Interferon γ limits the effectiveness of melanoma peptide vaccines

Abstract

The development of effective therapeutic vaccines to generate tumor-reactive cytotoxic T lymphocytes (CTLs) continues to be a top research priority. However, in spite of some promising results, there are no clear examples of vaccines that eradicate established tumors. Most vaccines are ineffective because they generate low numbers of CTLs and because numerous immunosuppressive factors abound in tumor-bearing hosts. We designed a peptide vaccine that produces large numbers of tumor-reactive CTLs in a mouse model of melanoma. Surprisingly, CTL tumor recognition and antitumor effects decreased in the presence of interferon γ (IFNγ), a cytokine that can provide therapeutic benefit. Tumors exposed to IFNγ evade CTLs by inducing large amounts of noncognate major histocompatibility complex class I molecules, which limit T-cell activation and effector function. Our results demonstrate that peptide vaccines can eradicate large, established tumors in circumstances under which the inhibitory activities of IFNγ are curtailed.

Figures

Figure 1

IFNγ inhibits the therapeutic activity of peptide vaccination and decreases the capacity of CD8 T cells to recognize tumor cells. (A) Therapeutic effects induced by Trp2180TriVax against 7-day-established subcutaneous B16 tumors in WT and IFNγ−/− mice. Mice (4/group) were inoculated with tumor and vaccinated intravenously with TriVax (200 μg of Trp2180 peptide, 100 μg of anti-CD40 mAb, and 50 μg of poly-IC) on days 7 and 18 (arrows). Nonvaccinated mice (No Vax) were included as controls. Tumor sizes are presented as mean tumor areas in square millimeters. Points, mean for each group; bars, SD. (B) Mice (4/group) received B16 cells intravenously and were vaccinated with Trp2180TriVax or Ova55TriVax, as described. On day 28, the numbers of B16 pulmonary nodules were evaluated in individual mice. Horizontal lines, means of each group. (C) Freshly isolated purified CD8 T cells from Trp2180TriVax-immunized WT mice were evaluated for cytolytic activity against various targets: Trp2180 peptide-pulsed and unpulsed EL4 cells (triangles), nontreated B16, B16 incubated with 100 U/mL IFNγ for 6 or 24 hours (circles), and B16/IFNγ cells (24 hours) pulsed with Trp2180 peptide (diamonds). (D) Cytokine-release EliSpot assays using purified CD8 T cells from Trp2180TriVax-immunized WT mice were performed using several stimulator cells: Trp2180 peptide-pulsed and unpulsed EL4, IFNγ-treated (100 U/mL, 24 hours), and nontreated B16. P values were calculated with 2-way ANOVA (A,C) or unpaired Student t tests (B,D). Experiments were repeated 3 times with similar results.

Figure 2

Expression of IFNγRDN in tumor cells overcomes the inhibitory effects of IFNγ. (A) Expression levels of IFNγR1, MHC-I (H-2Kb and H-2Db), MHC-II (I-Ab), and PD-L1 on B16 and 2 stable B16 clones expressing high or low levels of a IFNγRDN receptor. B16, B16-IFNγRDN/Hi, and B16-IFNγRDN/Lo cells were incubated (IFNγ) or not (No Tx) with 100 U/mL of IFNγ for 40 hours, and stained with specific antibodies as indicated, followed by flow-cytometry analysis. (B) Freshly isolated and purified CD8 T cells from Trp1455TriVax or Trp2180TriVax-immunized WT mice (as indicated) were tested for cytolytic activity against several target cells treated (IFNγ) and nontreated (No Tx) with IFNγ (100 U/mL, 24 hours): Parental B16 and 2 stable B16 clones expressing high (B16-IFNγRDN/Hi) or low (B16-IFNγRDN/Lo) levels of IFNγRDN. (C) Therapeutic effects induced by Trp2180TriVax in WT mice against 7-day-established subcutaneous B16 tumors expressing or not IFNγRDN. P value (2-way ANOVA) compares Trp2180TriVax-immunized mice bearing B16-IFNγRDN/Hi with Trp2180TriVax-immunized mice bearing B16. *50% of the Trp2180TriVax-immunized mice bearing B16-IFNγRDN/Hi rejected their tumors. These experiments were repeated twice with similar results.

Figure 3

PD1 blockade increases the therapeutic efficacy of TriVax. WT mice (4/group) were inoculated subcutaneously with B16, and vaccinated with Trp2180TriVax (A) or Trp1455TriVax (B). Anti-PD-L1 mAb (10F.9G2) was administered intraperitoneally on days 2, 4, 6, and 8 after each TriVax administration at 200 μg/dose. Nonvaccinated mice (No Vax) and Ova55TriVax were included as controls. Arrows, days when TriVax administered; gray bars, time when anti–PD-L1 mAb was administered. Tumor sizes were determined in individual mice by measuring 2 opposing diameters and are presented as tumor areas in square millimeters. Points indicate means for each group of mice; and bars, SD. P values were calculated using 2-way ANOVA test comparing with the TriVax alone with TriVax plus anti–PD-L1 mAb. These experiments were repeated twice with similar results.

Figure 4

Expression levels of noncognate peptide/MHC-I complexes dictate the antigenicity of B16 cells. (A) Expression levels of MHC-I (H-2Kb and H-2Db, top panels) and H-2Kb/Ova257 complexes (bottom panel) on a stable B16 clone expressing single-chain H-2Kb/Ova257 (B16-scKbOva) compared with B16 and IFNγ-treated B16 (100 U/mL, 24 hours) cells measured by flow cytometry. Levels of H-2Kb/Ova257 complexes were measured using antibody 25-D1.16. (B) Antigenicity of B16-scKbOva was evaluated with freshly isolated CD8 T cells from Trp1455TriVax- and Trp2180TriVax-immunized WT mice using cytokine release ELISA. IFNγ-treated (100 U/mL, 24 hours) and nontreated parental B16 cells were included for comparison. Cultures consisted of 3 × 105 CD8 T cells coincubated with 1 × 105 stimulator cells for 40 hours before removing culture supernatants for cytokine measurements. Results represent the average values of IFNγ (columns) and SD (error bars) from triplicate wells. These experiments were repeated twice with similar results.

Figure 5

IFNγ-treated H-2Kb-loss tumor variants are recognized by H-2Db–restricted CD8 T cells. (A) Expression levels of MHC-I and PD-L1 on H-2Kb–loss variants (B16Kb−, B16F1Kb−, and JB/RHKb−). The cells were incubated (IFNγ) or not (No Tx) with 100 U/mL of IFNγ for 24 hours, and stained with antibodies as indicated, followed by flow cytometric analysis. (B) CD8 T cells, which were freshly isolated from mice vaccinated with Trp1455TriVax (H-2Db–restricted), were evaluated for their capacity to recognize B16, 2 B16 H-2Kb-loss variants (B16Kb− and B16F1Kb−), and an H-2Kb− chemically induced melanoma (JB/RHKb−) using a cytokine-release ELISA assay. The cells were either incubated with IFNγ (100 U/mL, 24 hours) or not (No Tx). Cultures consisted of 3 × 105 CD8 T cells coincubated with stimulator cells (3:1 ratio) for 40 hours before removing culture supernatants for cytokine measurements. Results represent the average values of IFNγ (columns) and SD (error bars) from triplicate wells. Numbers above columns represent the percentage response of the IFNγ-treated cells compared with that of the nontreated group.

Figure 6

Up-regulation of CD8 coreceptors after in vitro culture restores the capacity of T cells to recognize B16 cells expressing high levels of noncognate MHC-I. Tissue-culture CD8 T cells were produced by placing the purified CD8 T cells from Trp1455TriVax-immunized mice in medium containing 50 U/mL IL-2 and 20 ng/mL IL-7 for 7 days. (A) Comparison of the levels of CD8α expression between freshly isolated and cultured CD8 T cells from Trp1455TriVax-immunized mice compared with naive CD8 T cells from nonvaccinated mice. MFI, mean fluorescence intensity of CD8α. Right panel shows histograms gating on the Trp1455 tetramer-positive populations. (B) Antigen-induced IFNγ production of cultured and freshly isolated CD8 T cells from Trp1455TriVax-immunized mice evaluated by ELISA. CD8 T cells were evaluated for their capacity to recognize B16 treated or not with IFNγ (100 U/mL, 24 hours) and B16-scKbOva. CD8 T cells (3 × 105) were incubated with tumor cells (1 × 105) for 40 hours, and supernatants were removed for cytokine measurements. Supernatants from T cells without tumor cells (T cell alone) were included as controls. Results represent the average amounts of IFNγ and SD (error bars) from triplicate cultures. These experiments were repeated twice with similar results.

Figure 7

Decreasing noncognate MHC-I levels together with PD1 blockade allows TriVax to eliminate advanced tumors in WT mice. (A) Expression levels of H-2Kb, H-2Db, and PD-L1 in B16-KbLo cells treated or not with IFNγ (100 U/mL, 24 hours). Results with IFNγ-treated parental B16 cells are included for comparison. (B) Responses (ELISPOT) of freshly isolated CD8 T cells from Trp1455TriVax- and Trp2180TriVax-immunized WT mice against IFNγ treated and nontreated B16-KbLo cells. Results represent the average number of spots from triplicate wells with SD (error bars) of the means. (C) Therapeutic effects induced by Trp1455TriVax against 7-day-established B16-KbLo tumors in WT mice. Mice (8/group) were inoculated with B16-KbLo cells and immunized with Trp1455TriVax or Ova55TriVax as indicated. Anti-PD-L1 mAb was administered intraperitoneally on days 2, 4, 6, and 8 after TriVax administration. Nonvaccinated mice (No Vax) were also included as controls. Arrows, days when the vaccines were administered; gray bars, period of anti–PD-L1 mAb treatment. (D-E). Tumor-growth curves are shown for individual mice from the Trp1455TriVax and Trp1455TriVax plus anti–PD-L1 groups. Tumor sizes were determined in individual mice by measurements of 2 opposing diameters and are presented as tumor areas in square millimeters. Points, mean for each group of mice; bars, SD. P < .0001, between the Trp1455TriVax and the Trp1455TriVax + anti–PD-L1 mAb group (obtained using a 2-way ANOVA analysis). *Seven of 8 mice in Trp1455TriVax + anti–PD-L1 mAb rejected their tumors; **1 mouse from Trp1455TriVax rejected its tumor. These experiments were repeated twice with similar results.