Inhibition of SIRT1 reactivates silenced cancer genes without loss of promoter DNA hypermethylation

Abstract

The class III histone deactylase (HDAC), SIRT1, has cancer relevance because it regulates lifespan in multiple organisms, down-regulates p53 function through deacetylation, and is linked to polycomb gene silencing in Drosophila. However, it has not been reported to mediate heterochromatin formation or heritable silencing for endogenous mammalian genes. Herein, we show that SIRT1 localizes to promoters of several aberrantly silenced tumor suppressor genes (TSGs) in which 5' CpG islands are densely hypermethylated, but not to these same promoters in cell lines in which the promoters are not hypermethylated and the genes are expressed. Heretofore, only type I and II HDACs, through deactylation of lysines 9 and 14 of histone H3 (H3-K9 and H3-K14, respectively), had been tied to the above TSG silencing. However, inhibition of these enzymes alone fails to re-activate the genes unless DNA methylation is first inhibited. In contrast, inhibition of SIRT1 by pharmacologic, dominant negative, and siRNA (small interfering RNA)-mediated inhibition in breast and colon cancer cells causes increased H4-K16 and H3-K9 acetylation at endogenous promoters and gene re-expression despite full retention of promoter DNA hypermethylation. Furthermore, SIRT1 inhibition affects key phenotypic aspects of cancer cells. We thus have identified a new component of epigenetic TSG silencing that may potentially link some epigenetic changes associated with aging with those found in cancer, and provide new directions for therapeutically targeting these important genes for re-expression.

Figure 1. siRNA Knockdown of SIRT1 Causes Re-Expression of Epigenetically Silenced TSGs

(A) RNAi-3 is most effective for reduction of SIRT1 in MCF7 cells. Retroviral expression vectors encoding SIRT1 cDNA that produce short hairpin loop RNA targeting either distinct regions of SIRT1 mRNA (RNAi-1, −2, or −3) or a control (ctrl) were used to infect MCF7. Western blot analysis for SIRT1 and β-actin was performed 48 h after two rounds of infection. (B) Both RNAi-2 and −3 are effective for reduction of SIRT1 protein in MDA-MB-231 cells as described in (A). (C) SIRT1 inhibition leads to TSG re-expression in MCF7 cells. RNA was isolated from parallel samples analyzed in (A), and RT-PCR was performed with intron-spanning primers specific for the genes SFRP1 and SFRP2. GAPDH was also analyzed as a control. Only the shRNA (RNAi-3) that caused substantial reduction in SIRT1 protein leads to gene re-expression. Control samples in which no reverse transcriptase was added were analyzed separately, and all were negative for amplification of the indicated genes. (D) SIRT1 inhibition leads to TSG re-expression in MDA-MB-231 cells. RT-PCR was performed for analysis of the genes SFRP1, SFRP2, and E-cadherin as described in (A). Only the shRNAs (RNAi-2 and −3) that caused substantial reduction in SIRT1 protein lead to gene re-expression (E) SIRT1 inhibition leads to TSG re-expression in RKO cells. SIRT1 protein reduction by RNAi-3 (top panel) as described in (A) leads to gene re-expression of SFRP1, SFRP2, and MLH1 as described in (C) (F) MDA-MB-231 and RKO cells infected with control or RNAi-3 shRNA as described in (A) were selected with puromycin for 3 d, and pooled colonies were harvested for Western blot analysis of protein re-expression that corresponded with the gene reactivation described in (D) and (E).

Figure 2. Pharmacologic and Dominant Negative Inhibition of SIRT1 Cause Re-Expression of TSGs and Synergize with 5-Deoxy-Azacytidine or TSA

(A) Pharmacologic inhibition of SIRT1 causes TSG re-expression. MDA-MB-231 cells were treated with 15 mM NIA or 300 μM SPT for 21 h, RNA was isolated, and RT-PCR was performed with intron−spanning primers specific for the indicated genes. Control samples in which no reverse transcriptase was added were analyzed separately, and all were negative for amplification of the indicated genes. (B) Combined treatment with low doses of Aza and SPT synergizes in the re-expression of TSGs. MDA-MB-231 cells were treated with either 50 nM Aza (+), 100 μM SPT (+) or with both Aza and SPT (++), and 34 h later, RT-PCR was performed for the indicated genes as described in (A). (C) Combined treatment with SPT and TSA synergize in the re-expression of genes. MDA-MB-231 cells were treated with either 0, 50, 100, or 120 μM SPT alone for 34 h, or the treatment was followed by treatment with 300 nM TSA for 3 h prior to RNA isolation and RT-PCR analysis. (D) SIRT1 protein knockdown synergizes with low doses of Aza for gene re-expression. MDA-MB-231 cells were infected with low titers of virus for shRNA specific for SIRT1. Aza (100 nM) was added 24 h prior to RNA isolation, and RT-PCR analysis was performed for the genes SFRP1, SFRP2, and GAPDH as described in (A). (E) Dominant negative inhibition of SIRT1 leads to TSG re-expression in MCF7 cells. MCF7 cells were infected with virus encoding either pBabe (vec) or the catalytically inactive SIRT1H363Y (HY) mutant, and RT-PCR was performed as described in (A). (F) Dominant negative inhibition of SIRT1 leads to TSG re-expression and synergizes with TSA and Aza. As shown in the left panel, MDA-MB-231 cells were infected with a control (vec) or mutant SIRT1 virus (HY), and RT-PCR was performed as described in (A). MDA-MB-231 cells were infected with low titers of pBabe or pBabe-SIRT1H363Y retrovirus and subsequently treated with 100 nM Aza for 24 h or with 300 nM TSA for 3 h prior to harvest, and RT-PCR was performed.

Figure 3. SIRT1 Inhibition Causes TSG Re-Expression without Changing Promoter DNA Hypermethylation

(A) TSG re-expression occurs without changes in the methylation profile of multiple clones analyzed for SFRP1 promoter methylation. Parallel samples analyzed in Figure 1D were subjected to bisulfite sequencing of the SFRP1 promoter from MDA-MB-231 cells stably infected with control vector or RNAi-2 or RNAi-3 retrovirus. Open circles indicate unmethylated cytosines, and closed circles indicate methylated cytosines. Numbers at the bottom show the position of cytosines relative to the transcription start site, which is at position 0, and those with a minus sign (−) are upstream from this start site. The region sequenced encompasses the CpG island in which methylation status correlates with gene expression status. (B) MSP analyses of DNA from MDA-MB-231 cells stably expressing vector control, RNAi-2, or RNAi-3 retrovirus. From left to right: (-) PCR Ctrl indicates H₂O only; (-) BS ctrl indicates bisulfite-treated H₂0; (+) M ctrl indicates the cell line in which SFRP1 is partially methylated and SFRP2 and GATA4 are fully methylated; and (+) U ctrl indicates the Tera-2 cell line in which each gene is unmethylated. All remaining lanes are for MDA-MB-231. From left to right: Aza indicates 1 μM Aza (24 h) treatment; Ctrl indicates empty vector infection; RNAi-2 indicates shRNA-2 infection alone; RNAi-3 indicates shRNA-3 infection alone; Aza indicates 1 μM Aza (24 h) treatment of control cells; Ctrl indicates empty vector infection + vehicle; RNAi-2 indicates shRNA-2 infection + 5 mM NIA treatment; and RNAi-3 indicates shRNA-3 infection + 5 mM NIA treatment.

Figure 4. SIRT1 Inhibition Causes Re-Expression of Epigenetically Silenced TSGs

(A) RKO cells were infected and stably selected to express short hairpin loop RNA targeting either a region unique to SIRT1 mRNA or a control (ctrl). To inhibit any residual SIRT1 protein, remaining RNAi-expressing cells were treated with 700 μM SPT and control samples were treated with DMSO for 24 h. For comparison, control RNA was isolated from parallel samples from HCT116 cells in which the two genes under study, CRB1 and E-cadherin, do not have promoter DNA hypermethylation and are basally expressed. RKO cells were also treated with 0.5 μM Aza (24 h), and samples were analyzed as described in Figure 1A; RT-PCR was performed with intron-spanning primers specific for the two genes. GAPDH was also analyzed as a control. Only the shRNA (RNAi-3) that caused substantial reduction in SIRT1 protein leads to gene re-expression. Control samples in which no reverse transcriptase was added were analyzed separately, and all were negative for amplification of the indicated genes. (B) Parallel samples described above were analyzed using real-time quantitative PCR. The level of TSG re-expression induced by Aza treatment or SIRT1 inhibition as described in (A) was compared to levels of expression in HCT116 cells in which the TSGs are basally expressed.

Figure 5. SIRT1 Inhibition Causes Increases in Histone H4-K16 Acetylation at the Promoter of Re-Expressed Genes

(A) Pooled populations of MDA-MB-231 cells stably selected to express RNAi constructs were analyzed via ChIP. These samples were isolated in parallel to those analyzed in Figure 3B. ChIP was performed with antibodies against SIRT1, acetylated histone H4, lysine 16 (H4-K16), or with no antibody (NAB) controls. Each promoter sequence was amplified by PCR under linear conditions for the genes SFRP1 and E-cadherin. (B) The average change in SIRT1 localization, acetylation of H4-K16, and acetylation of H3K9 at the SFRP1 and E-cadherin promoters as measured by ChIP was quantitated for multiple experiments. Error bars indicate the standard deviation for multiple experiments. (C) SIRT1 localizes to the promoters of silent genes whose DNA is hypermethylated, but not to these same promoters in cells in which the genes are expressed. ChIP was performed with antibodies against SIRT1 in RKO and SW480 colon cancer cells. As shown in the left panel, SIRT1 localizes to the MLH1 promoter in RKO cells in which the gene is silent, but not to the MLH1 promoter in SW480 cells in which it is expressed. As shown in the right panel, SIRT1 localizes to the E-cadherin promoter in RKO cells in which the gene is silent, but not to the E-cadherin promoter in SW480 cells where it is expressed.

Figure 6. SIRT1 Inhibition Affects Key Phenotypic Aspects of Cancer Cells

(A) MDA-MB-231 cells were infected for two rounds with RNAi-2 and −3 retrovirus, and puromycin-resistant colonies were counted after 3 d of selection. Error bars indicate standard deviation from the average of three experiments. (B) RKO cells were transfected with 500 ng of pGL3-OT, a TCF-LEF−responsive reporter, or pGL3-OF, a negative control with a mutated TCF-LEF binding site in combination with 10 ng of pRL-CMV vector. Twenty-four hours post-transfection, cells were treated with either vehicle (DMSO) control or with 700 μM SPT for 24 h. Firefly luciferase activity was measured and normalized to the Renilla luciferase activities. (C) As described in (A), pooled populations of MDA-MB-231 cells stably expressing RNAi-2 or RNAi-3 were harvested, protein concentrations were determined, and Western blot analysis was performed. An antibody that specifically recognizes the unphosphorylated (active) form of β-catenin was used, and on the same blot, β-actin was probed to ensure equal loading. (D) Western blot analysis was performed on RKO cells expressing control or SIRT1 RNAi. Antibodies against SIRT1, phospho-GSK3β (inactive), cyclin D1, p27, and β-actin were used for Western blotting. On the same blot, β-actin was probed to ensure equal loading.