Patent Application
20170202939
The Sequence Listing, which is a part of the present disclosure, includes a text file comprising primer nucleotide and/or amino acid sequences of the present invention. The subject matter of the Sequence Listing is incorporated herein by reference in its entirety. The information recorded in computer readable form is identical to the written sequence listing.
The incidence of malignant melanoma continues to rise worldwide. The number of new cases, in the US for 2012 is estimated to be 76,250 (8.6% increase compared to 2011) (Siegel, R., et al., Cancer statistics, 62, 10-29 2012). Despite recent advances in the treatment of metastatic melanoma with ipilimumab (anti-CTLA-4 antibody) and vemurafenib (BRAF V600E inhibitor), this disease remains an incurable malignancy with an expected survival of 12-14 months (Hodi, F. S., et al., N. Engl. J. Med. 363, 711-723, 2010; Chapman, P. B., et al., N. Engl. J. Med. 364, 2507-2516, 2011). Thus, metastatic melanoma represents a disease area of unmet medical need. Melanoma is distinguished for its association with early in life UV-light exposure, high mutational rate, and the ability to induce spontaneous anti-tumor immunity (Lennerz, V., et al., Proc. Nat'l. Acad. Sci. USA 102, 16013-16018, 2005; Garibyan, L., et al., Curr. Oncol. Rep. 12, 319-326, 2010; Pleasance, E. D. et al., Nature 463, 191-196, 2010; Berger, M. F., et al., Nature 485, 502-506, 2012; Hodis, E., et al., Cell 150, 251-263, 2012). The modest, yet reproducible, clinical activity of ipilimumab seen in patients with advanced melanoma provides strong evidence that immune targeting confers therapeutic benefit in this disease. Investigational cancer vaccines as well as adoptive T cell therapies while more technically demanding are now beginning to show efficacy in early phase clinical trials (Rosenberg, S. A. Science Translational Medicine 4, 127ps128, 2012).
However, a critical barrier facing investigators developing these cellular therapies is the paucity of validated melanoma antigens. New strategies are needed to identify patient-specific (unique) tumor antigens, which can serve as targets for immune intervention. Identification of the entire spectrum of unique antigens at the single tumor/patient level has been viewed historically as an unattainable goal.
The present inventors have developed anti-cancer vaccines, methods of constructing vaccines, methods of their use, and methods of identifying neoantigens create personalized vaccines to treat cancer. In various embodiments, the present teachings provide methods for identification of tumor-specific neoantigens and their incorporation in a vaccine, and adoptive T cell therapy for the treatment of cancers such as, without limitation, melanoma and lung cancer. Various embodiments involve patient-specific identification of tumor neo-antigens. In various configurations, such tumor neo-antigens, such as those arising during neoplastic transformation, can elicit T cell immunity capable of protecting the host from cancer progression. In various embodiments, the present teachings make use of next-generation sequencing technology, human leukocyte antigens (HLA) class I binding/stability prediction algorithms and in vitro assays to identify personalized tumor neoantigens. In various embodiments, these technologies can be incorporated into a vaccine/adoptive T cell therapy for treatment of cancer.
In some embodiments, the present teachings include strategies for personalized neoantigen-specific adoptive T cell therapy. In various aspects, DNA isolated from tumor and matched peripheral blood mononuclear cells (PBMC) can be subjected to exome sequencing to identify tumor somatic missense mutations. In some embodiments, RNA isolated from a tumor can be used for transcriptome analysis to identify those somatic mutations that are expressed. In some aspects, results can show that in cancers such as melanoma and lung cancer, a high number of missense mutations (>200) can be identified per tumor genome. In some embodiments, a combination of major histocompatibility complex (MHC) class I binding and stability prediction algorithms can be used to identify candidate neo-antigens among missense mutations, and expressed candidate neo-antigens can be selected for, peptide manufacturing. Biochemical and cellular assays can be performed to established binding and presentation of neo antigen-encoding peptides. Experimentally validated peptides can be selected for incorporation in a dendritic cell (DC) vaccine as described in Carreno, B. M., et al., J. Clin. Invest. 123, 3383-3394, 2013; after 3 vaccine doses patients can be subjected to apheresis and CD8+ T cells can be isolated from PBMC. These T cells can be expanded in an antigen-specific manner using a 2 step procedure as described in Carreno, B. M., et al., J. Immunology 188, 5839-5849, 2012. In various configurations, the 2 step procedure can take 10-30 days, such as, without limitation, 10 days, 11 days, 12 days, 13 days, 14 days, 15 days, 16 days, 17 days, 18 days, 19 days, 20 days, 21 days, 22 days, 23 days, 24 days, 25 days, 26 days, 27 days, 28 days, 29 days or 30 days for completion and can yield>104 fold antigen-specific T cell expansions. In various configurations, expanded neo-antigen specific T cells, can be infused into pre-conditioned patients as adoptive T cell therapy, by, for example, methods described by Linette, G. P. et al., Clin. Cancer Res. 11, 7692-7699, 2005.
In various configurations, the present teachings include a series of analytical steps for identification of neo-antigens from somatic tumor missense mutations, as illustrated in
Various embodiments of the present teachings include the following aspects: In some embodiments, a method of treating a cancer in a subject in need thereof can comprise: providing a neoantigen peptide encoded in DNA of a tumor of the subject, wherein the neoantigen peptide can consist of from 8 to 13 amino acids; transfecting at least one HLA class I positive cell with at least one tandem minigene construct that can comprise at least one sequence that can encode the at least one neoantigen; identifying a complex that can comprise the at least one HLA molecule and the at least one neoantigen peptide produced by the at least one HLA class I positive cell; forming a vaccine that can comprise the at least one neoantigen; and administering the vaccine to the subject, wherein at least one tumor cell of the cancer can comprise at least one polypeptide which can comprise at least one amino acid substitution. In some configurations, the at least one neoantigen peptide can consist of from 9 to 11 amino acids. In some configurations, the at least one neoantigen peptide can consist of 9 amino acids. In various configurations, the at least one neoantigen peptide can consist of 8, 9, 10, 11, 12, or 13 amino acids. In some configurations, the at least one neoantigen peptide can bind in silico to an HLA class I molecule with a stability>2 h. In some configurations, the at least one neoantigen peptide can bind in silico to an HLA class I molecule an affinity of <500 nM. In some configurations, the at least one neoantigen peptide can bind in silico to an HLA class I molecule with an affinity of <250 nM. In various configurations, the at least one neoantigen peptide can bind in silico to an HLA Class I molecule with an affinity of <550 nM, <500 nM, <450 nM, <400 nM, <350 nM, <300 nM, <250 nM, or <200 nM. In various configurations, the at least one neoantigen peptide can bind in vitro to an HLA class I molecule with an affinity of <4.7 log (IC50, nM), <4.6 log (IC50, nM), <4.5 log (IC50, nM), <4.4 log (IC50, nM), <4.3 log (IC50, nM), <4.2 log (IC50, nM), <4.1 log (IC50, nM), <4.0 log nM), <3.9 log (IC50, nM), <3.8 log (IC50, nM), or <3.7 log (IC50, nM). In some configurations, the at least one neoantigen peptide can bind in vitro to an HLA class I molecule with an affinity of <4.7 log (IC50, nM). In some configurations, the at least one neoantigen peptide can bind in vitro to an HLA class I molecule with an affinity of <3.8 log (IC50, nM). In some configurations, the at least one neoantigen peptide can bind in vitro to an HLA class I molecule with an affinity of <3.7 log (IC50, nM). In some configurations, the at least one neoantigen peptide can bind in vitro to an HLA class I molecule with an affinity of <3.2 log (IC50, nM). In some configurations, the vaccine can comprise at least seven neoantigen peptides. In various configurations, the HLA class I molecules can be selected from the group consisting of HLA-A*01:01, HLA-B*07:02, HLA-A*02:01, HLA-B*07:03, HLA-A*02:02, HLA-B*08:01, HLA-A*02:03, HLA-B*15:01, HLA-A*02:05, HLA-B*15:02, HLA-A*02:06, HLA-B*15:03, HLA-A*02:07, HLA-B*15:08, HLA-A*03:01, HLA-B*15:12, HLA-A*11:01, HLA-B*15:16, HLA-A*11:02, HLA-B*15:18, HLA-A*24:02, HLA-B*27:03, HLA-A*29:01, HLA-B*27:05, HLA-A*29:02, HLA-B*27:08, HLA-A*34:02, HLA-B*35:01, HLA-A*36:01, HLA-B*35:08, HLA-B*42:01, HLA-B*53:01, HLA-B*54:01, HLA-B*56:01, HLA-B*56:02, HLA-B*57:01, HLA-B*57:02, HLA-B*57:03, HLA-B*58:01, HLA-B*67:01, and HLA-B*81:01. In some configurations, the HLA class I molecules can be HLA-A*02:01 molecules. In some configurations, the HLA class I molecules can be HLA-A*11:01 molecules. In some configurations, the HLA class I molecules can be HLA-B*08:01 molecules. In some configurations, the at least one HLA class I positive cell can be at least one melanoma cell. In various configurations, the at least one melanoma cell can be selected from the group consisting of DM6 cell and an A375 cell. In some configurations, the tandem minigene can further comprise a ubiquitination signal and two mini-gene controls. In configurations where the neoantigens bind HLA-A*2:01 molecules, the tandem minigene can further comprise a ubiquitination signal and two mini-gene controls that encode HLA-A*02:01 peptides G280 and WNV SVG9. In various configurations, the cancer can be selected from the group consisting of skin cancer, lung cancer, bladder cancer, colorectal cancer, gastrointestinal cancer, esophageal cancer, gastric cancer, intestinal cancer, breast cancer, and a cancer caused by a mismatch repair deficiency. In various configurations, the skin cancer can be selected from the group consisting of basal cell carcinoma, squamous cell carcinoma, merkel cell carcinoma, and melanoma. In some configurations, the cancer can be a melanoma. In some configurations, the forming a vaccine can comprise: providing a culture comprising dendritic cells obtained from the subject; and contacting the dendritic cells with the at least one neoantigen peptide, thereby forming dendritic cells comprising the at least one neoantigen peptide. In some configurations, the forming a vaccine can further comprise maturing the dendritic cells. In some configurations, the maturing the dendritic cells can comprise administering CD40L and IFNγ. In various configurations, the maturing the dendritic cells can further comprise administering TLR agonist. In various configurations, the maturing the dendritic cells can further comprise administering a TLR3 agonist. In various configurations, the maturing the dendritic cells can further comprise administering a TLR8 agonist. In various configurations, the maturing the dendritic cells can further comprise administering TLR3 and TLR8 agonists. In various configurations, the maturing the dendritic cells can further comprise administering poly I:C and R848. In some configurations, the forming a vaccine can further comprise: administering to the subject the dendritic cells comprising the at least one neoantigen peptide; obtaining a population of CD8+ T cells from a peripheral blood sample from the subject, wherein the CD8+ cells recognize the at least one neoantigen; and expanding the population of CD8+ T cells that recognize the neoantigen. In some configurations, the forming a vaccine can further comprise administering to the subject the expanded CD8+ T cells. In various configurations, the forming a vaccine can comprise combining the neoantigen peptide with a pharmaceutically acceptable adjuvant.
In some embodiments, a method of treating a cancer in a subject in need thereof, can comprise: a) providing a sample of a tumor from a subject; b) performing exome sequencing on the sample to identify one or more amino acid substitutions comprised by the tumor exome; c) performing transcriptome sequencing on the sample to verify expression of the amino acid substitutions identified in b); and d) selecting at least one candidate neoantigen peptide sequence from amongst the amino acid substitutions identified in c) according to the following criteria: i) Exome VAF>10%; ii) Transcription VAF>10%; iii) Alternate reads>5; iv) FPKM>1. v) binds in silico to an HLA class I molecule with an affinity of <500 nM and a stability>2 h; e) performing an in vitro HLA class I binding assay; f) selecting at least one candidate neoantigen peptide sequence from amongst the amino acid substitutions identified in d) that bind HLA class one molecules with an affinity of <4.7 log (IC50, nM) in the assay performed in e); g) transfecting at least one HLA class I positive cell with at least one tandem minigene construct which can comprise at least one sequence encoding the at least one neoantigen; identifying a complex comprising the at least one HLA molecule and the at least one neoantigen peptide produced by the at least one HLA class I positive cell; i) forming a vaccine that can comprise the at least one neoantigen; and j) administering the vaccine to the subject, wherein at least one tumor cell of the cancer can comprise at least one polypeptide comprising the one or more amino acid substitutions. In some configurations, the Exome VAF can be ≧30%. In some configurations, the Exome VAF can be ≧40%. In some configurations, the Exome VAF can be ≧50%. In various configurations, the in vitro HLA class I binding assay can be selected from the group consisting of a T2 assay and a fluorescence polarization assay.
In some embodiments, a method of treating cancer in a subject in need thereof can comprise: a) providing a sample of a tumor from a subject; b) performing exome sequencing on the sample to identify amino acid substitutions comprised by the tumor exome; c) performing transcriptome sequencing on the sample to verify expression of the amino acid substitutions identified in b); d) performing a fluorescence polarization binding assay or a T2 assay of amino acid substitutions identified in c) to an HLA class I molecule; e) selecting at least one candidate neoantigen from amongst the amino acid substitutions identified in d) according to the following criteria: i) Exome variant allele fraction (VAF)>10%; ii) Transcriptome (seq capture data) VAF>10%; is iii) Alternate reads>5; iv) fragments per kilobase of exon per million fragments mapped (FPKM) (>1; v) Peptides comprise 9-11 amino acids; vi) Peptides are predicted in silico to bind to any HLA class I allele that meet the following criteria: A) Predicted MHC binding<250 nM; B) Predicted MHC stability>2 h; vii) MHC binding<3.2 log [IC50, nM] in fluorescence polarization binding assay; f) transfecting at least one HLA class I positive cell line such as a melanoma cell line with at least one tandem minigene construct comprising at least one sequence encoding the at least one candidate neoantigen identified in e); g) extracting from the at least one HLA class I positive cell line one or more HLA class I complexes comprising a HLA class I molecule and the one or more neoantigen peptides; h) identifying the sequence of at least one neoantigen peptide comprised by the soluble HLA class I complex using reverse phase HPLC and LC/MS; i) contacting dendritic cells obtained from the subject with the at least one neoantigen peptide of sequence identified in h), thereby forming dendritic cells comprising the at least one neoantigen peptide; j) administering to the subject the dendritic cells comprising the at least one neoantigen peptide; k) obtaining CD8+ T cells from a peripheral blood sample from the subject; l) enriching the CD8+ T cells that recognize the at least one neoantigen; m) administering to the subject the enriched CD8+ T cells. In some configurations of the present teachings, the HLA class I molecules can be selected from the group consisting of HLA-A*01:01, HLA-B*07:02, HLA-A*02:01, HLA-B*07:03, HLA-A*02:02, HLA-B*08:01, HLA-A*02:03, HLA-B*15:01, HLA-A*02:05, HLA-B*15:02, HLA-A*02:06, HLA-B*15:03, HLA-A*02:07, HLA-B*15:08, HLA-A*03:01, HLA-B*15:12, HLA-A*11:01, HLA-B*15:16, HLA-A*11:02, HLA-B*15:18, HLA-A*24:02, HLA-B*27:03, HLA-A*29:01, HLA-B*27:05, HLA-A*29:02, HLA-B*27:08, HLA-A*34:02, HLA-B*35:01, HLA-A*36:01, HLA-B*35:08, HLA-B*42:01, HLA-B*53:01, HLA-B*54:01, HLA-B*56:01, HLA-B*56:02, HLA-B*57:01, HLA-B*57:02, HLA-B*57:03, HLA-B*58:01, HLA-B*67:01, and HLA-B*81:01. In some configurations, the HLA class I molecules can be HLA-A*02:01 molecules. In some configurations, the HLA class I molecules can be HLA-A*11:01 molecules. In some configurations, the HLA class I molecules can be HLA-B*08:01 molecules. In various configurations, the melanoma cell line can be selected from the group consisting of DM6 and A375. In some configurations, the tandem minigene can further comprise a ubiquitination signal and two mini-gene controls. In configurations where the HLA-A molecules are HLA-A*02:01 molecules, the two mini-gene controls can encode G280 and WNV SVG9 peptides. In some configurations, the cancer can be a melanoma. In various configurations, the melanoma is a metastatic melanoma.
In some configurations, as many as 600 amino acid substitutions can be identified from any given tumor. In some configurations, each of these amino acid substitutions can be analyzed for predicted binding to HLA-A class I molecules. In various configurations, at least 1, at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11, at least 12, at least 13, at least 15, at least 16, at least 17, at least 18, at least 19, at least 20, at least 21, at least 22, at least 23, at least 24, at least 25, at least 26, at least 27, at least 28, at least 29, at least 30, at least 11, at least 32, at least 33, at least 34, at least 35, at least 36, at least 37, at least 38, at least 39, at least 40, at least 41, at least 42, at least 43, at least 44, at least 45, at least 46, at least 47, at least 48, at least 49 or at least 50 candidate neoantigens can be expressed in a tumor. In some configurations, at least 1, at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11, at least 12, at least 13, at least 15, at least 16, at least 17, at least 18, at least 19, at least 20, at least 21, at least 22, at least 23, at least 24, at least 25, at least 26, at least 27, at least 28, at least 29, at least 30, at least 31, at least 32, at least 33, at least 34, at least 35, at least 36, at least 37, at least 38, at least 39, at least 40, at least 41, at least 42, at least 43, at least 44, at least 45, at least 46, at least 47, at least 48, at least 49 or at least 50 candidate neoantigens can be selected to test their presentation to T cells. In some configurations, at least 1, at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11, at least 12, at least 13, at least 15, at least 16, at least 17, at least 18, at least 19, at least 20, at least 21, at least 22, at least 23, at least 24, at least 25, at least 26, at least 27, at least 28, at least 29, at least 30, at least 31, at least 32, at least 33, at least 34, at least 35, at least 36, at least 37, at least 38, at least 39, at least 40, at least 41, at least 42, at least 43, at least 44, at least 45, at least 46, at least 47, at least 48, at least 49 or at least 50 candidate neoantigens can be selected for incorporation into a vaccine. In some configurations, the tandem minigenes can comprise at least 1, at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11, at least 12, at least 13, at least 15, at least 16, at least 17, at least 18, at least 19, at least 20, at least 21, at least 22, at least 23, at least 24, at least 25, at least 26, at least 27, at least 28, at least 29, at least 30, at least 31, at least 32, at least 33, at least 34, at least 35, at least 36, at least 37, at least 38, at least 39, at least 40, at least 41, at least 42, at least 43, at least 44, at least 45, at least 46, at least 47, at least 48, at least 49 or at least 50 candidate neoantigen sequences. In some configurations, the dendritic cells can comprise at least 1, at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11, at least 12, at least 13, at least 15, at least 16, at least 17, at least 18, at least 19, at least 20, at least 21, at least 22, at least 23, at least 24, at least 25, at least 26, at least 27, at least 28, at least 29, at least 30, at least 31, at least 32, at least 33, at least 34, at least 35, at least 36, at least 37, at least 38, at least 39, at least 40, at least 41, at least 42, at least 43, at least 44, at least 45, at least 46, at least 47, at least 48, at least 49 or at least 50 neoantigen peptides. In some embodiments, the personalized neoantigen therapy can be paired with other forms of cancer therapy such as, but without limitation, chemotherapy. In some configurations, the chemotherapy can comprise ipilimumab and/or vemurafenib.
In some embodiments, the present teachings include a neoantigen peptide encoded in DNA of a tumor of the subject for use in the treatment of a cancer, wherein the neoantigen peptide consists of from 8 to 13 amino acids, binds in silico to an HLA class I molecule with an affinity of <500 nM and a stability>2 h and binds in vitro to an HLA class I molecule with an affinity of <4.7 log (IC50, nM).
In various embodiments of the invention, it includes the following aspects;
The present teachings describe methods of creating vaccines for personalized cancer treatment. As used herein, “a vaccine” is a preparation that induces a T-cell mediated immune response. As used in the present description and the appended claims, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context indicates otherwise.
In some embodiments, methods of the present teachings can comprise sequencing DNA from excised tumor tissue of a subject to identify amino acid substitutions, performing sequence capture to confirm the expression of the amino acid substitutions, selecting amino acid substitutions that bind or are likely to bind HLA molecules, transfecting nucleic acids encoding the selected amino acid substitutions into an HLA positive melanoma cell line, extracting HLA class I complexes from the transfected cells, identifying the sequence of neoantigens bound to the extracted HLA class one complexes, contacting dendritic cells obtained from the subject with the identified neoantigen peptides, thereby forming a dendritic cell vaccine, administering to the subject the dendritic cell vaccine, obtaining and enriching CD8+ T cells from the subject, and administering the enriched CD8+ T cells to the subject. In some embodiments, the neoantigen binding T cells can be used for adaptive T cell therapy. In some embodiments, a fluorescence polarization binding assay can be used to confirm the binding of neoantigen peptides to HLA molecules prior to selection for transfection.
In some configurations, the following criteria can be used to select the neoantigens for transfection into HLA class I positive cells; in the exome sequencing, the variant allele fraction of the neoantigen greater than 10%; in the transcript sequencing results the VAF greater than 10%, the alternate read counts greater than 5, and the FPKM greater than 1; the encoded peptides can be 9-11 amino acids in length; the predicted binding to any HLA class I allele can have following characteristics; the predicted MHC binding <250 nM (NetMHC3.4 algorithm), the predicted MHC stability>2 h (NetMHCStab, algorithm); the experimental MHC binding<3.2 log [IC50, nM] in the fluorescence polarization binding assay. In some embodiments, a personalized immunotherapy of the present teachings can be used in conjunction with check point inhibitors, such as but without limitation ipiplimumab therapy. In some configurations, a cancer vaccine can be generated by contacting dendritic cells obtained from the patient with at least one neoantigen peptide of the present teachings. In some configurations, the dendritic cell vaccine can then be administered to the subject. In some configurations, CD8+ T cells be obtained from PBMC samples from the subject, and CD8+ T cells that recognize the at least one neoantigen are isolated using cell sorting. In various configurations, the cell sorting can comprise using an affinity column or affinity beads. In some configurations, sorted CD8 + T cells that recognize neoantigens can be expanded using methods as described herein. In some configurations, the expanded T cells can then be administered to the subject.
In various configurations, the present teachings include a series of analytical steps for identification of neo-antigens from somatic tumor missense mutations, as illustrated in
The methods and compositions described herein utilize laboratory techniques well known to skilled artisans, and can be found in laboratory manuals such as Sambrook, J., et al., Molecular Cloning: A Laboratory Manual, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y., 2001; Methods In Molecular Biology, ed. Richard, Humana Press, NJ, 1995; Spector, D. L. et al, Cells: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y., 1998; and Harlow, E., Using Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y., 1999. Methods also are as described herein and in publications such as Linette, G. P. et al., Clin. Cancer Res. 11, 7692-7699, 2005; Carreno, B. M. et al., J. Immunol. 188, 5839-5849, 2012; and Carreno, B. M., et al., J. Clin. Invest. 123, 3383-3394, 2013.
In order to determine the safety, tolerability and immunological responses to Amino Acid Substitutions (AAS)-peptides formulated in an mDC vaccine, the following protocols were followed.
Human subjects, Eligible adult patients with newly diagnosed treatment naïve (ECOG performance status 0) stage IV cutaneous melanoma are enrolled in this clinical trial. All subjects are HLA-A*0201*, had gp100+ biopsy-proven (HMB45+, immunohistochemistry) melanoma metastases, have no evidence of autoimmune disorder, and are negative for HIV, HBV, and HCV. Leukapheresis was performed to obtain PBMCs from patients and healthy donors through the Barnes Jewish Hospital blood bank. For trial patients, leukapheresis is performed prior to treatment and after D3 and D6. Patients are not prescreened for IL-12p70 DC production prior to treatment. Prior to treatment, baseline imaging is performed by MRI scan of brain and CT scan of the chest/abdomen/pelvis with i.v. contrast.
All patients were enrolled in clinical trial (NCT00683670, BB-IND 13590) and signed informed consents that had been approved by the Institutional Review Board of Washington University. All subjects were HLA-A*02:01*, had no evidence of autoimmune disorder and were negative for HIV, HBV, and HCV. Leukapheresis was performed, prior to treatment and after the 3rd mature dendritic cell (DC) vaccination, at Barnes Jewish Hospital blood bank (Saint Louis, Mo.). Patients were not prescreened for interleukin (IL)-12p70 DC production prior to treatment. Prior to treatment, baseline imaging was performed by MRI scan of brain and CT scan of the chest, abdomen and pelvis with i.v. contrast. Toxicities and adverse effects were graded according to the National Cancer Institute Common Toxicity Scale (version 3.0). Informed consent for genome sequencing was obtained for all patients on protocols approved by the Institutional Review Board of Washington University.
Patient MEL21 was a 54-year-old man diagnosed with stage 3C cutaneous melanoma of the right lower extremity in 2010. The BRAF V600E mutation was detected. Surgery was performed to excise 2 cm inguinal lymph node and numerous in transit metastases. He developed recurrent in transit metastases and deep pelvic adenopathy in May 2012 and was given ipilimumab (3 mg/kg×4 doses) with stable disease until late 2013. Disease progression was noted with increasing 2 cm external iliac, 1.2 cm inguinal, and 7 mm retrocrural adenopathy. Three surgically resected melanoma lesions (inguinal lymph node Jan. 30, 2011, leg skin May 10, 2012, leg skin Jun. 6, 2013) and PBMC were submitted for genomic analysis in order to identity somatic missense mutations. The patient provided written'informed consent for the study and underwent apheresis, and received cyclophosphamide 4 days prior to administration of the first vaccine dose. He received a <total of three vaccine doses without side effect or toxicity. Re-staging CT showed stable disease and be remains in follow up 9 months later.
Patient MEL38 was a 47-year-old woman diagnosed with stage 3C cutaneous flank melanoma and underwent surgical resection of an axillary lymph node in 2012. The BRAF V600E mutation was detected. She developed recurrent disease in the skin and axilla that was surgically resected. A few months later, CT imaging confirmed metastatic disease in the right lung and axilla and she was given ipilimumab (3 mg/kg×4 doses) in May 2012 with complications of grade 2 autoimmune colitis requiring prednisone taper and later, grade 3 hypophysitis requiring replacement therapy with levothyroxine and hydrocortisone. Disease progression was noted 12 months later with new lung and skin metastases. Vemurafenib was administered for two months with no response in August 2013. Three surgically resected melanoma lesions (axilla lymph node Apr. 19, 2012, skin breast Feb. 14, 2013, skin abdominal wall Apr. 16, 2013) and PBMC were submitted for genomic analysis in order to identify somatic missense mutations. Further disease progression was evident with 3 lung nodules measuring 12 mm, 5 mm, and 5 mm in diameter. The patient provided written informed consent for the study and underwent apheresis, and received cyclophosphamide 4 days prior to the first vaccine dose. She received a total of three vaccine doses without side effect or toxicity. Re-staging CT showed 30% tumor reduction; however, the following CT examination 12 weeks later showed interval increase of tumor size back to baseline dimensions with no new sites of disease. The patient remains with stable disease for the past 8 months.
Patient MEL218 was a 52-year-old man diagnosed with stage 3C cutaneous melanoma on the left lower extremity in 2005. The BRAF mutation V600E mutation was detected when tested later on archived tumor. He underwent surgical resection and received adjuvant interferon for 6 months but had disease recurrence that was surgically resected on several occasions. In 2008, he developed disease progression with extensive in transit and subcutaneous metastases on the left leg with bulky inguinal nodal metastasis deemed unresectable. He received ipilimumab (10 mg/kg×14 doses) on clinical trial from 2008-2012 with complete response. One surgical specimen (inguinal lymph node Apr. 4, 2005) and PBMC were submitted for genomic analysis to identify somatic missense mutations. The patient provided written informed consent for the study and underwent apheresis, and received cyclophosphamide 4 days prior to the first vaccine dose. He received a total of three vaccine doses administered in the adjuvant setting without side effect or toxicity. Re-staging PET-CT imaging confirms no evidence of recurrent or metastatic disease. The patient remains in complete remission and continues in follow up.
Patient MEL69 was a 61-year-old man diagnosed with stage 3C cutaneous melanoma in 2012. Surgery was performed to excise the primary site and the axillary adenopathy. A total of 3 lymph nodes contained metastatic melanoma. The BRAF V600E mutation was detected. The patient received adjuvant Interferon for 5 months but this was discontinued after progression and development of metastatic disease. The patient was given vemurafenib for 10 months but progressed with new sites of disease. Dabrafenib and trametinib combination systemic therapy was administered for 7 additional months until progression. Several new sites of metastatic disease including a solitary brain lesion were resected. His subsequent course was complicated by malignant pericardial effusion and deep venous thrombosis. After appropriate treatment, he improved. Two surgically resected melanoma lesions (MEL69A2, limb and MEL69B2, scalp) and PBMC were submitted for genomic analysis in order to identify somatic missense mutations. The patient provided written informed consent, underwent apheresis, and then received cyclophosphamide 4 d prior to the first vaccine dose. He received a total of 2 vaccines doses without side effect or toxicity. Re-staging CT examination confirmed disease progression and the patient was removed from the study and enrolled in hospice care.
Patient MEL66 was a 43-year-old female diagnosed initially with stage 3B cutaneous melanoma in 2013. Surgery was performed to excise in transit metastases and the BRAF V600E mutation was detected. Subsequent imaging confirmed metastatic disease in the lung and retroperitoneal cavity deemed unresectable. She received several doses of ipilimumab and developed grade 3 autoimmune colitis treated with corticosteroids. After her recovery, disease progression was noted and combination therapy with dabrafenib/trametinib was begun. Disease progression was noted after 6 months of treatment. Surgical resection of several metastatic lesions was performed to render the patient disease-free. Two surgically resected melanoma lesions (ME1-66A, skin and. MEL66D, soft tissue) and PBMC were submitted for genomic analysis in order to identify somatic missense mutations. The patient provided written informed consent, underwent apheresis, and then received cyclophosphamide 4 d prior to the first vaccine dose. She received a total of 3 vaccine doses without side effect or toxicity. Re-staging Ct confirmed no evidence of disease recurrence and the patient remains in remission with no evidence of disease 4 months in follow up with no additional therapy.
Cyclophosphamide (300 mg/m2) was given 72 hours prior to D1 with the intention of eliminating Tregs (Hoons, D. S., et al., Cancer Res., 50, 5358-5364, 1990). All mature dendritic cell (mDC) vaccine doses were prepared at the time of immunization from either freshly isolated (D1) or cryopreserved (D2-D6) PBMCs (all derived from the same leukapheresis collection). A GMP-grade CD40Lexpressing K562 cell line (referred to as K463H), used for maturation of DCs, is generated, selected, and maintained under serum-free (Stemline, S1694 media) conditions. For each vaccine dose, monocyte-derived immature dendridic cells (iDCs) were generated as described previously (Linette, G. P., et al., Clin. Cancer Res., 11, 7692-7699, 2005) by culturing the PBMC adherent fraction in RPMI 1640 with 1% human AB-serum (DC media) supplemented with 100 ng/ml GM-CSF (Berlex) and 20 ng/ml IL-4 (CellGenix). 6 days after culture initiation, iDCs were harvested, washed in PBS, and cultured for an additional 24 hours in DC media (iDC control) or DC media with irradiated (100 Gy) K463H (5:1 DC/K463H ratio) and 100 U/ml IFN-γ (Actimmune; InterMune Inc.) to generate mDCs. 2 hours prior to infusion, mDCs were pulsed with (50 μg/106 cells/ml) peptide. For infusion, mDCs were resuspended in 50 ml normal saline supplemented with 5% human serum albumin and administered over 30 minutes by i.v. infusion after premedication with 650 mg acetaminophen.
mDC infusions were given i.v. every 3 weeks for 6 doses in the outpatient clinic. A restaging CT scan of the chest/abdomen/pelvis with i.v. contrast was performed after D3 and D6 and then every 2 months thereafter until disease progression. If clinical or radiographic disease progression was evident, the patient was removed from the study. For D1, patients received 1.5+107 DCs per peptide (6×107 DCs total); for D2-D6, patients received 5×106 DCs per peptide (2×107 DCs total). Patients underwent clinical evaluation prior to each mDC infusion. Toxicities and adverse effects were graded according to the National Cancer Institute Common Toxicity Scale (version 3.0). Clinical response was assessed by measurement of assessable metastatic deposits by CT scan, MRI scan, or direct measure of cutaneous deposits. The RECIST (v1.0) group system was used (Therasse, P., et al., J. Nat'l. Cancer Inst., 92, 205-216, 2000).
Immunologic monitoring (Examples 1-10). Immunologic analysis to evaluate the kinetics and magnitude of T cell response to gp100 peptides was performed using PBMCs collected weekly (prior to vaccination and until week 21. Fresh PBMCs obtained by Ficoll-Hypaque gradient centrifugation were adjusted to 2×106 cells/ml in Stemline media (Sigma-Aldrich) containing 5% human AB-serum, and dispersed at 1 ml/well in 24-well plates. Cultures were set up for the gp100 peptides and the CMV pp65 peptide (positive peptide control). Cultures were pulsed with 40 μg/ml peptide and 50 U/ml IL-2 fed starting at 48 hours and every other day thereafter. On day 12 (peak of response; the inventors' unpublished observation), cultures were harvested, counted, and stained for flow cytometry analysis. To assess the antigen-specific T cell frequency, cells were stained with HLAA*0201/peptide tetramers (Beckman Coulter) for 30 minutes at room temperature, followed by addition of FITC-conjugated CD4, CD14, CD19, and CD56 and allophycocyanin-conjugated CD8 (Invitrogen) for 15 minutes at 4° C. Cells were washed and resuspended in FACS buffer, and 7AAD was added 5 minutes before analysis. Control CMV pp65-specific CD8+ T cells were detected in all CMV-seropositive patients before and after immunization. A negative HLA-A*0201/HIV gag peptide tetramer control was included. 25,000 events in the CD8+ gate were collected using a hierarchical gating strategy that included FSC/SSC and excluded 7AAD+ (dead) cells and CD4+CD14+CD19+CD56+ cells. Data were acquired and analyzed using Flow-Jo software.
Cyclophosphamide (300 mg/m2) was given 96 h prior to the first DC dose with the intention of eliminating Tregs. All mature DC (mDC) vaccine doses were prepared at time of immunization from either freshly isolated (D1) or cryopreserved (D2-3) PBMC (all derived from same leukapheresis collection). For each vaccine, dose, monocyte-derived immature DCs were generated in 100 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF, Berlex) and 20 ng/mL IL-4 (Miltenyi Biotec) as described (Carreno, B. M., et al., J. Clin. Invest. 123, 3383-3394, 2013; Linette G P, et al., Clin. Cancer Res 11, 7692-7699, 2005) by culturing the PBMC adherent fraction in RPMI 1640 with 1% human AB-serum (DC media) supplemented with 100 ng/ml GM-CSF (Berlex) and 20 ng/ml IL-4 (CellGenix). Six days after culture initiation, immature DCs were cultured with irradiated (10,000 rad) GMP-grade CD40L-expressing K562 cells (Carreno, B. M., et al., J. Clin. Invest., 123, 3383-3394, 2013), 100 u/mL IFN-γ (Actimmune, InterMune Inc.), poly I:C (Invivogen, Inc) and R848 (Invivogen, Inc.) for 16 h to generate mDC. Two hours prior to infusion, mDC were pulsed (50 μg/106 cells/mL) separately with each peptide (7 AAS-peptides and 2 gp100 peptides, G209-2M and G280-9V) and, for dose 1 only, influenza virus vaccine (Fluvirin Novartis) was added to provide a source of recall antigen for CD4+ T cells. IL-12p70 production by vaccine DC was measured by ELISA (eBioscience) in accordance to the manufacturer's instructions. The initial priming dose was 1.5×107 DC per peptide (1.35×108 DC total), in remaining doses, patients received 5×106 DC per peptide (4.5×107 DC total). mDC were resuspended in 50 mL normal saline supplemented with 5% human serum albumin and administered over 30 min by intravenous infusion after premedication with acetaminophen 650 mg. Patients underwent clinical evaluation prior to each mDC infusion.
Cytokine Production
DC IL-12p70 and IL-12p40 production is measured by ELISA (eBioscience) according to the manufacturer's instructions. Production of additional cytokines and chemokines by DCs is determined using MILLIPLEX map Human Cytokine Panels I and II (EMD Millipore). For production of cytokines by T cells, G280-9V-specific T cells are expanded using mDCs and AT-SCT as described previously (infra and Carreno, B. M., et al., J. Immunol. 188, 5839-5849, 2012). The frequency of antigen-specific T cells after secondary stimulation is 2%-52%, as determined by HLA-A*0201/peptide tetramers (NIH tetramers Facility or Beckman Coulter). T cells are restimulated as described infra (Carreno, B. M., et al., J. Immunol. 188, 5839-5849, 2012), supernatants are collected at 24 hours, and production of cytokines is determined using MILLIPLEX® map Human Cytokine Panel I (EMD Millipore).
CD8+ T cells were isolated from PBMCs using a CD82 negative-selection kit (Miltenyi Biotec, Auburn, Calif.). Purified CD8+ T cells were cultured at a 20:1 ratio with irradiated (2500 rad) autologous mature DC (mDC) pulsed with peptide in Stemline media (S1694; Sigma-Aldrich, St. Louis, Mo.) supplemented with pooled human sera (Stemline-5), Human IL-2 (10-50 U/ml; Chiron, Emeryville, Calif.) was added every 2 d starting 48 h after culture initiation. Fourteen days after DC stimulation, T cell cultures were harvested, characterized for neo-antigen specific frequencies using HLA/peptide tetramers (see below), and restimulated with irradiated (10,000 rad) Single Chain Trimers (SCT; U.S. Pat. No. 8,518,697; U.S. Pat. No. 8,895,020; Carreno, B. M., et al., J. Immunol., 188, 5839-5849, 2012) or amino-terminal extended peptide MHC class I single-chain trimer (AT-SCT)-expressing K562 cells at a 1:1 ratio. Cultures were initiated in either six-well plates (106 each T and SCT or AT-SCT) or T25 flask (5×106 each) using Stemline-5. Twenty-four hours after stimulation, cultures were supplemented with IL-2 (500 U/ml), and viable cell counts were performed daily.
Cell concentrations were maintained at 5×105/ml throughout the culture period. For large-scale expansion, T cells were cultured in gas-permeable Lifecell bags (Nexell Therapeutics, Emeryville, Calif.). On days 10-14 of secondary stimulation, the percentage of tetramer+ cells and the number of viable cells were used to determine tetramer yields and tetramer folds.
For analysis of cytokines secreted by T cells upon SCT activation, cultures were activated 14 d after SCT or AT-SCT stimulation, T cells were restimulated with SCT at 1:1 ratio in RPMI 1640 supplemented with 5% pooled human sera (RPMI-5), supernatants were collected 24 h after activation and characterized using a MILLIPLEX® cytokine kit (Millipore, Billerica, Mass.), per the manufacturer's instructions.
qRT-PCR
qRT-PCR was performed as described previously (Carreno, B. M., et al., Immunol. Cell Biol. 87: 167-177, 2009). cDNAs were prepared (2 μg total RNA), and cDNA samples were amplified in triplicate using a GeneAmp 5700 sequencer detector (Applied Biosystems). Primers used are IL-12p35 (Hs00168405_m1) and ITGAX (integrin alpha X, referred to herein as CD11c; Hs01015070_m1). Transcript levels were calculated using the relative standard curve method, using CD11c transcript levels to normalize values.
51Cr release assays to measure specific lysis have been described previously (Carreno, B. M., et al., Immunol. Cell Biol., 87: 167-177, 2009; Linette, G. P. et al., Clin. Cancer Res. 11, 7692-7699, 2005). Melanoma cell lines DM6 (HLAA2+ gp100+) and A375 (HLA-A2+gp100−) were labeled with 25 μCi 51Cr for 1 hour, washed, and tested as targets in a standard 4-hour assay. Effectors were generated using PBMCs collected after D3 and cultured for 12 days in the presence of peptide (40 μg/ml) and IL2 (50 U/ml every other day). Vaccine-induced antigen-specific T cells were characterized using HLAA*0201/peptide dextramers (Immudex). To determine the avidity (effective concentration at 50% maximal lysis) of vaccine-induced T cells for antigen, T2 cells were pulsed with titrated G209-2M or G280-9V peptide concentrations for 1 hour in serum-free media followed by 51Cr (25 μCi) labeling for 1 hour, washed twice, and tested using vaccine-induced gp100-specific T cells in a standard 4-hour assay.
Student's t tests are 2-tailed (GraphPad Prism software, version 5.0). Data are presented as mean±1 SD, unless otherwise indicated. Cox regression analysis followed by likelihood-ratio test is used to evaluate whether (loge) IL-12p70 (sum) production added statistically significant information to a model of time to progression (TTP). Kaplan-Meier TTP model is used to test whether cytokine ratios added statistically significant information to a model of TTP. Wilcoxon matched-pairs analysis is used to compare IL-12p70 production between patients and healthy donors (GraphPad Prism software, version 5.0). All P values less than 0.05 were considered significant, except the Cox proportional hazard model, which used a lower threshold of significance (P<0.048) to adjust for 1 interim analysis of this endpoint.
Peptides were obtained lyophilized from American Peptide Company (>95% purity), dissolved in 10% DMSO in sterile water and tested for sterility, purity, endotoxin and residual organics. Peptide binding to HLA-A*02:01 was determined by T2 assay (Elvin et al. 1993 J. Immunol. Methods 158, 161) or using a fluorescence polarization assay (Pure Protein, L.L.C.) (Buchli, R., et al., Biochemistry 44, 12491-12507, 2005). The affinity scale of this latter assay is: high binders: log (IC50 nM)<3.7; intermediate binders: log (IC50 nM) 3.7-4.7; low binders: log (IC50 nM) 4.7-5.5; and very low binders: log (IC50 nM)≧6.0 (11).
Burrows-Wheeler Aligner (BWA; Li, H. and Durbin R., Bioinformatics 25, 1754-1760, 2009) is a reference-directed aligner that is used for mapping low-divergent sequences against a large reference genome, and consists of separate algorithms designed for handling short query sequences up to 100 bp, as well as longer sequences ranged from 70 bp to 1 Mbp.
Picard (Broad Institute, Cambridge, Mass.) is a set of Java-based command-line tools for processing and analyzing high-throughput sequencing data in both Sequence Alignment/Map (SAM) text format and SAM binary (BAM) format. The ‘MarkDuplicates’ utility within Picard examines aligned records in the supplied SAM or BAM file to locate duplicate molecule and can be used to flag and/or remove the duplicate records.
SAMtools (Li, H., et al., Bioinformatics, 25, 2078-2079, 2009) is a suite of programs for interacting with and post-processing alignments in the SAM/BAM format to perform a variety of functions like variant calling and alignment viewing as well as sorting, indexing, data extraction and format conversion.
Somatic Sniper (Larson, D. E., et al., Bioinformatics, 28, 311-317) is used to identify single nucleotide positions that are different between tumor and normal BAM files. It employs a Bayesian comparison of the genotype likelihoods in the tumor and normal, as determined by the germline genotyping algorithm implemented in the MAQ and then calculates the probability that the tumor and normal genotypes are different.
VarScan (Koboldt D. C., et al., Genome Research, 22, 568-576, 2012; Koboldt, D. C., et al., Bioinformatics 25, 2283-2285, 2009,) is a software program that detects somatic variants (SNPs and indels) using a heuristic method and a statistical test based on the number of aligned reads supporting each allele using an input SAMtools pileup/mpileup file. For tumor-normal pairs, it further classifies each variant as Germline, Somatic, or LOH, and also detects somatic copy number changes.
Strelka (Saunders, C. T., et al., Bioinformatics 28, 1811-1817, 2012) is an analysis package designed to detect SNVs and small indels from the sequencing data of matched tumor-normal samples. It is specifically designed to detect somatic variants at lower frequencies typically encountered in tumors due to high sample impurity or sub-clone variation, while maintaining sensitivity.
TopHat (Trapnell. C., et al., Bioinformatics, 25, 1105-1111, 2009; Kim, D., et al., Genome Biol., 14, R36, 2013) is a fast splice junction mapper for RNA-Seq reads that aligns reads to mammalian-sized genomes in order to identify exon-exon splice junctions. It uses the ultra high-throughput short read aligner Bowtie, and then analyzes the mapping results to identify splice junctions between exons.
Cufflinks (Trapnell, C., et al., Nat. Protoc., 7, 562-578, 2012) is a software program for transcriptome assembly and differential expression analysis for RNA-Seq data. It assembles transcripts from aligned RNA-Seq reads, estimates their abundances based on how many reads support each one, taking into account biases in library preparation protocols, and then tests for differential expression and regulation in RNA-Seq samples.
Flexbar (Dodt, M., et al., Biology (Basel), 1, 895-905, 2012) is a software package that preprocesses high-throughput sequencing data efficiently by demultiplexing barcoded runs and removing adapter sequences. Additionally, it supports trimming as well as filtering features; thereby aiming to increase read mapping rates and improve genome and transcriptome assemblies.
NetMHC 3.4 server (Nielsen, M., et al., Protein Sci., 12, 1007-1017, 2003; Lundegaard, C., et al., Nucleic Acids Res., 1, W509-512, 2008) makes high-accuracy predictions of major histocompatibility complex (MHC): peptide binding to a number of different HLA alleles. The predictions are based on artificial neural networks trained on different datasets (human and non-human) from several MHC alleles and position-specific scoring matrices (PSSMs).
In terms of additional filtering of variants from DNA/RNA data that would pass to analysis for identifying peptides, the following filters were used on coverage for tumor and normal, below which a variant is discarded from further consideration:
>=5× Normal coverage
>=10× Tumor coverage
<=2% Normal VAF
>=30% Tumor VAF
FPKM>1 (this is the only RNA-based filter).
In silico work flow.
The present inventors have developed an in silico automated pipeline for neoantigen prediction (pVAC-Seq) that can utilize several types of data input from next-generation sequencing assays. First a list of nonsynonymous mutations is identified by a somatic variant-calling pipeline using exomic sequencing and transcript sequencing of both normal and tumor tissue. This variant list can then be annotated with amino acid changes and transcript sequence. The HLA-haplotypes of the patient, can be derived through clinical genotyping assays or in silico approaches. These data can be input into the pVAC-Seq workflow which implements three steps: performing, epitope prediction, integrating sequencing-based information and lastly, filtering neoantigen candidates. The following paragraphs describe the analysis methodology from preparation of inputs to the selection of neoantigen vaccine candidates via pVAC-Seq.
Prepare Input Data: HLA-Typing, Alignment, Variant Detection and Annotation
As described above, pVAC-Seq utilizes input data generated from the analysis of next-generation sequence data that includes annotated nonsynonymous somatic variants that have been translated into mutant amino acid changes, as well as patient-specific HLA haplotypes. While these data could be obtained from any appropriate variant calling, annotation and HLA typing pipeline, the inventors' approach as disclosed herein utilized the following analysis methods for preparing these input data. In brief, BWA (version 0.5.9) (Li, H. and Durbin, R., Bioinformatics, 25, 1754-1760, 2009) was used as the aligner of choice with default parameters except the number of threads was set to 4 (−t 4) for faster processing, and the quality threshold for read trimming to 5 (−q 5). The resulting alignments were de-duplicated via Picard MarkDuplicates (version 1.46; Broad Institute, Cambridge, Mass.).
In cases where clinically genotyped HLA haplotyping calls were not available, the inventors used in silico HLA typing by HLAminer (Version1)(Warren, R. L., et al., Genome Med., 4, 95, 2012) to provide HLA haplotypes from either whole genome sequence data or RNA-seq data, or by Athlates (Liu, C., et al., Nucleic Acids Res, 41, e142, 2013) when exome data were available. Typing was performed on samples of the patient's normal cells, rather than cells from the tumor sample. The two software tools were >85% concordant in the inventors' test data; both algorithms were used in order to break ties reported by HLAminer (see below).
After alignments (and optional HLA typing) were completed, somatic mutation detection was performed using the following series of steps. (1) Samtools (Li, H., et al., Bioinformatics, 25, 2078-2079, 2009; Li, H. Bioinformatics, 27, 2987-2993, 2011) mpileup v0.1.16 was run with parameters ‘−A −B’ with default setting for the other parameters. These calls were filtered based on GMS ‘snp-filter v1’ and were retained if they met all of the following rules: (a) Site is greater than 10 bp from a predicted indel of quality 50 or greater, (b) The maximum mapping quality at the site is ≧40, (c) Fewer than 3 SNV calls are present in a 10 bp window around the site, (d) The site is covered by at least 3 reads and less than 1×109 reads, and (e) Consensus and SNP quality is ≧20. The filtered Samtools variant calls were intersected with those from Somatic Sniper version 1.0.2 (Larson, D. E., et al., Bioinformatics, 28, 311-317, 2012) (params: −F vcf q 1 −Q 15), and were further processed through the GMS ‘false-positive filter v1’ (params: -bam-readcount-version 0.4-bamreadcount-min-base-quality 15-min-mapping-quality 40-min-somatic-score 40). This filter used the following criteria for retaining variants: (a) ≧1% of variant allele support comes from reads sequenced on each strand, (b) variants have ≧5% Variant Allele Fraction (VAF) (c) more than 4 reads support the variant, (d) the average relative distance of the variant from the start/end of reads is greater than 0.1, (e) the difference in mismatch quality sum between variant and reference reads is less than 50, (f) the difference in mapping quality between variant and reference reads is less than 30, (g) the difference in average supporting read length between variant and reference reads is less than 25, (h) the average relative distance to the effective 3′ end of variant supporting reads is at least 0.2, and (i) the variant is not adjacent to 5 or more bases of the same nucleotide identity (e.g. a homopolymer run of the same base), (2) VarScan Somatic version 2.2.6 (Koboldt, D. C., et al., Bioinformatics, 25, 2283-2285, 2009; Koboldt, D. C., et al., Genome Res., 22, 568-576, 2012) was run with default parameters and the variant calls were filtered by GMS filter ‘varscan-high-confidence filter version v1’. The ‘varscan-high-confidence v1’ filter employed the following rules to filter out variants (a) p-value (reported by Varscan) is greater than 0.07, (5) Normal VAF is greater than 5%, (c) Tumor VAF is less than 10% or (d) less than 2 reads support the variant. The remaining variant calls were then processed through false-positive filter v1 (params: -bam-readcount-version 0.4-bamreadcount-min-base-quality 15) as described above. (3) Strelka version 1.0.10 (Saunders, C. T., et al., Bioinformatics, 28, 1811-1817, 2012) (params: isSkipDepthFilters=1).
The consolidated list of somatic mutations identified from these different variant-callers was then annotated using our internal annotator as part of the GMS pipeline. This annotator leverages the functionality of the Ensembl database (Flicek, P, et al., Nucleic Acids Res., 41, D48-55, 2013) and Variant Effect Predictor (VEP)(McLaren, W., et al., Bioinformatics, 26, 2069-2070, 2010).
From the annotated variants, there are two components that are needed for pVAC-Seq: amino acid change and transcript sequence. Even a single amino acid change in the transcript arising from missense mutations can alter the binding affinity of the resulting peptide with the MHC Class I molecule. Larger insertions and deletions, such as, for example, those arising from frameshift and truncating mutations, splicing aberrations or gene fusions can also result in potential neoantigens. However, for the present iterations of pVAC-Seq, the inventors chose to focus their analysis on only missense mutations.
One feature of the inventor's pipeline is the ability to compare the differences between tumor neo-antigens and normal peptides in terms of the peptide binding affinity. Additionally, it leverages RNA-Seq data to incorporate isoform-level expression information and to quickly cull variants that are not expressed in the tumor. To integrate RNA-Seq data, both transcript ID as well as the entire wild-type transcript amino acid sequence can be used as part of the annotated variant file.
One component of pVAC-Seq is predicting epitopes that result from mutations by calculating their binding affinity against the Class I MHC molecule. This process involves the following steps for effectively preparing the input data as well as parsing the output.
Peptide sequences are an input to the MHC binding prediction tool, and the existing process to compare the germline normal with the tumor can be very onerous. To streamline the comparison, the inventors first build a FASTA file that consists of two amino acid sequences per variant site—wild-type (normal) and mutant (tumor). The FASTA sequence can be built using approximately 8-10 flanking amino acids on each side of the mutated amino acid. However, if the mutation is towards the end or beginning of the transcript, then the preceding or succeeding 16-20 amino acids can be taken respectively, as needed, to build the FASTA sequence. Subsequently, a key file can be created with the header (name and type of variant) and order of each FASTA sequence in the file. This can be done to correlate the output with the name of the variant protein, as subsequent epitope prediction software strips off each name.
To predict high affinity peptides that bind to the HLA class I molecule, the standalone version of NetMHC 3.4 is used. The input to this software is the HLA type of the patient, determined via genotyping or using in silico methods, as well as the FASTA file generated in the previous step comprised of mutated and wild-type 17-21-mer sequences. Typically, antigenic epitopes presented by MHC class I molecules can vary in length from 8 to 13 or 8 to 11 amino acids. Therefore, specifying the same range when running epitope prediction software is recommended.
Starting with the output list of all possible epitopes from the epitope prediction software, the inventors apply specific filters to choose the best mutant peptide incorporating candidates. First, further consideration is restricted to strong to intermediate binding peptides by focusing on candidates with a mutant (MT) binding score of less than 500 nM or less than 250 nM. Second, epitope binding calls are evaluated only for those peptides that contain the mutant amino acid (localized peptides). This filter eliminates any wild-type (WT) peptides that may overlap between the two FASTA sequences. The pVAC-seq workflow enables screening across multiple lengths and multiple alleles very efficiently. If predictions are run to assess multiple epitope lengths (e.g., 9-mer, 10-mer, etc.), and/or to evaluate all different patient HLA allele types, the inventors review all localized peptides and choose the single best binding value representative across lengths (9aa, 10aa, etc.) based on lowest binding score for MT sequence. Furthermore, they choose the ‘best candidate’ (lowest MT binding score) per mutation between all independent HLA allele types that were used as input.
Subsequently several filters are applied to ensure that the predicted neoantigens are expressed as RNA variants, and are predicted correctly based on coverage depth in the normal and tumor tissue data sets. Specifically, gene expression levels from RNA-Seq data measured as Fragments per kilobase of exon per million reads mapped (FPKM) provide a method to filter only the expressed transcripts. We used the tuxedo suite—Tophat (Trapnell, C. et al., Bioinformatics, 25, 1105-1111, 2009; Kim, D., et al., Genome Biol., 14, R36, 2013) and Cufflinks (Trapnell, C., et al., Nat. Protoc., 7, 562-578, 2012) as part of the GMS to align RNA-Seq data and subsequently infer gene expression for our in-house sequencing data. Depending on the type of RNA prep kit, OVATION® RNA-Seq System V2 (NuGEN Technologies, Inc. San Carlos, Calif.) or TRUSEQ® Stranded Total RNA Sample Prep kit (ILLUMINA®, Inc. San Diego, Calif.), used, Tophat was run with the following parameters: Tophat v2.0.8 ‘-bowtie-version-2.1.0’ for OVATION®, and ‘-library-type fr-firststrand-bowtie-version=2.1.0’ for TRUSEQ®. For OVATION® data, prior to alignment, paired 2×100 bp sequence reads were trimmed with Flexbar version 2.21 (Dodt, M., et al. Biology (Basel), 1, 895-905, 2012.) (params: -adapter CTTTGTGTTTGA (SEQ. ID NO: 474)-adapter-trim-end LEFT-nono-length-dist-threads 4-adapter-min-overlap 7-maxuncalled 150-min-readlength 25) to remove single primer isothermal amplification adapter sequences. Expression levels (FPKM) were calculated with Cufflinks v2.0.2 (params-max-bundle-length=10000000-num-threads 4).
For selecting unique vaccine candidates, targeting the best ‘quality’ of mutations is an important factor for prioritizing peptides. Sequencing depth as well as the fraction of reads containing the variant allele (VAF) are used as criteria to filter or prioritize mutations. This information was added in our pipeline via bam-readcount (Larson, D., The Gnome Institute at Washington University). Both tumor (from DNA as well as RNA) and normal coverage are calculated along with the VAF from corresponding DNA and RNA-Seq alignments.
Since manufacturing antigenic peptides can be one of the most expensive steps in vaccine development and efficacy depends on selection of the best neoantigens, the inventors filter the list of predicted high binding peptides to the most highly confident set, primarily with expression and coverage based filters.
Depth based filters: any variants with normal coverage <=5× and normal VAF of >=2% can be filtered out. The normal coverage cutoff can be increased up to 20× to eliminate occasional misclassification of germline variants as somatic. Similarly, the normal VAF cutoff can be increased based on suspected level of contamination by tumor cells in the normal sample. For tumor coverage from DNA and/or RNA, a cutoff can be placed at >=10× with a VAF of >=10% or 30%. This can ensure that neoantigens from the major clones in the tumor are included, but the tumor VAF can be lowered to capture more variants, which may or may not be present in all tumor cells. Alternatively, if the patients are selected based on a pre-existing disease-associated mutation such as BRAF V600E in the case of melanoma, the VAF of the specific presumed driver mutation can be used as a guide for assessing clonality of other mutations.
Expression based filters: as a standard, genes with FPKM values of greater than zero are considered to be expressed. The inventors slightly increase this threshold to 1, to eliminate noise. Alternatively, the FPKM distribution (and the corresponding standard deviation) can be analyzed over the entire sample, to determine the sample-specific cutoffs for gene expression. Spike-in controls can also be added to the RNA-Seq experiment to assess quality of the sequencing library and to normalize gene expression data. This filtered list of mutations can be manually reviewed via visual inspection of aligned reads in a genome viewer like IGV (Robinson, J. T., et al., Nat Biotechnol., 29, 24-26, 2011; Thorvaldsdottir, H., et al. Brief Bioinform., 14, 178-192, 2013) to reduce the retention of obvious false positive mutations.
For functional characterization, neoantigen-specific T cell lines were generated using autologous mDC and antigen loaded artificial antigen presenting cells at a ratio of 1:1 as previously described (Carreno, B. M., et al., J. Immunol., 188, 5839-5849, 2012). To determine the peptide avidity (effective concentration at 50% maximal lysis, EC50) of neoantigen-specific T cells, T2 cells were pulsed with titrated peptide concentrations for 1 h, followed by 51Cr (25μCi) labeling for 1 h, washed twice and tested in a standard 4 h 51Cr release assay using neoantigen-specific cells as effectors. For production of cytokines, neoantigen-specific T cells were restimulated using artificial antigen presenting cells in the presence or absence of peptide, supernatants collected at 24 h and cytokine produced determined using MILLIPLEX® MAP Human Cytokine Panel I (EMD Millipore).
The present teachings it descriptions that are not intended to limit the scope of any aspect or claim. Unless specifically presented in the past tense, an example can be a prophetic or an actual example. The examples and methods are provided to further illustrate the present teachings. Those of skill in the art, in light of the present disclosure, will appreciate that many changes can be made in the specific embodiments that are disclosed and still obtain a like or similar result without departing from the spirit and scope of the present teachings.
This example illustrates the clinical use of common cancer antigen peptides and the difficulties of using matured dendritic cells in cancer vaccines.
Vaccination was performed with HLA-A*0201-restricted gp100 melanoma antigen-derived peptides (G209-2M, and G280-9V) (Carreno, B. M., et al., J. Clin. Investigation, 123, 3383-3394, 2013; Kawakami, Y., et al., J. Immunol., 154, 3961-3968, 1995; Skipper, J. C., et al., Int. J. Cancer, 82, 669-677, 1999) using autologous peptide-pulsed, CD40L/IFN-γ-activated mature DCs (mDCs). The top of
The bottom left of
This example illustrates techniques of maturing DC that overcome the limitations discussed in Example 1.
Based on the results obtained in Example 1, different DC maturation techniques were required to increase clinical response to cancer antigens. The inventors therefore tested maturation signals for dendritic cells. Immature DC were stimulated with a combination of CD40L/IFN-γ plus poly I:C (30 ug/mL, TLR3 agonist) and R848 (5 μg/mL, TLR8 agonist) (P8-P10) for 24 h and supernatants assayed for IL-12. As a control, data from immature dendritic cells stimulated with CD40L/IFN-γ(patients P1-P7; Carreno, B. M., et al., J. Clin. Invest. 123, 3383-3394, 2013) were plotted on the same graph. The results depicted in
A combination of innate and adaptive signals for DC maturation enhances the kinetics of the immune responses to gp100 (g209-2M and G280-9V) antigens.
This example illustrates in silico analysis of missense mutations found in melanoma tumors.
The diagram in
In various embodiments, the present teachings include analysis of missense mutations by prediction algorithms for binding to HLA-A*0201. Table 1 shows the chromosomal (CHR) location, genomic alignment position and nucleotide change encoding missense mutation in metastases (breast, abdominal wall) derived from a patient. Exomic variant allele fraction (under exome column) for each mutation as well as gene encoding mutation and amino acid change are shown. One mutation in OR5K2 is unique to breast metastasis, while mutations in CCDC57 and IL17Ra are unique to abdominal wall metastasis. Proteins encoding missense mutations were analyzed using the NetMHC and NetMHCstab algorithms in order to predict mutation-containing peptides (9-11 amino acid in length) that may bind to any of patient's HLA-class I molecules. Candidate peptides to consider for a vaccine are selected based on variant frequencies (exome, transcriptome>10), expression (FPKM>1) and HLA class I affinity (<250 nM0 and stability (>2 h). In Table 1, mutated peptides fulfilling these criteria are highlighted in bold. NR=not recorded.
This example illustrates the in vitro binding of neoantigen peptides to HLA class I molecules.
In some embodiments, the present teachings disclose HLA class I binding capacity of peptides containing tumor-specific missense mutations. The binding capacity of missense mutation-containing peptides is experimentally evaluated using a flow cytometric assay. Peptide binding to cell surface HLA class I can lead to stable peptide/HLA class I complexes that can be detected using a HLA-class I allele specific antibody. Four control peptides can be included in the assay, two known HLA-A*0201 binding peptides (FluM1,G280-9V) and 2 negative controls (G17, NP265). In the graph shown in
This example illustrates the translation of tumor missense imitations into patient-specific vaccines.
This example illustrates CD8+ T cell response to mutation containing peptides.
In some embodiments, the present teachings include vaccination with tumor-specific missense mutations to elicit CD8+ T cell immunity. As shown in
In some embodiments, predicted affinities (
In some embodiments, the present teachings include vaccine-induced CD8+ T cells directed at tumor missense mutations display high replicative potential. As shown in
This example illustrates the specificity of neoantigen peptide recognition by CD8+ T cells.
In various embodiments, the present teachings include disclosure of discrimination between mutated and wild-type sequences by vaccine-induced CD8+ T cells.
As illustrated in
For therapeutic use of vaccine-induced T cells, it can be important to determine whether responses elicited by MUT peptides can cross-react with WT sequences. T cell responses that cannot discriminate between MUT and WT sequences may have adverse effects if given to patients as part of adoptive cell therapy.
To examine cross-reactivity, T2 cells were pulsed with MUT or WT peptide at the indicated concentrations, labeled with 51CR-chromium and used as target in a cytotoxic assay. Vaccine-induced T cells were incubated with peptide-pulsed T2 cells and 51Cr-Chromium release measured at 4 h. Results obtained with T cell lines specific for 3 mutated peptides are shown in
This example illustrates that vaccine-induced mutation-specific T cells discriminate between mutated (MUT) and wild type (WT) sequences and recognized processed and presented antigens. Neoantigen-specific T cells recognition of mutated (closed circles) and wild type (open circles) peptides was determined in a standard 4 h 51Cr-release assay using peptide titrations on T2 (HLA-A*02:01) cells. Percent specific lysis of triplicates (mean±standard deviation) is shown in
This example illustrates cytokine production in response to neoantigen peptides.
In various embodiments, a vaccine of the present teachings can induce CD8+ T cells to display a Tc1 profile.
Substantial evidence supports the hypothesis that Th2/Tc2 immune polarization correlates with worse disease outcome in patients with cancer (Fridman, W. H., et al., Nat. Rev. Cancer, 12, 298-306, 2012). In our previous study (Carreno, B. M., et al., J. Clin. Invest., 123, 3383-3394, 2013) the inventors demonstrated that patients presenting vaccine-induced T cells displaying a Tc1 (high IFN-γ, low IL-4, -5, -13 production) benefited from vaccine as determined by an increased time to progression. Thus, we determined production of cytokines upon antigen stimulation as described above. In these studies, neo-antigen-specific AKAP13 (Q285K) T cells were incubated with peptide-pulsed SCD-expressing cells and supernatants collected 24 h after stimulation. Cytokine production was determined using a multi-plex bead assay. Results illustrated in
This example illustrates successful treatment of melanoma in mice using a vaccine of the present teachings.
In some embodiments, the present teachings disclose that adoptive transfer of human antigen-specific T cells can lead to melanoma rejection. In investigations by the inventors, humanized mice were inoculated i.v. with luciferase-expressing melanoma. Ten days later (indicated by vertical arrows
This example illustrates selection of neoantigens for further study.
Tumor missense mutations (MM), translated into amino acid substitutions (AAS), may provide a form of antigens that the immune system perceives as foreign, which elicits tumor-specific T cell immunity (Wölfel, T., et al., Science, 269, 1281-1284, 1995; Coulie, P. G., et al., Proc. Nat'l. Acad. Sci. USA 92, 7976-7980, 1995; van Rooij, N. et al., J. Clin. Oncol., 31, e439-e442, 2013; Robbins, P. F., et al., Nat. Med., 19, 747-752, 2013). In these experiments, three patients (MEL21, MEL38 and MEL218) with stage III resected cutaneous melanoma were consented for genomic analysis of their surgically excised tumors and subsequently enrolled in a phase 1 clinical trial with autologous, functionally mature, interleukin (IL)-12p70-producing dendritic cell (DC) vaccine (
All tumor samples were flash frozen except one from MEL 21 (skin, Jun. 6, 2013), which was formalin-fixed paraffin embedded. Peripheral blood mononuclear cells (PBMC) were cryopreserved as cell pellets. DNA samples were prepared using QIAAMP® DNA Mini Kit (Qiagen) and RNA using High Pure RNA Paraffin kit (Roche), DNA and RNA quality was determined by NANODROP® 2000 and quantitated by the QUBIT® Fluorometer (Life Technologies). For each patient, tumor/PBMC (normal) matched genomic DNA samples were processed for exome sequencing with one normal and two tumor libraries, each using 500 ng DNA input (Service, S. K. et al., P.L.o.S. Genet., 10, e1004147, 2014). Exome sequencing was performed to identify somatic mutations in tumor samples.
Tumor M M, translated as AAS-encoding nonamer peptides, were filtered through in silico analysis to assess HLA-A*02:01 peptide binding affinity (Nielsen, M. et al., Protein Sci., 12, 1007-1017, 2003). Alignment of exome reads was performed using the inventors' Genome Modeling System (GMS) processing-profile. This pipeline uses BWA (version 0.5.9) for alignment with default parameters except for the following: ‘−t 4 −q 5’. All alignments were against GRCh37-lite-build37 of the human reference genome and were merged and subsequently de-duplicated with Picard (version 1.46). Detection of somatic mutations was performed using the union of three variant callers: 1) SAMtools version r963 (params: −A −B) filtered by snp-filter v1 and further intersected with Somatic Sniper version 1.0.2 (params: −F vcf q 1 −Q 15) and processed through false-positive filter v1 (params: -bam-readcount-version 0.4-bamreadcount-min-base-quality 15 min-mapping-quality 40-min-somatic-score 40) 2) VarScan Somatic version 2.2.6 filtered by varscan-high-confidence filter version v1 and processed through false-positive filter v1 (params, -bam-readcount-version 0.4bamreadcount-min-base-quality 15), and 3) Strelka version 1.0.10 (params: isSkipDepthFilters=1). Amino acid substitutions (AAS) corresponding to each of the coding missense mutations (MM) were translated into a 21-mer amino acid FASTA sequence, with ideally 10 amino acids flanking the substituted amino acid on each side.
Each 21-mer amino acid sequence was then evaluated through the HLA class I peptide binding algorithm NetMHC 3.4 to predict high affinity HLA-A*02:01 nonamer peptides for the AAS—as well as the WT sequence to calculate differences in binding affinities (8, 32). Any peptides with binding affinity IC50 value<500 nM were considered for further analysis.
Experimental expression of genes encoding predicted HLA-A*02:01 peptide candidates was determined by cDNA capture. All RNA samples were DNase-treated with TURBO DNA-FREE™ kit (Invitrogen) according to the manufacturer's instructions; RNA integrity and concentration were assessed using Agilent Eukaryotic Total RNA 6000 assay (Agilent Technologies) and QUANT-IT™ RNA assay kit on a QUBIT™ Fluorometer (Life Technologies Corporation).
Given the dynamic nature of genomic technologies, multiple overlapping methods were tested. However, results for tumors within a patient (Tables 2-4) are consistent with one methodology: NuGen OVATION® V2 for MEL38 and MED218, Illumina TRUSEQ® Stranded for MEL21. The MicroPoly(A)PURIST™ Kit (Ambion) was used to enrich for poly(A) RNA from MEL218 and MEL38 DNAse-treated total RNA; MEL21 RNA was ribo-depleted using the RIBO-ZERO™ Magnetic Gold Kit (EpiCeture, Madison Wis.) following the manufacturer protocol. The inventors used either the OVATION® RNA-Seq System V2 (NuGen, 20 ng of either total or polyA RNA), or the OVATION® RNA-Seq FFPE System (NuGen, 150 ng of DNase-treated total RNA) or the TRUSEQ® Stranded Total RNA Sample Prep kit (Illumina, 20 ng ribosomal RNA-depleted total RNA) for cDNA synthesis. All NuGen cDNA sequencing libraries were generated using NEBNEXT® ULTRA™ DNA Library Prep Kit for ILLUMINA® with minor modifications.
All NuGEN generated cDNA was processed as described previously (Cabanski, C. R., et al., J. Mol. Diagn., 16, 440-451, 2014). Briefly, 500 ng of cDNA was fragmented, end-repaired, and adapter-ligated using IDT synthesized “dual same index” adapters. The TRUSEQ® stranded cDNA was also end-repaired and adapter-ligated using IDT synthesized “dual same index” adapters. These indexed adapters, similar to Illumina TRUESEQ® HT adapters, contain the same 8 bp index on both strands of the adapter. Binning reads requires 100% identity from the forward and reverse indexes to minimize sample crosstalk in pooling strategies. Each library ligation reaction was PCR-optimized using the Eppendorf Epigradient SqPCR instrument, and PCR-amplified for limited cycle numbers based on the Ct value in the optimization step.
Libraries were assessed for concentration using the QUANT-IT™ dsDNA HS Assay (Life Technologies) and for size using the BioAnalyzer 2100 and the Agilent DNA 1000 Assay (Agilent Technologies). The ILLUMINA®-ready libraries were enriched using the Nimblegen SeqCap EZ Human. Exome Library v3.0 reagent. The targeted genomic regions in this kit cover 63.5 Mb or 2.1% of the human reference genome, including 98.8% of coding regions, 23.1% of untranslated regions (UTRs), and 55.5% of miRNA bases (as annotated by Ensembl version 73 (Flicek. P., et al., Nucleic Acids Res., 41, D48-55, 2013)). Each hybridization reaction was incubated at 47° C. for 72 hours, and single-stranded capture libraries were recovered and PCR-amplified per the manufacturer's protocol. Post-capture library pools were sized and mixed at a 1:0.6 sample: Ampure XP magnetic head ratio to remove residual primer-dimers and to enrich for a library fragment distribution between 300 and 500 bp. The pooled capture libraries were diluted to 2 nM for Illumina sequencing.
For cDNA-capture data were aligned with Tophat v2.0.8 (params: version=2.1.0 for OVATION®; -library-type fr-firststrand-bowtie-version=2.1.0 for TRUSEQ®). For OVATION® data, prior to alignment, paired 2×100 bp sequence reads were trimmed with flexbar v 2.21 (params: -adapter CTTTGTGTTTGA (SEQ ID NO: 474-adapter-trim-end LEFT-nono-length-dist-threads 4-adapter-min-overlap 7-maxuncalled 150-min-readlength 25) to remove single primer isothermal amplification adapter sequences. In seqcap, the relative expression of a transcript is proportional to the number of cDNA fragments that originate from it. Therefore, expression levels expressed as fragments per kilobase of exon per million fragments mapped (FPKM) were calculated with Cufflinks v2.0.2 (Trapnell et al. 2010, Nature Biotechnology 28, 511; params-max-bundle-length=10000000-num-threads 4). A visual review step of cDNA capture data was performed to evaluate for expression of MM identified by exome data. Both cDNA-capture and FPKM values were considered for candidate prioritization.
Peptide candidates for experimental validation were selected according to the strategy described in
HLA-A*02:01 binding was evaluated using the T2 assay (See Analysis of T cell responses) (
This example illustrates the effectiveness of personalized dendritic vaccines.
To examine the kinetics and magnitude of T cell immunity to AAS peptides upon vaccination, peripheral blood mononuclear cells (PBMC) were collected prior to vaccination and weekly thereafter. The CD8+ T cell response to each peptide was analyzed using a HLA-A*02:01/AAS-peptide dextramer assay after a single round of in vitro stimulation.
Vaccination augmented the cell response to these neoantigens with observed frequencies of 23% TMEM48 F169L+ CD8+ T cells, 64% SEC24A P469L+ CD8+T cells and 89% EXOC8 Q656P+ CD8 T cells detected, upon culture, at the peak of response (
Analysis of T cell reactivity among the three patients indicated no preferential skewing towards AAS at specific positions in the peptide sequence—that is towards TCR, contact residues or primary anchor residues (Kim, Y., et al., J. Immunol. Methods, 374, 62-69, 2011). Rather, in each patient, T cell immunity appeared to focus on the 3 AAS candidates exhibiting the highest HLAA*02:01 binding affinity while the remaining medium-high affinity peptides were nonimmunogenic (Table 5) (Nielsen M., et al., Protein Sci., 12, 1007-1017, 2003; Buchli, R., et al., Biochemistry, 44, 12491-12507, 2005). Immunogenic AAS peptides (
To characterize the function of vaccine-induced neoantigen-specific T cells, short-term expanded CD8+ T cell lines were established and antigen specificity confirmed by dextramer assay (
The cytokine production profile of these cells was characterized as previously described (Carreno, B. M., et al., J. Clin. Invest., 123, 3383-3394, 2013; Fridman, W. H., et al., Nat. Rev. Cancer, 12, 298-306, 2012). This characterization is illustrated in
This example illustrates the in vitro detection of neoantigens that are presented to immune cells in vivo.
Tandem mini-gene constructs (TMC) were used for evaluating processing and presentation of neoantigens. The structure of a representative TMC (MEL21 AAS sequences) is shown in
TMC were cloned into pMX (GFP+), expressed as retrovirus and used to transfect the HLA-A*02:01+ melanoma lines DM6 (Darrow, T. L., et al., J. Immunol., 142, 3329-3335, 1989) or A375 (obtained from ATCC and mycoplasma free). TMC expressing cells were selected by sorting for GFP+ cells expressing cell surface HLA-A*02:01/SVG9 peptide complexes as detected by a T cell receptor mimic (TCRm) monoclonal antibody (Kim S., et al., J. Immunol., 184, 4423-4430, 2010). AAS- and WT-TMC reactivity with the HLA-A*02:01/SVG9 peptide complex specific TCRm monoclonal antibody validated expression of the mini-gene constructs.
DM6 cells expressing TMC were labeled with 25μCi 51Cr for 1 h, washed and tested as targets in a standard 4 h assay using neoantigen-specific T cells as effectors (Carreno B. M. et al. 2012 J Immunol 188, 5839). DM6 cells expressing AAS—(closed rectangles) or WT—(closed circles) TMC were co-cultured with neoantigen-specific T cells at a 1:1 ratio, supernatants harvest at 16 h and IFN-γ production evaluated by ELISA as described (Carreno, B. M., et al., J Immunol., 188, 58395849, 2014; plots in
This example illustrates the use of proteomic techniques to determine which neoantigens are presented to cells in vivo.
To validate neoantigen processing and presentation, proteomic analysis was performed on peptides eluted from soluble HLA-A*02:01 molecules isolated from melanoma cells expressing a TMC encoding AAS candidates from patient MEL218 tumor (Sercarz, E. E., et al., Annu. Rev. Imunol., 11, 729-766, 1993; Assarsson, E., et al., J. Immunol., 178, 7890-7901, 2007). TMC expressing A375 melanoma cells were transfected with soluble HLA-A*02:01(sHLA-A*02:01) and single cell sorted for a high (>1000 ng/ml in static culture) sHLAA*02:01 producing clone. The sHLA-A*02:01 construct includes a C-terminal VLDLr epitope purification tag (SVVSTDDDLA SEQ ID NO. 32) that is recognized by the anti-VLDLr mAb (ATCC CRL-2197). This antibody was also used for quantification of sHLA production as the capture antibody in a sandwich ELISA, with an antibody directed against β2-microglobulin (Dako Cytomation) as the detector antibody. Cells were grown in roller bottles and sHLA/peptide complexes were purified from supernatants by affinity chromatography with the anti-VLDLr antibody (Kaabinejadian, S., et al., P.L.o.S. One, 8, e66298, 2013). Eluate fractions containing sHLA/peptide complexes were brought to a final acetic acid concentration of 10%, pooled, and heated to 78° C. in a water bath. Peptides were purified through a 3 kDa molecular weight cutoff cellulose membrane (EMD Millipore) and lyophilized.
Synthetic peptides corresponding to the mutant sequences were resuspended in 10% acetic acid in water at 1 μM, and fractionated by RP-HPLC with an acetonitrile gradient in 10 mM ammonium formate at pH 10. Peptide-containing fractions were dried and resuspended in 25 ul of 10% acetic acid and subjected to nanoscale RP-HPLC at pH 2.5 utilizing an Eksigent nanoLC coupled to a TripleTOF 5600 (AB Sciex) quadrupole time-of-flight mass spectrometer (LC/MS). Information dependent acquisition (IDA) was used to obtain MS and MS/MS fragment spectra for peptide ions. The sequence of each peptide was determined by observed mass and fragment ions, and the 1st dimension fraction number and LC/MS retention times were recorded.
Next, peptides purified from TMC expressing A375 melanoma cells were resuspended in 10% acetic acid and HPLC fractionated under the same conditions and gradient method. Reverse phase HPLC was used to reduce the complexity and determine the elution profile of the pool of soluble HLA-A*02:01 restricted peptides presented by melanoma cells, as well as, the synthetic AAS peptide mixture.
Separation and sequencing of peptides were carried out by two-dimensional liquid chromatography, followed by information dependent acquisition (IDA) generated tandem MS (MS/MS). For the first dimension, the peptide sample was loaded on a reverse-phase C18 column (pore size, 110 Å; particle size, 5 μm; 2 mm i.d. by 150 mm long Gemini column; Phenomenex) with a Michrom BioResources Paradigm MG4 high performance liquid chromatograph (HPLC) with UV detection at 215 nm wavelength. Elution was at pH 10 using 10 mM ammonium formate in 2% acetonitrile/98% water as solvent A and 10 mM ammonium formate in 95% acetonitrile/5% water for solvent B. The 1st dimension HPLC column was preequilibrated at 2% solvent B, then the peptide sample, dissolved in 10% acetic acid/water, was loaded at a flow rate of ˜120 μl/min over an 18 minute period. Then a two segment gradient was performed at 160 μl/min; the 1st segment was a 40 minute linear gradient from 4% B to 40% B, followed by an eight minute linear gradient from 40% B to 80% B. Forty peptide-rich fractions were collected and dried by vacuum centrifugation.
For the second dimension chromatography, each dried fraction was resuspended in 10% acetic acid and subjected to nano-scale RP-HPLC (Eksigent nanoLC415, AB Sciex). The second dimension nano-HPLC setup included a C18 trap column (350 μm i.d. by 0.5 mm long; ChromXP (Eksigent) with 3 μm particles and 120 Å pores and a ChromXP, C18 separation column with dimensions of 75 μm i.d. by 15 cm long packed with the same medium. A two-solvent system was utilized, where solvent A is 0.1% formic acid in water and solvent B contains 0.1% formic acid in 95% acetonitrile/5% water. Samples were loaded at 5 μL/min flow rate on the trap column and at 300 nL/min flow rate on the separation column that was equilibrated in 2% solvent B. The separation was performed by a program with two linear gradients: 10% to 40% solvent B for 70 min and then 40% to 80% solvent B for 7 min. The column effluent was connected to the nanospray III ion source of an AB Sciex TripleTOF 5600 quadrupole-time of flight mass spectrometer with the source voltage set to 2400 v.
Extracted ion chromatograms revealed the presence of an eluted peptide with a retention time within 2 minutes of synthetic EXOC8 Q656P peptide in fraction 50.
This example illustrates characterization of the composition and diversity of neoantigen-specific T cells and the effect vaccination can have on these repertoires.
Short-term ex-vivo expanded neoantigen-specific T cells were purified to 97-99% purity by cell sorting in a Sony SY3200 BSC (Sony Biotechnology) fitted with a 100 um nozzle, at 30 psi, using 561 nm (585/40) and 642 nm (665/30) lasers and cell pellets were prepared. DNA isolation and TCRβ sequencing was performed by Adaptive Biotechnologies and The Genome Institute at Washington University. Sequencing was performed at either survey (for neoantigen-specific TCRβ reference libraries) or deep (for pre- and post-vaccine CD8+ T cell populations) level (Robins, H., et al., J. Immunol. Methods, 375, 14-19, 2012; Carlson, C. S., et al., Nat. Commun., 4, 2680, 2013). TCRβ V-, D-, J-genes of each CDR3 regions were defined using IMGT (ImMunoGeneTics)/Junctional algorithms and data uploaded into the ImmunoSeq Analyzer (Adaptive Biotechnologies) for analysis. Complete amino acid identity between the reference library and pre- and post-vaccine CD8 samples was required for assigning a TCRβ match. In the reference library, TCRβ clonotypes with frequencies above 0.1% (>100-fold sequencing depth) were set as a threshold for identification of neoantigen-specific TCRβ CDR3 sequences within pre- and post-vaccine CD8+ T cell populations.
Reference T cell receptor-β (TCRβ) complementarity-determining region 3 (CDR3) sequence libraries (shown schematically in
This example illustrates vaccination of patients using multiple HLA cell types.
Distribution of somatic (exomic and missense) mutations was identified in metachronous tumors of patients MEL66 is illustrated in
Distribution of somatic (exomic and missense) mutations identified in metachronous tumors of patients MEL69 is illustrated in
The vaccine for patient MEL66 included neoantigens that bound to HLA-A*02:01 and HLA-B*08:01 molecules. The vaccine for MEL69 included neoantigens that bound to HLA-A*03:01 and HLA-A*11:01 molecules. Both vaccines were prepared by contacting the neoantigens with the patient's own dendritic cells and maturing them prior to administration in accordance with the present teachings. Representative results (dextramer assay) to neoantigens restricted to these alleles are shown (
All cited publications are hereby incorporated by reference, each in its entirety.
8
ARFGEF1
QTIDNIVFL
SEQ ID. 37
QTIDNIVFF
SEQ ID. 38
387
10867
F1637L
9
CDKN2A
KMIGNHLWV
SEQ ID. 39
EMIGNHLWV
SEQ ID. 40
14
1044
E153K
5
GPX8
LLSIVPCTV
SEQ ID. 47
LLSIVLCTV
SEQ ID. 48
52
33
L27P
6
KDM1B
IIGAGPAEL
SEQ ID. 49
IIGAGPAGL
SEQ ID. 50
469
928
G394E
X
PHKA2
LLSIIFFPA
SEQ ID. 63
LLSIISFPA
SEQ ID. 64
23
25
S264F
3
TKT
AMFWSVPTV
SEQ ID. 69
AMFRSVPTS
SEQ ID. 70
4
1525
R438W
1
TMEM48
CLNEYHLFL
SEQ ID. 71
CLNEYHLFF
SEQ ID. 72
23
3442
F169L
8
ARFGEF1
21
129
14.00
64
240
20.98
31.37
9
CDKN2A
13
49
20.97
162
38
81.00
0.18
5
GPX8
7
63
10.00
20
62
24.39
15.02
6
KDM1B
15
55
21.43
23
24
48.94
7.33
X
PHKA2
13
25
34.21
13
21
38.24
4.60
3
TKT
10
45
18.18
124
190
39.49
0.64
1
TMEM48
7
40
14.89
292
382
43.13
0.17
8
ARFGEF1
109
154
41.44
140
177
44.03
34.67
9
CDKN2A
30
17
63.83
168
26
86.60
0.05
5
GPX8
35
27
56.45
30
12
71.43
6.92
6
KDM1B
35
51
40.70
34
28
54.84
12.67
X
PHKA2
31
5
86.11
47
11
81.03
6.98
3
TKT
36
25
59.02
129
122
51.19
128.54
1
TMEM48
20
24
45.45
430
263
61.52
0.24
8
ARFGEF1
31
103
23.13
69
195
25.84
34.23
9
CDKN2A
19
18
51.35
30
27
52.63
0.83
5
GPX8
18
45
28.57
17
47
26.56
0.16
6
KDM1B
17
37
31.48
10
37
21.28
12.01
X
PHKA2
11
12
45.83
41
26
61.19
7.46
3
TKT
21
21
50.00
173
338
33.86
0.93
1
TMEM48
12
15
44.44
40
72
34.19
0.43
aPredicted affinity as determined using NetMHC3.4 algorithm.
bVAF = Variant Allelic Fraction as determined from exome sequencing. BRAF VAF are reported as these were used as comparator to assess clonality of other mutations. Candidates formulated in vaccine are shown bolded.
cFPKM = Fragment Per Kilobase of transcript per Million per transcriptome as determined from cDNA-capture data
dBRAF VAF values are reported and were used as comparator to interpret frequencies of remaining MM-genes.
15
AKAP13
KLMNIQQKL
SEQ ID NO: 1
KLMNIQQQL
SEQ ID NO: 16
19
17
Q285K
8
ARPGEF1
FVSALCMFL
SEQ ID NO: 2
FVSALRMFL
SEQ ID NO: 17
19
88
R792C
1
HSD17B7
YISKCWDYA
SEQ ID NO: 7
YISKCWDHA
SEQ ID NO: 22
233
971
N108Y
11
OR8B3
QLSCISTYV
SEQ ID NO: 12
QLSCTSTYV
SEQ ID NO: 27
18
35
T190I
11
PSKCDBP
CLPPQTLAA
SEQ ID NO: 73
CLSPQTLAA
SEQ ID NO: 74
81
694
S153F
5
SEC24A
FLYNLLTRV
SEQ ID NO: 13
FLYNPLTRV
SEQ ID NO: 28
4
6
P469L
6
UTRN
QLDKCSAFV
SEQ ID NO: 15
QLDQCSAFV
SEQ ID NO: 30
21
22
Q1058K
15
AKAP13
20
50
28.57
4
13
23.53
54.3
8
ARPGEF1
23
81
22.12
60
161
27.15
7.1
1
HSD17B7
52
183
22.13
68
228
22.97
29.7
11
OR8B3
15
0
100
13
1
92.88
0.6
11
PSKCDBP
13
0
100
24
0
100.00
0.0
5
SEC24A
22
25
46.81
50
56
46.73
2.6
6
UTRN
22
0
100
44
1
97.78
6.9
15
AKAP13
19
117
14.0
31
101
23.5
1.47
8
ARPGEF1
46
194
19.2
206
1161
15.1
23.73
1
HSD17B7
102
467
17.9
411
1644
20.0
0.11
11
OR8B3
40
21
65.8
3
0
100.0
0.35
11
PSKCDBP
21
6
77.8
161
11
93.6
0.64
5
SEC24A
33
55
37.5
127
172
42.5
1.34
6
UTRN
35
25
58.3
207
46
81.5
15.94
15
AKAP13
39
116
25.16
13
37
26.00
0.14
8
ARPGEF1
29
219
11.65
56
460
10.79
17.51
1
HSD17B7
100
443
18.42
195
896
17.86
0.20
11
OR8B3
17
52
24.64
1
2
33.33
0.25
11
PSKCDBP
17
9
65.38
112
13
88.89
1.94
5
SEC24A
19
60
24.05
34
134
20.12
0.39
6
UTRN
23
36
38.98
58
42
57.43
12.56
aPredicted affinity as determined using NetMHC3.4 algorithm.
bVAF = variant Allelic Fraction as determined from exome sequencing. BRAF VAF are reported as tehse were used as comparator to assess clonality of other mutations.
cFPKM = Fragment Per Kilobase of transcript per Million per transcriptome as determined from cDNA-capture data.
dBRAF VAF values are reported and were used as comparator to interpret frequencies of remaining MM-genes
1
EXOC8
IILVAVPHV
SEQ ID NO: 91
IILVAVQHV
SEQ ID NO: 92
25
143
Q656P
4
LARP7
AVIDAYTEI
SEQ ID NO: 103
AVINAYTEI
SEQ ID NO: 104
213
775
N515D
7
MRPS17
VLLRALPVL
SEQ ID NO: 105
VLLRALPVP
SEQ ID NO: 106
24
11696
P167L
2
MRPS5
HLYASLSRA
SEQ ID NO: 107
HPYASLSRA
SEQ ID NO: 108
116
23536
P59L
8
PABPC1
MLGEQLFPL
SEQ ID NO: 111
MLGERLFPL
SEQ ID NO: 112
3
3
R520Q
3
PLA1A
FIWGDAPPT
SEQ ID NO: 113
SIWGDAPPT
SEQ ID NO: 114
41
744
S6F
20
SMOX
KLANPLPYT
SEQ ID NO: 117
KLAKPLPYT
SEQ ID NO: 118
38
63
K499N
1
SRP9
IMAHCILDL
SEQ ID NO: 119
IIAHCILDL
SEQ ID NO: 120
22
250
I64M
1
EXOC8
7
26
21.21
145
300
32.37
4.83
4
LARP7
6
36
14.29
30
50
37.5
10.15
7
MRPS17
5
71
6.58
29
75
27.88
1.48
2
MRPS5
10
58
14.49
60
125
32.43
14.55
8
PABPC1
16
44
26.67
4073
11235
26.6
1180.59
20
SMOX
131
0
100
11
50
18.03
3.01
1
SRP9
0
58
0*
43
41
51.19
2.31
aPredicted affinity as determined using NetMHC3.4 algorithm.
dBRAF VAF values are reported and were used as comparator to interpret frequencies of remaining MM-genes.
K
MIGNHLWV
AMF
W
SVPTV*
CLNEYHLF
L
KLMNIQQ
K
L
QLSC
I
STYV
FLYN
L
LTRV
IILVAV
P
KV
H
L
YASLSRV*
MLGE
Q
LFPL
aMutated residues are underlined and peptides that elicited immune responses are itacized (naturally-occurring) and bold (vaccine-induced).
bAffinity experimentally determined using fluorescence polarization-based competitive peptide-binding assay, high affinity binding peptides in this assay are log(IC50; nM) <3.7 (11).
cAs determined by immune monitoring assay (FIG. 31, FIG. 30B),
dAntigenic determinant classification according to Sercarz et al. Annu. Rev. Immunol. 11, 729-766 (1993).
2
MPV17
VLDGFIPGT
SEQ ID NO: 127
VLDRFIPGT
SEQ ID NO: 128
51
233
R75G
5
RUFY1
KLADYLNVL
SEQ ID NO: 129
KLADYLKVL
SEQ ID NO: 130
5
15
K225N
2
GPC1
RLFGEAPREIL
SEQ ID NO: 141
RPFGEAPREL
SEQ ID NO: 142
83
21700
P201L
19
SIPA1L3
ILGIFNEFV
SEQ ID NO: 147
ILGISNEFV
SEQ ID NO: 148
45
118
S893F
11
SCYL1
FLFELIPEP
SEQ ID NO: 153
FPFELIPEP
SEQ ID NO: 154
21
12401
P13L
5
PRRC1
QMIYSAARV
SEQ ID NO: 155
QMIYSAARA
SEQ ID NO: 156
79
1783
A431V
7
BRAF
LATEKSRWS
SEQ ID NO: 163
LATVKSRWS
SEQ ID NO: 164
24853
27478
V600E
2
MPV17
34.78
31.51
44.1711
36.59
37.87
44.5254
5
RUFY1
32.5
17.95
10.8626
23.81
42.05
12.321
2
GPC1
28.12
30
7.40362
33.33
38.89
9.89646
19
SIPA1L3
8.33
29.41
1.41955
30
64.71
3.27408
11
SCYL1
15.38
27.54
29.3756
46.15
37.41
48.8269
5
P44C1
11.11
26.14
26.921
30.56
36.17
31.9828
7
BRAF
30
67.67
13.3533
56.25
56.1
14.5002
12
SMARCC2
KVFEHVGSR
KVSEHVGSR
69
390
S624F
19
PIP5K1C
FISNTVFRK
FMSNTVFRK
21
25
M439I
X
ERCC6L
KIYRRQIFK
KIYRRQVFK
12
13
V476I
7
BRAF
LATEKSRWS
LATVKSRWS
24853
27478
V600E
12
SMARCC2
27.66
17.78
14.734
26.83
41.77
20.227
19
PIP5K1C
22.5
23.81
6.1374
24
38.57
11.467
X
ERCC6L
55.56
69.23
2.4877
43.24
63.64
2.4041
7
BRAF
30
67.67
13.353
56.25
56.1
14.6
This application claims the benefit of and priority to PCT application PCT/US15/49836, filed Sep. 11, 2015. which claims benefit of and priority to U.S. Provisional Application 62/050,195 filed on Sep. 14, 2014. PCT/US15/49836 also claims the benefit of and priority to U.S. Provisional Application 62/141,602 filed Apr. 1, 2015. Each of these applications are hereby incorporated by reference, each in their entirety.
This invention was made with government support under CA179695 awarded by the National Institutes of Health. The government has certain rights in the invention.
| Number | Date | Country | |
|---|---|---|---|
| 62141602 | Apr 2015 | US | |
| 62050195 | Sep 2014 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/US15/49836 | Sep 2015 | US |
| Child | 15458149 | US |
