Identification of the SUMO E3 ligase PIAS1 as a potential survival biomarker in breast cancer

Abstract

Metastasis is the ultimate cause of breast cancer related mortality. Epithelial-mesenchymal transition (EMT) is thought to play a crucial role in the metastatic potential of breast cancer. Growing evidence has implicated the SUMO E3 ligase PIAS1 in the regulation of EMT in mammary epithelial cells and breast cancer metastasis. However, the relevance of PIAS1 in human cancer and mechanisms by which PIAS1 might regulate breast cancer metastasis remain to be elucidated. Using tissue-microarray analysis (TMA), we report that the protein abundance and subcellular localization of PIAS1 correlate with disease specific overall survival of a cohort of breast cancer patients. In mechanistic studies, we find that PIAS1 acts via sumoylation of the transcriptional regulator SnoN to suppress invasive growth of MDA-MB-231 human breast cancer cell-derived organoids. Our studies thus identify the SUMO E3 ligase PIAS1 as a prognostic biomarker in breast cancer, and suggest a potential role for the PIAS1-SnoN sumoylation pathway in controlling breast cancer metastasis.

Fig 1. Characterization of the PIAS1 antibody.

(A) PIAS1 and actin-immunoblots of lysate of 293T cells expressing one of two different concentrations (1X or 10X) of FLAG-tagged PIAS1, PIAS2-xα, PIAS2-xβ, PIAS3 or, PIAS4, or transfected with the vector control. Actin immunoblotting was used as loading control. Only PIAS1 antibody-immunoreactive bands corresponding to endogenous PIAS1 and exogenous FLAG tagged-PIAS1 were detected. Immunoblots are from an experiment that was repeated twice with similar results. (B) PIAS1 or actin immunoblots of lysate of MDA-MB-231 cells transiently transfected with an RNAi vector control or receiving a pool of plasmids expressing short hairpin RNAs against two distinct regions of PIAS1 [18,19]. Control and PIAS1 RNAi plasmids also express CMV-driven green fluorescence protein (GFP). Immunoblots are from an experiment that was repeated twice with similar results. (C) Representative PIAS1 (red), GFP (green) and nuclei (blue) fluorescence microscopy micrographs of MDA-MB-231 cells transfected as in B, and subjected to anti-PIAS1 indirect immunofluorescence and counterstained with Hoechst 33342 fluorescent nucleotide dye to visualize nuclei. GFP signal indicate control vector or PIAS1 RNAi plasmid-transfected cells. Arrow shows an example of each a vector transfected cell (upper row) and a PIAS1 RNAi transfected cell (lower row) to highlight the knockdown of endogenous PIAS1 in PIAS1i-transfected cell. These experiments were repeated two times with similar outcomes. (D) Representative PIAS1 immunoblots of serially-diluted lysate of 293T cells (upper panel), and protein abundance of PIAS1 quantified by densitometry (y-axis) plotted versus the protein amount of lysate (x-axis) (lower panel). The abundance of PIAS1 for each point in the XY graph is the mean ± SEM from three independent experiments including the one shown in upper panel. Regression analysis indicated that the protein abundance of PIAS1 follows a linear relationship with total protein amount in cells lysates.

Fig 2. PIAS1 protein abundance analysis in reference TMA and breast cancer TMA.

(A) PIAS1 and actin immunoblots of lysate of MDA-MB-231 cells expressing four increasing concentrations of PIAS1 (samples 1 to 4). (B) Bar graph of actin-normalized PIAS1 protein abundance in samples 1 to 4 shown in A and expressed relative to actin-normalized PIAS1 abundance in sample 3. (C) PIAS1 immunoblot of serially diluted lysate of MDA-MB-231 cells overexpressing PIAS1 in sample 4. (D) XY-graph plot of the lysate's total protein on the x-axis versus PIAS1 protein abundance, quantified from PIAS1immunoblot in C, on the y-axis. (E) Representative PIAS1 (red), and nuclei (blue) fluorescence micrographs of sections of Histogel-embedded MDA-MB-231 cells reference TMA expressing increasing abundance of PIAS1, corresponding to samples 1 to 4 in panel A, which were subjected to anti-PIAS1 indirect immunofluorescence and DAPI dye staining to visualize nuclei. (F) Bar graph depicts AQUA analysis software-quantified PIAS1 abundance in the reference TMA shown in E. (G) The XY-graph shows the relationship between Log of relative abundance of PIAS1 in samples 1 to 4 of MDA-MB-231 cell lysates quantified by immunoblotting on the x-axis versus Log of abundance of PIAS1 in MDA-MB-231 samples 1 to 4 of reference TMA quantified by AQUA analyses of immunocytochemistry on the y-axis. (H) Representative fluorescence microscopy micrographs of histogel-MDA-MB-231 cell reference TMA blocks. (I) Representative fluorescence microscopy micrographs of paraffin-embedded normal breast tissue and examples of three breast cancer tissues expressing different amounts of PIAS1. For both H and I, TMA were subjected to anti-PIAS1 (red) and anti-Pan cytokeratin (green) antibodies indirect immunofluorescence, and nuclear counterstaining with DAPI (blue). PIAS1-Cytokeratin-Nuclei merged fluorescence micrograph panels show relative abundance of PIAS1 in reference breast cancer TMA, normal breast and breast cancer tissue array.

Fig 3. PIAS1 as a potential positive predictor of overall survival in breast cancer.

(A) Kaplan Meier survival curves showing univariate analysis with number of individuals at risk for each group detailed in table below for whole tumor PIAS1 expression by AQUA, no expression vs any expression (logrank p-value = 0.0127) suggest that patients with any PIAS1 tumor expression were less likely to have a DSOS event. (B) Kaplan Meier survival curves showing univariate analysis with number at risk for each group detailed in table below for PIAS1 tumor nuclear expression relative to PIAS1 tumor cytoplasmic expression by HALO (logrank p-value = 0.0246) suggest that patients who had higher PIAS1 tumor nuclei expression, relative to the tumor cytoplasm, had a more favorable DSOS. Number at risk refers to the number of patients who are still alive in each group at the specified time point as indicated by the x-axis of the Kaplan Meier graph.

Fig 4. Endogenous SnoN mediates the ability of PIAS1 to suppress the invasive growth of breast cancer cell-derived organoids.

(A) SnoN and actin immunoblots of lysate of MDA-MB-231 cells cotransfected with a plasmid encoding a short hairpin RNA targeting a specific region of SnoN mRNA or a control plasmid, together with an expression plasmid encoding an RNAi-resistant SnoN (SnoN(r)) or wild type RNAi-sensitive SnoN, or the corresponding vector control. (B) Representative DIC light microscopy micrographs of untreated or 100pM TGFβ-treated 8-day old three-dimensional organoids derived from MDA-MB-231 cells stably expressing PIAS1 or the corresponding vector control, and co-transfected with SnoN RNAi, SnoN(r), wild type SnoN, or respective control plasmids. (C) Bar graph depicts mean ± SEM proportion of non-deformed organoids expressed as a percentage of total colonies counted for each experimental condition from three independent experiments including the one shown in B. Non-deformed organoids represents non-invasive growth phenotype. Knockdown of endogenous SnoN promoted invasive growth of organoids as compared to vector control or PIAS1 stably expressing cells in the absence or presence of TGFβ. Expression of SnoN(r) but not wild type SnoN suppressed the ability of SnoN RNAi to promote invasive growth of organoid in vector control or PIAS1 expressing cells in the absence or presence of TGFβ. Significant difference, ANOVA: ***P≤0.001. Scale bar indicates 50 μm. Arrows and arrowheads indicate non-deformed and invasive organoids, respectively.

Fig 5. TGFβ regulates PIAS1 protein stability and SnoN sumoylation abundance in breast cancer cells.

(A) PIAS1 and actin immunoblots of lysates of MDA-MB-231 cells pre-incubated without or with 100pM TGFβ for 12 hours prior to treatment with 10μg/ml cycloheximide (CHX) for 0, 1, 2, 4, 8 and 12 hours (h). (B) An XY-graph of CHX treatment (h) on x-axis versus mean ± SEM (n = 3 experiments including one shown in A) of relative actin-normalized PIAS1 abundance in cells pre-incubated without (blue line) or with TGFβ (red line) on the y-axis. TGFβ increases the PIAS1 protein abundance turnover. (C) Bar graph represents the mean ± SEM of PIAS1 protein half-life in MDA-MB-231 cells pre-incubated without (blue) or with TGFβ (red) quantified from three-independent experiments including from the experiment shown in A. Half-life of PIAS1 protein was determined from graph shown in B by interpolation. TGFβ reduced PIAS1 protein half-life by 57%. (D) SUMO1 and SnoN immunoblots (panels 1 and 2) of NEM-treated SnoN immunoprecipitation (SnoN IP) of lysate of SnoN-expressing MDA-MB-231 cells incubated for 12 hours without or with TGFβ, KI, alone or together. Unmodified SnoN has a MW of 77 KDa (arrow), and sumoylated SnoN protein species run as 100 KDa and higher as detected in SUMO and SnoN immunoblots of SnoN immunoprecipitation (SUMO(n)SnoN, vertical lines) [12,18]. SnoN and actin immunoblots (panels 3 and 4) of lysates of cells treated as described above for panels 1 and 2 are also shown. (E) The bar graph represents the mean ± SEM of proportion of sumoylated SnoN relative to unmodified SnoN quantified from SnoN immunoblots of SnoN immunoprecipitation and expressed relative to the proportion of sumoylated SnoN in the untreated control from three independent experiments including the one shown in D. TGFβ-signalling reduces the proportion of sumoylated SnoN by 68% in MDA-MB-231 cells. Statistical difference, (C) Student t-test: ***P≤ 0.001, (E) ANOVA: **P≤0.01, *P≤0.05.

Fig 6. Sumoylation of SnoN supresses TGFβ-induced invasiveness of breast cancer organoids.

(A) SnoN and actin immunoblots of lysates of MDA-MB-231 cells expressing SnoN (WT), SnoN (KdR), SUMO-SnoN or transfected with a control vector. (B) Representative DIC light microscopy micrographs of untreated or 100pM TGFβ-treated 8-day old three dimensional organoids derived from MDA-MB-231 cells transfected as in A. (C) Bar graph represents mean ± SEM proportion of non-deformed organoids expressed as a percentage of total colonies counted for each experimental condition from four independent experiments including the one shown in B. Non-deformed organoids represents non-invasive growth phenotype. SnoN (KdR) promoted an invasive growth of breast cancer cell-derived organoids even in the absence of TGFβ. SUMO-SnoN suppressed TGFβ-induced invasive growth of breast cancer cell-derived organoids. Significant difference, ANOVA: ***p<0.001. Scale bar indicates 50 μm. Arrows and arrowheads indicate non-deformed and invasive organoids, respectively.

Fig 7. PIAS1 suppresses TGFβ-induced budding and disruption of breast cancer cell-derived organoids via sumoylation of SnoN.

(A) Representative DIC light microscopy micrographs of untreated or 100pM TGFβ-treated 8-day old organoids derived from MDA-MB-231 cells stably expressing SUMO-SnoN or the control vector, and transfected with a pool of plasmids encoding short hairpin RNAs targeting distinct regions of PIAS1 mRNA, or a control RNAi plasmid. (B) Bar graph represents mean ± SEM proportion of non-deformed organoids expressed as a percentage of total colonies counted for each experimental condition from three independent experiments including the one shown in A. SUMO-SnoN suppressed the ability of PIAS1 knockdown to promote invasive growth of MDA-MB-231 cell-derived organoids in absence or presence of TGFβ. (C) Representative DIC light microscopy micrographs of untreated or 100pM TGFβ-treated 8-day old organoids derived from MDA-MB-231 cells stably expressing SUMO-SnoN or the control plasmid, and transfected with a vector expressing PIAS1 (CS) or the control vector. (D) Bar graph represents mean ± SEM proportion of non-deformed organoids expressed as a percentage of total colonies counted for each experimental condition from three independent experiments including the one shown in C. SUMO-SnoN suppressed PIAS1 (CS)-induction of invasive growth of MDA-MB-231 cell-derived organoids in absence or presence of TGFβ. (E) Representative DIC light microscopy micrographs of untreated or 100pM TGFβ-treated 8-day old organoids derived from MDA-MB-231 cells stably expressing SnoN (WT), SnoN (KdR) or a vector control, and transfected with PIAS1 (WT) or the control vector. (F) Bar graph depicts mean ± SEM proportion of non-deformed organoids expressed as a percentage of total colonies counted for each experimental condition from three independent experiments including the one shown in E. SnoN (KdR) antagonized the ability of PIAS1 (WT) to suppress the invasive growth of MDA-MB-231 cell-derived organoids in absence or presence of TGFβ. Significant difference, ANOVA: ***P≤0.001, **P≤0.01. Scale bar indicates 50 μm. Non-deformed organoids represents non-invasive growth phenotype. Arrows and arrowheads indicate intact non-deformed and invasive organoids, respectively.