Mutant MHC class II epitopes drive therapeutic immune responses to cancer

Abstract

Tumour-specific mutations are ideal targets for cancer immunotherapy as they lack expression in healthy tissues and can potentially be recognized as neo-antigens by the mature T-cell repertoire. Their systematic targeting by vaccine approaches, however, has been hampered by the fact that every patient's tumour possesses a unique set of mutations ('the mutanome') that must first be identified. Recently, we proposed a personalized immunotherapy approach to target the full spectrum of a patient's individual tumour-specific mutations. Here we show in three independent murine tumour models that a considerable fraction of non-synonymous cancer mutations is immunogenic and that, unexpectedly, the majority of the immunogenic mutanome is recognized by CD4(+) T cells. Vaccination with such CD4(+) immunogenic mutations confers strong antitumour activity. Encouraged by these findings, we established a process by which mutations identified by exome sequencing could be selected as vaccine targets solely through bioinformatic prioritization on the basis of their expression levels and major histocompatibility complex (MHC) class II-binding capacity for rapid production as synthetic poly-neo-epitope messenger RNA vaccines. We show that vaccination with such polytope mRNA vaccines induces potent tumour control and complete rejection of established aggressively growing tumours in mice. Moreover, we demonstrate that CD4(+) T cell neo-epitope vaccination reshapes the tumour microenvironment and induces cytotoxic T lymphocyte responses against an independent immunodominant antigen in mice, indicating orchestration of antigen spread. Finally, we demonstrate an abundance of mutations predicted to bind to MHC class II in human cancers as well by employing the same predictive algorithm on corresponding human cancer types. Thus, the tailored immunotherapy approach introduced here may be regarded as a universally applicable blueprint for comprehensive exploitation of the substantial neo-epitope target repertoire of cancers, enabling the effective targeting of every patient's tumour with vaccines produced 'just in time'.

Extended Data Figure 1. Non synonymous cancer-associated mutations are frequently immunogenic and pre-dominantly recognized by CD4⁺ T cells

T-cell responses obtained by vaccinating C57BL/6 mice with antigen-encoding RNA in the B16F10 tumour model (n = 5). Left, prevalence of non-immunogenic, MHC-class-I- or class-II-restricted mutated epitopes. Right, detection and typing of mutation-specific T cells (individual epitopes shown in Extended Data Table 1).

Extended Data Figure 2. Mutant epitope-specific T cells induced by RNA vaccination control tumour growth

a, Splenocytes of mice (n = 5) vaccinated with B16-M30 RNA were tested by ELISpot for recognition of mutated peptides as compared to the corresponding wild-type (B16-WT30) sequence. Right, testing of truncated variants of B16-M30 (mean + s.e.m.). b, Mean ± s.e.m. tumour growth (left) and survival (right) of C57BL/6 mice (n = 10) inoculated s.c. with B16F10 and left untreated (control) or injected i.v. with irrelevant RNA. c, Lungs of B16F10-Luc tumour bearing mice shown in Fig. 2b (day 27 after tumour inoculation). d, Therapeutic antitumour activity against B16F10 tumours in mice (B16-M27, Trp2 n = 8; B16-M30 n = 7; others n = 10) conferred by immunization with epitopes encoding immunogenic B16F10 mutations or an immunodominant wild type Trp2 epitope. The area under the tumour growth curve at day 30 after tumour inoculation was normalized to untreated control mice and depicted as mean ± s.e.m. Red and black columns represent mutations recognized by CD8⁺ or CD4⁺ T cells, respectively. e, Spontaneous immune responses in splenocytes of irrelevant RNA treated B16F10 tumour bearing C57BL/6 mice (n = 3) were tested by ELISpot for recognition of peptides (mean + s.e.m.).

Extended Data Figure 3. Mechanism of antitumour activity of mutation specific poly-epitope vaccines in CT26 tumour-bearing mice

a, BALB/c mice (n = 5) were vaccinated either with pentatope (35 μg) or the corresponding mixture of five RNA monotopes (7 μg each). T-cell responses in peptide-stimulated splenocytes of mice were measured ex vivo via ELISpot (medium control subtracted mean ± s.e.m.). b, c, BALB/c mice (n = 10) were inoculated i.v. with CT26 tumour cells and left untreated or injected with irrelevant, CT26-M19 or pentatope1 or 2 RNA in absence (b) or presence of a CD8 T cell depleting antibody or a CD40L blocking antibody (c). Mean ± s.e.m. of tumour nodules per lung are shown. d, Immunofluorescence analyses of tumour-infiltrating lymphocytes in pentatope2-vaccinated mice. Upper panel, lung tumour tissue stained for CD4 and CD8 or CD4 and FoxP3. Scale bar, 50 μm. Lower panel left, proportion of infiltrating cells in sections of irrelevant (CD4: n = 13; CD8 = 9; FoxP3: n = 13) or pentatope (CD4: n = 17; CD8: n = 6; FoxP3: n = 10) RNA-treated animals. Lower panel right, tumour area in sections of control (n = 22) and pentatope2-treated (n = 20) animals (mean ± s.e.m.).

Extended Data Figure 4. Immunogenicity testing of P_ME pentatope-encoded mutations

Splenocytes of P_ME RNA vaccinated BALB/c mice (n = 6) were tested ex vivo for recognition of peptides representing the mutated 27mer sequences represented in P_ME pentatopes with or without addition of an MHC class II-blocking antibody. Mean + s.e.m. of background (medium control) subtracted responses are shown.

Figure 1. Cancer-associated mutations are frequently immunogenic and pre-dominantly recognized by CD4⁺ T cells

a, Schematic describing mutation discovery and immunogenicity testing. b–d, Splenocytes of mice vaccinated with peptides and polyinosinic:polycytidylic acid (polyI:C) (b, B16F10, n = 5) or immunized with antigen-encoding RNA (c, CT26, n = 5; d, 4T1, n = 3) were tested for recognition of mutated peptides by ELISpot. Subsequent subtyping was performed via MHC II blockade or intracellular cytokine and CD4/CD8 surface staining. Pie charts represent the prevalence of non-immunogenic, MHC class I- or II-restricted mutated epitopes. b, Right, subtyping of mutation-specific T cells. c, Right, MHC restriction of neo-epitopes prioritized based on either good (0.1–2.1) or poor (>3.9) MHC I binding scores.

Figure 2. Efficient tumour control and survival benefit in B16F10 melanoma with an RNA vaccine encoding a single mutated CD4⁺ T-cell epitope

a, Tumour growth (mean + s.e.m.) and survival (±CD4- or CD8-depleting antibodies) in untreated (control) or B16-M30 immunized C57BL/6 mice (n = 10) inoculated subcutaneously (s.c.) with B16F10. b, B6 albino mice (n = 8) developing lung metastases upon intravenous (i.v.) injection of B16F10-Luc were treated with B16-M30 or irrelevant RNA (control). Median tumour growth was determined by BLI as photons per second. c, Single-cell suspensions of B16F10 tumours of irrelevant (control) or B16-M30 RNA immunized mice (n = 4) were restimulated with B16-M30 or irrelevant peptide (vesicular stomatitis virus nucleoprotein, VSV-NP_52–59) and tested by ELISpot (mean + s.e.m.). Data pooled from two experiments. d, Frequency of infiltrating cells in s.c. B16F10 tumours (n = 3) left untreated (control) or vaccinated with B16-M30 RNA.

Figure 3. RNA pentatope immunization confers disease control and survival benefit in murine tumours

a, Engineering of a poly-neo-epitope RNA. b, BALB/c mice (n = 10) developing lung metastases upon i.v. injection of CT26-Luc were treated with a mixture of two pentatopes or left untreated (control). The median tumour growth by BLI, survival data and lungs from treated animals are shown. c, Upper panel, CD3 stained lung tissue sections. Scale bars: 1,000 μm (scan), 100 μm (top), 50 μm (bottom). Lower panel, proportional lymphocyte areas in lung tumour tissue of control (n = 6) or pentatope-treated (CD3: n = 14; CD4, CD8, FoxP3: n = 12) animals. Lower panel right, tumour area (mean ± s.e.m.) in sections of control (n = 18) and pentatope112-treated (n = 39) mice.

Figure 4. RNA pentatope vaccines with mutations selected for in silico predicted favourable MHC class II binding and abundant expression confer potent antitumour control

a, Comparison of median MHC II binding scores of immunogenic (Response) and non-immunogenic (No response) mutated 27mers. b, Highly expressed mutations were selected with (‘ME’) or without (‘E’) considering MHC class II binding score. Ten mutations (two pentatopes) per category were used for vaccination of CT26-Luc tumour-bearing mice (n = 10). Tumour growth, area under the curve (AUC) at day 40 and ink-treated lungs are shown. c, Mice (n = 5) were analysed for T-cell responses against the RNA pentatopes via ELISpot (mean ± s.e.m. subtracted by an irrelevant RNA control). d, CT26 tumour nodules per lung of untreated mice or mice (n = 10) injected with irrelevant or P_ME (±CD8 depletion or CD40L blocking) RNA. e, T-cell responses against gp70_423–431 (gp70-AH1) determined via ELISpot in blood (pooled from 5 mice at day 20) and spleen (n = 5). Background (medium control) subtracted mean ± s.e.m. shown. f, Genomic, expressed and predictively presented (HLA-DRB1, IEDB rank <10) non-synonymous single nucleotide variations (nsSNVs) derived from human cancer samples (TCGA). SKCM, skin cutaneous melanoma; COAD, colon adenocarcinoma; BRCA, breast invasive carcinoma.