Oncogene (2012) 31, 68–79 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE The tumor-suppressor ARHI (DIRAS3) suppresses cell migration through inhibition of the Stat3 and FAK/Rho signaling pathways

DB Badgwell1,3,ZLu1,3,KLe1, F Gao1, M Yang1, GK Suh1, J-J Bao1, P Das1, M Andreeff 2, W Chen1,YYu1, AA Ahmed1, W S-L Liao1 and RC Bast Jr1

1Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, TX, USA and 2Department of Blood and Marrow Transplantation, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

Ovarian cancers migrate and metastasize over the surface Keywords: tumor-suppressor gene ARHI; migration of the peritoneal cavity. Consequently, dysregulation of suppression; Stat3; RhoA GTPase; cytoskeleton mechanisms that limit cell migration may be particularly important in the pathogenesis of the disease. ARHI is an imprinted tumor-suppressor gene that is downregulated in 460% of ovarian cancers, and its loss is associated Introduction with decreased progression-free survival. ARHI encodes a 26-kDa GTPase with homology to Ras. In contrast Epithelial ovarian cancers have been thought to arise to Ras, ARHI inhibits cell growth, but whether it also from cells that cover the ovarian surface or that line sub- regulates cell motility has not been studied previously. serosal inclusion cysts. Recently, cancers that resemble Here we report that re-expression of ARHI decreases the ovarian primaries have been shown to arise from the motility of IL-6- and epidermal growth factor (EGF)- fimbriae of the Fallopian tube, deposits of endometriosis stimulated SKOv3 and Hey ovarian cancer cells, inhibit- and from the peritoneal surface. Whatever the site of ing both chemotaxis and haptotaxis. ARHI binds to and origin, ovarian cancers spread through the abdominal sequesters Stat3 in the cytoplasm, preventing its trans- cavity, forming multiple implants on the peritoneal location to the nucleus and localization in focal adhesion surface. Consequently, genetic and epigenetic changes complexes. Stat3 siRNA or the JAK2 inhibitor AG490 that dysregulate motility are likely to be important for produced similar inhibition of motility. However, the the pathogenesis of ovarian cancer. Although ovarian combination of ARHI expression with Stat3 knockdown cancer can be cured in up to 90% of cases while still or inhibition produced greatest inhibition in ovarian confined to the ovary, approximately 70% are diag- cancer cell migration, consistent with Stat3-dependent nosed after the occurrence of peritoneal dissemination, and Stat3-independent mechanisms. Consistent with when the cure rate reduces to less than 30% (Jemal two distinct signaling pathways, knockdown of Stat3 et al., 2011). selectively inhibited IL-6-stimulated migration, whereas Ovarian surface epithelial cells are generally quies- knockdown of focal adhesion kinase (FAK) preferentially cent, with a low rate of proliferation. Following inhibited EGF-stimulated migration. In EGF-stimulated ovulation and rupture of the mature follicle, ovarian ovarian cancer cells, re-expression of ARHI inhibited surface epithelial cells proliferate and migrate to the site FAKY397 and SrcY416 phosphorylation, disrupted focal of ovulation in order to repair the wound left by release adhesions, and blocked FAK-mediated RhoA signaling, of the ovum. Once integrity of the ovarian surface is resulting in decreased levels of GTP-RhoA. Re-expression restored, normal ovarian epithelial cells cease to of ARHI also disrupted the formation of actin stress fibers migrate. Thus, the ability to proliferate and migrate is in a FAK- and RhoA-dependent manner. Thus, ARHI has an intrinsic characteristic of normal ovarian cells that is a critical and previously uncharacterized role in the regu- under tight regulation (Katabuchi and Okamura, 2003; lation of ovarian cancer cell migration, exerting inhibitory Okamura and Katabuchi, 2005). During ovarian onco- effects on two distinct signaling pathways. genesis, ovarian cancer cells lose regulatory constraints Oncogene (2012) 31, 68–79; doi:10.1038/onc.2011.213; on motility and invasion (Liotta et al., 1987; Katabuchi published online 6 June 2011 and Okamura, 2003). A number of factors can stimulate the migration of ovarian cancer cells (Bast et al., 2009), including Stat3, epidermal growth factor (EGF) and Correspondence: Dr RC Bast Jr, Department of Experimental fibronectin, but relatively little is known regarding the Therapeutics, The University of Texas MD Anderson Cancer Center, genetic or epigenetic factors that are dysregulated to 1515 Holcombe Boulevard, Box 355, Houston, TX 77030, USA. increase motility and migration. E-mail: [email protected] 3These authors contributed equally to this work. Cell migration is a highly regulated process that Received 23 September 2010; revised 13 March 2011; accepted 11 April involves continuous formation and turnover of 2011; published online 6 June 2011 complexes within focal adhesions that serve both as ARHI suppresses ovarian cancer cell migration DB Badgwell et al 69 points of traction and as signaling centers (Ridley et al., cycle arrest. Growth of ovarian and 2003; Romer et al., 2006). Regulation of cell migration xenografts is reversibly suppressed by ARHI and during embryonic development, inflammation and survival of dormant cells appears to depend, at least in tumorigenesis is mediated by cytokines, growth factors part, on autophagy (Lu et al., 2008). The impact of and integrins through activation of intracellular signal- ARHI on cell motility and migration has not been ing molecules that include Stat3 and focal adhesion explored previously. kinase (FAK) (Naora and Montell, 2005; Jang et al., In the present study, using conditional expression of 2007; Tomar and Schlaepfer, 2009). physiological levels of ARHI in SKOv3 and Hey Stat3 is constitutively phosphorylated and activated ovarian cancer cell lines, we show that re-expression of in 70% of ovarian cancers (Rosen et al., 2004). ARHI inhibits ovarian cancer cell motility by interfering Frequently, this is associated with autocrine stimulation with JAK/Stat signal transduction, and inhibiting the of the interleukin-6 (IL-6) receptor recruiting JAK2 and FAK signaling pathway, decreasing the formation of phosphorylating Stat3. In addition to its nuclear role as focal adhesion complexes and stress fibers. a transcription factor, phosphorylated Stat3 has also been found to localize to focal adhesions, interact with focal adhesion and contribute to ovarian cancer cell motility (Silver et al., 2004). Inhibition of Stat3 Results activation with chemical inhibitors or small interfering RNA (siRNA) reduced the motility of ovarian cancer ARHI reduces chemotactic and haptotactic responses, and cells. Further, the Stat3-associated increase in cell decreases the speed of cell migration motility was concomitant with upregulation of epithe- We have generated stable sub-lines of SKOv3 and Hey lial-mesenchymal transition-associated N-cadherin and ovarian cancer cells by Tet-on-inducible expression of vimentin expression (Colomiere et al., 2009). ARHI (SKOv3-ARHI and Hey-ARHI). These two FAK is also activated in ovarian cancers. FAK is ovarian cancer cell lines were chosen because SKOv3 localized in focal adhesions and regulates the cycle of cells express low levels of endogenous ARHI, whereas focal contact formation and disassembly required for Hey cells have no detectable ARHI (Feng et al., 2008; efficient cell movement (Geiger and Bershadsky, 2001). Lu et al., 2008). Incubation of each sub-line with 1 mg/ml FAK activation correlates with paxillin phosphorylation DOX induced moderate ARHI expression comparable and subsequent actin stress fiber formation (Schaller, to that observed in cultured normal surface epithelial 2001; Parsons, 2003). These events are often coupled to cells (Lu et al., 2008), as well as to the normal ovarian Rho-family that cycle between inactive, GDP- epithelium tissue as determined by immunohistochem- bound and active, GTP-bound forms. FAK can enhance ical staining (Figure 1a). To examine the effects of the activation of RhoA GTPase (Chikumi et al., 2002; ARHI on cell motility, SKOv3-ARHI and the parental Zhai et al., 2003), contributing to the regulation of the SKOv3 cell lines were treated with DOX and then actin cytoskeletal structure, focal adhesion complexes inoculated (1 Â 105 cells for chemotaxis and 2 Â 104 for and cell polarity, as well as cell–cell communication haptotaxis) in Boyden chambers and allowed to migrate (Van Aelst and D’Souza-Schorey, 1997; Kaibuchi et al., toward 10% fetal bovine serum or 5 mg/ml fibronectin 1999; Etienne-Manneville and Hall, 2002). for 16 h. Induction of ARHI expression significantly Dysregulation of Stat3 and FAK could relate, in part, decreased the fetal bovine serum- and fibronectin- to loss of function of critical tumor-suppressor . stimulated migration of SKOv3-ARHI cells when The maternally imprinted growth-regulatory gene ARHI compared with un-induced cells (Figure 1b, middle is a tumor-suppressor gene whose function is down- panels). As expected, DOX treatment had no effect on regulated or lost in 460% of ovarian cancers by several the chemotactic or haptotactic migration of parental different mechanisms, including loss of heterozygosity, SKOv3 cells (Figure 1b, left panels). Similar inhibition hypermethylation, transcriptional regulation and shor- in cell migration by ARHI was also observed in Hey- tened mRNA half-life (Yu et al., 2006). Loss of ARHI ARHI ovarian cancer cells after treatment with DOX expression is associated with tumor progression in (Figure 1b, right panels). To address whether the expres- breast cancers and decreased disease-free survival in sion of endogenous ARHI may also affect ovarian ovarian cancers (Rosen et al., 2004). ARHI encodes a cancer cell migration, we knocked down ARHI in the 26-kDa protein with 55–62% homology to Ras. In parental SKOv3 and Hey cells, and assessed their effects contrast to Ras, ARHI inhibits, rather than stimulates on cell migration. As shown in Figure 1c, knockdown of cell growth. This growth-inhibitory function has been ARHI in ARHI-expressing SKOv3 cells resulted in an attributed to a 34-amino-acid N-terminal extension that increase in cell migration, whereas had no effect on Hey is unique to ARHI (Luo et al., 2003). Re-expression of cells that do not express endogenous ARHI. To confirm ARHI in cancer cells inhibits signaling through Ras/ that ARHI has an inhibitory effect on cell migration MAPK and phosphatidylinositol-3-kinase, upregulates that is independent from its growth-inhibitory effect, we p21WAF1/CIP1, downregulates cyclin-D1, induces JNK performed live-cell time-lapse imaging of migrating cells activation and inhibits Stat3 signaling (Yu et al., 1999; in a scratch assay over a 4.5-h interval. SKOv3-ARHI Luo et al., 2003). When ARHI is expressed at cells that express ARHI showed a 40% reduction in physiological levels from a doxycycline (DOX)-inducible average migration speed when compared with those promoter, autophagy is induced and cells undergo cell- of un-induced cells (Figure 1d).

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 70 ARHI inhibits cell migration induced by the JAK/Stat3 has been reported to promote ovarian cancer cell signaling pathway motility (Huang et al., 2000; Silver et al., 2004; Debidda Many ovarian cancer cell lines have constitutively et al., 2005). We therefore sought to examine whether activated Stat3 (Huang et al., 2000) and 470% of this Stat3-induced cell motility could be inhibited by ovarian cancers showed a higher nuclear localization of ARHI. SKOv3-ARHI cells were stimulated with IL-6 phosphorylated Stat3, which correlated with poor over- and assayed for chemotaxis or haptotaxis. Treatment all prognosis (Rosen et al., 2004). Furthermore, Stat3 of SKOv3-ARHI cells with IL-6 induced both chemo-

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 71 tactic and haptotactic cell migration by 2- to 3-fold chemotactic cell migration by 25–45% when compared (Figure 2a). IL-6-induced cell migration was signifi- with that in siControl-transfected cells (Figure 2d). cantly reduced when ARHI was expressed. This ARHI- Expression of ARHI in siStat3-transfected cells further mediated inhibition of cell migration, however, is not reduced cell motility. Together, these data indicated that because of inhibition of Stat3 tyrosine phosphorylation chemotaxis is IL-6-dependent whereas haptotaxis is not. (Figure 2b). As control, SKOv3-ARHI cells were treated Moreover, expression of ARHI inhibited the basal and with AG490 to block IL-6-mediated Stat3 activation. Stat3-mediated motility of ovarian cancer cells by Stat3- Treatment with AG490 dramatically inhibited both dependent and Stat3-independent mechanisms. basal and IL-6-induced cell migration. Interestingly, in the presence of AG490, ARHI expression can further reduce SKOv3-ARHI cell migration, suggesting that ARHI complexes with Stat3 and prevents its localization ARHI’s inhibitory effects on cell migration may not in the focal adhesions and in the nucleus be limited to its inhibition of the JAK/Stat3 signaling Our earlier work showed that Stat3 is an ARHI- pathway. interacting protein (Nishimoto et al, 2005); however, To further evaluate the influence of Stat3 and ARHI these studies were performed with adeno-ARHI-infected on cell migration, SKOv3-ARHI cells were transfected breast cancer cells that produced supra-physiological with Stat3 siRNA to knockdown endogenous Stat3 levels of ARHI. To document that ARHI expressed and then treated with DOX to induce ARHI. Treatment at physiological levels can interact with Stat3, we of SKOv3-ARHI cells with different siStat3 effec- co-immunoprecipitated ARHI and Stat3 from cell tively knocked down Stat3 (Figure 2c) and reduced lysates prepared from DOX-induced SKOv3-ARHI cells.

Figure 2 ARHI inhibits cell migration by interfering with JAK/Stat signaling. (a) Stat3 activation enhances and Stat3 inhibition decreases ovarian cancer cell migration. SKOv3-ARHI cells in Boyden chambers were treated with IL-6 and AG490, individually or together, and allowed to migrate toward the lower chambers containing either 10% fetal bovine serum (chemotaxis) or 5 mg/ml fibronectin (haptotaxis). The columns indicate the mean, and the bars indicate the s.e. (**Po0.01). (b) ARHI expression does not affect Stat3 phosphorylation at tyrosine-705. (c) Knockdown of Stat3 in SKOv3-ARHI cells with siStat3 as indicated by western blotting. Two individual Stat3 siRNAs (1 and 2) as well as a pooled Stat3 siRNA (P) were used in the knockdown. (d) Knockdown of Stat3 and induction of ARHI produce additive inhibition of ovarian cancer cell migration. Migration assays of cells transfected with three different Stat3 siRNAs or a control siRNA with or without pretreatment with DOX for 48 h. The columns indicate the mean, and the bars indicate the s.e. (*Po0.05; **Po0.01). DOX, doxycycline; IL, interleukin; siRNA, small interfering RNA.

Figure 1 ARHI expression inhibits the chemotaxis, haptotaxis and migration speed of ovarian cancer cells. (a) Comparable ARHI expression in DOX-induced SKOv3-ARHI cells and the normal ovarian epithelium (NOE) tissue. Immunohistochemical staining for ARHI expression in NOE and SKOv3-ARHI cells with and without DOX. (b) Chemotaxis and haptotaxis. Parental SKOv3, SKOv3- ARHI and Hey-ARHI cells were treated with or without 1 mg/ml DOX for 24 h before addition to the upper compartment of Boyden chambers. The cells were allowed to migrate for 16 h toward the lower chambers, which contained either 10% fetal bovine serum (chemotaxis) or 5 mg/ml fibronectin (haptotaxis). The cells that had migrated through the membrane were counted and expressed as the mean number of cells in triplicates from three independent experiments. The columns indicate the mean, and the bars indicate the s.e. (*Po0.05; **Po0.01). (c) Knockdown of endogenous ARHI increases cell motility. Left panel: Parental SKOv3 and Hey cells were transfected with siARHI and assayed for cell motility (**Po0.01). Right panel: Endogenous ARHI expression was determined by real- time quantitative reverse transcription–PCR. (d) Migration speed in scratch assays. SKOv3-ARHI cells were grown to confluence with or without DOX. Following a scratch wound, wound closure was recorded over a 4.5-h time period by live-cell time-lapse microscopy. Representative images at the indicated time points are shown. The speeds of migration are calculated as distance over time and expressed as the mean of cell migration speed (n ¼ 6 in each group; **Po0.01). DOX, doxycycline.

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 72

Figure 3 ARHI interacts with Stat3 and prevents its localization to the focal adhesion complex and to the nucleus. (a) Stat3 and ARHI co-immunoprecipitation. SKOv3 cells were incubated with or without DOX for 48 h. Cell lysates were incubated with anti- ARHI or anti-Stat3 antibody, and analyzed by western blotting. IgG was included as a negative control. (b) ARHI expression disrupts the colocalization of Stat3, vinculin and paxillin on immunofluorescence staining. SKOv3-ARHI cells were treated with fibronectin and stained with an anti-Stat3 antibody in the absence or presence of a Stat3-blocking peptide (upper panels 1–3). DOX-induced SKOv3- ARHI cells were stained with anti-vinculin and anti-Stat3 (middle panels), or anti-paxillin and anti-Stat3 antibodies (lower panels), and examined by fluorescence confocal microscopy. (c) ARHI and Stat3 colocalize in the cytoplasm. SKOv3-ARHI cells were pretreated with DOX to induce ARHI expression and then with IL-6 for 30 min to activate Stat3. Treated cells were stained using anti-ARHI and anti-Stat3, and the subcellular localization of Stat3 was examined by confocal microscopy. (d) ARHI reduces N-cadherin and vimentin expression. SKOv3-ARHI cells were treated with DOX for 24 h before addition of IL-6 (10 ng/ml). mRNA levels were normalized with GAPDH mRNA and the fold changes are relative to the untreated controls to which a value of 1.0 was assigned (*Po0.05; **Po0.01). DOX, doxycycline; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IL, interleukin.

Consistent with our previous results, Stat3 was co- As an integral component of cell motility, focal immunoprecipitated with an anti-ARHI antibody and adhesion complexes form at the leading edge of moving anti-Stat3 co-immunoprecipitated ARHI (Figure 3a), cells and permit the cells to spread in the forward showing that ARHI, expressed at physiological levels, direction (Carragher and Frame, 2004). Silver et al. can indeed interact with Stat3. (2004) reported localization of activated Stat3 in focal

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 73 adhesions and showed a crucial role for Stat3 in the growth factor- and integrin-mediated cell adhesion. motility of ovarian cancer cells. Consistent with their Activation of FAK by EGF or integrin clustering leads observation, we also observed specific punctate Stat3 to autophosphorylation at Tyr397 (p-FAKY397), a localization at the focal adhesion following stimulation binding site for Src family kinases (Schlaepfer and with fibronectin (Figure 3b). As ARHI can interact with Mitra, 2004; Long et al., 2010). We examined the effects Stat3, we hypothesized that one mechanism by which of ARHI on FAK and Src activation by monitoring the ARHI may interfere with Stat3’s ability to promote cell phosphorylation of FAKY397 and the phosphorylation of motility is by sequestering Stat3 away from the focal SrcY416, respectively, in response to EGF or fibronectin. adhesion complexes as well as from the nucleus. As SKOv3-ARHI cells were treated with or without DOX such, ARHI may prevent Stat3 from affecting the in the presence or absence of fibronectin or EGF, and function of focal adhesions or inducing genes that then the levels of p-FAKY397 and p-SrcY416 were promote cell motility. To test this hypothesis, we analyzed. As shown in Figures 4a and b, expression of performed immunofluorescence staining of ARHI, Stat3 ARHI reduced p-FAKY397 and p-SrcY416 levels induced and focal adhesion proteins to determine whether ARHI by either fibronectin or EGF. Reduction in p-FAKY397 disrupts Stat3’s participation at the focal adhesion but not total FAK levels by ARHI was further complexes and in the nucleus. SKOv3-ARHI cells were confirmed by immunofluorescence staining (Figure 4c). grown on fibronectin-coated slides with or without As ARHI interacts with and sequesters Stat3 in the DOX and stained for Stat3, ARHI and two focal cytoplasm, it raises the possibility that inhibition of adhesion markers, vinculin and paxillin. In the absence FAK activation by ARHI might be because of reduced of ARHI, Stat3 staining was observed at the focal Stat3-FAK colocalization at the focal adhesion. Inter- adhesions and colocalized with vinculin and paxillin estingly, knockdown of Stat3 did not reduce p-FAKY397 (Figure 3b). This colocalization was, however, disrupted levels (Figure 4d). To further evaluate the independent when ARHI expression was induced. Our data thus roles of Stat3 and FAK in cell migration, SKOv3-ARHI show that expression of ARHI not only prevented cells were transfected with Stat3 siRNA and FAK Stat3 localization at the focal adhesions, but also siRNA to knockdown endogenous Stat3 and FAK reduced dramatically the formation of focal adhesion (Figure 4e), followed by DOX treatment to induce complexes as reflected by the reduced vinculin and ARHI in the presence of IL-6 or EGF. Stimulation of paxillin staining. SKOv3-ARHI cells with either IL-6 or EGF increased To examine whether ARHI also affects Stat3 nuclear cell migration by 1.5- to 2-fold (Figure 4f). Interestingly, localization, SKOv3-ARHI cells were stimulated with the IL-6-stimulated increase was more dramatically IL-6 after DOX treatment. In control cells, IL-6 induced affected by Stat3 knockdown (31% decrease; Po0.01) rapid Stat3 nuclear translocation within 30 min. How- than by FAK knockdown. By contrast, the EGF- ever, expression of ARHI prevented Stat3 nuclear mediated increase in cell migration was strongly affected localization in response to IL-6, resulting in their by FAK knockdown (57% decrease; Po0.01) than by colocalization in the cytoplasm (Figure 3c). To confirm Stat3 knockdown (Figure 4f). Finally, knockdown of that the cytoplasmic sequestration of Stat3 by ARHI both Stat3 and FAK resulted in most dramatic results in reduced Stat3-dependent transcription, we reduction in chemotactic cell migration. Additional analyzed the expression of two Stat3 target genes, N- expression of ARHI in these double-knockdown cadherin and vimentin, known to participate in the SKOv3-ARHI cells did not inhibit cell migration any motility of ovarian cancer cells (Cheng et al., 2008; further (Figure 4f). Together, these data indicate that Vij et al., 2008; Colomiere et al., 2009). Treatment of expression of ARHI inhibited ovarian cancer cell SKOv3-ARHI cells with IL-6 significantly increased N- motility through two distinct pathways, one dependent cadherin and vimentin expression. This increase was on Stat3 signaling and the other on FAK-mediated inhibited in ARHI-expressing cells (Figure 3d). How- pathways. ever, in parental SKOv3 cells incapable of expressing ARHI, DOX treatment did not change N-cadherin and vimentin expression (data not shown). Taken together, ARHI inhibits FAK-mediated focal contact formation sequestration of Stat3 in the cytoplasm by ARHI FAK activation is associated with both focal contact reduces the transcription of Stat3 target genes important formation and the formation of actin stress fibers for motility and also prevents translocation of Stat3 to (Orr et al., 2000; Orr and Wang, 2001; Parsons, 2003). focal adhesions where it is required for optimal cell Furthermore, the FAK signaling complex regulates migration. fibronectin-associated RhoA activation and focal adhe- sion formation during cell migration (Lim et al., 2008). As EGF activates RhoA and stimulates motility ARHI regulates cell migration by inhibiting FAK (Marcoux and Vuori, 2005; Kakinuma et al., 2008), activation we asked whether inhibition of EGF-stimulated migra- FAK is a major component of the focal adhesion tion by ARHI may be mediated through inhibition complex that regulates cell motility. We therefore of RhoA. Chemotaxis of SKOv3-ARHI cells was examined whether ARHI can also inhibit the migration measured with or without EGF, the RhoA inhibitor of SKOv3-ARHI cells through regulation of the FAK C3 transferase and a combination of the two agents. and Src signaling pathway. FAK activity is regulated by Addition of EGF increased the motility of SKOv3-ARHI

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 74

Figure 4 ARHI regulates cell migration by inhibiting FAK activation. (a) ARHI inhibits FAKY397 phosphorylation. SKOv3-ARHI cells were treated with DOX for 24 h and then stimulated with fibronectin (FN) (1 mg/ml for 1 h) or EGF (10 ng/ml for 30 min). The levels of FAKY397 phosphorylation are expressed as the ratio of p-FAKY397 to total FAK, to which the ratio for control cells is assigned a value of 1.0. (b) ARHI inhibits Src phosphorylation. SKOv3-ARHI cells were treated with DOX for 24 h and then stimulated with FN (1 mg/ml for 1 h) or EGF (10 ng/ml for 30 min). The levels of Src phosphorylation are expressed as the ratio of p-Src to total Src, to which the ratio for control cells is assigned a value of 1.0. (c) ARHI inhibits p-FAKY397 at focal adhesions. Cells were plated on fibronectin-coated cover slips, treated with or without DOX for 48 h, and stained with anti-p-FAKY397, anti-FAK and anti-ARHI antibodies. Fluorescence intensities were quantified by ImageJ. The columns indicate the mean, and the bars indicate the s.e. (**Po0.01). (d) Stat3 knockdown does not reduce p-FAKY397 levels. (e) Knockdown of Stat3 and FAK in SKOv3-ARHI cells by siRNAs. (f) Migration assays of cells transfected with Stat3 and FAK siRNAs, or control siRNA, with or without pretreatment with DOX for 48 h, and with or without IL-6 or EGF stimulation for 16 h. The columns indicate the mean, and the bars indicate the s.e. (*Po0.05; **Po0.01). In IL-6-stimulated cells, siStat3 significantly inhibited migration (Po0.01), whereas siFAK did not (P40.11). Similarly, in EGF-stimulated cells, siFAK significantly inhibited migration (Po0.01), whereas siStat3 did not (P40.08). DOX, doxycycline; EGF, epidermal growth factor; FAK, focal adhesion kinase; IL, interleukin; siRNA, small interfering RNA.

cells. Cell motility, however, was reduced significantly whether ARHI can inhibit EGF-induced RhoA activa- when ARHI was expressed (Figure 5a). As expected, tion. A green fluorescent protein (GFP)-RhoA plasmid treatment with the RhoA inhibitor C3 transferase was transfected into cells treated with or without DOX completely abrogated cell motility (Figure 5a). Stimula- and EGF. As predicted, the level of active RhoA was tion with EGF could only restore the migration slightly induced by EGF but was significantly reduced when in both control and DOX-treated cells, suggesting that ARHI was expressed (Figure 5b). ARHI also reduced ARHI may regulate SKOv3-ARHI cell migration by stress fiber formation and GFP-RhoA-stimulated cell inhibiting RhoA activation. We therefore tested directly migration significantly (Figures 5c and d). Together,

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 75

Figure 5 ARHI decreases RhoA activity and inhibits actin stress fiber formation. (a) SKOv3-ARHI cells were treated with DOX for 24 h before they were treated with EGF and C3 transferase, individually or in combination (n ¼ 9 in each group; **Po0.01). (b) ARHI inhibits RhoA GTPase activity. SKOv3-ARHI cells were induced with DOX and then transfected with the GFP-RhoA plasmid. The cells were treated with DOX for an additional 24 h before stimulation with EGF (30 min). (c) ARHI reduces actin stress fiber formation. SKOv3-ARHI cells were treated with DOX for 24 h before they were transfected with the GFP-RhoA plasmid for 24 h. Actin stress fibers were stained with rhodamine-labeled phalloidin. The yellow arrowheads indicate the GFP-RhoA-transfected cells and the blue arrowheads indicate the non-transfected cells. (d) ARHI inhibits RhoA-stimulated cell migration. SKOv3-ARHI cells were treated with DOX and transfected with GFP-RhoA before the migration assay. The columns indicate the mean, and the bars indicate the s.e. (*Po0.05). DOX, doxycycline; EGF, epidermal growth factor; GFP, green fluorescent protein. these data indicate that ARHI expression leads to re-expression of ARHI inhibited SKOv3 and Hey decreased RhoA activity and inhibition of stress fiber ovarian cancer cell migration, and suggests that loss formation, both of which contribute to ARHI-mediated of ARHI expression may be one factor that contributes inhibition of SKOv3-ARHI cell migration. to the increased motility and metastasis of ovarian cancer cells. Although ARHI has been implicated in tumor proliferation, dormancy and autophagy, this is Discussion the first report that re-expression of ARHI inhibits the motility, chemotaxis and haptotaxis of human ovarian Epithelial ovarian carcinoma is characterized by wide- cancer cells. The underlying mechanism that accounts spread intra-abdominal metastases mediated through for the differential inhibitory effects between SKOv3 surface shedding of tumor cells, migration and peri- and Hey cells is unclear. However, factors such as the toneal implantation. Our current study shows that lower levels of ARHI expression in Hey-ARHI cells as

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 76 well as differences in their genetic cell background are association of paxillin and vinculin with focal adhesions likely to influence the extent of inhibition. (Figure 3b). Indeed, correlation between FAK activa- Interaction of ARHI and Stat3 is of particular interest tion, paxillin and vinculin phosphorylation, and the in that Stat3 signaling is important for both prolifera- subsequent actin stress fiber formation has been tion and motility. A significant fraction of ovarian reported (Schaller, 2001; Parsons, 2003). These events cancers not only secrete IL-6 but also express the IL-6 are often associated with Rho-family GTPases that act receptor, resulting in high cell proliferation and motility as switches between an inactive, GDP-bound form and owing to autocrine stimulation (Martinez-Maza and an active, GTP-bound form. Importantly, FAK pro- Berek, 1991; Lidor et al., 1993; Rustin et al., 1993). motes p190RhoGEF tyrosine phosphorylation and Quiescent Stat3 exists mostly in the cytosol, whereas enhances the activation of RhoA (Zhai et al., 2003). activated Stat3 translocates to the nucleus to induce ARHI decreased the EGF-induced chemotaxis of gene transcription (Zhong et al., 1994). Nuclear Stat3 SKOv3-ARHI cells. Chemotaxis in response to EGF has been found in 70% of ovarian cancers and was was reduced by FAK knockdown (Figure 4f) and also associated with poor prognosis. Using SKOv3-ARHI impaired by the RhoA inhibitor, C3 transferase; and Hey-ARHI inducible ovarian cancer cells, we have however, chemotaxis was not inhibited further by ARHI examined the effect of re-expression of ARHI at (Figure 5a), indicating that ARHI may regulate cell physiological levels on Stat3 signaling and motility. migration by reducing FAK-mediated RhoA activation. Although Stat3 tyrosine phosphorylation was unaf- We confirmed that RhoA activity, which is critical for fected by ARHI (Figure 2b), ARHI expression was able the formation of actin stress fiber and cell migration to inhibit Stat3 translocation to the nucleus as well as to (Kurokawa et al., 2005), is largely abrogated upon focal adhesion complexes (Figures 3b and c). This ARHI induction (Figure 5b). Additionally, cell migra- inhibition is presumably because of physical interaction tion requires the integration of specific focal adhesion between ARHI and Stat3, resulting in the sequestration dynamics, including formation of actin stress fibers, of Stat3 in the cytoplasm. Stat3 has been shown recently which is inhibited by ARHI (Figure 5c). Together, our to accumulate in mitochondria and contribute to Ras- results suggest that ARHI regulates ovarian cell migra- dependent cellular transformation (Gough et al., 2009). tion by interfering with the function of Stat3, by As such, ARHI may potentially regulate mitochondrial inhibiting FAK and RhoA activation, and by decreasing function by sequestering Stat3 in the cytoplasm. In the formation of stress fibers (Figure 6). addition, failure of Stat3 to translocate to the nucleus was associated with decreased expression of Stat3 target genes, N-cadherin and vimentin. Thus ARHI can inhibit both proliferation and motility by inhibiting a single critical mediator. Moreover, ovarian oncogenesis may require not only autocrine stimulation by IL-6, but also loss of ARHI-mediated inhibition of Stat3 signaling. Our study confirms earlier reports that Stat3 has a critical role in the motility of ovarian cancer cells (Takeda et al., 1997; Yamashita et al., 2002). Localiza- tion of Stat3 at focal adhesions suggested a direct role in motility through protein–protein interaction. As such, Stat3 may serve as an adapter protein in integrin- mediated cell adhesion or could function as a sensor of adhesion, becoming activated in focal adhesion and translocating to the nucleus to alter gene expression in response to cell adhesion (Roger et al., 2006). Alter- natively, the nuclear translocation of activated Stat3 from focal adhesions may induce critical proteins needed for motility. In either case, sequestration by ARHI inhibits Stat3 function as it relates to motility. FAK is activated by a variety of growth factors receptors and integrins, and transmits signals down- stream to a variety of target molecules to regulate the cycle of focal contact formation and disassembly required for efficient cell movement. Thus, FAK acts as an integrator of cell motility-associated signaling events. Growth factor stimulates cell motility by inducing the phosphorylation of FAK; however, FAKY397 phosphorylation is inhibited when ARHI was Figure 6 A molecular model. The tumor-suppressor gene ARHI induced (Figure 4). Given that Stat3 knockdown did not functions as a regulatory molecule in controlling ovarian cancer Y397 alter p-FAK levels, we reasoned that the effect of cell migration through modulation of Stat3- and FAK-mediated ARHI on FAK activity might relate to decreased signaling pathways. FAK, focal adhesion kinase.

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 77 Integrative functions that include control of cell chamber. Cells migrated to the lower surface were counted 16 h motility have been reported for other genes known to later and migration was expressed as the mean number of cells have a critical role in oncogenesis. They include the in triplicate wells from three independent experiments. tumor-suppressor proteins TP53 (Roger et al.,2006), PTEN (Raftopoulou et al., 2004) or LKB-1 (Forcet et al., Live-cell time-lapse microscopy 2005). Our data indicate that ARHI’s tumor-suppressor SKOv3-ARHI cells (4 Â 105) were plated in glass-bottom function is not limited to its antiproliferative activity, but dishes (MatTek, Ashland, MA, USA) and grown to confluence may also rely on its combined effects in the cell migration with or without DOX. The cells were scratched with a 10-ml of ovarian cancer cells. Considering that Stat3 and FAK pipette tip and placed onto a microscope stage in a 37 1C are constitutively activated in many ovarian cancers, the chamber with 5% CO2. Scratch-wound closure was monitored by DIC microscopy and an Olympus IX-81 inverted microscope loss of ARHI expression in the majority of ovarian (Olympus, Center Valley, PA, USA). Images were captured cancers may result in the upregulation of Stat3 and FAK every 15 min using the Slidebook software (http://www.intelligent- activity, and thereby contribute to oncogenesis. imaging.com/home.php) to manage the Hamamatsu Orca II ER camera and microscope settings.

Materials and methods Immunohistochemical staining of ARHI Paraffin-embedded normal ovarian epithelium tissue and cell Antibodies and reagents pellets from DOX-treated or untreated SKOv3-ARHI cells Antibodies against Stat3, p-SrcY416, paxillin, FAK and b-actin were sectioned. After initial deparaffinization, endogenous were purchased from Cell Signaling Technology (Beverly, MA, peroxidase activity was blocked using 0.3% hydrogen peroxide USA), and the anti-p-FAKY397 antibody was purchased from and steamed in 1 Â Diva Decloaker (Biocare Medical, Millipore (Worcester, MA, USA). Antibodies against ARHI Concord, CA, USA) for 1 h to restore latent epitopes. The were generated in our laboratory. AG490 was purchased from slides were then incubated with an anti-ARHI antibody Calbiochem (La Jolla, CA, USA). The cell-permeable RhoA (15E11) at a final concentration of 15 mg/ml at 4 1C overnight. inhibitor C3 transferase was purchased from Cytoskeleton The slides were washed and incubated for 30 min with each (Denver, CO, USA). IL-6, EGF and anti-vinculin were with biotin-labeled secondary antibody, and then with purchased from Sigma (St Louis, MO, USA), and siRNAs streptavidin/peroxidase. The slides were then incubated with and the Dharmafect #4 transfection reagent were from the diaminobenzidine substrate (Biocare Medical), counter- Dharmacon Research (Lafayette, CO, USA). stained with hematoxylin and mounted in Permount.

Cell culture Immunofluorescence staining Cells were maintained in McCoy’s 5A (SKOv3 and SKOv3- SKOv3-ARHI cells were plated onto slides with or without ARHI) or RPMI (Hey-ARHI) medium supplemented with fibronectin and treated with DOX for 24 h, fixed in 4% 10% Tet-system-approved fetal bovine serum, 200 mg/ml paraformaldehyde and permeabilized using 0.5% Triton G418, and 0.25 mg/ml puromycin, 100 mML-, X-100. Washed cells were blocked with 3% bovine serum 100 mg/ml streptomycin and 100 U/ml penicillin, and grown albumin and incubated with antibodies against Stat3, vinculin, paxillin or p-FAKY397. After washing, the cells were incubated at 37 1C in 95% humidified air and 5% CO2. Physiological levels of ARHI were achieved 16–24 h after addition of 1 mg/ml with secondary antibodies conjugated with Alexa Fluor-488 or DOX to the culture medium (Lu et al., 2008). Alexa Fluor-594 (Molecular Probes, Invitrogen, Eugene, OR, USA). Actin stress fibers were stained with rhodamine-labeled phalloidin (1/1000; Molecular Probes, Invitrogen). Finally, Chemotaxis and haptotaxis cells were mounted and examined by confocal microscopy Briefly, 1 Â 105 cells (SKOv3 and SKOv3-ARHI) or 5 Â 104 (Olympus FluoView 500 or 1000; Olympus Inc., Melville, NY, cells (Hey-ARHI) in 0.5 ml of serum-free medium were USA). introduced into the upper compartment of Boyden chambers (BD Discovery Labware, Franklin Lakes, NJ, USA) fitted with Immunoprecipitation membranes of 3.0 mm porosity separating the upper and lower 6 compartments. The lower compartment was filled with normal SKOv3-ARHI cells (1 Â 10 ) were incubated with or without culture medium, medium supplemented with IL-6 (chemotaxis) DOX for 24 h and then treated with lysis buffer (50 mM or serum-free media containing 5 mg/ml fibronectin (hapto- HEPES (pH 7.0), 150 mM NaCl, 1.5 mM MgCl2,1mM EGTA, taxis). After 16 h of incubation, cells were wiped from the 100 mM NaF, 10 mM sodium pyrophosphate, 10% glycerol, upper surface of each membrane. The cells on the lower 1% Triton X-100, 1 mM phenylmethylsulphonyl fluoride, 1 mM surface were stained with Diff-Quick (Siemens, Deerfield, IL, Na3VO4,10mg/ml leupeptin and 10 mg/ml aprotinin). For immuno- USA) and counted in 10 representative fields. Each condition precipitation, 3 mg of anti-Stat3 or anti-ARHI antibody was added 1 was assayed in triplicate and each experiment was repeated at to 2 mg of protein lysate. After overnight incubation at 4 C, 1 least three times. ARHI expression was induced with DOX for protein-G beads were added and incubated for 1.5 h at 4 C. 24 h before performing chemotaxis and hapotaxis assays, and After washing with wash buffer (10 mM Tris (pH 7.4), 150 mM DOX treatment was maintained throughout the assay. For NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM Na3VO4,1mM phenyl- treatment with Stat3 inhibitors or activators, cells were incu- methylsulphonyl fluoride, 0.5% Triton X-100 and 0.2% NP-40), the immunoprecipitated proteins were eluted for western blotting. bated with 100 mM AG490, a JAK2 inhibitor, for 2 h followed by 30-min incubation with AG490, AG490 plus IL-6, IL-6 plus dimethylsulfoxide or dimethylsulfoxide in serum-free medium. Western blot analysis For knockdown experiments, cells were transfected with siRNA Proteins were separated on 12 or 15% sodium dodecyl sulfate– and treated with DOX for 48 h; 1 Â 105 cells were inoculated into PAGE, transferred to polyvinylidene difluoride membranes the upper chamber and with complete medium in the bottom and subjected to western blotting using an ECL chemilumi-

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 78 nescence reagent (GE/Amersham, Piscataway, NJ, USA). quantified by ImageJ (Bongard, 2005). Briefly, cells in a region Protein band intensities were quantified using the Java-based of interest were encircled using the program’s freehand image-processing program ImageJ developed at the NIH. selection tool. The total fluorescence corresponding to staining for p-FAKY397 or total FAK was quantified and the average siRNA transfection fluorescence intensity was calculated by dividing the total A mixture of siRNA (100 nM final concentration) and the fluorescence intensity by the cell number. Dharmafect #4 transfection reagent was incubated for 20 min at room temperature. This mixture was then added to cells and Statistical analysis allowed to incubate for 48 h before cells were harvested for All experiments were repeated independently at least two protein expression measurements or for motility assays. times and the data have been expressed as the mean±s.e. Statistical analysis was performed using Student’s t-test Real-time quantitative reverse transcription–PCR (two-tailed). The criterion for statistical significance was Expression of Stat3-responsive genes was measured using real- taken as Po0.05 (two-sided). time quantitative reverse transcription–PCR. Total cDNA was synthesized using 2 mg of total RNA. The reverse transcriptase reaction was performed according to the manufacturer’s instruc- Conflict of interest tions using oligo(dT)16 and SuperScript II reverse transcriptase (Invitrogen) followed by SYBR Green RT–PCR (ABI Prism The authors declare no conflict of interest. 7000 Sequence Detection System; Applied Biosystems, Foster City, CA, USA). The primers for N-cadherin were 50-GCCTG CAGATTTTAAGGTGG-30 and 50-CTCTTGAGGAAAAG Acknowledgements GTCCCC-30, and for vimentin were 50-GAGAACTTTGCCG TTGAAGC-30 and 50-TTCAGGGAGGAAAAGTTTGG-30. mRNA levels were normalized with a concurrent determination for This work was supported by Grants NCI P01 CA064602 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA. 1RO1 CA135354-01 from the National Institutes of Health. We also acknowledge the support of the MD Anderson SPORE in Ovarian Cancer 5P50 CA83639 and the CCSG Transient transfection and RhoA GTPase activation assay shared resources funded, in part, by CA 5 P30 CA016672. 6 SKOv3-ARHI cells (1 Â 10 ) were transfected with 16 mgof Support was also provided by the Ovarian Cancer Research GFP-RhoA plasmid (Addgene plasmid 12 965; Cambridge, Fund through a Program Project Award. DBB was supported MA, USA) and harvested 48 h later. RhoA activity was by an Excellence Award from the Ovarian Cancer Research measured with a RhoA activation assay kit (Cell Biolabs, San Fund. AAA was supported by a Cancer Research UK Diego, CA, USA). Levels of active RhoA were determined by Clinician Scientist fellowship. We thank Jodie Polan for western blot analysis. excellent technical support with confocal microscopy and Jared Burks from the Flow Cytometry and Cell Imaging core Quantitation of band intensity and average fluorescence intensity laboratory for the live-cell time-lapse microscopy partially Band intensity from western blots was quantified by ImageJ supported by the MD Anderson Cancer Center CCSG NCI (Ferreira and Rasband, 2011). Fluorescence intensity was P30 CA16672.

References

Bast Jr RC, Hennessy B, Mills GB. (2009). The biology of ovarian Etienne-Manneville S, Hall A. (2002). Rho GTPases in cell biology. cancer: new opportunities for translation. Nat Rev Cancer 9: Nature 420: 629–635. 415–428. Feng W, Marquez RT, Lu Z, Liu J, Lu KH, Issa JP et al. (2008). Bongard M. (2005). Visual Guide IV: measuring cells with ImageJ, Imprinted tumor suppressor genes ARHI and PEG3 are the most http://naranja.umh.es/Batg/tutorials/VGIV-MeasuringCellsImageJ.pdf. frequently downregulated in human ovarian cancers by loss of Carragher NO, Frame MC. (2004). Focal adhesion and actin heterozygosity and promoter methylation. Cancer 112: 1489–1502. dynamics: a place where kinases and proteases meet to promote Ferreira T, Rasband W. (2011). The ImageJ User Guide; p 144, http:// invasion. Trends Cell Biol 14: 241–249. rsb.info.nih.gov/ij/docs/user-guide.pdf. Cheng GZ, Zhang WZ, Sun M, Wang Q, Coppola D, Mansour M Forcet C, Etienne-Manneville S, Gaude H, Fournier L, Debilly S, et al. (2008). Twist is transcriptionally induced by activation of Salmi M et al. (2005). Functional analysis of Peutz–Jeghers STAT3 and mediates STAT3 oncogenic function. J Biol Chem 283: mutations reveals that the LKB1 C-terminal region exerts a crucial 14665–14673. role in regulating both the AMPK pathway and the cell polarity. Chikumi H, Fukuhara S, Gutkind JS. (2002). Regulation of G protein- Hum Mol Genet 14: 1283–1292. linked guanine nucleotide exchange factors for Rho, PDZ-Rho- Geiger B, Bershadsky A. (2001). Assembly and mechanosensory GEF, and LARG by tyrosine phosphorylation: evidence of a role function of focal contacts. Curr Opin Cell Biol 13: 584–592. for focal adhesion kinase. J Biol Chem 277: 12463–12473. Gough DJ, Corlett A, Schlessinger K, Wegrzyn J, Larner AC, Levy Colomiere M, Findlay J, Ackland L, Ahmed N. (2009). Epidermal DE. (2009). Mitochondrial STAT3 supports Ras-dependent onco- growth factor-induced ovarian carcinoma cell migration is asso- genic transformation. Science 324: 1713–1716. ciated with JAK2/STAT3 signals and changes in the abundance and Huang M, Page C, Reynolds RK, Lin J. (2000). Constitutive activation localization of alpha6beta1 integrin. Int J Biochem Cell Biol 41: of stat 3 oncogene product in human ovarian carcinoma cells. 1034–1045. Gynecol Oncol 79: 67–73. Debidda M, Wang L, Zang H, Poli V, Zheng Y. (2005). A role of Jang AC, Starz-Gaiano M, Montell DJ. (2007). Modeling migration STAT3 in Rho GTPase-regulated cell migration and proliferation. and metastasis in Drosophila. J Mammary Gland Biol Neoplasia 12: J Biol Chem 280: 17275–17285. 103–114.

Oncogene ARHI suppresses ovarian cancer cell migration DB Badgwell et al 79 Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. (2011). Parsons JT. (2003). Focal adhesion kinase: the first ten years. J Cell Sci Global cancer statistics. CA Cancer J Clin 61: 69–90. 116: 1409–1416. Kaibuchi K, Kuroda S, Fukata M, Nakagawa M. (1999). Regulation Raftopoulou M, Etienne-Manneville S, Self A, Nicholls S, Hall A. of cadherin-mediated cell–cell adhesion by the Rho family GTPases. (2004). Regulation of cell migration by the C2 domain of the tumor Curr Opin Cell Biol 11: 591–596. suppressor PTEN. Science 303: 1179–1181. Kakinuma N, Roy BC, Zhu Y, Wang Y, Kiyama R. (2008). Kank Ridley AJ, Schwartz MA, Burridge K, Firtel RA, Ginsberg MH, regulates RhoA-dependent formation of actin stress fibers and Borisy G et al. (2003). Cell migration: integrating signals from front cell migration via 14-3-3 in PI3K–Akt signaling. J Cell Biol 181: to back. Science 302: 1704–1709. 537–549. Roger L, Gadea G, Roux P. (2006). Control of cell migration: a Katabuchi H, Okamura H. (2003). Cell biology of human ovarian tumour suppressor function for p53? Biol Cell 98: 141–152. surface epithelial cells and ovarian carcinogenesis. Med Electron Romer LH, Birukov KG, Garcia JG. (2006). Focal adhesions: Microsc 36: 74–86. paradigm for a signaling nexus. Circ Res 98: 606–616. Kurokawa K, Nakamura T, Aoki K, Matsuda M. (2005). Mechanism Rosen DG, Wang L, Jain AN, Lu KH, Luo RZ, Yu Y et al. and role of localized activation of Rho-family GTPases in growth (2004). Expression of the tumor suppressor gene ARHI in factor-stimulated fibroblasts and neuronal cells. Biochem Soc Trans epithelial ovarian cancer is associated with increased expression 33: 631–634. of p21WAF1/CIP1 and prolonged progression-free survival. Lidor YJ, Xu FJ, Martinez-Maza O, Olt GJ, Marks JR, Berchuck A Clin Cancer Res 10: 6559–6566. et al. (1993). Constitutive production of macrophage colony- Rustin GJ, van der Burg ME, Berek JS. (1993). Advanced ovarian stimulating factor and interleukin-6 by human ovarian surface cancer. Tumour markers. Ann Oncol 4(Suppl 4): 71–77. epithelial cells. Exp Cell Res 207: 332–339. Schaller MD. (2001). Biochemical signals and biological responses Lim Y, Lim ST, Tomar A, Gardel M, Bernard-Trifilo JA, Chen XL elicited by the focal adhesion kinase. Biochim Biophys Acta 1540: et al. (2008). PyK2 and FAK connections to p190Rho guanine 1–21. nucleotide exchange factor regulate RhoA activity, focal adhesion Schlaepfer DD, Mitra SK. (2004). Multiple connections link FAK to formation, and cell motility. J Cell Biol 180: 187–203. cell motility and invasion. Curr Opin Genet Dev 14: 92–101. Liotta LA, Wewer U, Rao NC, Schiffmann E, Stracke M, Guirguis R Silver DL, Naora H, Liu J, Cheng W, Montell DJ. (2004). Activated et al. (1987). Biochemical mechanisms of tumor invasion and signal transducer and activator of transcription (STAT)3: localiza- metastasis. Anticancer Drug Des 2: 195–202. tion in focal adhesions and function in ovarian cancer cell motility. Long W, Yi P, Amazit L, LaMarca HL, Ashcroft F, Kumar R et al. Cancer Res 64: 3550–3558. (2010). SRC-3Delta4 mediates the interaction of EGFR with FAK Takeda K, Noguchi K, Shi W, Tanaka T, Matsumoto M, Yoshida N to promote cell migration. Mol Cell 37: 321–332. et al. (1997). Targeted disruption of the mouse Stat3 gene leads Lu Z, Luo RZ, Lu Y, Zhang X, Yu Q, Khare S et al. (2008). The tumor to early embryonic lethality. Proc Natl Acad Sci USA 94: suppressor gene ARHI regulates autophagy and tumor dormancy 3801–3804. in human ovarian cancer cells. J Clin Invest 118: 3917–3929. Tomar A, Schlaepfer DD. (2009). Focal adhesion kinase: switching Luo RZ, Fang X, Marquez R, Liu SY, Mills GB, Liao WS et al. between GAPs and GEFs in the regulation of cell motility. Curr (2003). ARHI is a Ras-related small G-protein with a novel N- Opin Cell Biol 21: 676–683. terminal extension that inhibits growth of ovarian and breast Van Aelst L, D’Souza-Schorey C. (1997). Rho GTPases and signaling cancers. Oncogene 22: 2897–2909. networks. Genes Dev 11: 2295–2322. Marcoux N, Vuori K. (2005). EGF receptor activity is essential for Vij N, Sharma A, Thakkar M, Sinha S, Mohan RR. (2008). PDGF- adhesion-induced stress fiber formation and cofilin phosphoryla- driven proliferation, migration, and IL8 chemokine secretion in tion. Cell Signal 17: 1449–1455. human corneal fibroblasts involve JAK2–STAT3 signaling path- Martinez-Maza O, Berek JS. (1991). Interleukin 6 and cancer way. Mol Vis 14: 1020–1027. treatment. In vivo 5: 583–588. Yamashita S, Miyagi C, Carmany-Rampey A, Shimizu T, Fujii R, Naora H, Montell DJ. (2005). Ovarian cancer metastasis: integrating Schier AF et al. (2002). Stat3 controls cell movements during insights from disparate model organisms. Nat Rev Cancer 5: 355–366. zebrafish gastrulation. Dev Cell 2: 363–375. Nishimoto A, Yu Y, Lu Z, Mao X, Ren Z, Watowich SS et al. (2005). Yu Y, Luo R, Lu Z, Wei Feng W, Badgwell D, Issa JP et al. (2006). A Ras homologue member I directly inhibits signal transducers and Biochemistry and biology of ARHI (DIRAS3), an imprinted tumor activators of transcription 3 translocation and activity in human suppressor gene whose expression is lost in ovarian and breast breast and ovarian cancer cells. Cancer Res 65: 6701–6710. cancers. Methods Enzymol 407: 455–468. Okamura H, Katabuchi H. (2005). Pathophysiological dynamics of Yu Y, Xu F, Peng H, Fang X, Zhao S, Li Y et al. (1999). NOEY2 human ovarian surface epithelial cells in epithelial ovarian (ARHI), an imprinted putative tumor suppressor gene in ovarian carcinogenesis. Int Rev Cytol 242: 1–54. and breast carcinomas. Proc Natl Acad Sci USA 96: 214–219. Orr FW, Wang HH. (2001). Tumor cell interactions with the Zhai J, Lin H, Nie Z, Wu J, Canete-Soler R, Schlaepfer WW et al. microvasculature: a rate-limiting step in metastasis. Surg Oncol (2003). Direct interaction of focal adhesion kinase with p190Rho- Clin N Am 10: 357–381, ix-x. GEF. J Biol Chem 278: 24865–24873. Orr FW, Wang HH, Lafrenie RM, Scherbarth S, Nance DM. (2000). Zhong Z, Wen Z, Darnell Jr JE. (1994). Stat3: a STAT family member Interactions between cancer cells and the endothelium in metastasis. activated by tyrosine phosphorylation in response to epidermal J Pathol 190: 310–329. growth factor and interleukin-6. Science 264: 95–98.

Oncogene