5-Azacytidine Potentiates Anti-tumor Immunity in a Model of Pancreatic Ductal Adenocarcinoma

Abstract

Tumors evolve a variety of mechanisms to escape immune detection while expressing tumor-promoting molecules that can be immunogenic. Here, we show that transposable elements (TE) and gene encoded, tumor-associated antigens (TAA), which can be both highly immunogenic and tumor-promoting, are significantly upregulated during the transition from pre-malignancy to malignancy in an inducible model of pancreatic ductal adenocarcinoma (PDAC). Coincident with the increased presence of TEs and TAAs was the downregulation of gene transcripts associated with antigen presentation, T cell recruitment and intrinsic anti-viral responses, suggesting a unique strategy employed by PDAC to possibly augment tumorigenesis while escaping detection by the immune system. In vitro treatment of mouse and human PDAC cell lines with the DNA methyltransferase inhibitor 5-azacytidine (Aza) resulted in augmented expression of transcripts for antigen presentation machinery and T cell chemokines. When immunocompetent mice implanted with PDAC were therapeutically treated with Aza, we observed significant tumor regression that was not observed in immunocompromised mice, implicating anti-tumor immunity as the principal mechanism of tumor growth control. Analysis of PDAC tumors, immediately following Aza treatment in immunocompetent mice, revealed a significantly greater infiltration of T cells and various innate immune subsets compared to control treatment, suggesting that Aza treatment enhances tumor immunogenicity. Thus, augmenting antigen presentation and T cell chemokine expression using DNA methyltransferase inhibitors could be leveraged to potentiate adaptive anti-tumor immune responses against PDAC.

Keywords: 5-azacytidine; anti-tumor immunity; immune evasion; pancreatic ductal adenocarcinoma; transposable element; tumor-associated antigens.

Figure 1

Expression of transposable elements (TEs) and non-TE, gene-encoded transcripts during transition to malignancy. (A) Waterfall plots of TE expression, using RNA-seq, during the transition from healthy pancreas (HP) to acinar-to-ductal metaplasia (ADM) (pre-malignant) and (B) from ADM to pancreatic ductal adenocarcinoma (PDAC) (malignant). (C) Waterfall plots of non-TE, protein coding gene transcripts displayed in order from highest to lowest fold change for ADM/HP and (D) PDAC/ADM.

Figure 2

Antigen presentation and viral response-associated genes decrease following malignant transformation to PDAC. Log₂ fold-change in gene expression from RNA-seq analysis, calculated as the ratio of ADM fold-change over HP (white) or PDAC fold-change over ADM (gray) (n = 3 mice per group). (A) Fold-change in antigen presentation-related genes. (B) Fold-change in IFN-γ and innate immune-related genes. (C) Human microarray dataset (E-GEOD-2019) was analyzed for signal for CXCL9 and CXCL10 in various cancer types. p-values shown are for significance in fold change of PDAC over ADM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 3

Aza treatment augments antigen presentation- and T cell recruitment-related gene expression in mouse and human PDAC cells. qRT-PCR was performed on total RNA to assess gene expression for (A) KPT277 cells treated with DMSO for 96 h in vitro vs. healthy pancreas from a C57Bl/6 mouse, and (B) KPT277 cells treated in vitro with DMSO or 10 μM Aza for 96 h. (C) qRT-PCR for the mouse PDAC cell line Pan02 treated in vitro with DMSO or 5 μM Aza. (D–F) Human PDAC cell lines treated in vitro with DMSO or Aza (100 μM for PANC1; 10 μM for MiaPaCa-2; and 50 μM for Hs-766-T) for 96 h. In (A), bars represent Log base 2-fold change relative to “Healthy Pancreas.” n = 3 biological replicates. In (B) and (C–F), bars represent quantity of transcripts relative to vehicle (DMSO) for each gene. n = 3 biological replicates. p-values were assessed by[[Inline Image]] two-way ANOVA followed by Sidak's multiple comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. n.s. = not significant.

Figure 4

Aza treatment potentiates splenocyte killing of KPT277 cells that is not enhanced by PD-1 checkpoint blockade. (A) KPT277 cells were treated with Aza at 20 μM in vitro for 96 h and analyzed by flow cytometry using a fluorescent, fixable viability dye and an antibody recognizing PD-L1. Bars represent living, %PD-L1 positive cells out of total cells. n = 3 biological replicates. (B) KPT277 cells pre-treated for 72 h with DMSO or Aza (20 μM) were implanted in mice subcutaneously for 48 h. Tumors were then removed, digested, and analyzed by flow cytometry for PD-1, CD-8, and CD45 expression as well as live/dead cells using a fixable viability dye. Bars represent the percent of PD-1⁺/CD8⁺ cells out of live, CD45⁺ cells. N = 2 tumors per group. (C) KPT277 cells treated with DMSO or 20 μM Aza (72 h) were incubated with or without splenocytes (100:1) from an Aza pre-treated (10 μM for 96 h), KPT277 tumor-burdened mouse. Splenocytes pre-treated with anti-PD-1 antibody (2.5 μg/mL) in vitro for 1 h are indicated. KPT277 cell viability after 24 h of co-incubation was assayed by MTT. Bars represent absorbance values at 595 nm. n = 3 biological replicates. Error bars represent SEM. p-values were calculated using an unpaired t-test. *p < 0.05 and **p < 0.01.

Figure 5

Aza treatment controls PDAC tumor growth through an immune-dependent mechanism and induces significant infiltration by CD4⁺ and CD8⁺ T cells. (A) KPT277 tumor growth curves from immunocompetent C57Bl//6 mice treated with vehicle (DMSO) or 1 mg/kg 5-Azacytidine (Aza) once weekly. Measurements were taken three times weekly. Curves represent mean measurements for n = 4 mice in the DMSO group and n = 5 in Aza-treated group. (B) Tumors after removal from C57Bl/6 mice euthanized on Day 28. Ruler (cm) shows approximate tumor lengths. (C) KPT277 tumor growth curves from immunocompromised NSG mice treated with vehicle (DMSO) or 1 mg/kg 5-Azacytidine (Aza) once weekly. Measurements were taken three times weekly. Curves represent mean measurements for n = 5 mice in the DMSO group and n = 6 in Aza-treated group. Error bars represent standard error of the mean (SEM). *p < 0.05, t-test comparing data between groups on Day 10. (D) Western blots of tumor lysates (n = 3 per group) from Day 8 Aza and DMSO-treated tumors (48 h after first treatment) using anti-DNMT1, anti-IRF5, and anti-survivin antibodies. GAPDH is shown as a loading control. DNMT1 and survivin were assayed using the same blot. Relative quantifications of band intensity normalized to GAPDH are shown to the right of blots (individual values plotted, error bars = SEM). Increased frequency and infiltration of (E) CD4⁺ and CD8⁺ T cells, and (F) CD11c⁺ and F4/80⁺ cells as determined by IHC in tumor tissue fixed 48 h after initial dosing with DMSO or Aza. Scale bars = 75 μm. Dab signal (brown) indicates CD4+, CD8+, CD11c+, or F4/80+ staining. Nuclei were counterstained with hematoxylin. Positive-staining cells were enumerated for 30–40 fields approximating the entire tumor (n = 2 tumors DMSO, n = 3 tumors Aza) and are presented as cells per 5,000 μm² tumor area (bar graph). Error bars represent SEM. ****p < 0.0001, Mann-Whitney test.