Inhibition of miR-20a by pterostilbene facilitates prostate cancer cells killed by NK cells via up-regulation of NKG2D ligands and TGF-β1down-regulation

Abstract

Natural killer (NK) cells play a potent role in antitumor immunity via spontaneously eliminating tumor directly. However, some tumors such as prostate cancer constantly escape this immune response by down-regulating cell surface molecule recognition and/or secreting immune impressive cytokines. Here, we found pterostilbene, a natural agent with potent anticancer activity, could enhance expression of major histocompatibility complex class I chain-related proteins A and B (MICA/B) on prostate cancer cells surface, which are ligands of the natural killer group 2 member D (NKG2D) expressed by NK cells, and inhibit TGF-β1 secretion by prostate cancer cells. Further, we discovered that these effects were caused by inhibition of miR-20a in prostate cancer cells by pterostilbene. MiR-20a could target the 3' untranslated region (UTR) of MICA/B, resulting in their expression down-regulation. Inhibition of TGF-β1 function by its specific antibody attenuated its impairment to NKG2D on NK cells. Finally, we observed that pterostilbene-treated prostate cancer cells were more easily to be killed by NK cells. Taken together, our findings demonstrated inhibition of miR-20a by pterostilbene in prostate cancer cells could increase MICA/B expression and decrease TGF-β1 secretion, which enhanced NK cell-mediated cytotoxicity againt prostate cancer cells, suggesting a potential approach for increasing anti-prostate cancer immune.

Keywords: MICA/B; NKG2D; Natural killer cell; Prostate cancer; Pterostilbene; TGF-β1; miR-20a.

Fig. 1

Pterostilbene enhanced the cytotoxicity of NK cells against prostate cancer cells. (A, B) PC3 and DU145 cells were treated with Pte at 50 μmol/L or without Pte as control for 24 h. Then cells were collected and co-incubated with NK92 cells at an indicated E: T ratio of 10 : 1, 3 : 1 and 1 : 1 for another 4 h. The cytotoxic activity of NK92 cells were evaluated using the CytoTox 96 Non-Radioactive Cytotoxicity Assay. (C, D) PC3 and DU145 cells were treated with Pte at 50 μmol/L for 24 h. Then cells were collected and co-incubated with NK92 cells pretreated for 30 min with anti-NKG2D blocking mAbs or a control Ab at an indicated E: T ratio of 10 : 1, 3 : 1 and 1 : 1 for another 4 h. CytoTox 96 Non-Radioactive Cytotoxicity Assay was used to detect the cytotoxic activity of NK92 cells. (E, F) Primary prostate cancer cells derived from patient were treated with Pte at 50 μmol/L or without Pte as control for 24 h. NK cells isolated from PBMCs of patients (E) and healthy donors (F) by negative selection using the MACS NK-cell isolation kit were stimulated by IL-2. Then primary prostate cancer cells were collected and co-incubated with NK cells at an E: T ratio of 10 : 1, 3 : 1 and 1 : 1 for another 4 h. The cytotoxic activity of NK cells were evaluated using the CytoTox 96 Non-Radioactive Cytotoxicity Assay. (G) Gating strategy for NK cells and representative histogram or dot plot for each NK cell marker. Lymphocytes were gated according to forward scatter and side scatter. CD3 and CD56 staining was used to identify NK92 cells. (H–K) PC3 and DU145 cells treated with or without Pte were co-incubated with NK92 cells at an E: T ratio of 10 : 1 for another 4 h. The cells were then collected for flow cytometry to measure the expression of IFN-γ and CD107a in NK92 cells. Data were presented as the mean ± SD of 3 independent experiments, each performed in triplicate (*P < 0.05).

Fig. 1

Pterostilbene enhanced the cytotoxicity of NK cells against prostate cancer cells. (A, B) PC3 and DU145 cells were treated with Pte at 50 μmol/L or without Pte as control for 24 h. Then cells were collected and co-incubated with NK92 cells at an indicated E: T ratio of 10 : 1, 3 : 1 and 1 : 1 for another 4 h. The cytotoxic activity of NK92 cells were evaluated using the CytoTox 96 Non-Radioactive Cytotoxicity Assay. (C, D) PC3 and DU145 cells were treated with Pte at 50 μmol/L for 24 h. Then cells were collected and co-incubated with NK92 cells pretreated for 30 min with anti-NKG2D blocking mAbs or a control Ab at an indicated E: T ratio of 10 : 1, 3 : 1 and 1 : 1 for another 4 h. CytoTox 96 Non-Radioactive Cytotoxicity Assay was used to detect the cytotoxic activity of NK92 cells. (E, F) Primary prostate cancer cells derived from patient were treated with Pte at 50 μmol/L or without Pte as control for 24 h. NK cells isolated from PBMCs of patients (E) and healthy donors (F) by negative selection using the MACS NK-cell isolation kit were stimulated by IL-2. Then primary prostate cancer cells were collected and co-incubated with NK cells at an E: T ratio of 10 : 1, 3 : 1 and 1 : 1 for another 4 h. The cytotoxic activity of NK cells were evaluated using the CytoTox 96 Non-Radioactive Cytotoxicity Assay. (G) Gating strategy for NK cells and representative histogram or dot plot for each NK cell marker. Lymphocytes were gated according to forward scatter and side scatter. CD3 and CD56 staining was used to identify NK92 cells. (H–K) PC3 and DU145 cells treated with or without Pte were co-incubated with NK92 cells at an E: T ratio of 10 : 1 for another 4 h. The cells were then collected for flow cytometry to measure the expression of IFN-γ and CD107a in NK92 cells. Data were presented as the mean ± SD of 3 independent experiments, each performed in triplicate (*P < 0.05).

Fig. 2

Pterostilbene up-regulated MICA/B expression and decreased TGF-β1 secretion in PC3 and DU145 cells. (A–H) PC3 and DU145 cells were treated for 24 h with Pte at 50 μmol/L or without Pte as control, and then collected for flow cytomery detecting the expression of MICA/B. Gray-filled profiles, control isotype staining. (I–J) PC3 and DU145 cells were treated for 24 h with Pte at 50 μmol/L or without Pte as control. The supernatant was collected and subjected to the measurement of TGF-β1 by ELISA. (K) Gating strategy for NK cells. Lymphocytes were gated according to forward scatter and side scatter. CD3 and CD56 staining was used to identify NK92 cells. (L–M) PC3 and DU145 cells were treated with neutralizing TGF-β1 antibody (10 μg/mL) for 30 min and then co-cultured with NK92 cells for another 24 h at ratio of 10 : 1 (NK 92 cells: PC3 or DU145). The expression of NKG2D was analyzed by flow cytometry. Data were presented as the mean ± SD of 3 independent experiments, each performed in triplicate (*P < 0.05).

Fig. 2

Pterostilbene up-regulated MICA/B expression and decreased TGF-β1 secretion in PC3 and DU145 cells. (A–H) PC3 and DU145 cells were treated for 24 h with Pte at 50 μmol/L or without Pte as control, and then collected for flow cytomery detecting the expression of MICA/B. Gray-filled profiles, control isotype staining. (I–J) PC3 and DU145 cells were treated for 24 h with Pte at 50 μmol/L or without Pte as control. The supernatant was collected and subjected to the measurement of TGF-β1 by ELISA. (K) Gating strategy for NK cells. Lymphocytes were gated according to forward scatter and side scatter. CD3 and CD56 staining was used to identify NK92 cells. (L–M) PC3 and DU145 cells were treated with neutralizing TGF-β1 antibody (10 μg/mL) for 30 min and then co-cultured with NK92 cells for another 24 h at ratio of 10 : 1 (NK 92 cells: PC3 or DU145). The expression of NKG2D was analyzed by flow cytometry. Data were presented as the mean ± SD of 3 independent experiments, each performed in triplicate (*P < 0.05).

Fig. 3

Inhibition of miR-20a induced MICA/B up-regulation and TGF-β1 decrease in PC3 and DU145 cells by pterostilbene. (A–B) PC3 and DU145 cells were treated with Pte at 50 μmol/L or without Pte as control for 24 h, and then the expression of MICA/B was detected by qRT-PCR. (C–D) PC3 and DU145 cells were treated with Pte (50 μmol/L) in the presence of miR-20a mimics (50 nm) or miR-20a scramble control for 24 h, and then collected to measure the expression of MICA/B using flow cytometry. (E) PC3 and DU145 cells were treated with Pte (50 μmol/L) in the presence of miR-20a mimics or miR-20a scramble control for 24 h. The concentration of TGF-β1 in the supernatant was measured by ELISA. (F–G) The predicted binding sites of miR-20a in the 3′-UTR of MICA/B was identified using TargetScanHuman 7.1 and the relative luciferase activity was measured after transfection of the indicated reporter plasmids (MICA/B 3′-UTR or the mutant MICA/B 3′-UTR) into DU145 cells. Data were presented as the mean ± SD of 3 independent experiments, each performed in triplicate (*P < 0.05).