PKCε promotes oncogenic functions of ATF2 in the nucleus while blocking its apoptotic function at mitochondria

Abstract

The transcription factor ATF2 elicits oncogenic activities in melanoma and tumor suppressor activities in nonmalignant skin cancer. Here, we identify that ATF2 tumor suppressor function is determined by its ability to localize at the mitochondria, where it alters membrane permeability following genotoxic stress. The ability of ATF2 to reach the mitochondria is determined by PKCε, which directs ATF2 nuclear localization. Genotoxic stress attenuates PKCε effect on ATF2; enables ATF2 nuclear export and localization at the mitochondria, where it perturbs the HK1-VDAC1 complex; increases mitochondrial permeability; and promotes apoptosis. Significantly, high levels of PKCε, as seen in melanoma cells, block ATF2 nuclear export and function at the mitochondria, thereby attenuating apoptosis following exposure to genotoxic stress. In melanoma tumor samples, high PKCε levels associate with poor prognosis. Overall, our findings provide the framework for understanding how subcellular localization enables ATF2 oncogenic or tumor suppressor functions.

Figure 1. ATF2 localizes at the mitochondrial outer membrane in response to genotoxic stress. (See also Figure S1)

A) Control, 5 μM etoposide-treated (ETO, overnight treatment), UVC-treated (20 J/m²), ionizing radiation-treated (IR, 5Gy), or 40 ng/ml leptomycin B (LMB) and ETO co-treated SCC9 cells grown on coverslips were immunofluorescently stained for HSP60 (red), ATF2 (green) or DNA (blue). Control and ETO panels represent 89±4% and 64±11%, respectively, of the cells from three independent replicate coverslips per condition (n > 100 cells counted per replicate). B) Control- or ETO-treated SCC9 cells were harvested for whole cell lysate (W) or biochemically fractionated to cytoplasmic (C) and mitochondrial (M) fractions, and subject to immunoblot analysis with indicated antibodies. C) Jun2 promoter-luciferase-transfected SCC9 cells were assayed for luciferase activity prior to (Control) or after ETO treatment. *: p < 0.0001. D) Mitochondria (M) purified from ETO-treated SCC9 cells were subject to proteolysis protection assay with titrated concentrations of trypsin: 0.2 (1), 0.1 (2), 0.05 (3), 0.025 (4), or 0.0125 (5) μg in the absence or presence of 1% Triton X-100 (TX) detergent. Scale bars represent 10 μm.

Figure 2. ATF2 interacts with VDAC1 and HK1 complex at the mitochondria. (See also Figure S2)

A) Control or 5 μM etoposide (ETO, overnight)-treated SCC9 cells were lysed and subject to size exclusion chromatography separation. Every other fraction was subject to immunoblot analysis with indicated antibodies. Fractions are arranged, left to right, in order of descending molecular weight, with control protein weight markers indicated above (in kDa). B) Control- or ETO-treated SCC9 cells were crosslinked and lysed. ATF2 (upper) or HK1 (lower) were immunoprecipitated and subject to immunoblot analysis with indicated antibodies. C) Coverslip-grown SCC9 cells knocked-down for either HK1 or HK2 and subject to ETO treatment were immunofluorescently stained with indicated antibodies. D) Quantitation of mitochondrial ATF2 (MtATF2, grey bars) versus displaced ATF2 (red bars) before or after knockdown of HK1 or HK2, n > 50 cells per replicate experiment over three independent experiments. Scale bars= 10 μm.

Figure 3. PKCε phosphorylation of ATF2 on Thr52 negatively regulates its mitochondrial localization. (See also Figure S3)

Coverslip-grown SCC9 cells were immunofluorescently stained with indicated antibodies after: A) overnight treatment with 1 or 5 μM PKCε translocation inhibitor (PKCε-i); B) overnight treatment with 20 nM scrambled control (siSC) or PKCε-targeted (siPKCε) siRNA; C) transfection with HIS-tagged constitutively active PKCε (caPKCε, red) and treatment with 5 μM etoposide (ETO, overnight). D) SCC9 cells treated overnight with ETO or 5 μM PKCε translocation inhibitor (PKCε-i) were lysed and subject to immunoblot analysis with indicated antibodies. E) In vitro PKCε kinase assay with Millipore control peptide (CTL), purified GST-ATF2 50–100aa WT (WT), GST-ATF2 50–100aa WT + PKCε catalytic inhibitor, Gö6850 (WT+Gö) (3) or GST-ATF2 50–100aa T52A (52A) as substrate. *: non-specific bands. Coomassie-stained gel (left) and autoradiograph (right) are displayed as indicated. F) Coverslip-grown SCC9 cells were transfected with HA-tagged wild-type ATF2 (WT), ATF2^T52A or ATF2^T52E, and subsequently immunofluorescently stained with indicated antibodies. G) SCC9 control or ATF2 knocked-down cells (shATF2) were transfected with Jun2 promoter-luciferase construct and either empty vector (EV) or caPKCε were assayed for luciferase activity before and after overnight ETO or Gö6850 (2 μM) treatment.•: p = 0.0034; #: p = 0.0157. H) SCC9 cells stably knocked-down for ATF2 with a 3′-UTR-targeted shRNA were reconstituted with Jun2 promoter-luciferase construct and ATF2^WT, ATF2^T52A or ATF2^T52E and were subsequently assayed for luciferase activity. *: p = 0.0047. Scale bars represent 10 μm.

Figure 4. ATF2^T52A displaces HK1 from mitochondria. (See also Figure S4)

Coverslip-grown SCC9 cells were transfected with empty vector (EV) or HA-tagged ATF2 (WT), ATF2^T52A or ATF2^T52E in the absence (A, C, E, E*, G) or presence (B, D, F, F*, H) of 5 μM etoposide (ETO, overnight), and subsequently immunostained with indicated antibodies and MitoTracker. Yellow arrowheads indicate HA-positive cells. Scale bars represent 10 μm.

Figure 5. ATF2^T52A abrogates HK1:VDAC association, promoting mitochondrial leakage, cytochrome C release and cell death. (See also Figure S5)

A) SCC9 cells were transfected with empty vector (EV) or HA-tagged ATF2 (W), ATF2^T52A (A) or ATF2^T52E (E) in the presence or absence of 5 μM etoposide (ETO, overnight), crosslinked and lysed for whole cell lysate (input) or immunoprecipitated for VDAC1 and subject to immunoblot analysis with indicated antibodies. B) SCC9 cells were treated with ETO (E) or PKCε translocation inhibitor (i, 5 μM) or both (i-E), lysed for whole cell lysate (input) or immunoprecipitated for VDAC1 and subject to immunoblot analysis with indicated antibodies. C) SCC9 cells transfected with EV, W, A or E in the presence (*, red bars) or absence (grey bars) of 5 μM etoposide overnight were pulse labeled with tetramethylrhodamine ethyl ester (TMRE) or nonylacridine orange (NaO), and subsequently analyzed by FACS analysis. Histogram bars represent ratios of TMRE/NaO uptake. See Figure S4 for individual TMRE and NaO values. N = 10,000 cells per replicate; 3 replicates per condition were performed. •: p < 0.0001; #: p = 0.01 D) Coverslip-grown SCC9 cells transfected with EV, W, A or E were immunostained for cytochrome C and MitoTracker, and analyzed for cytochrome C release by fluorescence microscopy (n>100 cells per replicate; 3 replicates per condition were performed).•: p = 0.0008 E) Right, SCC9 cells treated with PKCε-i (5μM) alone or in the presence of ETO (overnight) were stained with Annexin-V and propidium iodide (PI) and subject to FACS analysis. •: p < 0.0001; #: p = 0.6018. Representative FACS plots (left) and corresponding Annexin V-negative and Annexin V-positive (AV−, AV+) percentages are as follows: control (97%, 3%); PKCε-i (92%, 8%); ETO (89.5%, 11.5%); and ETO+PKCε-i (87%, 13%). Averaged value histograms (right) are displayed as indicated. F) SCC9 cells were transfected with EV, W, A or E in the presence (*, red bars) or absence (grey bars) of ETO. Cells were stained with Annexin-V and subject to FACS analysis. n = 10,000 cells per replicate; three replicates per condition were performed. •: p = 0.0001; #: p<0.0001

Figure 6. PKCε phosphorylation of ATF2 on Thr52 and resistance of melanoma cells to genotoxic stress-induced cell death. (See also Figure S6)

A) SCC9 cells were transfected with empty vector or constitutively active PKCε (caPKCε), and subject to immunoblot analysis with indicated antibodies. Densitometric ratios of bands for (−caPKCε, +caPKCε) are as follows: pS729/β-actin (0.193, 0.636); pT52/Total ATF2 (0.215, 0.662). B) Control or ATF2-knocked down LU1205 cells were subject to immunoblot analysis with indicated antibodies. C) Indicated cell lines were subject to (10 μM, ETO, overnight) etoposide treatment and subject to immunoblot analysis with indicated antibodies. Densitometric band ratios for pT52/Total ATF2 (−ETO, +ETO) are as follows: SCC9 (1, 0.4), UACC903 (1, 0.7), 501Mel (1, 0.9), LU1205 (1, 1). Densitometric band ratios for pS729/Total PKCε (−ETO, +ETO) are as follows: SCC9 (1, 0.7), UACC903 (1, 0.7), 501Mel (1, 0.97), LU1205 (1, 1). D) Control or ETO-treated coverslip-grown LU1205, 501Mel or UACC903 cells were immunofluorescently stained with indicated antibodies. E) 501Mel cells treated with 10 μM PKCε translocation inhibitor (PKCε-i) alone or in the presence or absence of 5 μM etoposide (ETO, overnight) were stained with Annexin-V and subject to FACS analysis. n = 10,000 cells per replicate; three replicates per condition were performed. •: p = 0.0009; #: p = 0.0032

Figure 7. PKCε and phospho-T52 ATF2 expression in normal cells and malignant tumors. (See also Figure S7)

A) Primary keratinocytes (NHEK) and melanocytes (HEM), squamous carcinoma (M7, P9, SCC9) and melanoma (501Mel, LU1205, WM793 and UACC903) cell lines were subject to immunoblot analysis with indicated antibodies. Histogram represents ratio A/B, where A = densitometric ratio values of pT52/total ATF2 bands, and B = densitometric ratio values of pS729/total PKCε bands. B) Box plots showing distribution of PKCε in metastatic (left) and primary (right) specimens. PKCε levels are denoted on the Y-axis (P < 0.0001, t-statistic – 4.257). The mean +/− one standard deviation is depicted by the horizontal bars and is higher in the metastatic specimens. C) Kaplan Meier survival curves showing significantly shorter survival in higher PKCε expressers in primary tumors. D) Alignment of amino acid sequences flanking Thr52 in ATF2, as compared to consensus sequence used as PKCε inhibitor peptide. E) Schematic representation of PKCε-mediated regulation of the subcellular localization, and therefore, the oncogene or tumor suppressor activities of ATF2.