Impaired interferon signaling is a common immune defect in human cancer

Abstract

Immune dysfunction develops in patients with many cancer types and may contribute to tumor progression and failure of immunotherapy. Mechanisms underlying cancer-associated immune dysfunction are not fully understood. Efficient IFN signaling is critical to lymphocyte function; animals rendered deficient in IFN signaling develop cancer at higher rates. We hypothesized that altered IFN signaling may be a key mechanism of immune dysfunction common to cancer. To address this, we assessed the functional responses to IFN in peripheral blood lymphocytes from patients with 3 major cancers: breast cancer, melanoma, and gastrointestinal cancer. Type-I IFN (IFN-alpha)-induced signaling was reduced in T cells and B cells from all 3 cancer-patient groups compared to healthy controls. Type-II IFN (IFN-gamma)-induced signaling was reduced in B cells from all 3 cancer patient groups, but not in T cells or natural killer cells. Impaired-IFN signaling was equally evident in stage II, III, and IV breast cancer patients, and downstream functional defects in T cell activation were identified. Taken together, these findings indicate that defects in lymphocyte IFN signaling arise in patients with breast cancer, melanoma, and gastrointestinal cancer, and these defects may represent a common cancer-associated mechanism of immune dysfunction.

Fig. 1.

Real-time quantitative PCR analysis of ISG expression in lymphocytes from breast cancer patients and healthy controls. The expression levels of ISGs STAT1, IFI44, IFIT1, IFIT2, and MX1 were measured in unstimulated lymphocytes from breast cancer patients (BC filled circles) and age-matched healthy controls (H open squares) by real-time quantitative PCR. Expression of each gene was normalized to GAPDH. Medians are indicated by the bar in each data set. Two-sided Wilcoxon-Mann-Whitney tests were performed to compare ISG expression values from breast cancer patients with healthy controls; STAT1 P = 0.0381, IFI44 P = 0.0303, IFIT1 P = 0.0480, IFIT2 P = 0.0177, MX1 P = 0.019.

Fig. 2.

IFN-α- and IFN-γ-stimulated fold-change in pSTAT1 (Y701) in peripheral blood mononuclear cells (PBMC) subsets from breast cancer patients, melanoma patients and GI cancer patients versus healthy controls. PBMCs were stimulated with 1,000 IU/ml IFN-α, IFN-γ, or unstimulated and pSTAT1 was measured by Phosflow. The IFN-induced fold-change in pSTAT1 was measured in T, B, and NK cells from healthy controls (H open squares), patients with breast cancer (BC filled circles), melanoma (Mel filled triangles), and GI cancer (GI x). (A) IFN-α- BC, (B) IFN-α- Mel, (C) IFN-α- GI, (D) IFN-γ- BC, (E) IFN-γ- Mel, (F) IFN-γ- GI. Medians are indicated by the bar in each data set. Two-sided Wilcoxon-Mann-Whitney tests were used to compare values from cancer patients with age-matched healthy controls; P-values less than 0.05 are denoted; #1 P = 0.00077, #2 P = 0.00012, #3 P = 0.00040, #4 P = 0.00420, #5 P = 0.04408, #6 P = 0.00270, #7 P = 0.00003, #8 P = 0.00012, #9 P = 0.00570, #10 P = 0.00190. P-values and estimated differences with 95% confidence intervals (CI) for each comparison are shown in Table S1.

Fig. 3.

Effect of cancer stage, chemotherapy, and T-cell phenotype on IFN-α- and IFN-γ-stimulated fold-changes in pSTAT1 in cancer patients and healthy controls. PBMCs were stimulated with IFN-α, IFN-γ, or unstimulated and pSTAT1 was measured by Phosflow in T, B, and NK cells from healthy controls (H open squares), BC patients with stage II (BC II filled diamonds), stage III (BC III open circles), stage IV (BC IV filled circles) disease, BC patients not treated (BC –), and BC patients treated with chemotherapy (BC +) and stage III–IV melanoma patients (Mel filled triangles). Naive, effector, and memory T-cell subsets were gated based on CD27 and CD45RA expression. (A) IFN-α- T cells; (B) IFN-α- B cells; (C) IFN-α- NK cells; (D) IFN-γ-T cells; (E) IFN-γ- B cells; (F) IFN-γ- NK cells; (G) IFN-α- T, B, and NK cells from BC patients not treated (BC–) or treated (BC+) with chemotherapy; (H) IFN-γ- T, B, and NK cells from BC– or BC+ patients; (I) IFN-α- naive, effector, and memory T cells; (J) IFN-γ- naive, effector, and memory T cells [(A–H) BC patients, (I–J) Mel patients). The median is indicated by the bar in each data set. Two-sided Wilcoxon-Mann-Whitney tests were performed to compare values from cancer patients with age-matched healthy controls; P-values less than 0.05 are denoted; #1 P = 0.00447, #2 P = 0.03327, #3 P = 0.03985, #4 P = 0.00077, #5 P = 0.02690, #6 P = 0.01180, #7 P = 0.01107, #8 P = 0.00108, #9 P = 0.00286, #10 P = 0.01019, #11 P = 0.01349, #12 P = 0.00069, #13 P = 0.04490, #14 P = 0.00037, #15 P = 0.00764, #16 P = 0.00339, #17 P = 0.00527, #18 P = 0.00233, #19 P = 0.00131, #20 P = 0.01806, #21 P = 0.03107. There was no significant difference in the frequency of T-cell phenotypes between melanoma patients and healthy controls. P-values and estimated differences with 95% CI for each comparison are shown in Table S1A.

Fig. 4.

Expression of activation markers and apoptosis of T cells stimulated with anti-CD3/CD28 antibodies and IFN-α or IFN-γ in breast cancer patients and healthy controls. Lymphocytes from BC patients (filled circles) and healthy controls (open squares) were stimulated with anti-CD3 and anti-CD28 antibody-coated beads alone or with IFN-α or -γ, or unstimulated. The % T cells positive for CD25, HLA-DR, CD54, CD95, Annexin V (Anx), and ViViD (Viv) were measured by FACS at 48 h. (A) % CD25+; (B) % HLA-DR+; (C) % CD54+; (D) % CD95+; (E) IFN-induced change in % CD95+ activated T cells (% CD95+ in conditions stimulated with anti-CD3 anti-CD28 plus IFN-α or -γ minus the % CD95+ in conditions stimulated only with anti-CD3 anti-CD28); (F) % Anx+ Viv+ (all T cells that were positive for either or both Anx and Viv). The median is indicated by the bar in each data set. Two-sided Wilcoxon-Mann-Whitney tests were used to compare data from breast cancer patients and healthy controls; P-values ≤0.05 are denoted; #1 P = 0.004, #2, P = 0.020, #3 P = 0.018, #4 P = 0.016, #5 P = 0.018, #6 P = 0.053. Multivariate analyses are shown in Table S3.