PARP-1 regulates metastatic melanoma through modulation of vimentin-induced malignant transformation

Abstract

PARP inhibition can induce anti-neoplastic effects when used as monotherapy or in combination with chemo- or radiotherapy in various tumor settings; however, the basis for the anti-metastasic activities resulting from PARP inhibition remains unknown. PARP inhibitors may also act as modulators of tumor angiogenesis. Proteomic analysis of endothelial cells revealed that vimentin, an intermediary filament involved in angiogenesis and a specific hallmark of EndoMT (endothelial to mesenchymal transition) transformation, was down-regulated following loss of PARP-1 function in endothelial cells. VE-cadherin, an endothelial marker of vascular normalization, was up-regulated in HUVEC treated with PARP inhibitors or following PARP-1 silencing; vimentin over-expression was sufficient to drive to an EndoMT phenotype. In melanoma cells, PARP inhibition reduced pro-metastatic markers, including vasculogenic mimicry. We also demonstrated that vimentin expression was sufficient to induce increased mesenchymal/pro-metastasic phenotypic changes in melanoma cells, including ILK/GSK3-β-dependent E-cadherin down-regulation, Snail1 activation and increased cell motility and migration. In a murine model of metastatic melanoma, PARP inhibition counteracted the ability of melanoma cells to metastasize to the lung. These results suggest that inhibition of PARP interferes with key metastasis-promoting processes, leading to suppression of invasion and colonization of distal organs by aggressive metastatic cells.

Figure 1. PARP inhibition down-regulates vimentin expression and inhibits endothelial-to-mesenchymal transition in HUVECs.

Cell extracts from HUVEC either treated with vehicle or 40 µM DPQ were subjected to 2D electrophoresis as described in Materials and Methods. Image analysis software (DeCyder) indicated that seven proteins exhibited decreased expression in HUVEC treated with DPQ compared to untreated cells. Proteins were identified using MALDI-TOF. Spots labeled with arrows indicate proteins that were identified by mass spectrometry (see Figure 2). (A) The spot with the arrow is vimentin. (B) PARP inhibition reduced the expression of both vimentin and Snail1 and up-regulated VE-cadherin in human endothelial cells (HUVEC) as determined by immunoblotting, indirect immunofluorescence (C), and mRNA levels (D). PARP inhibition decreased HUVEC cell migration (E). (**P<0.01, ***P<0.001 PARP inhibitor groups versus DPQ).

Figure 2. Proteins differentially expressed and identified by mass spectrometry analysis in HUVEC.

The level of expression of various proteins in HUVEC was altered following PARP inhibition as determined by 2D-DIGE, and the proteins were positively identified using mass spectrometry analysis. Of particular interest for this study was vimentin, the major structural protein of intermediary filaments (spot 1). Expression of this protein was decreased in HUVEC following PARP inhibition. The proteins were identified by MALDI-TOF. Sequence coverage (%) and number of peptides were identified with  = 1% FDR (false discovery rate cut-off against decoy-concatenated randomized database). Coverage and score was determined using the MASCOT algorithm. The average ratio of protein expression between the control and cells treated with the PARP inhibitor DPQ was determined in HUVEC.

Figure 3. PARP inhibition inhibits the acquisition of an EMT phenotype in malignant melanoma cells.

Human melanoma G361 cells and murine B16- F10 melanoma cells (Figure S3) were used for these experiments. Cells were treated with either DPQ (40 µM), PJ-34 (10 µM) or KU0058948 (100 nM) for 22 hours. IF, western blot or qPCR assays were performed to evaluate the effects of PARP inhibition on EMT markers. PARP inhibition reduced the expression of vimentin and Snail1 and up-regulated E-cadherin in human melanoma cells as determined by immunoblotting (A), indirect immunofluorescence (B), and mRNA levels (C). (*P<0.05, ***P<0.001, PARP Inhibitor groups versus the control). β-actin was used as an internal control for protein loading. (D) Snail1 and E-cadherin promoter activity are regulated by PARP inhibitors. Luciferase activity was determined after transfecting the constructions into G361 cells. Firefly Luciferase was standarized to the levels of Renilla Luciferase. Cells were cotransfected with 0.5 µg renilla as a transfection control and 0.5 µg of Snail1 or E-cadherin using jetPEI cationic polymer transfection reagent according to the manufacturer's instructions. Cells were compared in the presence or absence of serum (***P<0.001 control versus PJ-34). The expression of both Firefly and Renilla luciferase was analyzed 48 h after transfection. Cloning of the human Snail1 promoter (−869/+59) into pGL3 basic (Promega) was described previously (41). The E-Cadherin promoter was cloned into pGL3-basic (Promega) to generate pGL3-E-cadherin (−178/+92). (E) Inhibitory effect of PARP on B16F10 motility. Treatment with the PARP inhibitor PJ-34 (10 µM) decreased cell migration in vitro. Migration was quantified as distance between Wound Healing limits (*** P<0.001 control versus DPQ).

Figure 4. PARP-1 or vimentin is sufficient to reverse EMT and confer increased cell motility.

(A) Melanoma (G361) and endothelial (HUVEC) (B) cells were silenced for PARP-1 or vimentin and the expression levels of Axl, E-/VE-cadherin, Snail1, ILK, β-catenin, GSK-3β, PARP-1, and vimentin were determined by immunoblot. (C) HUVEC were silenced for vimentin and wound healing was measured. After over-expression of vimentin wound healing closure was measured in HUVEC cells (D) or B16-F10 (E). (F) Cell migration was analyzed in epithelial cell line Madin Darby canine kidney (MDCK) cells transfected with either GFP or GFP-vimentin using video-microscopy and MetaMorph Image Analysis software. While vimentin was able to increase the length of the trajectories in the absence or presence of hepatocyte growth factor (HGF), treatment with PARP inhibitor resulted in a sustained reduction in cell motility (*P<0.05 PJ-34 or olaparib versus control).

Figure 5. Interaction between vimentin over-expression and the activation of EMT signaling pathway.

(A) Over-expression of vimentin in G361 cells. (B) Forced expression of vimentin drives human breast tumor epithelial cells (MCF7) to a mesenchymal phenotype through the integrin-linked-kinase/GSK-3β axis. 5 mM LiCl was used to inhibit GSK-3β, as detected by the accumulation of beta-catenin. (C) ILK was knocked down to analyze the significance of the interaction between vimentin and ILK in promoting the transition to a mesenchymal phenotype.

Figure 6. Vasculogenic mimicry is reduced by PARP inhibition in cells and in xenogafts of malignant melanoma.

(A) Western-blot and immunofluorescence of VE-cadherin and pVE-cadherin in B16-F10 cells treated with PJ-34 or KU0058948. (B, C) B16-F10 cells were cultured on polystyrene-treated culture slides and treated with the PARP inhibitor PJ-34 at 20 µM or left untreated. Following treatment, pictures were taken and analyzed using Wimasis image analysis software. “Branching points”: crossroads from at least three “branches”. “Loops”: Closed areas surrounded by cells. Four independent experiments were performed (*P<0.05; **P<0.01).

Figure 7. Decreased melanoma-induced lung metastasis following PARP inhibition.

(A) Mice were inoculated with the murine melanoma cell line B16-F10-luc. Localization and the intensity of luciferase expression were monitored by in vivo bioluminescence imaging (dpi, days post cells injection). At the bottom of Figure A two lungs from vehicle (left) or DPQ (right) treated mice are shown. Lungs were extracted to analyze the number of melanoma foci. Quantification of luciferase activity over time shows the average light (photons) emission in photons/s (B) (**P<0.01; ***P<0.001 versus DPQ). (C) The number of metastatic foci/lung were counted macroscopically (***P<0.001). (D) Angiogenesis was measured using a specific endothelial cell marker (tomato lectin) and measured as blood vessels per mm² in tumor sections of lung metastasis (Columns, mean ± SE. *P<0.05, with respect to control and DPQ–treated mice. (E) Immunohistochemistry staining of Snail1 and E-Cadherin in lung metastasis and quantitation using ImageJ , colour deconvolution plugin (F and G) Kaplan-Meyer survival curve shows the survival advantage of DPQ-treated mice following intravenous tail injection of melanoma cells as previously described in mice treated with DPQ (F) or injected with B16F10 stably silenced for PARP-1 (G) (** P<0. 01).

Figure 8. PARP inhibitors interfere with EndoMT, EMT and vasculogenic mimicry in melanoma cells.

Vimentin down-regulation is pivotal in driving this effect of PARP inhibitors, acting through the ILK/GSk-3β (see the text). While VE-cadherin is upregulated by PARP inhibitors in endothelial cells, contributing to vascular normalisation, its levels are down-regulated in malignant melanoma cells (Figure 5C). The ultimate reason for this cell-specific regulation of VE-cadherin expression by PARP is being studying currently in our laboratory.