ETS-1-mediated Transcriptional Up-regulation of CD44 Is Required for Sphingosine-1-phosphate Receptor Subtype 3-stimulated Chemotaxis*

Background: S1P3-mediated chemotaxis plays a pivotal role in various physiological and pathophysiological activities. Results: S1P/S1P3 signaling activates ROCK/JNK/ETS-1/CD44 pathway, and inhibition of this pathway abrogates S1P3-stimulated chemotaxis. Conclusion: ETS-1/CD44 signaling mediates S1P/S1P3-regulated chemotaxis. Significance: Therapeutically manipulating S1P3-mediated chemotaxis requires a molecular understanding of its regulated signaling pathway. Sphingosine-1-phosphate (S1P)-regulated chemotaxis plays critical roles in various physiological and pathophysiological conditions. S1P-regulated chemotaxis is mediated by the S1P family of G-protein-coupled receptors. However, molecular details of the S1P-regulated chemotaxis are incompletely understood. Cultured human lung adenocarcinoma cell lines abundantly express S1P receptor subtype 3 (S1P3), thus providing a tractable in vitro system to characterize molecular mechanism(s) underlying the S1P3 receptor-regulated chemotactic response. S1P treatment enhances CD44 expression and induces membrane localization of CD44 polypeptides via the S1P3/Rho kinase (ROCK) signaling pathway. Knockdown of CD44 completely diminishes the S1P-stimulated chemotaxis. Promoter analysis suggests that the CD44 promoter contains binding sites of the ETS-1 (v-ets erythroblastosis virus E26 oncogene homolog 1) transcriptional factor. ChIP assay confirms that S1P treatment stimulates the binding of ETS-1 to the CD44 promoter region. Moreover, S1P induces the expression and nuclear translocation of ETS-1. Knockdown of S1P3 or inhibition of ROCK abrogates the S1P-induced ETS-1 expression. Furthermore, knockdown of ETS-1 inhibits the S1P-induced CD44 expression and cell migration. In addition, we showed that S1P3/ROCK signaling up-regulates ETS-1 via the activity of JNK. Collectively, we characterized a novel signaling axis, i.e., ROCK-JNK-ETS-1-CD44 pathway, which plays an essential role in the S1P3-regulated chemotactic response.

Sphingosine-1-phosphate (S1P), 4 a critical serum-borne lipid mediator, regulates a wide array of biological processes, such as cell proliferation and survival (1)(2)(3), immune cell trafficking (4), suppression of apoptosis (5,6), and chemotaxis (7)(8)(9). S1P can function either as an extracellular ligand or as an intracellular mediator (10 -12). When functioning as an extracellular ligand, S1P-regulated biological activities are mediated by the S1P family of G-protein-coupled receptors (10,13,14). Five members of the S1P family G-protein-coupled receptors (S1P 1-5 receptors) have been identified. We and others have shown that S1P 1 and S1P 3 receptor-mediated signaling pathways play critical roles in endothelial cell chemotaxis, adherens junction assembly, endothelial morphogenesis, and angiogenic responses (10,15,16). The balance between S1P 1 and S1P 2 signaling is important in the regulation of endothelial integrity (17,18) and vascular inflammation (19). In addition, S1P can also function as an intracellular lipid mediator to regulate Ca 2ϩ mobilization and suppress apoptosis (20 -22). It was recently demonstrated that intracellular S1P interacts with histone deacetylases, HDAC1 and HDAC2, and modulates enzymatic activity of HDACs, as well as gene expression regulated by HDACs (11). Moreover, S1P was shown to bind to the N-terminal RING domain of TNF receptor-associated factor 2, leading to the activation of the E3 ligase activity of TNF receptorassociated factor 2 (12). S1P is an important regulator of cell chemotactic response (16,24,25). S1P-regulated chemotactic response has been shown to play critical roles in various physiological and pathophysiological conditions, including angiogenesis (15,16), vascular maturation (26), atherosclerosis (19), lymphocyte egress from lymphoid organs (27,28), multiple sclerosis (29), and the invasion and metastasis of tumor cells (30,31). Chemotactic response regulated by S1P is mediated by the cell membrane S1P family of G-protein-coupled receptors. It was shown that S1P treatment can either stimulate or inhibit cell migration (16,25). The stimulatory and inhibitory effects of S1P are mediated by the S1P 1 /S1P 3 and S1P 2 receptor subtypes, respectively (16,24,25,32). S1P 1 -stimulated chemotaxis requires the activation of the Rho family small GTPases (16,32). In contrast, S1P 2 signaling was shown to suppress the activity of Rac GTPase, leading to the inhibition of cell migration (25). However, molecular details of the S1P 3 -mediated chemotactic response are poorly understood and remain to be elucidated. S1P/S1P 3 -mediated chemotaxis was shown to play important roles in the invasion/migration phenotype of MCF10A human breast epithelial cells (33), prostate migration of carcinoma cells (30), B-cell development, egress and positioning within the bone marrow (34), homing of bone marrow-derived cells (35), dissemination of inflamed dendritic cells (36), shuttle of B cells from splenic follicular areas to marginal zone (37), and high density lipoprotein-stimulated endothelial cell migration (38), among others. We recently showed that S1P 3 receptors are abundantly expressed in a panel of human lung adenocarcinoma cell lines, and S1P 3 signaling plays an essential role in stimulating the migration/invasion of these cell lines (39). In the present study, we investigated the molecular mechanism underlying the S1P/S1P 3 -mediated cell chemotaxis by utilizing these lung adenocarcinoma cell systems. We observed that S1P/S1P 3 signaling markedly increased CD44 expression and membrane localization of CD44 polypeptides. Knockdown of CD44 abrogated the S1P/S1P 3 -mediated chemotaxis. Moreover, we demonstrated that the S1P/S1P 3 -induced CD44 expression depends on ETS-1 (v-ets erythroblastosis virus E26 oncogene homolog 1) activity. Collectively, our data elucidate for the first time that the novel ETS-1/CD44 signaling pathway plays a critical role in S1P 3 -stimulated chemotactic response.

EXPERIMENTAL PROCEDURES
Reagents-Sphingosine-1-phosphate (Biomol) was dissolved in methanol, aliquoted, vacuum-dried, and stored at Ϫ20°C. When needed, an aliquot was resuspended in 4% fatty acid-free BSA (Sigma) by sonication to make a stock solution of 200 M. RPMI 1640, keratinocyte serum-free medium, trypsin, FBS, goat anti-mouse IgG, and goat anti-rabbit IgG were obtained from Invitrogen. CD44, c-Jun, and phospho-JNK antibodies were purchased from Cell Signaling. ETS-1 antibody was obtained from Santa Cruz Biotechnology. RNeasy Mini-Kit, si-ROCK1, and nontargeting siRNA control were purchased from Qiagen. si-JNK1 was from Ambion. ROCK inhibitor Y-27632 and PI3K inhibitor LY 294002 were purchased from EMD Chemicals. NFB inhibitor BAY 11-7085 was obtained from Biomol. Unless otherwise specified, all chemicals and reagents were purchased from Sigma.
HBEC2-KT and HBEC3-KT cells were cultured in keratinocyte serum-free medium. Cells were serum-starved overnight followed by the treatment of S1P or vehicle for various times. Then the cells were collected for RNA or protein extraction or subjected to functional analysis as described below.
RNA Isolation, RT-PCR, and Real Time PCR-Total RNAs were isolated from cells using an RNeasy mini-kit (Qiagen) according to the manufacturer's instructions. RNA quality and concentration were assessed with a NanoDrop ND-1000 spectrophotometer.
Total RNAs were reverse transcribed with an oligo(dT) primer (Promega) by Moloney murine leukemia virus reverse transcriptase (Promega) for the first strand cDNA synthesis. For real time PCR quantitation, 50 ng of reverse transcribed cDNAs were amplified with the ABI 7500 system (Applied Biosystems) in the presence of TaqMan DNA polymerase. The sense and antisense primers of CD44, ETS-1, ROCK1, S1P receptors, and GAPDH were purchased from Applied Biosystems. Real time PCRs were performed by using a universal PCR Master Mix (Applied Biosystems) according to the manufacturer's instructions. Relative quantification (RQ) was calculated using the Applied Biosystems SDS software based on the equation RQ ϭ 2 Ϫ⌬⌬Ct , where Ct is the threshold cycle to detect fluorescence. Ct data were normalized to the internal standard GAPDH.
shRNA-mediated Gene Knockdown-Stable knockdown of S1P 3 receptor in cultured cells was performed essentially as we described (39). For knocking down CD44 and ETS-1, cells were plated in 6-well plates (2 ϫ 10 5 cells/well) and cultured at 37°C for 20 h in a humidified atmosphere of 5% CO 2 . Cells were transfected with human GIPZ lentiviral shRNAmir vector, RHS4430-99158569 and RHS4430-100995224 (Open Biosystems) specific to silence CD44 and ETS-1, respectively. Transfection with nontargeting GIPZ lentiviral shRNAmir RHS4346 vector was used as a control. Transfection was performed by using Lipofectamine 2000 reagent (Invitrogen). Seventy-two hours later, stably transfected cells were selected with puromycin (1 g/ml). The efficacy and specificity of CD44 and ETS-1 knockdown were assessed by both real time PCR and Western blot analysis.
Chemotaxis Analysis-Cell chemotaxis was measured by using the Neuro Probe A series 96-well chamber with standard framed filters (8-m pore size) (Neuro Probe), as previously described (39,40). The cells were grown to confluence, washed three times with PBS, and serum-starved in plain medium supplemented with 0.01% FBS for 16 h. The cells were collected by brief trypsinization, washed, and resuspended in plain RPMI 1640 medium (2 ϫ 10 5 cells/ml). Standard framed filters were precoated with fibronectin (5 g/ml) (39,40) at 37°C for 1 h and then air-dried. Cell suspensions (400 l) were plated in upper chambers, and chemoattractants were added to lower chambers. Cells were allowed to migrate for 8 h at 37°C. Subsequently, the cells that remained on the upper surface of filters were removed by gently wiping with a cotton swab. The cells that migrated to the lower surface were fixed with 4% paraformaldehyde and stained with 0.5% crystal violet for 30 min. After washing, crystal violet dye was eluted with 10% acetic acid, and absorbance was measured at 595 nm.
ChIP Assay-ChIP assays were performed using a Pierce chromatin prep module and Pierce agarose ChIP kit following the manufacturer's instructions (Thermo Scientific). Briefly, cells were cross-linked with 1% formaldehyde and collected into lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8.0, 1ϫ protease inhibitor mixture). Cell lysates were digested with micrococcal nuclease, followed by immunoprecipitating with rabbit ETS-1 or c-Jun antibody. Immunoprecipitation with a normal rabbit IgG (Thermo Scientific) was used as a negative control. After incubation with the protein A/G Plus agarose resin, immunoprecipitates were washed and then heated at 65°C for 1.5 h to reverse the formaldehyde crosslinking. DNA fragments were purified with the DNA clean-up column and reagents included in the Pierce agarose ChIP kit.
Immunostaining Analysis-Cells cultured in glass-bottomed Petri dishes (MatTek) were treated with or without S1P (200 nM), followed by fixation with 4% paraformaldehyde for 30 min at room temperature. Cultures were permeabilized with 0.05% Triton X-100 and blocked with 1% bovine serum albumin for 30 min. Subsequently, cells were incubated with the indicated primary antibody (1:100) followed by the FITC-conjugated secondary antibody (1:500). Nuclei were stained with DAPI (Sigma-Aldrich). Fluorescent images were analyzed by the Leica TCS SP5 confocal system (Leica, Wetzlar, Germany).
microRNA Quantification-Levels of miR-34a were quantified using TaqMan microRNA assays (Applied Biosystems) (41,42). Total RNAs were isolated from cells, and small RNA fractions (Ͻ200 nucleotides) were recovered using the mirVANA PARIS miRNA isolation kit (Ambion). RQ was calculated using the Applied Biosystems SDS software based on the equation RQ ϭ 2 Ϫ⌬⌬Ct , where Ct is the threshold cycle to detect fluorescence. Ct data were normalized to the internal standard, miR-103.
Statistical Analysis-The results are presented as means Ϯ S.D. The difference between various treatments was analyzed by Student's t test with p values Ͻ 0.05 considered significant. A conventional two-way analysis of variance (GraphPad Prism 5) was performed for migration data to compare the migratory capability before and after the silence of ETS-1 and CD44.

S1P Stimulates CD44 Expression and Induces Membrane
Localization of CD44 Proteins-Many of CD44-mediated biological responses, such as cell-cell and cell-matrix interactions (43) as well as cell migration (44,45), are regulated by S1P signaling pathways. Thus, we examined whether S1P stimulation activates CD44 and whether CD44 mediates S1P-regulated responses. H1793 cells, a human lung adenocarcinoma cell line abundantly expressing S1P 3 receptor subtype of the S1P family of G-protein-coupled receptors (39), were treated with or without S1P (200 nM) for various times. The expression of CD44 at the mRNA and protein levels was measured. As shown in Fig.  1A, S1P treatment increased levels of CD44 mRNA in a timedependent manner. The CD44 mRNA expression increased 1.5-fold at 30 min after S1P treatment and 3.3-fold at 24 h after S1P treatment. Levels of CD44 proteins were also profoundly increased after S1P treatment (Fig. 1B). Immunostaining analysis showed that CD44 proteins significantly increased and were located at plasma membrane regions following S1P treatment (Fig. 1C). In addition, we examined the S1P-induced CD44 expression in several other human lung-derived cell lines, including H1792, H1650, and H23 lung adenocarcinoma cells as well as HBEC2-KT immortalized normal bronchial epithelial cells (46,47). We found that S1P stimulated CD44 expression in H1792, H1650, and H23 lung adenocarcinoma cells. In contrast, S1P was unable to enhance CD44 expression in HBEC2-KT normal lung epithelial cells (Fig. 1D). S1P-stimulated CD44 Expression Is Mediated by the S1P 3 Receptor/ROCK Signaling Pathway-Previously, we showed that S1P 3 receptors are abundantly expressed in H1793, H1792, H23, and H1650 human lung adenocarcinoma cell lines, whereas S1P 3 receptors are barely detected in HBEC2-KT normal lung epithelial cells (39). S1P was able to induce CD44 expression in lung adenocarcinoma cells and not in normal lung epithelial cells (Fig. 1), suggesting that S1P-stimulated CD44 expression is mediated by S1P 3 receptors. Therefore, we employed the shRNA-mediated gene silencing technique to specifically knockdown S1P 3 receptors in H1793 cells. As shown in Fig. 2A, knockdown of S1P 3 completely abolished CD44 induction following S1P stimulation. In contrast, cells stably transfected with pRS control vector (sh-Ctrl) had no effect on S1P-stimulated CD44 expression. The sh-S1P 3 -RNA is highly specific, because it only knocked down S1P 3 receptors and had no effect on the other S1P receptor subtypes present in H1793 cells (Fig. 2B). Moreover, immunostaining analysis showed that S1P induced a significant increase of CD44 proteins in the membrane regions of H1793 cells stably transfected with sh-Ctrl vector, whereas the S1P-induced CD44 up-regulation was completely inhibited in the S1P 3 knockdown H1793 cells (Fig. 2C). These results indicate that the S1P 3 -transduced signaling pathway plays an essential role in the S1P-stimulated CD44 expression.
We next utilized pharmacological inhibitors to characterize signaling pathways involved in the S1P 3 -mediated CD44 upregulation. S1P was shown to activate signaling molecules such as PI3K (48,49) and NFB (12,50), among others. However, treatment with LY294002 (inhibitor of PI3K) or BAY 11-7085 (inhibitor of NFB) did not significantly inhibit the S1P-in-duced CD44 up-regulation (Fig. 2D). In contrast, treatment of H1793 cells with Y-27632, a specific inhibitor of ROCK, completely abrogated the S1P-increased CD44 expression (p Ͻ 0.01, t test) (Fig. 2D). Moreover, we used siRNA to specifically knockdown ROCK1 in H1793 cells (Fig. 2E). The S1P-stimulated CD44 expression was significantly diminished in si-ROCK1 transfected cells (Fig. 2F). In agreement, we and others have shown that S1P 3 signaling leads to the activation of ROCK, which plays a critical role in S1P 3 -regulated biological activities (39,51). In addition, immunostaining analysis showed that pretreatment of H1793 cells with a ROCK inhibitor completely abolished the S1P-increased CD44 proteins in membrane regions (Fig. 2G). Collectively, these data suggest that S1P-induced CD44 expression is mediated by the S1P 3 /ROCK signaling pathway.
CD44 Proteins Play a Critical Role in S1P/S1P 3 -stimulated Cell Migration-We have shown that levels of S1P 3 receptors are significantly increased in a panel of lung adenocarcinoma cell lines, compared with that in immortalized normal lung epithelial cells (39). We have also shown that S1P-stimulated cell migration and invasion is mediated by S1P 3 receptors in these lung adenocarcinoma cells (39). Recently, it was elegantly demonstrated that CD44 proteins play a key role in tumor invasion and metastasis (52). Therefore, we investigated whether S1Penhanced CD44 expression plays an essential role in S1P/S1P 3 signaling-stimulated cell migration. We employed an shRNAmediated gene knockdown technique to specifically diminish CD44 expression. As shown in Fig. 3A, 78% of CD44 mRNA was successfully knocked down in H1793 cells stably transfected with sh-CD44, compared with that in cells stably transfected with sh-Ctrl vector. Next, H1793 cells stably transfected with sh-CD44 or sh-Ctrl vector were treated with or without S1P (200 nM) for 4 h. Levels of CD44 proteins were measured by Western blotting analysis. As shown in Fig. 3B, S1P treatment substantially increased levels of CD44 proteins in the cells transfected with the sh-Ctrl vector. In contrast, S1P stimulation failed to increase CD44 proteins in sh-CD44 transfected cells. Transfection of sh-CD44 had no effect on endogenous ␤-actin proteins (Fig. 3B), suggesting that sh-CD44 is specific in knocking down CD44 molecules.
Subsequently, we examined the role of CD44 in S1P/S1P 3 signaling-stimulated cell migration. As shown in Fig. 3C, S1P treatment dose-dependently stimulated migration of sh-Ctrl vector transfected cells. In sharp contrast, S1P-induced chemotaxis was completely inhibited in CD44 knocked down H1793 cells (p Ͻ 0.01, sh-Ctrl versus sh-CD44, analysis of variance) (Fig. 3C). Also, we showed that S1P 3 -mediated CD44 upregulation requires ROCK1 activity (Fig. 2F). In agreement, knockdown of ROCK1 significantly diminished S1P-stimulated cell migration (Fig. 3D). These results suggest that the S1P/ S1P 3 /ROCK1 signaling axis induces the expression of CD44, which plays an essential role in S1P/S1P 3 -mediated cell chemotactic response. S1P Treatment Stimulates ETS-1 Binding to CD44 Promoter-It was recently shown that miR-34a inhibits metastasis of prostate cancer cells by directly suppressing CD44 expression (52). Thus, we examined whether S1P/S1P 3 signaling diminishes miR-34a expression, consequently leading to the up-regulation A and B, levels of CD44 mRNAs (A) and proteins (B) were measured by qPCR and Western blot analysis, respectively. Note that S1P treatment significantly enhances CD44 expression. Lower panel in B, levels of CD44 proteins were quantitated by the National Institutes of Health ImageJ software. C, H1793 cells were treated with or without S1P (200 nM) for 4 h. CD44 proteins (green) were detected by immunofluorescence staining. Nuclei were stained with DAPI (lower panels). Note that S1P treatment markedly increases membrane localization of CD44 proteins. D, HBEC2-KT immortalized normal lung epithelial (2KT) and H1792, H23, and H1650 lung adenocarcinoma cells were treated with or without S1P (200 nM) for 4 h, followed by qPCR quantitation of CD44 expression. Note that CD44 expression was significantly elevated in lung adenocarcinoma cells after S1P stimulation, whereas S1P was unable to stimulate CD44 expression in HBEC2-KT cells. A and D, data represent means Ϯ S.D. from three individual experiments performed in triplicate. *, p Ͻ 0.05; **, p Ͻ 0.01 (S1P versus control vehicle treatment), t test. Ctrl, control; IB, immunoblotting.  . S1P-induced CD44 expression is mediated by S1P 3 /ROCK signaling pathway. A, S1P-increased CD44 expression was measured in H1793 cells transfected with sh-S1P 3 or sh-Ctrl (control) vector. B, expression of S1P receptor subtypes was measured in H1793 cells stably transfected with sh-S1P 3 or sh-Ctrl vector by real time PCR. C, immunofluorescence staining of CD44 (green, panels a, c, e, and g) in sh-S1P 3 (panels e-h) or sh-Ctrl (panels a-d) transfected H1793 cells, treated with or without S1P for 4 h. Cell nuclei were stained with DAPI (panels b, d, f, and h). D, H1793 were pretreated for 1 h with or without indicated pharmacological inhibitors, followed by stimulating in the presence or absence of S1P for 4 h. The expression of CD44 was quantitated by qPCR. LY, LY294002 for PI3K; BAY, Bay 11-7085 for NFB; Y, Y-27632 for ROCK. Note that pretreatment with ROCK inhibitor completely diminished the S1P-induced CD44 expression. E, H1793 were transfected with nontargeting siRNA control or si-ROCK1 (Qiagen). Levels of ROCK1 were quantitated by qPCR. F, si-Ctrl or si-ROCK1 transfected H1793 cells were treated with or without S1P for 4 h. Levels of CD44 were measured by qPCR. G, H1793 cells were pretreated with control vehicle (panels a-d) or Y-27632 (Y) (panels e-h) for 1 h, followed by stimulating with or without S1P for 4 h. CD44 proteins were analyzed by immunofluorescence staining (panels a, c, e, and g), and cell nuclei were stained with DAPI (panels b, d, f, and h). Note that the S1P-increased CD44 at membrane regions was inhibited by pretreatment of ROCK inhibitor. The data represent means Ϯ S.D. from three (A, B, and D) or two (E and F) individual experiments performed in triplicate. *, p Ͻ 0.01; **, p Ͻ 0.05; t test. Ctrl, control. of CD44 proteins. H1793 cells were treated with or without S1P for various times. Levels of miR-34a were quantified by using a TaqMan microRNA assay kit (Applied Biosystems) (41,42). We observed that miR-34a levels were significantly increased (ϳ2-3-fold; p Ͻ 0.05, t test, n ϭ 3), in a time-dependent manner, after S1P stimulation (data not shown). This result suggests that the S1P/S1P 3 -increased CD44 expression is not directly regulated by miR-34a-mediated suppression mechanism.
Next, we employed the cis-element Cluster Finder software (53) to analyze the CD44 promoter region for potential binding sites of transcriptional factors. Four potential binding sites of ETS-1 (v-ets erythroblastosis virus E26 oncogene homolog 1) transcriptional factor were revealed by this promoter analysis. ETS-1 was shown to be involved in VEGF-stimulated endothelial chemotaxis (54), however, the underlying mechanism is completely unknown. Therefore, we investigated whether ETS-1 activity is required for S1P/S1P 3 -stimulated CD44 upregulation and chemotaxis.
ETS-1 binding site 1 is located at nucleotides Ϫ1384 to Ϫ1374, site 2 at nucleotides Ϫ1359 to Ϫ1349, site 3 at nucleotides Ϫ1148 to Ϫ1138, and site 4 at nucleotides Ϫ1107 to Ϫ1097 in the 5Ј-up-stream region of the CD44 translational initiation site (designated as nucleotide ϩ1) (Fig. 4A). Sites 1 and 2 are separated by only 15 nucleotides, and sites 3 and 4 are separated by only 31 nucleotides. Therefore, three pairs of primers were designed to assess the binding of ETS-1 to these candidate sites by chromatin immunoprecipitation analysis. P1 amplified sites 1 and 2 (amplicon size 220 bp), P2 amplified sites 3 and 4 (amplicon size 230 bp), and P3 amplified sites 1-4 (amplicon size 429 bp). Initially, we examined the binding capa-bilities of ETS-1 to sites 1 and 2 and/or sites 3 and 4 by using primer pair P1 or P2, respectively, following chromatin immunoprecipitation with anti-ETS-1. PCR amplification with P2 primer pair showed an amplicon of 230 bp, whereas no PCR amplicon was observed by using P1 primer pair (Fig. 4B, upper  panel). Anti-ETS-1 ChIP assay is specific, because there was no amplicon when the ChIP assay was performed using normal IgG as a control (Fig. 4B, lower panel). These results suggest that sites 3 and 4 may be the bona fide binding sites of ETS-1.
Next, we examined whether S1P treatment stimulates the binding of ETS-1 to sites 3 and 4 in the CD44 promoter region. H1793 cells were treated with S1P (200 nM) for various times. The binding of ETS-1 to CD44 promoter region was assessed by ChIP analysis. By utilizing the P3 primer pair, we found a detectable increase of ETS-1 binding at 1 h after S1P stimulation (Fig. 4C). Also, the S1P-enhanced ETS-1 binding was increased in a time-dependent manner. The same kinetics of S1P-increased ETS-1 binding was observed by using the P2 primer pair, which specifically amplifies sites 3 and 4 (Fig. 4D). These data together suggest that S1P treatment stimulates the binding of ETS-1 to sites 3 and 4, which may contribute to the S1P-increased CD44 expression. S1P Induces ETS-1 Expression via S1P 3 /ROCK Pathway-ChIP analysis suggests that S1P treatment activates ETS-1. Thus, we investigated the signaling pathway in S1P-mediated ETS-1 activation. H1793 cells were treated with or without S1P (200 nM) for various times, and the expression of ETS-1 at the mRNA level was measured by qPCR analysis. We observed that ETS-1 mRNA was significantly increased (1.7-fold) at 1 h after S1P treatment (Fig. 5A). The S1P-mediated ETS-1 up-regulation steadily increased up to 24 h of S1P stimulation. Also, S1Pincreased ETS-1 expression was observed in H1792, H23, and H1650 lung adenocarcinoma cells (Fig. 5B), which were also shown to abundantly express S1P 3 receptors (39). In contrast, ETS-1 up-regulation was not observed in HBEC2-KT immortalized normal lung epithelial cells (Fig. 5B), which express low levels of S1P 3 receptors (39). Moreover, Western blot analysis showed that ETS-1 proteins were profoundly increased (5.0 Ϯ 2.5-fold) at 4 h after S1P stimulation. The increment of ETS-1 up-regulation was sustained at 24 h after treatment (Fig. 5C). Furthermore, analysis with anti-ETS-1 immunostaining technique showed that S1P treatment not only increased ETS-1 expression but also stimulated the nuclear localization of ETS-1 proteins (Fig. 5D, arrows).
We next examined the role of S1P 3 receptors in S1P-stimulated ETS-1 up-regulation. H1793 cells stably transfected with sh-S1P 3 or sh-Ctrl vector were stimulated with or without S1P (200 nM) for 4 h, and ETS-1 expression was measured by qPCR analysis. As shown in Fig. 5E, knockdown of S1P 3 receptors completely abrogated the S1P-stimulated ETS-1 up-regulation. Similar to signaling molecules involved in S1P-induced CD44 expression (Fig. 2D), pharmacological inhibition of ROCK, not PI3K and NFB, diminished 85% of the S1P-increased ETS-1 up-regulation (Fig. 5F). The requirement of ROCK is further supported by a specific gene knockdown technique, in which si-ROCK1 transfected cells significantly diminished the S1Pincreased ETS-1 expression (Fig. 5G). Collectively, our results suggest that the S1P/S1P 3 /ROCK signaling axis stimulates FIGURE 4. S1P stimulates ETS-1 binding to CD44 promoter. A, CD44 promoter contains four candidate ETS-1 binding sites. P1, P2, and P3 primer pairs were used to differentially amplify these candidate sites. B, ETS-1 binds to sites 3 and/or 4. ChIP analysis was performed as described under "Experimental Procedures." Note that a specific amplicon (expected size, 230 bp) was detected by using the P2 primer pair, whereas no amplicon was detected by using the P1 primer pair (top panel). Middle panel, PCR amplification of total input chromatin, used as a loading control. Bottom panel, ChIP analysis was performed with normal IgG (nIgG) or ETS-1 immunoprecipitates, followed by PCR amplification with P2 primer pair. C, cells were treated with or without S1P for the indicated times and analyzed for ETS-1 ChIP assay with P3 primer pair. PCR amplification of anti-ETS-1 immunoprecipitates and total input chromatin (expected size, 429 bp) are shown in the upper and lower panels, respectively. D, PCR amplification of anti-ETS-1 immunoprecipitates and total input DNA with P2 primer pair is shown in the upper and lower panels, respectively. Note that S1P treatment increased ETS-1 binding to sites 3 and/or 4 in the CD44 promoter region. B-D are images of a representative experiment that was repeated two times with similar results. NOVEMBER 8, 2013 • VOLUME 288 • NUMBER 45

JOURNAL OF BIOLOGICAL CHEMISTRY 32131
ETS-1 expression and nuclear translocation, leading to transcriptional up-regulation of CD44.
ETS-1 Plays an Essential Role in S1P-stimulated CD44 Expression and S1P-promoted Chemotaxis-We present evidence showing that S1P treatment enhances ETS-1 expression (Fig. 5) and stimulates ETS-1 binding to the CD44 promoter (Fig. 4). Next, we employed the shRNA-mediated gene silencing technique to investigate the role of ETS-1 in S1P-stimulated CD44 expression and cell migration. We were able to knock down ϳ50% of ETS-1 at both the mRNA and protein levels (Fig.  6, A and B). Transfection of sh-ETS-1 vector had no effects on the expression of GAPDH (Fig. 6A, used for internal control of qPCR analysis) and actin polypeptides (Fig. 6, B-D), suggesting that sh-ETS-1 is highly specific in knocking down ETS-1 molecules. Also, cells stably transfected with sh-ETS-1 or sh-Ctrl vector were treated with or without S1P (200 nM) for 4 h. The S1P-induced ETS-1 expression was completely inhibited in sh-ETS-1 transfected cells (Fig. 6C). Importantly, the S1P-stimulated CD44 expression was significantly diminished in ETS-1 knockdown cells (Fig. 6, D and E). Moreover, ETS-1 proteins were increased 1.4 Ϯ 0.2-and 2.2 Ϯ 0.1-fold (n ϭ 3) at 15 and 30 min, respectively, after S1P stimulation. In contrast, we did not observe significant alterations in CD44 levels at 15 and 30 min of S1P treatment (Fig. 6F, upper panels). Transfection of sh-ETS-1 profoundly diminished the S1P-increased ETS-1 and CD44 (Fig. 6F, lower panels). These results support the notion that S1P treatment stimulates the expression of ETS-1 proteins, which lead to the transcriptional up-regulation of CD44 molecules. In addition, the S1P-stimulated chemotactic responses were measured in cells stably transfected with sh-ETS-1 or sh-Ctrl vector. S1P treatment dose-dependently stimulated chemotaxis in cells stably transfected with sh-Ctrl vector. In con-trast, S1P was incapable of stimulating a migratory response in H1793 cells stably transfected with sh-ETS-1 vector (Fig. 6G). All these data together suggest that ETS-1 plays an essential role in the S1P-stimulated CD44 expression and chemotaxis. S1P 3 /ROCK1 Signaling Up-regulates ETS-1 via the JNK/c-Jun Pathway-It has been shown that JNK is a downstream signaling molecule of ROCK1 (55)(56)(57). Also, promoter analysis found four candidate binding sites of AP-1 (a heterodimeric complex composed of proteins including c-Jun and c-Fos) transcriptional factor in the ETS-1 promoter region (found in a EpiTect ChIP qPCR Primers search at the SABiosciences website). Thus, we examined whether S1P 3 /ROCK1 signaling-mediated ETS-1 up-regulation is controlled by JNK/c-Jun pathway. Western blotting with phospho-JNK antibody indicated that S1P treatment was capable of activating JNK (Fig. 7A). S1P-stimulated JNK activation was inhibited in S1P 3 knocked down cells (Fig. 7B), as well as in cells pretreated with the ROCK inhibitor (Fig. 7C), indicating the involvement of S1P 3 /ROCK1 pathway in S1P-mediated JNK activation. ChIP analysis with the c-Jun antibody showed an increase of c-Jun binding to the AP-1 site in the ETS-1 promoter region following S1P stimulation (Fig. 7D). Pharmacological inhibition of JNK activity with SP600125 diminished the S1P-increased protein levels of ETS-1 and CD44 (Fig. 7E). Transfection with si-JNK1 knocked down 42 Ϯ 9 and 22 Ϯ 11% (n ϭ 9) of p54 and p46, respectively, JNK1 (Fig. 7F). Moreover, si-JNK1 transfection substantially abrogated the S1P-increased ETS-1 and CD44 proteins (Fig. 7,  F and G). In addition, SP600125 treatment (Fig. 7H) and si-JNK1 transfection (Fig. 7I) significantly inhibited the S1P-stimulated cell migration. These data together suggest that S1P 3 / ROCK1 activates the JNK/AP-1 pathway, ultimately leading to transcriptional up-regulation of ETS-1. . S1P enhances ETS-1 expression via S1P 3 /ROCK pathway. A, ETS-1 mRNAs were quantitated in H1793 cells treated with S1P for various times. B, S1P increased ETS-1 expression in H1792, H23, and H1650 lung adenocarcinoma cells and not in HBEC2-KT cells. C, Western blotting showed that S1P treatment increased ETS-1 proteins. D, H1793 cells were treated with or without S1P for 4 h. ETS-1 was detected by immunostaining with ETS-1 antibody (upper panels), and nuclei were stained with DAPI (lower panels). Note that S1P treatment stimulated the expression and nuclear localization of ETS-1 proteins (white arrows). E, knockdown of S1P 3 completely abrogated S1P-increased ETS-1 expression. H1793 cells transfected with sh-S1P 3 or sh-Ctrl vector were treated with or without S1P for 4 h. ETS-1 mRNAs were quantified by qPCR. F, inhibition of ROCK significantly diminished S1P-induced ETS-1 expression. H1793 were pretreated with or without pharmacological inhibitors, followed by stimulating with S1P for 4 h. ETS-1 mRNAs were measured by qPCR.

DISCUSSION
S1P, a serum-borne bioactive lipid, plays an important role in the regulation of cell migration. S1P was shown to be a potent chemoattractant, because the migration stimulatory effect of S1P is mediated by S1P 1 or S1P 3 receptors (16,24,59). In addition, S1P was shown to exhibit the migration inhibitory effect, which is mediated by S1P 2 receptors (25,60). Mechanistically, S1P/S1P 1 signaling activates AKT kinase, leading to the activation of Rac/Cdc42 small GTPases and ultimately stimulating the formation of membrane ruffling and chemotaxis (16,61). Also, the Rho family of small GTPases mediated translocation of cortactin to the membrane ruffling area (62)(63)(64), and activation of integrin molecules (32,65,66) was shown to be important in the S1P/S1P 1 -stimulated chemotaxis. In contrast, S1P/ S1P 2 signaling was shown to stimulate Rac GTPase-activating protein (Rac-GAP) and thus abrogate growth factor-induced Rac activation and chemotaxis (25,60). S1P/S1P 3 -mediated chemotactic response was shown to play important roles in various physiological and pathophysiological responses (30,(33)(34)(35)(36)(37)(38). However, molecular details of the S1P 3mediated chemotactic response are poorly understood and remain to be elucidated. We found that S1P 3 receptors are abundantly expressed in a panel of lung adenocarcinoma cell lines (39). Therefore, we utilized these lung carcinoma cell lines to characterize the mechanism underlying S1P 3 -mediated cell migration. We observed that S1P treatment markedly enhanced CD44 expression at both mRNA and protein levels and increased membrane localization of CD44 proteins (Fig. 1). We further demonstrated that S1P-induced CD44 expression is mediated by the S1P 3 /ROCK signaling pathway (Fig. 2). Knockdown of CD44 completely abrogated S1P/S1P 3 -stimulated migratory response (Fig. 3), indicating that CD44 polypeptides play an essential role in S1P 3 -mediated chemotaxis. Moreover, we demonstrated that the S1P/S1P 3 /ROCK/JNK/c-Jun signaling axis stimulated the expression of ETS-1 transcriptional factor and nuclear translocation of ETS-1 (Fig. 5). ChIP analysis suggests that S1P treatment increased the binding of ETS-1 to the CD44 promoter region (Fig. 4). This observation leads to the speculation that S1P-stimulated ETS-1 binding to the CD44 promoter may trans-activate the expression of CD44 molecules. Indeed, knockdown of ETS-1 not only abrogated S1Pinduced CD44 expression but also inhibited S1P 3 -stimulated cell migration (Fig. 6). Collectively, our data elucidate a novel chemotaxis stimulatory signaling pathway, i.e., ETS-1/CD44 axis, which plays an essential role in S1P 3 -stimulated cell migration.
The ETS family of transcription factors plays important roles in various biological processes such as development, differentiation, proliferation, apoptosis, migration, tissue remodeling, and angiogenesis in various cell types. ETS-1, the prototype of the ETS family of transcription factors, controls a wide array of cellular activities via its transcriptional regulation of the expression of specific genes (84). Many genes (e.g., matrix metalloproteinases and urokinase plasminogen activator) that are important in extracellular matrix remodeling and cell migration are known targets of ETS-1 (85)(86)(87)(88), suggesting a critical role for ETS-1 in promoting cell migration (89). It should be noted that the ETS family of transcription factors has been shown to be associated with tumorigenic processes (23,58,90). Also, it has been elegantly demonstrated that CD44 polypeptides play critical roles in tumor "stemness," invasion, and metastasis. S1P is a potent chemoattractant. Recently, we reported that S1P 3 expression is markedly up-regulated in a panel of cultured lung adenocarcinoma cell lines (39). All these data together highlight an exciting possibility that the novel S1P 3 -ETS-CD44 signaling axis, characterized in this study, may contribute to invasion or metastasis of lung cancer progression.
In summary, we have characterized a novel S1P-regulated signaling cascade, i.e., S1P 3 activates ETS-1 transcription factor via the ROCK/JNK/c-Jun pathway, leading to the transcriptional up-regulation of CD44 molecules, which consequently promotes cell chemotactic response. Moreover, our study implies that the S1P 3 /ETS-1/CD44 axis may represent a novel chemotaxis stimulatory signaling pathway for the migration and invasion of lung adenocarcinoma cells. FIGURE 7. The JNK/c-Jun signaling pathway mediates the S1P 3 /ROCK1 up-regulated ETS-1 expression. A, H1793 cells were treated with S1P (200 nM) for various times. Protein lysates were immunoblotted with phospho-JNK antibody. Extracts of HEK293 cells treated without and with phorbol 12-myristate 13-acetate ("Ϫ"ve and "ϩ"ve, respectively) were used as negative and positive control. B, sh-Ctrl or sh-S1P 3 stably transfected H1793 cells were treated with or without S1P (200 nM, 15 min), followed by Western blotting with anti-phospho-JNK. C, H1793 cells were pretreated in the presence or absence of a ROCK inhibitor (Y27632) for 30 min, followed by stimulating with or without S1P (200 nM, 15 min). Extracts were probed with antiphospho-JNK. D, H1793 cells were treated with S1P (200 nM) for indicated times. AP-1 ChIP assays were performed by amplifying the anti-c-Jun precipitates with primer pairs specific for AP-1 site in the ETS-1 promoter region (SABiosciences; GPH1016833(Ϫ)01A). E, H1793 cells were pretreated with JNK inhibitor (SP600125, 10 M) for 30 min, followed by stimulating with S1P for 4 h. Extracts were blotted with anti-ETS-1 or anti-CD44. F, levels of ETS-1 and CD44 were measured in H1793 cells, transfected with si-Ctrl or si-JNK1 (Ambion), following S1P stimulation. G, immunoblot was quantified by National Institutes of Health ImageJ software and normalized to actin. H and I, S1P-stimulated chemotaxis was measured in H1793 cells pretreated with or without JNK inhibitor (SP600125) (H) or transfected with si-Ctrl or si-JNK1 (I). Cell migration induced by FBS (10%) was used as a control. *, p Ͻ 0.05, t test (n ϭ 6). M.W., molecular mass.