0261] Flow cytometric analysis was used to determine the subtype of mutation reactive T cells. In the presence of Brefeldin A (Sigma-Aldrich) 2×10.sup.6 splenocytes were stimulated with 2×10.sup.5 RNA transfected BMDC or 2 μg/mL peptide. As a positive control splenocytes were treated with phorbol 12-myristate 13-acetate (PMA, 0.5 μg/ml, Sigma-Aldrich) and Ionomycin (1 μg/ml, Sigma-Aldrich). Cells were incubated 5 h at 37° C. and subsequently stained for CD4.sup.+ and CD8.sup.+ cell surface marker. Cells were permeabilized and fixated using BD Cytofix/Cytoperm according to the manufacturer's protocol and thereafter stained for INF-γ, TNF-α and IL-2 cytokines (BD Biosciences). Cytokine secretion among CD4.sup.+ or CD8.sup.+ T cells in stimulated samples was compared to control samples (medium, irrelevant RNA or irrelevant peptide) in order to determine the responding T cell subtype (n=5). Tumor infiltrating leucocytes were prepared from subcutaneous B16F10 tumors as described previously (PMID:2071934). The resulting cell suspension was stained for CD4, CD8, Gr-1 and CD11b surface marker. Intracellular FoxP3 staining was performed according to the manufacturer's protocol (Mouse Foxp3 Buffer Set, BD). Samples were acquired on a BD FACSCanto II.

[0262] Immune histochemistry. Lungs of CT26 tumor bearing mice were fixated overnight in 4% phosphate buffered formaldehyde solution (Carl Roth) and embedded in paraffin. 50 μm consecutive sections (3 per mouse) were stained for CD3 (clone SP7, Abcam), CD4 (clone 1, cat #50134-M08H, Sino Biologinal) and FoxP3 (polyclonal, cat #NB100-39002, Novus Biologicals) following detection by a HRP-conjugated antibody (Poly-HRP-anti-rabbit IgG, ImmunoLogic) and the corresponding peroxidase substrate (Vector Nova Red, Vector Laboratories) and counterstained with hematoxylin. CD3.sup.+, CD4.sup.+, FoxP3.sup.+ and tumor areas were captured on an Axio Scan.Z1 (Zeiss) and manually pre-defined tumor and lung regions were quantified via computerized image analysis software (Tissue Studio 3.6.1, Definiens).

[0263] Immunofluorescence staining. Cryoconserved organs were cut in 8 μm sections and attached on Superfrost slides. Sections were dried overnight at room temperature (RT) and fixed in 4% para-formaldehyde (PFA) for 10 min at RT in the dark. Sections were washed 3 times with PBS and blocked using PBS supplemented with 1% BSA, 5% mouse serum, 5% rat serum and 0.02% Nonident for 1 h at RT in the dark. Fluorescent labeled antibodies (FoxP3, clone FJK-16s, eBioscience; CD8, clone 53-6.7, BD; CD4, clone RM4-5, BD) were diluted in staining buffer (PBS supplemented with 1% BSA, 5% mouse serum and 0.02% Nonident) and sections were stained overnight at 4° C. After washing twice with washing buffer (PBS supplemented with 1% BSA and 0.02% Nonident) and once with PBS, slides were stained for 3 min with Hoechst (Sigma), washed 3 times with PBS, once with distilled water and mounted using Mounting Medium Flouromount G (eBioscience). Immunofluorescence images were acquired using an epifluorescence microscope (ApoTome, Zeiss). Tumor, CD4, CD8 and FoxP3 stained areas were quantified within manually pre-defined tumor regions via computerized image analysis software (Tissue Studio 3.6.1., Definiens)

[0264] Statistics. Means were compared by using Student's t-test for two groups. For comparison of means in more than two groups one-way ANOVA with Tukey's test was applied. The area under the curve (AUC) for comparison of tumor growth dynamics was determined for single mice per group and was displayed as median. Statistical differences in medians between two groups were calculated with a nonparametric Mann-Whitney U test. Survival benefit was determined with the log-rank test. All analyses were two-tailed and carried out using GraphPad Prism 5.03. ns: P>0.05, *: P≤0.05, **: P≤0.01, ***: P≤0.001. Grubb's test was used for identification of outliers (alpha=0.05).

Example 2: MHC Class II Restricted T Cell Epitopes in Neo-Epitope Vaccines

[0265] A. Characterization of T Cell Subtypes Reactive Against Mutated Epitopes

[0266] Recently, we described a workflow for comprehensive mapping of non-synonymous mutations of the B16F10 tumor by NGS (FIG. 1a) (Castle, J. C., et al., Cancer Res 72, 1081 (2012)). Tumor-bearing C57BL/6 mice were immunized with synthetic 27mer peptides encoding the mutated epitope (mutation in position 14), resulting in T cell responses which conferred in vivo tumor control. In continuation of that work, we now characterized the T cell responses against the mutated epitopes starting with those with a high likelihood of MEW I binding. Mice were vaccinated with synthetic 27mer mutated epitope peptides (FIG. 1b upper right). Their splenocytes were tested in IFN-γ ELISpot to identify immunogenic mutations for further analysis of subtype and cytokine expression (FIG. 1a). About 30% of mutated epitopes were found to induce mutation reactive cytokine secreting T cells in mice (FIG. 1b). Surprisingly, responses against nearly all mutated epitopes ( 16/17, 95%) were of CD4.sup.+ T cell type (FIG. 1b, Table 1).

TABLE-US-00001 TABLE 1 Immunogenic B16F10 mutations. B16F10 mutations determined to be immunogenic upon peptide or RNA immunization (as described in FIG. 1). (WT, wild type; AA#, number of mutated amino acid; Mut, Mutation) Substi- Response tution Reactive MHC I score after Mu- Mutated sequence  (WT, AA#, T cell  (best pre- vaccination with tation Gene used for vaccination  Mut) subtype diction) Peptide RNA B16-M05 Eef2 FVVKAYLPVNESFAFTADLRSNTGGQA G795A CD4.sup.+ 1.1 X (SEQ ID NO: 1) B16-M08 Ddx23 ANFESGKHKYRQTAMFTATMPPAVERL V602A CD4.sup.+ 1.3 X (SEQ ID NO: 2) B16-M12 Gnas TPPPEEAMPFEFNGPAQGDHSQPPLQV S111G CD4.sup.+ 1.2 X (SEQ ID NO: 3) B16-M17 Tnpo3 VVDRNPQFLDPVLAYLMKGLCEKPLAS G504A CD4.sup.+ 1.0 X (SEQ ID NO: 4) B16-M20 Tubb3 FRRKAFLHWYTGEAMDEMEFTEAESNM G402A CD4.sup.+ 1.9 X (SEQ ID NO: 5) B16-M21 Alp11a SSPDEVALVEGVQSLGFTYLRLKDNYM R552S CD4.sup.+ 0.1 X (SEQ ID NO: 6) B16-M22 Asf1b PKPDFSQLQRNILPSNPRVTRFHINWD A141P CD4.sup.+ 1.7 X (SEQ ID NO: 7) B16-M24 Dag1 TAVITPPTTTTKKARVSTPKPATPSTD P425A CD4.sup.+ 2.2 X (SEQ ID NO: 8) B16-M25 Plod1 STANYNTSHLNNDVWQIFENPVDWKEK F530V CD4.sup.+ 0.1 X X (SEQ ID NO: 9) B16-M27 Obsl1 REGVELCPGNKYEMRRHGTTHSLVIHD T1764M CD8.sup.+ 2.3 X X (SEQ ID NO: 10) B16-M28 Ppp1r7 NIEGIDKLTQLKKPFLVNNKINKIENI L170P CD4.sup.+ 3.2 X X (SEQ ID NO: 11) B16-M29 Mthfd1I IPSGTTILNCFHDVLSGKLSGGSPGVP F294V CD4.sup.+ 1.7 X (SEQ ID NO: 12) B16-M30 Kif18b PSKPSFQEFVDWENVSPELNSTDQPFL K739N CD4.sup.+ 1.2 X X (SEQ ID NO: 13) B16-M33 Pbk DSGSPFPAAVILRDALHMARGLKYLHQ V146D CD8.sup.+ 0.1 X (SEQ ID NO: 14) B16-M36 Tm9sf3 CGTAFFINFIAIYHHASRAIPFGTMVA Y382H CD4.sup.+ 0.2 X (SEQ ID NO: 15) B16-M44 Cpsf3I EFKHIKAFDRTFANNPGPMVVFATPGM D314N CD4.sup.+ 0.5 X X (SEQ ID NO: 16) B16-M45 Mkm1 ECRITSNFVIPSEYWVEEKEEKQKLIQ N346Y CD4.sup.+ 1.4 X (SEQ ID NO: 17) B16-M46 Actn4 NHSGLVTFQAFIDVMSRETTDTDTADQ F835V CD4.sup.+ 0.2 X X (SEQ ID NO: 18) B16-M47 Rpl13a GRGHLLGRLAAIVGKQVLLGRKVVVVR A24G CD4.sup.+ 0.5 X (SEQ ID NO: 19) B16-M48 Def8 SHCHWNDLAVIPAGVVHNWDFEPRKVS R255G CD4.sup.+ 3.8 X X (SEQ ID NO: 20) B16-M50 Sema3b GFSQPLRRLVIHVVSAAQAERLARAEE L663V CD4.sup.+ 2.9 X X (SEQ ID NO: 21)

[0267] To exclude any bias associated with a peptide-based vaccine format, this experiment was repeated using in vitro transcribed (IVT) mRNA encoding the mutated epitopes (FIG. 1c upper graph right hand side). T cell reactivities determined with these RNA monotopes were largely comparable to the data obtained with synthetic peptides (FIG. 1c, Table 1), with somewhat lower numbers of immunogenic epitopes (about 25%). Importantly, also in this setting the majority of mutation-specific immune responses ( 10/12, ˜80%) were conferred by CD4.sup.+ T cells.

[0268] We extended our study to the chemically induced colon carcinoma model CT26 (Griswold, D. P. and Corbett, T. H., Cancer 36, 2441 (1975)) in BALB/c mice, in which we recently identified over 1680 non-synonymous mutations (Castle, J. C., et al., BMC Genomics 15, 190 (2014)). We selected 96 mutations based on their predicted MHC class I binding properties. In analogy to the B16F10 study, half of the candidates were good binders (‘low score’ 0.1-2.1). The other half was deliberately chosen for poor MHC I binding (‘high score’>3.9). In total, about 20% of mutated epitopes were immunogenic in mice immunized with the respective RNA monotopes (FIG. 1d pie chart, Table 2). It is noteworthy that in the ‘low’ MHC I score subgroup a couple of CD8.sup.+ T cells inducing epitopes were identified, which was not the case in the ‘high’ score subgroup (FIG. 1d right). This apparently did not bias against MHC class II restricted epitopes, as these were represented in similar frequency in both subgroups constituting the majority of CT26 immunogenic mutations ( 16/21, 80%).

TABLE-US-00002 TABLE 2 Immunogenic CT26 mutations. CT26 mutations determined to be immunogenic upon RNA immunization (as described in FIG. 1). (WT, wild type; AA#, number of mutated amino acid; Mut, Mutation) Reactive MHC I score Mutated sequence  Substitution T cell (best pre- Mutation Gene used for vaccination (WT, AA#, Mut) subtype diction) CT26-M03 Slc20a1 DKPLRRNNSYTSYIMAICGMPLDSFRA T425I CD4.sup.+  0.3 (SEQ ID NO: 22) CT26-M12 Gpc1 YRGANLHLEETLAGFWARLLERLFKQL E165G CD8.sup.+  1.9 (SEQ ID NO: 23) CT26-M13 Nphp3 AGTQCEYWASRALDSEHSIGSMIQLPQ G234D CD4.sup.+  0.1 (SEQ ID NO: 24) CT26-M19 Tmem87a QAIVRGCSMPGPWRSGRLLVSRRWSVE G63R CD8.sup.+  0.7 (SEQ ID NO: 25) CT26-M20 Slc4a3 PLLPFYPPDEALEIGLELNSSALPPTE T373I CD4.sup.+  0.9 (SEQ ID NO: 26) CT26-M24 Cxcr7 MKAFIFKYSAKTGFTKLIDASRVSETE L340F CD4.sup.+  1.8 (SEQ ID NO: 27) CT26-M26 E2f8 VILPQAPSGPSYATYLQPAQAQMLTPP I522T CD8.sup.+  0.1 (SEQ ID NO: 28) CT26-M27 Agxt2I2 EHIHRAGGLFVADAIQVGFGRIGKHFW E247A CD4.sup.+  0.2 (SEQ ID NO: 29) CT26-M35 Nap1I4 HTPSSYIETLPKAIKRRINALKQLQVR V63I CD4.sup.+  0.7 (SEQ ID NO: 30) CT26-M37 Dhx35 EVIQTSKYYMRDVIAIESAWLLELAPH T646I CD4.sup.+  0.1 (SEQ ID NO: 31) CT26-M39 Als2 GYISRVTAGKDSYIALVDKNIMGYIAS L675I CD8.sup.+  0.2 (SEQ ID NO: 32) CT26-M42 Deptor SHDSRKSTSFMSVNPSKEIKIVSAVRR S253N CD4.sup.+  0.3 (SEQ ID NO: 33) CT26-M43 Tdg AAYKGHHYPGPGNYFWKCLFMSGLSEV H169Y CD4.sup.+  0.3 (SEQ ID NO: 34) CT26-M55 Dkk2 EGDPCLRSSDCIDEFCCARHFWTKICK G192E CD4.sup.+  9.7 (SEQ ID NO: 35) CT26-M58 Rpap2 CGYPLCQKKLGVISKQKYRISTKTNKV P113S CD4.sup.+ 11.3 (SEQ ID NO: 36) CT26-M68 Steap2 VTSIPSVSNALNWKEFSFIQSTLGYVA R388K CD4.sup.+  6.8 (SEQ ID NO: 37) CT26-M75 Usp26 KTTLSHTQDSSQSLQSSSDSSKSSRCS S715L n.d.  5.8 (SEQ ID NO: 38) CT26-M78 Nbea PAPRAVLTGHDHEIVCVSVCAELGLVI V576I CD4.sup.+  6.3 (SEQ ID NO: 39) CT26-M90 Aldh18a1 LHSGQNHLKEMAISVLEARACAAAGQS P154S CD4.sup.+  8.3 (SEQ ID NO: 40) CT26-M91 Zc3h14 NCKYDTKCTKADCLFTHMSRRASILTP P497L CD4.sup.+  8.8 (SEQ ID NO: 41) CT26-M93 Drosha LRSSLVNNRTQAKIAEELGMQEYAITN V1189I CD4.sup.+  9.9 (SEQ ID NO: 42)

[0269] On a similar note, when analyzing all immune responses to RNA monotopes representing all 38 mutations we identified in the 4T1 mammary carcinoma model, nearly 70% of the recognized epitopes were recognized by CD4.sup.+ T cells (data not shown; Table 3).

TABLE-US-00003 TABLE 3 Immunogenic 4T1 mutations. 4T1 mutations determined to be immunogenic upon RNA immunization (as described in FIG. 1). (WT, wild type; AA#, number of mutated amino acid; Mut, Mutation) Reactive Mutated sequence  Substitution T cell Mutation Gene used for vaccination (WT, AA#, Mut) subtype 4T1-M2 Gen1 IPHNPRVAVKTTNNLVMKNSVCLERDS K707N CD4 (SEQ ID NO: 43) 4T1-M3 Polr2a LAAQSLGEPATQITLNTFHYAGVSAKN M1102I CD4 (SEQ ID NO: 44) 4T1-M8 Tmtc2 QGVTVLAVSAVYDIFVFHRLKMKQILP V201I CD8 (SEQ ID NO: 45) 4T1-M14 Zfr AHIRGAKHQKVVTLHTKLGKPIPSTEP K411T CD4 (SEQ ID NO: 46) 4T1-M16 Cep120 ELAWEIDRKVLHQNRLQRTPIKLQCFA H68N CD4 (SEQ ID NO: 47) 4T1-M17 Malt1 FLKDRLLEDKKIAVLLDEVAEDMGKCH T534A CD4 (SEQ ID NO: 48) 4T1-M20 Wdr11 NDEPDLDPVQELIYDLRSQCDAIRVTK T340I CD8 (SEQ ID NO: 49) 4T1-M22 Kbtbd2 DAAALQMIIAYAYRGNLAVNDSTVEQL T91R CD4 (SEQ ID NO: 50) 4T1-M25 Adamts9 KDYTAAGFSSFQKLRLDLTSMQIITTD I623L CD4 (SEQ ID NO: 51) 4T1-M26 Pzp AVKEEDSLHWQRPEDVQKVKALSFYQP G1199E CD8 (SEQ ID NO: 52) 4T1-M27 Gprc5a FAICFSCLLAHALNLIKLVRGRKPLSW F119L CD8 (SEQ ID NO: 53) 4T1-M30 Enho MGAAISQGAIIAIVCNGLVGFLL L10I CD4 (SEQ ID NO: 54) 4T1-M31 Dmrta2 EKYPRTPKCARCGNHGVVSALKGHKRY R73G CD4 (SEQ ID NO: 55) 4T1-M32 Rragd SHRSCSHQTSAPSPKALAHNGTPRNAI L263P CD4 (SEQ ID NO: 56) 4T1-M35 Zzz3 KELLQFKKLKKQNLQQMQAESGFVQHV K311N CD8 (SEQ ID NO: 57) 4T1-M39 Ilkap RKGEREEMQDAHVSLNDITQECNPPSS 127S CD4  (SEQ ID NO: 58) 4T1-M40 Cenpf RVEKLQLESELNESRTECITATSQMTA D1327E CD4 (SEQ ID NO: 59)

[0270] Thus, we have found in three independent mouse tumor models on different MHC backgrounds that a considerable fraction of non-synonymous cancer mutations are immunogenic and that quite unexpectedly the immunogenic mutanome is pre-dominantly recognized by CD4.sup.+ T cells.

[0271] B. MHC Class II Restricted Cancer Mutations as Vaccine Targets

[0272] To investigate whether MHC class II restricted cancer mutations are good vaccine targets in vivo, we proceeded to use synthetic RNA as vaccine format. Antigen-encoding synthetic RNA is emerging as promising vaccine technology due to its advantages including its capability to deliver more than one epitope, its selective uptake by antigen presenting cells (APC) and its intrinsic adjuvanticity (Diken, M., et al., Gene Ther 18, 702 (2011); Kreiter, S., et al., Curr Opin Immunol 23, 399 (2011); Pascolo, S., Handb Exp Pharmacol, 221 (2008); Sahin, U., et al., Nat Rev Drug Discov 13, 759 (2014); Van, L. S., et al., Hum Vaccin Immunother 9 (2013)). Our group has developed pharmacologically optimized RNA (stabilizing elements in RNA sequence and liposomal formulation), which meanwhile has reached the stage of clinical testing (NCT01684241) (Holtkamp, S., et al., Blood 108, 4009 (2006); Kreiter, S., et al., J Immunol 180, 309 (2008); Kuhn, A. N., et al. Gene Ther 17, 961 (2010)). We engineered RNA encoding B16-M30, one of the epitopes identified in the B16F10 tumor model. B16-M30 elicited strong CD4.sup.+ T cell responses, which did not recognize the wild type peptide (FIG. 2a left) as the mutated amino acid was shown to be essential for T cell recognition (FIG. 2a right). When B16F10 tumor-bearing C57BL/6 mice were repeatedly vaccinated with the B16-Mt30 RNA monotope, tumor growth was profoundly retarded (FIG. 2b). Half of the B16-M30 RNA treated mice were still alive 120 days after tumor vaccination, while all the control RNA treated mice died within 65 days.

[0273] Similarly, repeated vaccination in a lung metastasis model with luciferase transduced B16F10 cells revealed efficient eradication of metastases with B16-M30 RNA but not control synthetic RNA in the vast majority of mice as shown by bioluminescence imaging (BLI) (FIG. 2c). Consistently, tumor infiltrating leukocytes purified from B16F10 tumors of B16-M30 RNA immunized mice showed strong reactivity against B16-M30 (FIG. 2d).

[0274] Taking together, these data establish B16-M30 as a novel major rejection antigen in B16F10 tumors. They also exemplify that immunizing with RNA encoding a single immunogenic mutated epitope may give rise to functional T cells. These cells appear to be capable to target into the cancer lesion triggering control and even cure in murine tumor models. Our findings are in agreement with recent reports supporting the pivotal role of CD4.sup.+ T cell immunity in the control of cancer (Schumacher, T., et al., Nature 512, 324 (2014); Tran, E., et al., Science 344, 641 (2014)).

[0275] As the vast majority of mutations are unique to the individual patient, tapping the mutanome as a source for vaccine antigens requires an actively individualized approach (Britten, C. M., et al., Nat Biotechnol 31, 880 (2013)). In this respect, one of the major challenges is instant manufacturing of a tailored on-demand vaccine. This can be viably addressed by RNA vaccine technology. RNA manufacturing based on in vitro transcription usually takes a few days (FIG. 3a). At present, the GMP-grade material could be made ready for release within three weeks and this process is continuously being optimized to reduce the duration. On another note, though we have shown tumor eradication in mouse models with a single mutation, one would ideally prefer to combine several mutations in a poly-neo-epitope vaccine. This would allow us to address several factors that counteract the clinical success of vaccines in humans such as tumor heterogeneity and immunoediting (Gerlinger, M., et al., N Engl J Med 366, 883 (2012); Koebel, C. M., et al., Nature 450, 903 (2007)).

[0276] In light of these considerations, we explored how to use our insights on immunogenic epitopes to develop a cancer vaccine concept which we call “mutanome engineered RNA immunotherapy” (MERIT) (FIG. 3a). To test this concept, we selected four MHC class II (CT26-M03, CT26-M20, CT26-M27, CT26-M68) and one MHC class I (CT26-M19) restricted mutations that were derived from the CT26 model (see Table 2) and engineered RNA monotopes encoding each of them. In addition, a synthetic RNA pentatope was engineered encoding all five mutated epitopes connected by 10 mer non-immunogenic glycine/serine linkers to avoid the generation of junctional epitopes (FIG. 3a). By immunizing naive BALB/c mice we found that the quantity of IFN-producing T cells elicited by the pentatope was comparable to that evoked by the respective monotope for three of these mutations (FIG. 3b). However, for two of these mutations the pentatope RNA was significantly superior in robustly expanding mutations-specific T cells.

[0277] We assessed the anti-tumour efficacy of immune responses elicited by RNA pentatope vaccines in a lung metastasis model of CT26 luciferase transfectant (CT26-Luc) tumors. Tumor-bearing BALB/c mice were vaccinated repeatedly with a mixture of two RNA pentatopes (3 MHC class I and 7 class II restricted epitopes) including the mutations tested in the previous experiment. Tumor growth in vaccinated mice was significantly inhibited as measured by BLI of the lung (FIG. 3c left). At day 32 all mice in the RNA pentatope group were alive whereas 80% of the control mice had already died (FIG. 3c mid). Post mortem macroscopic (FIG. 3c right), histological (FIG. 3d right) and computerized image analysis (data not shown) of tissue sections revealed significantly lower tumor load in the vaccinated mice as compared to untreated controls. Tumor lesions of pentatope RNA vaccinated mice were briskly infiltrated with CD3.sup.+ T cells, whereas the number of CD3+ T cells was significantly lower in their surrounding lung tissues. Tumors of untreated controls displayed CD3.sup.+ cells staining which was not much different to that of the surrounding lung tissue in terms of quantity and mainly at the tumor border but not within the tumor. (FIG. 3d).

[0278] Altogether, these findings indicate that T cells against each single epitope are elicited with a MERIT approach employing a poly-neo-epitope encoding RNA vaccine. These T cells target tumor lesions, recognize their mutated targets and result in efficient tumor control in vivo.

[0279] C. Selection of Mutations Having Anti-Tumor Immunity

[0280] One of the key questions is how to select the mutations with the highest probability of inducing efficient anti-tumor immunity. We (FIG. 1d right) and others (Matsushita, H., et al., Nature 482, 400 (2012); Robbins, P. F., et al., Nat Med 19, 747 (2013); van, R. N., et al., J Clin Oncol 31, e439-e442 (2013)) have shown that MHC class I binding scores enable enrichment for mutated epitope candidates which elicit CD8.sup.+ responses and tumor rejection (Duan, F., et al., J Exp Med 211, 2231 (2014)). Our findings described above indicate that MHC class II presented mutated epitopes may even be of higher interest for a MERIT approach. In fact, a correlation analysis revealed that immunogenic mutations have a significantly better MHC class II binding score as compared to non-immunogenic ones (FIG. 4a). Most cancers lack MHC class II expression. Effective recognition of neo-epitopes by CD4.sup.+ T cells in MHC class II negative tumors should depend on release of tumor antigens to be taken up and presented by antigen presenting cells (APCs). This should be most efficient for antigens with highly abundant expression. To test this hypothesis, we implemented an algorithm combining good MEW class II binding with abundant expression of the mRNA encoding the mutated epitope. For the latter we used confirmed mutated RNA sequencing reads normalized to the overall read count (NVRC: normalized variant read counts). We ranked CT26 mutanome data with this algorithm and selected the top ten mutations (‘ME’ mutations in Table 4) predicted to be good MEW class II binders among the most abundant candidate epitopes (NVRC ≥60). As control we chose ten mutations based on abundant expression only (‘E’ mutations in Table 4). Most importantly, these epitopes were used without any further pre-validation or immunogenicity testing to engineer two RNA pentatopes for each group (P.sub.ME and P.sub.E pentatopes). When mice with established CT26-Luc lung tumors were vaccinated with these epitopes, P.sub.ME as compared to P.sub.E pentatopes induced a much stronger T cell response (FIG. 4c). Established lung metastases were completely rejected in almost all mice whereas P.sub.E pentatopes were not able to confer tumor growth control (FIG. 4b).

TABLE-US-00004 TABLE 4 In silico prediction of CT26 mutations with abundant expression and favorable MHC class II binding properties. CT26 mutations selected for high expression with (ME) or without (E) consideration of the MHC II percentile rank (IEDB consensus version 2.5). (WT, wild type; AA#, number of mutated amino acid; Mut, Mutation) MHC II score Mutated sequence  Substitution Expression (best pre- Mutation Gene used for vaccination (WT, AA#, Mut) (NVRC) diction) CT2-E16 Asns DSVVIFSGEGSDEFTQGYIYFHKAPSP L370F 1428.05 45.45 (SEQ ID NO: 60) CT26-E2 Cd34 PQTSPTGILPTTSNSISTSEMTWKSSL D120N 1150.85 23.76 (SEQ ID NO: 61) CT26-E3 Actb WIGGSILASLSTFHQMWISKQEYDESG Q353H  974.16  8.30 (SEQ ID NO: 62) CT26-E4 Tmbim6 SALGSLALMIWLMTTPHSHETEQKRLG A73T  825.51  2.96 (SEQ ID NO: 63) CT26-E5 Glud1 DLRTAAYVNAIEKIFKVYNEAGVTFT V5461  619.54  8.04 (SEQ ID NO: 64) CT26-E16 Eif4g2 KLCLELLNVGVESNLILKGVILLIVDK K108N  327.79 20.99 (SEQ ID NO: 65) CT26-E17 Sept7 NVHYENYRSRKLATVTYNGVDNNKNKG A314T  316.98  6.47 (SEQ ID NO: 66) CT26-E18 Fn1 YTVSVVALHDDMENQPLIGIQSTAIPA S1710N  303.62 17.41 (SEQ ID NO: 67) CT26-E19 Brd2 KPSTLRELERYVLACLRKKPRKPYTIR S703A  301.83  7.86 (SEQ ID NO: 68) CT26-E20 Uchl3 KFMERDPDELRFNTIALSAA A224T  301.78  9.75 (SEQ ID NO: 69) CT26-ME1 Aldh18a1 LHSGQNHLKEMAISVLEARACAAAGQS P154S   67.73  0.05 (SEQ ID NO: 70) CT26-ME2 Ubqln1 DTLSAMSNPRAMQVLLQIQQGLQTLAT A62V   84.08  0.24 (SEQ ID NO: 71) CT26-ME3 Ppp6r1 DGQLELLAQGALDNALSSMGALHALRP D309N  139.80  0.44 (SEQ ID NO: 72) CT26-ME4 Trip12 WKGGPVKIDPLALMQAIERYLVVRGYG V1328M   83.09  0.49 (SEQ ID NO: 73) CT26-ME5 Pcdhgc3 QDINDNNPSFPTGKMKLEISEALAPGT E139K   86.16  0.54 (SEQ ID NO: 74) CT26-ME6 Cad SDPRAAYFRQAENDMYIRMALLATVLG G2139D  152.86  0.55 (SEQ ID NO: 75) CT26-ME7 Smarcd1 MDLLAFERKLDQTVMRKRLDIQEALKR I161V  125.85  0.60 (SEQ ID NO: 76) CT26-ME8 Ddx27 ITTCLAVGGLDVKFQEAALRAAPDILI S297F   61.82  0.62 (SEQ ID NO: 77) CT26-ME9 Snx5 KARLKSKDVKLAEAHQQECCQKFEQLS T341A  120.27  0.73 (SEQ ID NO: 78) CT26-ME10 Lin7c GEVPPQKLQALQRALQSEFCNAVREVY V41A   71.24  1.09 (SEQ ID NO: 79)

[0281] Antigen specific T.sub.H cells promote the cross-priming of tumor specific CTL responses by CD40 ligand mediated licensing of dendritic cells. This may result in antigen spread if T.sub.H cells recognize their antigen on the same APC that cross-presents an unrelated CTL epitope (Bennett, S. R., et al., Nature 393, 478 (1998); Schoenberger, S. P., et al., Nature 393, 480 (1998)). Congruently, in the blood and spleen of mice immunized with P.sub.ME but not P.sub.E pentatopes we detected strong CD8.sup.+ T cell responses against gp70-AH1, a well characterized immunodominant CTL epitope derived from the endogenous murine leukemia virus-related cell surface antigen (FIG. 4d). This indicates that cancer neo-epitope specific T.sub.H cells, in analogy to viral neo-antigen specific T cells (Croxford, J. L., et al., Autoimmun Rev 1, 251 (2002)), may exert their anti-tumour function by antigen spreading and augmentation of CTL responses

[0282] D. Summary

[0283] In summary, our data indicate that MHC class II restricted T cell epitopes are abundant in the cancer mutanome and can be used to customize RNA-based poly-neo-epitope vaccines with substantial therapeutic effect in mouse tumor models.

[0284] The mechanism responsible for the high rate of CD4.sup.+ T cell recognition of mutations is unclear yet. A simple explanation may be the longer and variable size of peptides presented on MHC class II molecules as compared to MHC class I epitopes increasing the likelihood that a mutation is covered by the respective peptide. T cell epitopes presented by MHC class I molecules are typically peptides between 8 and 11 amino acids in length with well-defined N- and C-termini. MHC class II molecules present longer peptides of 13-17 amino acids in length with a 9 amino acid MHC II core binding region and variable number of additional flanking amino acids both contributing to the recognition by CD4.sup.+ T cells (Arnold, P. Y., et al., J Immunol 169, 739 (2002)).

[0285] While the first evidence of the spontaneous CD8.sup.+ and CD4.sup.+ T-cell responses directed against mutated gene-products in cancer patients was generated in the 1990s (Dubey, P., et al., J Exp Med 185, 695 (1997); Lennerz, V., et al., Proc Natl Acad Sci USA 102, 16013 (2005); Wolfel, T., et al., Science 269, 1281 (1995)), only the recent high level publications have created broad acceptance for the enormous potential of mutation-specific T cells to confer anti-tumor activity in cancer patients (Lu, Y. C., et al., J Immunol 190, 6034 (2013); Schumacher, T., et al., Nature 512, 324 (2014); Tran, E., et al., Science 344, 641 (2014)). To assess whether the principles we unraveled in the mouse models for melanoma, colon and breast cancer are true in the human setting, we analyzed mutation and RNA-Seq data in the same three human cancer types provided by The Cancer Genome Atlas (TCGA). For all three human cancer types we confirmed the abundance of mutations predicted to bind to MHC class II we revealed in mouse models (FIG. 4.e). The MERIT approach we presented here integrates advances in the field of next generation sequencing, computational immunology and synthetic genomics and thereby provides the integrated technology for comprehensive exploitation of the neo-epitope target repertoire. Targeting multiple mutations at once may at least in theory pave the way to solve critical problems in current cancer drug development such as clonal heterogeneity and antigen escape (Kroemer, G. and Zitvogel, L., Oncoimmunology 1, 579 (2012); Mittal, D., et al., Curr Opin Immunol 27, 16 (2014)).

[0286] Meanwhile, based on this study and our prior work clinical translation has been initiated and a first-in-concept trial in melanoma patients (Castle, J. C., et al., Cancer Res 72, 1081 (2012); Castle, J. C., et al., Sci Rep 4, 4743 (2014); Lower, M., et al., PLoS Comput Biol 8, e1002714 (2012)) is actively recruiting (NCT02035956) and confirms that “just in time” production of a poly-neo-epitope mRNA cancer vaccine is in fact feasible.

Example 3: Selection of Mutations Having Anti-Tumor Immunity

[0287] For selecting/ranking amino acid sequence modifications one may proceed as follows: [0288] 1. Within a given list of non-synonymous point mutations, compute a peptide sequence which has the mutated amino acid in the middle and is flanked by up to 13 amino acids on the N and C-terminal end, respectively; this will be called 27mer in the following text (the length for each flanking sequence may be smaller than 13 amino acids when the mutation is close to the N or C-terminus of the whole protein) [0289] 2. Compute MHC class II binding prediction consensus scores (e.g. using the IEDB T-cell prediction tools [Wang P, et al. (2010) BMC Bioinformatics. 11:568. PMID: 21092157. http://tools.immuneepitope.org/mhcii/]) for each overlapping 15 nt long subsequence of each 27mer; the best (=lowest) score is assigned to the whole 27mer [0290] 3. Compute the expression (preferably in RPKM units [Ali Mortazavi, et al. (2008) Nature Methods 5, 621-628]) of the genes to which the 27mers are associated [0291] 4. Compute the variant allele frequency (VAF) of each mutation in the RNA: [0292] input are short read alignments of an RNA-Seq experiment done with the same tumor sample as used for mutation detection [0293] look up the alignments and reads overlapping the mutation site [0294] tally the nucleotides mapped to the mutation site using the reads aggregated a step before [0295] compute the sum of mutant-allele nucleotides divided by the sum of all nucleotides mapped to the genomic site of the mutation (FIG. 5) [0296] 5. Multiply the respective gene expression with the VAF to get the mutation expression (preferably in RPKM units) [0297] 6. Rank all 27mers by the MHC binding score (as computed in step 2, lowest score is best) and remove 27mers with an associated mutation expression of less than a given threshold

[0298] Application to Murine Data Set:

[0299] For testing the algorithm, 185 mutations were selected from the murine tumor models 4T1, CT26 and B16F10 were tested for their antigenicity. Then we first tried to test the influence of the level of gene and mutation expression on the predictive performance of the algorithm (FIG. 6). Here we can observe that the maximum area under the curve of the receiver operating characteristic (ROC AUC [Fawcett T., Pattern Recogn Lett. 2006; 27:861-874. doi: 10.1016/j.patrec.2005.10.010]) is higher when the mutation expression is filtered instead of the gene expression (FIG. 6 left (gene expression) vs. right (mutation expression) plot).

[0300] FIG. 7 shows the ROC curves for the optimum thresholds, indicating a pronounced influence of the mutation expression for binders with only a mediocre relative binding affinity (FIG. 7, right panel, values between a false positive rate of about 0.3 and 0.6).