The Epithelial-Mesenchymal Transition (EMT) Regulatory Factor SLUG (SNAI2) Is a Downstream Target of SPARC and AKT in Promoting Melanoma Cell Invasion

During progression of melanoma, malignant melanocytes can be reprogrammed into mesenchymal-like cells through a process similar to epithelial-mesenchymal transition (EMT), which is associated with downregulation of the junctional protein E-cadherin and acquisition of a migratory phenotype. Recent evidence supports a role for SLUG, a transcriptional repressor of E-cadherin, as a melanocyte lineage transcription factor that predisposes to melanoma metastasis. However, the signals responsible for SLUG expression in melanoma are unclear and its role in the invasive phenotype is not fully elucidated. Here, we report that SLUG expression and activation is driven by SPARC (also known as osteonectin), a secreted extracellular matrix-associated factor that promotes EMT-like changes. Ectopic expression or knockdown of SPARC resulted in increased or reduced expression of SLUG, respectively. SLUG increase occurred concomitantly with SPARC-mediated downregulation of E-cadherin and P-cadherin, and induction of mesenchymal traits in human melanocytes and melanoma cells. Pharmacological blockade of PI3 kinase/AKT signaling impeded SPARC-induced SLUG levels and cell migration, whereas adenoviral introduction of constitutively active AKT allowed rescue of SLUG and migratory capabilities of SPARC knockdown cells. We also observed that pharmacological inhibition of oncogenic BRAFV600E using PLX4720 did not influence SLUG expression in melanoma cells harboring BRAFV600E. Furthermore, SLUG is a bona fide transcriptional repressor of E-cadherin as well as a regulator of P-cadherin in melanoma cells and its knockdown attenuated invasive behavior and blocked SPARC-enhanced cell migration. Notably, inhibition of cell migration in SPARC-depleted cells was rescued by expression of a SLUG transgene. In freshly isolated metastatic melanoma cells, a positive association between SPARC and SLUG mRNA levels was also found. These findings reveal that autocrine SPARC maintains heightened SLUG expression in melanoma cells and indicate that SPARC may promote EMT-associated tumor invasion by supporting AKT-dependent upregulation of SLUG.


Introduction
Epithelial to mesenchymal transition (EMT) is a highly conserved developmental program activated during mesoderm formation and neural crest development. This program has also been implicated in promoting dissemination of single malignant cells from primary epithelial tumors [1]. During EMT, cells discard their epithelial characteristics, including cell adhesion and polarity, reorganize their cytoskeleton and acquire a mesenchymal morphology and the ability to migrate. One of the hallmarks of EMT is the functional loss of the cell-cell junction protein Ecadherin. E-cadherin is considered a suppressor of tumor invasion and consistently, loss or partial loss of E-cadherin has been associated with metastatic dissemination and poor prognosis in several solid tumors [1]. Several transcription factors have been identified that can repress E-cadherin expression including SNAIL/SNAI1, SLUG/SNAI2, ZEB1, ZEB2/SIP1, Twist proteins and E47 [2]. These EMT transcription factors bind to E-box elements in the promoter region of E-cadherin leading to transcriptional repression of junctional complexes and induction of the mesenchymal phenotype.
Cutaneous melanoma is an aggressive and potentially fatal form of cancer that derives from melanin-producing melanocytes in the epidermis. Melanocytes originate in the neural crest, a population of highly migratory embryonic cells [3]. Melanoma is a neoplasm of neuroectodermal origin and because of this melanoma cells may not undergo classic EMT-like changes. However, their ability to invade into the dermis is associated with an EMT-like phenotype characterized by changes in expression of cell-cell adhesion molecules of the cadherins family [4,5]. In normal skin, Ecadherin mediates contacts between melanocyte and adjacent keratinocytes. During melanoma progression, the transition from radial growth phase (RGP) to invasive or vertical growth phase (VGP) is characterized by decreased E-cadherin expression that results in the loss of keratinocyte-mediated growth and motility control [6]. In addition to the loss of E-cadherin, downregulation of other members of classical cadherins such as P-or H-cadherin as well as generation of a truncated secreted form of P-cadherin are frequently observed during progression of melanomas [7][8][9]. Over the past several years, major advances have been made in the identification of genetic factors that contribute to melanoma initiation such as activating mutations in the oncogenes BRAF and NRAS, however the molecular mechanisms that govern the transition of RGP to VGP with the capacity to metastasize and EMT-related events are still poorly characterized. SLUG (or SNAI2) belongs to the SNAIL superfamily of zingfinger transcription factors. These developmental proteins are central regulators of EMT during neural crest cell migration and cancer [10]. SLUG is critical for the normal development of neural crest-derived cells [11] and loss-of-function SLUG mutations result in piebaldism and Waardenburg syndrome type 2, two human disorders associated with defective functioning of the embryonic neural crest [12,13]. SLUG is also required for epithelial cell motility during wound healing and SLUG null mice have retarded epithelial migration rates [14]. More interesting, in melanoma, it was shown that SLUG functions as a melanocytespecific factor required for the strong metastatic propensity of this tumor [15]. This study also revealed that a gene relevant to embryonic neural crest development might also play a role in the acquisition of the metastatic phenotype. More recently, SLUG was also shown to regulate the transcriptional activity of ZEB1, another important EMT regulator [16]. Apart from its role as an E-cadherin repressor and EMT inducer, SLUG is a survival molecule that promotes resistance to apoptosis in some cell contexts [17][18][19][20]. In addition, SLUG is a p53 target that antagonizes p53-mediated apoptosis [21]. Whereas SLUG is recognized to drive melanoma metastasis and a key EMT inducer, an important question remains concerning the mechanisms that contribute to the expression and regulation of SLUG during melanoma progression.
Among the proteins whose expression is associated with metastasis and EMT-like changes is the secreted matricellular protein SPARC (also known as osteonectin), which regulates diverse cellular functions, including cell adhesion, migration, cell cycle and survival in a cell-type-and context-dependent manner [22,23]. Suppression of SPARC expression in human melanoma cells reduced their tumorigenic potential in mouse xenograft assays and inhibited migratory and invasive abilities of human melanoma cells in vitro [22,24,25]. Conversely, the expression of SPARC in human melanocytes was sufficient to repress E-and P-cadherin and promote epithelial-mesenchymal like transition with acquisition of a migratory phenotype [22]. Our previous study also found that SPARC regulates SNAIL expression in melanocytes and melanoma cells [22]. However, it remains to be addressed whether and how SPARC activates other EMT-inducing transcription factors to suppress E-cadherin and promote cell migration. In this work, we studied further the interplay between SPARC, E-and Pcadherin repression and EMT-like induction in melanoma cells. Our results show SPARC/AKT-dependent regulation of SLUG as an important mechanism underlying EMT-induced cell migration in melanoma.

Plasmids
The Myc-tagged human SPARC expression plasmid was described previously [22]. Wild-type (2178 wt) and E2 box mutant (mE-pal) mouse E-cadherin promoter constructs fused to luciferase were the generous gifts of Dr. Amparo Cano and have been described previously [26]. Flag-tagged SLUG expression plasmid (Addgene plasmid 25696) was from Dr. Eric R. Fearon.

Cells and Reagents
Human 501mel cells, kindly supplied by Ruth Halaban were described elsewhere [27]. Human WM9 and WM793 cell lines were the kind gift of Meenhard Herlyn and have been previously described [28]. MeWo and SK-mel28 cell lines were purchased from ATCC. Human melanoma cells were cultured in Dulbecco's modified Eagle Medium (DMEM) supplemented with 7% fetal bovine serum (Hyclone). Radial growth phase (SGM3 and SGM4) and vertical growth phase, (SGM5 and ME1402) melanoma lines were provided by the Wellcome Trust Functional Genomics Cell Bank at St George's and described previously [29]. M234, Melan20, M110 and M253b cells were derived from metastatic melanoma tumors by the Dermato-Oncology Department of Nantes University Hospital. A written consent was obtained from the patients for the use of samples and studies were approved by the local ethics committees ''Comité de Protection des Personnes Ouest IV-Nantes'' and the ''Agence Française de Sécurité Sanitaire des Produits de Santé''. Human primary epidermal melanocytes were isolated from foreskin and cultured as described previously [22]. 501mel cells expressing a Myc-tagged SPARC (501mel SPARC clones #1 and #2) or carrying an empty expression cassette of pcDNA3 vector (501mel CTRL) were described earlier [25]. 501mel cells were stably transfected with SLUG expression plasmid and bulk selected in 2 mg/ml puromycin prior to protein analysis and migration assays. Primers and culture reagents were purchased from Invitrogen. AKT inhibitor IV (AIIV) and PI3 kinase inhibitor LY29004 were from Calbiochem. BRAF inhibitor PLX4720 was from Synkinase. Cells were exposed to the various inhibitors at the indicated concentrations for 3 hours. An equal amount of DMSO was used as vehicle control. All other reagents were obtained from Sigma unless stated otherwise.

Real-time Quantitative PCR
Total RNA was extracted from cell samples using Nucleospin RAII kit (Macherey-Nagel) and following the manufacturer's instructions. Recovered RNA samples were quantified using NanoDrop spectrophotometer ND1000 (Thermo Fisher Scientific). Reverse transcription was performed on 0.5 mg of total RNA in a volume of 20 mL using High capacity cDNA Reverse Transcription kit (Applied Biosystems) according to the manufacturer's instructions. Quantitative PCR was performed on 25 ng cDNA samples, in sealed 384-well microtiter plates using the SYBR Green TM PCR Master Mix (Applied Biosystems) with the 7900HT Fast Real-Time PCR System (Applied Biosystems). Transcript relative amounts were determined using the deltadelta-Ct method [30] and human 18S transcript level was used as internal normalizer for each sample. Primer pairs for each cDNA were designed using Primer Express Software (Applied Biosystems). Primer sequences are shown in Table 1.

Adenoviral Gene Transduction
Adenoviruses carrying an empty expression cassette of pcDNA3 vector, used as control (AdCTRL) or expressing SPARC with a Myc-tag at its carboxyl terminus were described previously (AdSPARC) [22]. Adenovirus expressing the constitutively active mutant of AKT1, Myr-HA-AKT1 (AdAKT ca ), was purchased from Vector Biolabs. Adenoviruses were amplified as earlier described [31]. Cells were infected at multiplicity of infection of 5 and assayed 3 days post-infection.

siRNA Transfection
The control and StealthH SPARC siRNA duplexes (sequences #1 and #2) were designed by Invitrogen and described previously [25,32]. Human SLUG and p53 StealthH siRNAs were purchased from Invitrogen. Transfection of siRNA was carried out using Lipofectamine RNAiMAX (Invitrogen), at a final concentration of 50 nM. Sequences are shown in Table 2.

Fluorescence and Confocal Microscopy
Cells grown on glass coverslips were rinsed briefly in PBS, fixed in PBS containing 4% formaldehyde, and incubated in a solution containing 0.2% saponin and 1% BSA in PBS for 1 h at room temperature. Cells were incubated with primary antibodies (anti-E-cadherin 1:100, BD Biosciences; anti-SPARC 1:50, R&D Systems; anti-SLUG 1:100, Santa Cruz Biotechnology; anti-Fibronectin 1:100, BD Biosciences) for 1 h at room temperature. Cells were washed, incubated with Alexa Fluor-conjugated secondary antibodies (1:1000; Molecular probes), washed again, and mounted in Prolong antifade (Invitrogen). For actin staining, fixed cells were incubated with Texas Red-X phalloidin (1:100, Molecular Probes) for 1 h at room temperature. Images were captured on a Zeiss LSM 510 META laser scanning confocal microscope, with sequential fluorophore excitation.

Cell Migration and Three-dimensional (3D) Spheroid Invasion Assay
Chemotaxis assays were monitored using modified Boyden chambers as previously described [25]. Melanoma spheroids were generated using the liquid overlay technique and implanted into a collagen type 1 matrix in growth medium as described [34]. Tumor cell outgrowth was visualized by phase contrast microscopy.

E-cadherin Promoter Analysis
Transfection of 501mel cells was carried out by using the FuGENE 6 reagent (Roche Applied Sciences). Cells were transfected in 24-well plates with 200 ng of the wild-type (2178 wt) or mutant (mE-pal) luciferase reporter constructs and 50 ng of pCMV-b-galactosidase as a control for transfection efficiency. Cells were harvested 72 hours after transfection and assayed for luciferase activity using the Luciferase assay system (Promega). All experimental values were determined from triplicate wells. Each experiment was repeated at least twice [22,25].

Cell-cell Adhesion Assay
The cell aggregation assay was performed essentially as described in [35] with the minor following modifications. Cells were detached by a treatment with HyQTase (Hyclone), plated on bacterial Petri dishes and incubated at 37uC on a gyratory shaker

Fluorescent-based Cell Adhesion Assay
The adhesion assays were performed in black 96-well culture dishes in triplicate. Plates were coated overnight at 4uC with 10 mg/mL Fibronectin (BD Biosciences). Cells were collected by HyQTase treatment (Hyclone), labeled with 5 mM CellTracker Green CMFDA (Molecular probes), a cell permeant fluorogenic esterase substrate. After 30 min at 37uC cells were allowed to adhere in serum-free DMEM containing 0.1% fatty-acid-free BSA. At indicated time points, non-adherent cells were removed by washing with PBS supplemented with 1 mM CaCl 2 and 1 mM MgCl 2 and fluorescence was measured using a Fluoroskan microplate reader (excitation, 485 nm; emission, 530 nm). Data are expressed as the percentage of adherent cells per total cells.

Cell Proliferation, Cell Cycle and Apoptosis Analysis
Cell proliferation was determined by a colorimetric assay that is based on the cleavage of yellow tetrazolium salt (XTT) to form an orange formazan dye by mitochondrial dehydrogenases (Cell Proliferation Kit II; Roche Diagnostics). The absorbance of the formazan dye was measured at 490 nm. All experimental values were determined from quadruplicate wells.
Cell cycle profiles were determined by flow cytometric analysis of propidium iodide (PI)-stained cells as previously described [25]. Cells were analyzed using a FACScan (Becton-Dickinson) and the CellQuest software. For flow cytometric analysis of apoptotic death, cells were stained with the Annexin-V-fluos staining kit (Roche Applied Science) according to the manufacturers' protocol.

Statistical Analysis
Unless otherwise stated all experiments were repeated at least three times and representative data/images are shown. Error bars indicate 6 SD. Student's t and Spearman tests were performed to determine statistical significance. P,0.05 was considered statistically significant.

Tumor Cell-derived SPARC Controls SLUG during EMTlike Transition in Melanoma Cells and Melanocytes
We previously showed that SPARC induces E-cadherin repression and EMT-like processes in melanocytes and melanoma cells [22]. To further explore the mechanism of SPARC-mediated E-cadherin silencing, we examined the mRNA expression levels of known E-cadherin repressors that initiate EMT-like transition: SNAIL, SLUG, Twist1, Twist2, E47 [2]. We also examined the levels of Tbx2 and Tbx3, which have been associated with Ecadherin downregulation in melanoma cells [35]. We increased SPARC expression in E-cadherin positive 501mel cells by adenoviral delivery. Real-time Q-PCR analyses showed that overexpression of SPARC in 501mel led to an induction of SLUG, SNAIL, Tbx2 and Tbx3 transcripts ( Figure 1A). In contrast, Twist1 and E47 mRNA levels were unaffected by SPARC overexpression. Twist2 mRNA, however, was slightly decreased in cells transduced by AdSPARC compared to control cells. SNAIL mRNA induction following SPARC expression was consistent with our previous observations [22] that SPARC induced SNAIL in human primary melanocytes. At protein level, we confirmed upregulation of SLUG and SNAIL by adenoviraldelivered SPARC expression ( Figure 1B, left; Figure S1A), whereas levels of Tbx2 and Tbx3 were unaffected ( Figure S1A). Interestingly, SLUG protein induction followed the expression of SPARC-Myc transgene in the culture media, and was temporally correlated with decreased levels of E-cadherin and P-cadherin and increased expression of the mesenchymal marker Fibronectin ( Figure 1B, left). Induction of SLUG mRNA and protein levels as well as alterations of EMT-associated proteins upon SPARC overexpression were also observed in clones of 501mel stably transfected with SPARC ( Figure 1B, right; Figure S1C). These EMT-like morphological changes induced by SPARC were confirmed by immunofluorescent staining of SLUG, E-cadherin and Fibronectin. In addition, F-actin staining revealed more stress fibers in SPARC overexpressing cells than in control cells ( Figure 1C).
We next asked whether SLUG was regulated by SPARC in normal human melanocytes (NHM). Primary melanocytes were infected with an adenovirus control or expressing SPARC for 4 days and expression of SLUG was analyzed by immunoblotting ( Figure 1D). We found that melanocytes modified to produce and secrete SPARC showed a marked increase in SLUG expression compared with control cells. Upregulation of SLUG was associated with loss of E-cadherin and P-cadherin expression and with a significant increase in Fibronectin. The latter finding was confirmed by densitometric analyses of the immunoblots ( Figure S1B).
Collectively, these data indicate that SPARC-induced EMT-like transition in melanocytes and melanoma cells is accompanied by upregulation of SLUG, and suggest that SPARC functions by maintaining the augmented expression of SLUG in melanomas.
We next examined whether silencing of SPARC in melanoma cells could affect SLUG protein levels. As we recently showed that siRNA-mediated depletion of SPARC after a 5-day period promotes spontaneous apoptosis [36], we investigated the effect of SPARC depletion in a suitable time window, in which SPARC knockdown did not induce cell death ( Figure S1D). To diminish the likelihood of RNAi off-targets effects, we used two independent non-overlapping siRNAs to silence SPARC in 501mel cells. We found that reduction of SPARC in culture media was concomitant with a time-dependent decrease in SLUG and Fibronectin expression and increase in E-and P-cadherins protein levels (Figure 2A, left). A similar robust decrease of SLUG expression upon SPARC knockdown was observed in WM9 melanoma cells (Figure 2A, right). Immunofluorescence microscopy confirmed decreased levels of SLUG and Fibronectin and increased levels of E-cadherin in SPARC knockdown cells compared with control cells ( Figure 2B), suggesting that SPARC knockdown cells reverses the EMT-like phenotype. However, F-actin staining revealed no significant differences in cell morphology between control and siSPARC-transfected cells, but an increase in stress fibers formation in SPARC knockdown cells.

The Regulation of SLUG Protein Levels by SPARC is Independent of p53
We recently demonstrated that SPARC suppresses p53 functions in melanoma cells. Notably, we found that p53 is stabilized and activated upon RNAi-mediated knockdown of SPARC [36]. As p53 was shown to induce MDM2-mediated SLUG degradation in lung cancer cells [37], we asked whether downregulation of SLUG was mediated through p53 in SPARC knockdown cells. We used siRNA to target p53 in wild-type p53 501mel cells and melanoma cells containing mutant copies of p53. Treatment with p53 siRNA led to an abolition of p53 in 501mel cells and prevented SPARC siRNA-mediated accumulation of p53 ( Figure 3A). Blocking p53 activation in SPARC-depleted cells had no effect on the reduction of SLUG protein levels mediated by SPARC knockdown. Accordingly, in p53-mutated MeWo and SKmel28 cells, we still observed the downregulation of SLUG upon siRNA-mediated SPARC knockdown and the regained expression of E-and P-cadherin along with reduction of Fibronectin levels ( Figure 3B). These findings indicate that the effects of SPARC knockdown on suppression of SLUG and reversion of the EMT-like process did not require activation of the p53 pathway.

Role of PI3 Kinase and AKT in SPARC-induced SLUG Expression and Melanoma Cell Migration and Invasion
We sought to identify the signaling pathways mediating SPARC-induced SLUG expression. The recent observation that SPARC stimulates PI3 kinase/AKT-dependent pathway in A375 melanoma cells [36] prompted us to examine the role of this pathway in SPARC-driven SLUG expression and promotion of melanoma cell migration. We first examined the effect of pharmacological inhibition of AKT on SLUG expression in control and SPARC-overexpressing 501mel cells. As shown in Figure 4A, SPARC overexpression was associated with increased levels of AKT phosphorylated on Ser473. Addition of AKT inhibitor IV (AIIV) efficiently blocked phosphorylation of AKT and of its downstream target GSK3b on Ser9 and lowered levels of SLUG in both control and SPARC overexpressing cells. Consistently, pharmacologic blockade of PI3 kinase activity by LY294002 also reduced SLUG protein levels in 501mel SPARC cells. Like the majority of melanoma tumors, 501mel cells harbor the oncogenic BRAF V600E mutation and consequently constitutive activation of the MAP kinase/ERK signaling pathway. We thus examined the effect of a specific BRAF inhibitor (PLX4720) [38] on SLUG protein levels. Treatment with PLX4720 for 3 hours inhibited constitutive MAP kinase/ERK phosphorylation but had no major effect on SLUG protein levels ( Figure 4A, right pannel). Similar observations were made after prolonged exposure to PLX4720 (data not shown). This suggests that oncogenic BRAF signaling does not decrease SLUG expression in BRAF V600E melanoma cells. In migration assays, pharmacologic blockade of PI3 kinase and AKT by LY294002 and AIIV, respectively inhibited basal levels of cell migration and blocked SPARCenhanced migration ( Figure 4B).
To further assess the contribution of AKT, we tested whether a constitutively active form of AKT can rescue SLUG and the migratory phenotype of SPARC knockdown cells. 501mel cells were infected with an adenovirus control (AdCMV) or expressing Myr-AKT (AdAKT ca ) and then transfected with control or SPARC siRNA. SPARC knockdown significantly reduced phosphorylation of AKT and SLUG expression ( Figure 4C). Forced expression of constitutively active AKT into 501mel cells resulted in its constitutive phosphorylation and increased cell migration ( Figure 4C and D). Interestingly, adenoviral delivery of constitutively active AKT allowed rescue of SLUG expression and migratory capabilities of SPARC knockdown cells. Thus, SPARCmediated SLUG expression and promotion of cell migration requires AKT activation. These data also reveal a potential link between PI3 Kinase/AKT-regulated cell migration and SLUG expression in melanomas.
To assess the role of SLUG in the control of the cell-cell adhesion molecules E-and P-cadherins in melanoma cells, we carried out knockdown experiments using two differents nonoverlapping siRNAs specifically targeting SLUG, and gain-offunction experiments where 501mel cells were stably transfected with SLUG. As shown in Figure 5A, knockdown of SLUG led to regained expression of both E-and P-cadherins in a timedependent manner. A slight decrease in Fibronectin protein expression was also observed in SLUG silenced 501mel cells compared to control siRNA-treated cells. Increased expression of E-cadherin in SLUG knockdown cells was also confirmed by immunofluorescence analysis ( Figure 5B). In addition, SLUGdeficient cells adopted a rounded morphology with an altered actin cytoskeleton and a loss of the characteristic fibroblastic-like morphology, as revealed by fluorescent F-actin staining ( Figure 5B). In SLUG-overexpressing 501mel cell populations, we found reduced E-and P-cadherins protein levels and increased Fibronectin expression ( Figure 5C). We next tested if SLUG depletion would have a functional impact on Ca 2+ -dependent cellcell adherence. 501mel cells were transfected with control or SLUG siRNA and an in vitro cell-cell adhesion assay was performed in presence or absence of Ca 2+ ions (+ EDTA). The results shown in Figure 5D revealed an increase in cell aggregates from SLUG-depleted cells compared to control cells, and that formation of cell aggregates was prevented upon treatment with Figure 2. siRNA-mediated knockdown of SPARC decreases SLUG and restores E-cadherin and P-cadherin expression. (A) Expression levels of EMT markers in SPARC-depleted cells: immunoblots of 501mel cells transfected with control siRNA (siCTRL) for 4 days or two SPARC siRNAs for the indicated times (left), and of WM9 cells transfected with siCTRL or siSPARC #2 for 4 days (right). Total protein lysates and cell supernatants were analyzed for expression of the indicated proteins. (B) Morphology of SPARC-depleted cells: F-actin and immunofluorescent staining of 501mel cells transfected with control siRNA (siCTRL) or SPARC siRNA (siSPARC). After 4 days, cells were co-stained with Texas Red-X phalloidin, anti-SPARC (green) and anti-Fibronectin (cyan) and imaged with confocal microscopy. Merged images of the corresponding three channels are presented. Bottom panels, expression of SLUG (green) and E-cadherin (cyan) was analyzed by immunofluorescence staining. Bars, 10 mm. doi:10.1371/journal.pone.0040378.g002 the chelating agent EDTA. Thus, knockdown of SLUG increases Ca 2+ -dependent cell-cell adhesion. In the next series of experiments, we confirmed that SLUG is a transcriptional repressor of E-cadherin in 501mel cells [26,39]. Analysis of E-cadherin mRNA levels by real-time Q-PCR upon SLUG depletion or ectopic expression showed an increase or decrease of E-cadherin transcript, respectively ( Figure 5E). We also examined the effect of SLUG on the activity of the mouse E-cadherin promoter in 501mel cells ( Figure 5F). Luciferase reporter assays show that depletion of SPARC or SLUG by siRNA increased the activity of the E-cadherin promoter and conversely, expression of SPARC or SLUG repressed promoter activity. No repression was seen when the E-cadherin promoter is mutated in the E-box elements (mEpal construct). Similar responses to SLUG knockdown or SLUG ectopic expression were seen in other melanoma cell lines ( Figure  S2).   test). (E) SLUG regulates E-cadherin mRNA levels: RNAs were prepared from 501mel cells transfected with siCTRL or siSLUG for 4 days, and from bulk selected control or SLUG-overexpressing 501mel cells. The relative mRNA expression levels of SLUG and E-cadherin were measured by SYBR greenbased real-time Q-PCR. Columns, mean of three independent amplifications performed in duplicate; error bars, SD. *P,0.05 (Student's test). (F) Ecadherin promoter activity: 501mel cells were transfected with siCTRL, siSPARC or siSLUG, and 24 hours later with wild-type E-cadherin promoter reporter construct (left). 501mel cells were co-transfected with an empty vector (mock) or vectors expressing SPARC or SLUG and wild-type (2178 wt/luc) or mutant (mE-pal/luc) E-cadherin promoter reporter constructs (right). After 3 days, luciferase activities were measured and normalized to b-galactosidase activities. Columns, mean of triplicates; errors bars, SD. doi:10.1371/journal.pone.0040378.g005 Because siRNA SPARC-induced migratory inhibition coincides with SLUG reduction, we next analyzed the effect of SLUG knockdown on melanoma cell migration and invasive behavior. Knockdown of SPARC or SLUG with siRNAs in several melanoma cells strongly inhibited melanoma cell migration in Boyden chambers assays, compared with control cells ( Figure S3). These data confirm previous observations and indicate that SPARC and SLUG are both important regulators of melanoma cell migration.
Although SLUG is recognized as a pro-survival factor in some cell types and melanocytes, some studies showed that its depletion in melanoma cells was not associated with significant changes in proliferation and survival [15,20]. Similarly, we found that siRNAmediated depletion of SLUG did not alter cell proliferation and cell cycle profiles, and was not associated with significant apoptosis over a 4-day period ( Figure S4). Therefore, apoptosis does not account for the inhibitory effects on cell migration observed in SLUG knockdown cells.
Tumor migration was then analyzed in a more physiological context; 501mel, WM9 or WM793 cells were grown as spheroids embedded in collagen, an assay that mimics 3-dimensional growth and invasion by melanoma cells [40]. Knockdown of SPARC or SLUG had no effect on spheroid growth rate but significantly reduced cell invasion into collagen ( Figure 6A). Thus, SPARC and SLUG expression is required for tumor invasion in 3-dimensional cultures. Consistent with the regulation of SLUG by the PI3 kinase/AKT pathway (see Figure 4), pharmacologic inhibition of this signaling pathway by LY294002 or AIIV inhibited the invasive phenotype of melanoma spheroids into collagen ( Figure 6B).
It has remained unclear how SLUG affects cell migration. We observed that Slug-deficient cells showed altered actin cytoskeletal organization ( Figure 5B). To determine if the migration defect of SLUG-depleted melanoma cells was related to an alteration in cell adhesion, a fluorescent adhesion assay was performed with Fibronectin. Suppression of SLUG led to decreased attachment of 501mel, MeWo and SKmel28 cells on Fibronectin ( Figure S5). Thus, some migratory defects of SLUG knockdown cells may occur through alterations in cell-substratum interactions and cytoskeleton reorganization.

SLUG is Required for SPARC-mediated Melanoma Cell Migration
To show that SLUG is a downstream target of SPARC in melanoma cells, we knocked down SLUG in SPARC-overexpressing 501mel cells. As shown in Figure 7A, knockdown of SLUG abrogated the effect of SPARC-enhanced cell migration as well as basal migration of 501mel cells. Immunoblot analysis shows the efficient depletion of SLUG in control cells and cells stably transfected with SPARC ( Figure 7B). As a control, we also analyzed SNAIL expression and found its upregulation in SPARC-overexpressing cells, which is consistent with our early study showing that SPARC increased SNAIL expression in human primary melanocytes [22].
To provide further functional link between SPARC and SLUG in controlling EMT-induced cell migration, we next examined if ectopic expression of SLUG was able to protect from the migratory phenotype of SPARC knockdown cells. Control 501mel cells or stably transfected with SLUG were treated with SPARC siRNA and cell migration assays were performed 4 days later. Ectopic expression of SLUG was able to protect from the knockdown phenotype on migration ( Figure 7C) and reinduction of E-cadherin ( Figure 7D). Figure 7D also shows that SPARC level was dramatically reduced following siRNA treatment, and that endogenous SLUG level was reduced upon SPARC depletion. Thus, SLUG functionally contributes to SPARC-enhanced melanoma cell migration and E-cadherin repression.

Relationship between SPARC and SLUG Levels in Melanoma Cells at Various Stages of Tumor Development
Finally, we have examined the expression profiles of SPARC and SLUG in a small panel of cultures of melanoma cells derived from RGP or VGP primary tumors or from lymph node melanoma metastases (Figure 8). Real-time Q-PCR analysis revealed a positive association between SPARC and SLUG transcripts in freshly isolated melanoma samples tested. The positive expression of SPARC mRNA observed in the two VGP melanoma samples confirmed our previous observations that SPARC is induced during the RGP to VGP transition of melanomas [22]. The correlation (R) and P values between SLUG and SPARC levels were determined by regression analysis. These data provide independent support for a potential functional link between SPARC and SLUG during melanoma progression.

Discussion
SPARC is widely expressed in cancer and is thought to promote tumor progression through regulation of cell survival, invasiveness and tumor-host interactions [23,41]. Several studies from our laboratory and others have shown that inhibition of SPARC in both malignant cells in vitro [22,24] and tumors in vivo [24,25,42] reduces melanoma growth, invasiveness, and induces spontaneous apoptotic cell death and anti-tumor cytotoxic capacity. One mechanism underlying SPARC function in melanoma appears to be E-cadherin repression and promotion of an EMT-like transition [22,43], a process having a central role during RGP to VGP progression. In this study, we show that SPARC regulates the EMT regulatory factor SLUG and demonstrate the critical role of the PI3 kinase/AKT pathway in this process.
Much evidence exists to support an important role for the protein kinase B/AKT in melanomagenesis. Activated AKT was found in 70% of human melanomas [44]. Several mechanisms may contribute to this event including PTEN deletion, AKT3 amplification and aberrant growth factors signaling. Recently, we showed that autocrine production of SPARC by melanoma cells participates to elevated levels of AKT activation and increased tumor cell survival [36]. Besides the involvement of AKT in survival, growth and metabolic-related pathways, AKT affects cell motility and invasion, prerequisite processes for metastasis [45,46]. Interestingly, AKT was shown to downregulate E-cadherin expression and to promote EMT-like transition and invasiveness in carcinoma cells by inducing SNAIL [47]. Our study showing that AKT also contributes to SLUG regulation downstream SPARC-induced EMT-like changes in melanomas emphasizes the crucial role of AKT in controlling SNAIL family factors and tumor-associated EMT processes. However, the mechanisms by which AKT activates SLUG expression and transcriptional activity remain unclear and need further investigations.
We recently described an alternate role for SPARC in melanoma progression, namely its suppression of p53-dependent responses. Inactivation of p53 by SPARC resulted from AKTdependent activation of MDM2 and we demonstrated that p53 is activated upon knockdown of SPARC in melanoma cells [36]. Recently, p53 was shown to promote SLUG degradation in lung carcinoma cells [37]. We therefore examined the regulation of SLUG by p53 in SPARC knockdown cells and found that SLUG regulation by SPARC is independent of both expression and mutational status of p53. Accordingly, we observed that reversion of the EMT-like process, including E-and P-cadherin derepression, did not require activation of p53. Consistently, we also reported that decreased motility and invasion of SPARC deficient cells is independent of p53 genetic status [25]. These observations support the notion that SPARC controls two AKT-dependent pathways involved in (i) melanoma cell survival through p53; and (ii) EMT-associated cell migration through SLUG. Interestingly, our analysis of actin cytoskeleton revealed that suppression or overexpression of SPARC results in a similar phenotype i.e. an increase of stress fibers formation. The molecular mechanisms that SLUG is a potent inducer of cell movement during development [48], epithelial cell migration [14] and tumor cell motility [1]. Here we found that SLUG knockdown cytoskeletal organization, abrogates melanoma cell migration and tumor invasiveness, further supporting the role of SLUG in tumor motility. However, it remains still unclear how SLUG controls these processes. It has been reported that SLUG controls integrin expression in human epidermal keratinocytes [49]. Integrindependent cell-matrix substratum interaction is essential for directing cell movement. Interestingly, our findings show that several integrin-dependent events are altered in SLUG knockdown cells, such as adhesion on Fibronectin, tumor invasiveness in 3D collagen matrix, cell morphology and actin cytoskeleton, suggesting that abnormal cell migration observed in SLUG silenced cells result from impaired integrin signaling and adhesion. Whether this results from a direct transcriptional repression of integrin expression or indirect changes in integrin inside-out signaling is currently under investigation in our laboratory.
SLUG knockdown increases Ca 2+ -dependent cell-cell adhesion and, in addition to E-cadherin, our data demonstrate that SLUG regulates P-cadherin expression in melanoma cells. P-cadherin was described as a regulator of cell-cell adhesion and melanoma invasiveness [50]. Down-regulation of P-cadherin is a frequent event observed during melanoma progression [7]. However, the regulatory mechanism of P-cadherin suppression is unknown. To our knowledge, our findings that SLUG controls P-cadherin levels in melanomas cells are the first description of such a mechanism. In addition, we provide here evidence that SPARC constitutes an important upstream regulator of P-cadherin in melanocytes and melanoma cells. However, it remains to be shown whether Pcadherin is a transcriptional target of SLUG in EMT-associated events.
In melanoma, SLUG functions as a melanocyte-specific factor required for the promotion of the metastatic phenotype [15]. The signaling pathways controlling SLUG expression in melanoma cells are beginning to be unraveled. Recently, SLUG was reported to be regulated by HGF-dependent signaling pathways in human melanoma cells [51], and by TGF-beta and activin A in B16 murine melanoma cells [52]. Here, we show that SLUG is controlled by the PI3 kinase/AKT signaling pathway downstream of the matricellular SPARC protein, connecting for the first time SLUG levels to extracellular environment cues critical for the acquisition of the invasive phenotype.
The transcriptional factors of the SNAIL family SNAIL and SLUG were recently shown to collaborate on local tumor invasion and promotion of distant metastasis [53]. In addition, a sequential, hierarchical action of SNAIL factors during tumor progression has been proposed. However, it was reported that SNAIL and SLUG play distinct roles in breast carcinoma progression [54]. Thus, it would be of great importance to investigate whether SNAIL and SLUG cooperate or act independently in the acquisition of invasiveness downstream of SPARC. We previously observed an inversed correlation between SPARC and E-cadherin levels in melanocytes and melanoma cells [22]. Interestingly, in some melanomas SNAIL and SPARC levels were not correlated, suggesting that mechanisms other than SNAIL induction may be involved in SPARC-mediated E-cadherin repression. Although the melanoma sample size analyzed here is relatively small, the results indicate that SPARC and SLUG expression is positively associated during melanoma progression, further underlining the importance of our present observations. Additional analyses on a larger clinical collection of the expression of SPARC, SNAIL, SLUG and Ecadherin are clearly warranted.
In conclusion, we report here the first demonstration of a regulation of the lineage-specific developmental factor SLUG by SPARC through an AKT-dependent pathway and the importance of this mechanism in EMT-induced cell invasion in melanoma.  501mel cells overexpressing SPARC (cl. SPARC #1 and #2) or control vector (cl. CTRL). Slug mRNA expression was measured by SYBR green-based real-time Q-PCR. Relative expression level of SLUG mRNA was normalized for RNA concentrations with four different housekeeping genes. Columns, mean of two independent amplifications performed in duplicate; error bars, SD. *P,0.05 (Student's test). (D) Viability of melanoma cells after 4 days of SPARC knockdown by siRNA. The indicated melanoma cells were transfected with control siRNA (siCTRL, open bars) or SPARC siRNA (siSPARC, filled bars) at 50 nM. 4 days after transfection, cell proliferation was measured by XTT assay. Results are expressed in percent of control. Columns, mean of 4 independent determinations; error bars, SD. NS, not significant (Student's test). (TIF) Figure S2 siRNA-mediated SLUG knockdown restores E-cadherin expression through transcriptional derepression of the promoter. (A) Analysis of E-and Pcadherin protein levels in SLUG-depleted cells. The indicated melanoma cells were transfected with control siRNA (siCTRL) for 4 days or SLUG siRNA (siSLUG) for the indicated times. Expression levels of SLUG, E-cadherin and P-cadherin were analyzed by immunoblotting. HSP60 was used as loading control. (B) E-cadherin mRNA levels following SLUG or SPARC depletion. RNAs were prepared from WM9 cells transfected with siCTRL, siSPARC or siSLUG for 3 days. The relative mRNA expression levels of SPARC, SLUG and Ecadherin were measured by SYBR green-based real-time Q-PCR. *P,0.05 (Student's test). (C) E-cadherin promoter activity following SLUG or SPARC depletion. Cells were transfected with siCTRL, siSPARC or siSLUG, and 24 hours later with wildtype E-cadherin promoter reporter construct. After 3 days, luciferase activities were measured and normalized to b-galactosidase activities. Columns, mean of triplicates; errors bars, SD. (D) Ecadherin promoter activity following SLUG or SPARC expression. Cells were co-transfected with an empty vector (mock) or vectors expressing SPARC or SLUG, and wild-type Ecadherin promoter reporter construct as indicated. Measurement of luciferase activities was assessed as described above. (TIF) Figure S3 siRNA-mediated SPARC or SLUG knockdown impairs melanoma cell migration. 501mel (A), WM9 (B), MeWo (C) and SKmel28 (D) cells were transfected with control siRNA (siCTRL), two SPARC siRNAs (siSPARC #1 and #2) or two SLUG siRNAs (siSLUG #1 and #2) for 4 days. Serumstimulated cell migration was assessed using Boyden chamber assays. Cells were left to migrate for 20 hours, then fixed, stained and counted. Results are expressed in percent of control. Columns, means of triplicates from two independent experiments; error bars, SD. Representative images of lower surface of membranes are shown. (TIF) Figure S4 siRNA-mediated SLUG knockdown shows no effect on melanoma cell proliferation, cell cycle pro-gression or apoptosis. (A) Analysis of cell proliferation. The indicated melanoma cells were transfected with control siRNA (siCTRL, open bars) or SLUG siRNA (siSLUG, filled bars) at 50 nM. 4 days after transfection, cell proliferation was measured by XTT assay (B) Analysis of cell cycle distribution. The indicated melanoma cells were transfected with siCTRL or siSLUG, stained with PI and analyzed for DNA content by flow cytometry. Histograms represent the percentage of cells in different phases of the cell cycle. (C) Analysis of PARP cleavage. 501mel cells were transfected with control siRNA (siCTRL) or SLUG siRNA (siSLUG) for 4 days. Expression levels of SLUG and cleaved PARP were analyzed by immunoblotting. Treatment with Staurosporine for 10 hours was used as a positive control of apoptotic cell death. HSP60 was used as loading control (D) Analysis of cell apoptosis. WM9 and SKmel 28 cells were transfected with control siRNA (siCTRL) or SLUG siRNA (siSLUG) for 4 days or treated with Staurosporine as above.