βArrestin1 Regulates the Guanine Nucleotide Exchange Factor RasGRF2 Expression and the Small GTPase Rac-mediated Formation of Membrane Protrusion and Cell Motility*

Background: G protein-coupled receptors (GPCRs) and βarrestins have been shown to regulate cell motility. Results: βArrestin1 regulates cell migration through RasGRF2 (a dual guanine nucleotide exchange factor) gene expression and the small GTPase Rac activity. Conclusion: βArrestin1 may regulate cellular functions at the gene expression level. Significance: βArrestins may function at transcriptional and post-translational levels to regulate cell migration. βArrestin proteins shuttle between the cytosol and nucleus and have been shown to regulate G protein-coupled receptor signaling, actin remodeling, and gene expression. Here, we tested the hypothesis that βarrestin1 regulates actin remodeling and cell migration through the small GTPase Rac. Depletion of βarrestin1 promotes Rac activation, leading to the formation of multipolar protrusions and increased cell circularity, and overexpression of a dominant negative form of Rac reverses these morphological changes. Small interfering RNA library screen identifies RasGRF2 as a target of βarrestin1. RasGRF2 gene and protein expression levels are elevated following depletion of βarrestin1, and the consequent activation of Rac results in dephosphorylation of cofilin that can promote actin polymerization and formation of multipolar protrusions, thereby retarding cell migration and invasion. Together, these results suggest that βarrestin1 regulates rasgrf2 gene expression and Rac activation to affect membrane protrusion and cell migration and invasion.

␤Arrestin proteins shuttle between the cytosol and nucleus and have been shown to regulate G protein-coupled receptor signaling, actin remodeling, and gene expression. Here, we tested the hypothesis that ␤arrestin1 regulates actin remodeling and cell migration through the small GTPase Rac. Depletion of ␤arrestin1 promotes Rac activation, leading to the formation of multipolar protrusions and increased cell circularity, and overexpression of a dominant negative form of Rac reverses these morphological changes. Small interfering RNA library screen identifies RasGRF2 as a target of ␤arrestin1. RasGRF2 gene and protein expression levels are elevated following depletion of ␤arrestin1, and the consequent activation of Rac results in dephosphorylation of cofilin that can promote actin polymerization and formation of multipolar protrusions, thereby retarding cell migration and invasion. Together, these results suggest that ␤arrestin1 regulates rasgrf2 gene expression and Rac activation to affect membrane protrusion and cell migration and invasion.
Cell migration and invasion are multistep processes that require actin cytoskeleton remodeling and the establishment and maintenance of cell polarity (1). Rac is a Rho GTPase family member that regulates actin remodeling and protrusive structures at the leading edge of migrating cells. Activation of Rac through GTP binding is catalyzed by guanine nucleotide exchange factors (GEFs), 2 which can be divided into two subfamilies as follows: those that contain the Dbl homology-pleckstrin homology and those that contain Dock homology region domains (2,3). RasGRF2 is a multidomain protein that has dual Ras GEF and Rac GEF activities. The Cdc25 homology domain at the C terminus catalyzes exchange of GDP to GTP on Ras family proteins, whereas the Dbl homology-pleckstrin homology domain functions as a GEF for Rac proteins (4). RasGRF2 was initially identified in neuronal tissues, and its known functions in the nervous system include regulation of NMDA receptor signaling and synaptic plasticity (5). Although mechanisms underlying the restricted tissue distribution of RasGRF2 are not fully understood, recent studies have suggested that RasGRF2 expression is suppressed in different human cancers, implying a tumor suppressive role (4), but by as yet unclear mechanisms.
␤Arrestins, including ␤arrestin1 (␤Arr1 also known as Arr2) and ␤Arr2 (also known as Arr3), were initially characterized based on their ability to regulate G protein-coupled receptor (GPCR) signaling (6), and they have been implicated in human cancers (7)(8)(9)(10). Emerging evidence shows that ␤Arr1 and ␤Arr2 also function as adaptors to transduce signals and to regulate a wide array of cellular functions, including actin remodeling and cell migration (11,12). One mechanism by which ␤Arrs regulate actin remodeling and cell migration is through regulation of monomeric G proteins, including RhoA, Cdc42, and RalA. For example, activation of angiotensin II type 1A receptor resulted in RhoA activation that is dependent upon ␤Arr1, leading to stress-fiber formation (13). The type III transforming growth factor receptor, whose expression is lost in many human cancers, inhibits cell migration through ␤Arr2mediated activation of Cdc42 (14). ␤Arr1 was also shown to form a complex with Ral-guanine dissociation stimulator in the cytosol, and upon activation of formylmethionylleucylphenylalanine or lysophosphatidic acid receptors, RalA was activated resulting in actin remodeling and cell migration (15). GPCRs and ␤Arrs have been shown to regulate Rho GTPases through either RhoGEFs or RhoGAPs (GTPase-activating proteins). Activation of the angiotensin II type 1A receptor facilitated association of ␤Arr1 with the RhoGAP ARHGAP21, and consequent inhibition of the GTPase-activating proteins contributed to higher RhoA activity (16). Angiotensin II type 1A receptor was also shown to mediate RhoA activation through both ␤Arr1 and G q (13), as well as through JAK2-mediated phosphorylation of p115RhoGEF (17). We recently reported that ␤Arr2 formed a complex with p115RhoGEF in the cytosol and that activation of the ␤ 2 adrenergic receptor resulted in membrane translocation and activation of p115RhoGEF and RhoA (18).
Here, we report the regulation of RasGRF2 expression by ␤Arr1. Depletion of ␤Arr1 resulted in higher expression of RasGRF2 and higher Rac activity, leading to the formation of multipolar protrusions and cell rounding. Consistent with increased actin polymerization, depletion of ␤Arr1 was associated with lower levels of cofilin phosphorylation, reflecting higher cofilin activity. Knockdown of RasGRF2 expression with siRNA increased cofilin phosphorylation in ␤Arr1 Ϫ/Ϫ , but not ␤Arr1 ϩ/ϩ , mouse embryonic fibroblasts (MEFs), which was reversed by Rac inhibition. Overexpression of a RasGRF2 mutated form with a deleted Ras GEF domain was sufficient to cause cofilin dephosphorylation, supporting the idea that higher Rac activity is responsible for the ␤Arr1-dependent regulation of cofilin. Chromatin immunoprecipitation results indicated that ␤Arr1 bound to the promoter region of rasgrf2, and treatment with the demethylating agent decitabine enhanced rasgrf2 gene expression. In agreement with the formation of multipolar protrusions and loss of unipolarity, depletion of ␤Arr1 resulted in the reduced rate of directional cell migration and invasion. Therefore, ␤Arr1 regulates Rac activity and membrane protrusions through, at least in part, RasGRF2-related mechanisms.
Transfection and Immunofluorescence-␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs (kind gift from Dr. R. J. Lefkowitz) were transfected using GenJet (SignaJen), and RCC7 cells were transfected using PEI (Polysciences Inc.). Gene knockdown using SMARTpool siRNAs (Dharmacon) targeting specific Rac GEFs was performed using Lipofectamine RNAiMax (Invitrogen). Stable knockdown of ␤Arr1 in RCC7 cells was achieved by transfection of shRNA constructs (Open Biosystems) in lentiviral pLKO vector plus an equal concentration of vesicular stomatitis virus G and ⌬8.9 vector into packaging HEK293T cells for 24 and 48 h. Lentivirus containing medium was harvested, mixed with Polybrene, and used to infect RCC7 cells. The infected polyclonal cells were selected with 2 g/ml puromycin for 2 weeks. To restore ␤Arr1 expression, ␤Arr1 Ϫ/Ϫ MEFs were infected with adenoviruses encoding ␤Arr1, and infection with adenoviruses encoding RFP was used as a control.
For immunofluorescence staining, cells were trypsinized and replated onto fibronectin-coated coverslips, incubated in Opti-MEM or other medium (as indicated) for 6 h, and fixed with 2% formaldehyde. Actin cytoskeleton was visualized by staining with rhodamine-conjugated phalloidin. Slides were examined using an epifluorescence microscope (DM 6000B, Leica) equipped with a ϫ63/1.4-0.6 oil immersion lens or a Leica confocal microscope (TCS SP5) equipped with ϫ63/1.4 NA oil immersion lens. Images were captured and analyzed using the Volocity software 5.5 (PerkinElmer Life Sciences) or the application suite Advanced Fluorescence 2.0.2 software (Leica). For protrusion numbers, at least 100 control or knock-out MEFs were counted for each assay, and the experiments were repeated three times. For circularity measurement, the short axis and the long axis of each MEF were measured, and the circularity was expressed as the quotient of the short axis divided by the long axis. Hence, the lower value reflects elongated morphology, and the higher value indicates cell rounding.
GST Pulldown-GST, GST-CRIB (Cdc42/Rac-interactive binding domain of PAK), and GST-RBD (Rho binding domain of rhotekin) fusion proteins were expressed in BL21 cells. After induction with isopropyl 1-thio-␤-D-galactopyranoside, cells were harvested by centrifugation and lysed in 1% Triton X-100 in PBS with protease inhibitors using a French pressure cell press. Cell lysates were centrifuged at 100,000 ϫ g at 4°C for 1 h, and the supernatants were incubated with glutathioneconjugated agarose beads at 4°C for 1 h followed by washing with PBS. Fresh cell lysates (in 20 mM Tris, pH 8.0, 100 mM NaCl, 10% (v/v) glycerol, 1% (v/v) Triton X-100, 1 mM EDTA, 5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, and 2 g/ml pepstatin A) were incubated with GST fusion proteins bound to beads with rotation at 4°C for 1 h. The beads were washed three times and boiled in SDS-PAGE sample buffer. Precipitated proteins (i.e. GTP-bound Rac, Cdc42, or RhoA) were determined by Western blot.
Cell Migration and Invasion-Cell migration and invasion were examined using Boyden chambers. MEFs or renal cell carcinoma (RCC) cells were starved overnight in 0.2% (v/v) FBS medium and then detached using HyQTase (Thermo). A total of 2.5 ϫ 10 4 MEFs or Caki-1 cells and 1.0 ϫ 10 5 SN12C or RCC7 cells in 100 l of medium containing 0.2% FBS were loaded into the top chamber. After 3 h of seeding, medium containing 0.5% FBS or PDGF (10 ng/ml) was added into the bottom chamber. For invasion assay, 1 mg/ml Matrigel (BD Biosciences) was thawed at 4°C overnight and added onto the chamber filter at 37°C for 1 h. Cell migration was terminated at 8 h for MEFs and 18 h for RCC, and cell invasion proceeded for 24 h. Cells were fixed and stained with Diff-Quik staining solution (Siemens).
Cells that remained in the top chamber were removed using a cotton swab, and the cells on the underside of the filter were photographed in four randomly chosen fields using the Leica DMI 4000B inverted microscope and counted using ImageJ software.
Real Time PCR-Decitabine (Sigma) treatment was performed as follows. Cells were cultured at low density for 24 h before decitabine was added at 1 M for an additional 24 h. Cells were washed three times with PBS, and fresh medium was added for an additional 24 h. A total of three decitabine treatment cycles were performed prior to harvesting cells for quantitative RT-PCR. Total RNA was isolated using TRIzol (Invitrogen), and 2 g of RNA was reverse-transcribed to generate cDNA using Superscript III (Invitrogen). Real time PCR was performed on Bio-Rad iQ5 thermal iCycler detection system using iQ TM SYBR Green SuperMix (Bio-Rad). Relative expression levels were calculated using the ⌬⌬C t method, and actin was used as the internal control (7). Expression level of target genes was presented relative to the expression in control cells. The sense and antisense primers used were as follows: Chromatin Immunoprecipitation (ChIP)-ChIP assay was performed using EZ ChIP kit (Millipore) following the manufacturer's protocol. Briefly, both ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs were treated with 1% formaldehyde to cross-link the DNA and proteins. Glycine (1.25 M) was added into the medium to quench formaldehyde. Cells were lysed and sonicated (Covaris E210, set up to generate 400-bp DNA fragments) to generate cross-linked DNA fragments with an average size of 400 bp. Immunoprecipitation was done by incubating chromatin with A1CT antibody at 4°C overnight. ChIP complex was eluted with elution buffer (1% SDS, 0.1 M NaHCO 3 ), and the cross-link was reversed using 5 M NaCl. DNA was purified using spin columns, and PCR amplification was performed using the purified DNA template and hot start Taq polymerase (Qiagen). The primers used were as follows: rasgrf2 promoter forward 5Ј-ATA GGC GCC CTA GGT CTG G-3Ј and rasgrf2 promoter reverse 5Ј-GAG TCG TAG CCT CAG CTT CTT G; p21 promoter forward 5Ј-CCA GAG GAT ACC TTG CAA GGC-3Ј and p21 promoter reverse 5Ј-TCT CTG TCT CCA TTC ATG CTC CTC C-3Ј.
Statistical Analysis-Experiments were repeated at least three times, and data are expressed as mean Ϯ S.E. Statistical analysis was performed using Student's t test or one-way analysis of variance with Tukey's post hoc test. Graphs were generated using Prism (version 5.0), and axis labels were generated using Adobe Illustrator (version CS5) software, respectively.

RESULTS
␤Arr1 Regulates Cell Morphology and Rac Activity-Deletion of ␤Arr1 resulted in morphological changes of MEFs (18). As shown in Fig. 1A, panels a and b, ␤Arr1 ϩ/ϩ MEFs were elongated with monopolar protrusions, whereas ␤Arr1 Ϫ/Ϫ MEFs were less elongated and more flattened with multipolar protrusions (Fig. 1A, panels c and d). Quantification of protrusion numbers indicated that there were roughly three protrusions per ␤Arr1 ϩ/ϩ MEF and six protrusions per ␤Arr1 Ϫ/Ϫ MEF (Fig. 1B). We measured cell dimensions and found that ␤Arr1 ϩ/ϩ MEFs exhibited low circularity in comparison with the ␤Arr1 Ϫ/Ϫ MEFs (Fig. 1C), suggesting that deletion of ␤Arr1 resulted in loss of unipolar morphology in a manner comparable with the loss of polarity of migratory T cells upon activation of T cell receptors (19). Similar results were obtained when MEFs were seeded on noncoated surfaces or cultured in the presence or absence of serum (data not shown). Hence, these morphological alterations are likely acquired properties irrespective of serum growth factor receptor or integrin receptor activation.
As membrane protrusions are regulated, at least in part, by Rho GTPases, we measured the level of active Rho GTPase in MEFs. Knock-out of ␤Arr1 did not affect the average level of Cdc42⅐GTP ( Fig. 1, D and E) with no noticeable effect on RhoA⅐GTP levels, as we have reported previously (18). Distinctly, ␤Arr1 Ϫ/Ϫ MEFs showed elevated levels of Rac⅐GTP (Fig. 1D), and densitometry analysis indicated that deletion of ␤Arr1 increased Rac activity by roughly 3-fold (Fig. 1F). To test the idea that the increased Rac activity was a direct result of ␤Arr1 loss of expression, we restored expression of ␤Arr1 in ␤Arr1 Ϫ/Ϫ MEFs through infection with ␤Arr1-encoding adenovirus (Fig. 1G). Rac activity decreased in MEFs with rescued ␤Arr1 expression, as compared with ␤Arr1 Ϫ/Ϫ MEFs infected with control RFP-encoding adenoviruses (Fig. 1H).
To provide further evidence that Rac activity increases upon depletion of ␤Arr1, we compared the spreading of ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs on fibronectin-coated surfaces. ␤Arr1 ϩ/ϩ MEFs were elongated with membrane ruffles after 30 min of spreading (Fig. 1I, panels a and b, small arrows), with very few MEFs forming lamellipodia (Fig. 1I, panel b, big arrow). In contrast, most of the ␤Arr1 Ϫ/Ϫ MEFs formed lamellipodia at 30 min of spreading (Fig. 1I, panels c and d), suggesting higher Rac activity following deletion of ␤Arr1.
As Cdc42 also plays critical roles in the regulation of cell morphology, we examined whether overexpression of dominant negative Cdc42 affects ␤Arr1 Ϫ/Ϫ MEFs. Following overexpression of GFP-(T17N)Cdc42, ␤Arr1 Ϫ/Ϫ MEFs remained rounded with multiple protrusions (Fig. 2F, panels a-d), suggesting that Cdc42 is less likely to be involved in the regulation of ␤Arr1 Ϫ/Ϫ MEF morphology.
␤Arr1/Rac Signaling Negatively Regulates Activity of Cofilin-Cofilin is an actin-binding and -severing protein that increases the number of free barbed ends to promote actin polymerization and formation of actin-rich protrusions. Cofilin activity is regulated by phosphorylation; phosphorylation of the serine 3 residue inhibits cofilin's actin-severing capacity, although its dephosphorylation results in the cofilin activation. ␤Arr1 Ϫ/Ϫ MEFs formed multipolar protrusions, suggesting a high rate of actin polymerization. Accordingly, knock-out of ␤Arr1 resulted in dephosphorylation of cofilin (Fig. 3A). Densitometry analysis revealed that there was a 75% Cofilin is known to be activated by, among others, Rac (22). Higher activities for both Rac and cofilin in ␤Arr1 Ϫ/Ϫ MEFs suggest that they function in the same pathway following ␤Arr1 depletion to regulate actin polymerization. We compared the effects of platelet-derived growth factor (PDGF) on cofilin dephosphorylation in ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs. Treatment with PDGF significantly increased Rac activity in ␤Arr1 ϩ/ϩ MEFs, and in ␤Arr1 Ϫ/Ϫ MEFs whose basal Rac activity is high, PDGF stimulation failed to further increase the Rac activation (Fig. 3B). The similar treatment with PDGF resulted in cofilin dephosphorylation in ␤Arr1 ϩ/ϩ MEFs (Fig. 3C). In ␤Arr1 Ϫ/Ϫ MEFs whose basal cofilin phosphorylation level is low, PDGF stimulation only slightly augmented the dephosphorylation signal (Fig. 3C). To further implicate Rac in cofilin phosphorylation, we transfected HEK293 cells with cDNAs encoding (WT)Rac, (T17N)Rac, or (Q61L)Rac (to overcome the low transfection efficiency of MEFs). Expression of constitutively active (Q61L)Rac resulted in dephosphorylation of cofilin in unstimulated cells, whereas no difference in cofilin dephosphorylation was observed in cells overexpressing (WT)Rac or (T17N)Rac (Fig. 3D). These results suggest that cofilin serves as a Rac effector.
To test the possibility that elevated Rac activity contributes to the lowered phosphorylation level of cofilin in ␤Arr1 Ϫ/Ϫ MEFs, we treated MEFs with NSC23766. Rac inhibition slightly increased the cofilin phosphorylation signal in ␤Arr1 ϩ/ϩ MEFs (Fig. 3, E and F) and elevated cofilin phosphorylation level in ␤Arr1 Ϫ/Ϫ MEFs to that of untreated ␤Arr1 ϩ/ϩ MEFs (Fig. 3, E  and F). These results link the higher Rac activity to increased cofilin activity in ␤Arr1 Ϫ/Ϫ MEFs.
To further support the idea that morphological characteristics of ␤Arr1 Ϫ/Ϫ MEFs are a direct result of ␤Arr1 loss of expression, we restored ␤Arr1 protein expression in ␤Arr1 Ϫ/Ϫ MEFs by transient transfection with HA-tagged ␤Arr1. Overexpression of ␤Arr1 altered morphology of ␤Arr1 Ϫ/Ϫ MEFs, which were now more elongated with fewer protrusions (Fig.  3H, panels a-d). In contrast, overexpression of ␤Arr1 protein did not affect morphology of ␤Arr1 ϩ/ϩ MEFs (data not shown). These data suggest that loss of ␤Arr1 expression results in higher Rac activity that functions, at least in part, through cofilin to regulate the cell morphology.
Depletion of ␤Arr1 Yields Higher rasgrf2 Gene Expression-Elevated Rac activity in ␤Arr1 Ϫ/Ϫ MEFs suggests that depletion of ␤Arr1 activates Rac GEFs or inactivates RacGAPs. We used a SMARTpool siRNA library that targets known Rac GEFs to screen for changes in membrane protrusion and circularity of ␤Arr1 Ϫ/Ϫ MEFs. Among the 23 Rac GEFs examined, knockdown of Tiam1, ARHGEF6, RasGRF2, and DOCK4 reduced protrusion numbers (Fig. 4A) and, concordantly, promoted elongation of the ␤Arr1 Ϫ/Ϫ MEFs (Fig. 4B) as determined by decreased cell circularity. Next, we performed real time PCR to examine whether gene levels of these four Rac GEFs were up-regulated in ␤Arr1 Ϫ/Ϫ MEFs. Results show that mRNA levels of tiam1 and arhgef6 were lower in ␤Arr1 Ϫ/Ϫ than in ␤Arr1 ϩ/ϩ MEFs (Fig. 4C). However, dock4 mRNA levels were about four times higher, and rasgrf2 mRNA levels were 120 times higher in ␤Arr1 Ϫ/Ϫ compared with ␤Arr1 ϩ/ϩ MEFs (Fig. 4C). Because rasgrf2 levels were the most changed among the ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs, we examined the RasGRF2 protein level and consistently observed it to be higher in ␤Arr1 Ϫ/Ϫ than in ␤Arr1 ϩ/ϩ MEFs (Fig. 4D). To rule out the possibility of cell type-specific effects, we measured RasGRF2 protein expression in human kidney cancer RCC7 cells with transient knockdown of ␤Arr1. Western blot results showed that RasGRF2 protein was hardly detectable in control RCC7 cells, and knockdown of ␤Arr1 significantly increased the RasGRF2 levels (Fig. 4E). Mirroring our observations with ␤Arr1 Ϫ/Ϫ MEFs (Fig. 3), the knockdown of ␤Arr1 in RCC7 cells also resulted in cofilin activation as reflected by its dephosphorylation (Fig. 4E).
␤Arr1 Regulates Cofilin Activity through RasGRF2-RasGRF2 is a dual exchange factor for both Rac and Ras, and therefore, the elevated expression of RasGRF2 may increase Ras activity, which promotes ERK phosphorylation. Basal ERK phosphorylation appeared higher in ␤Arr1 Ϫ/Ϫ than in ␤Arr1 ϩ/ϩ MEFs (Fig. 5A), consistent with the idea that higher RasGRF2 expression associates with concomitantly higher Ras activity. PDGF treatment elicited strong ERK phosphorylation that was slightly higher in ␤Arr1 Ϫ/Ϫ than in ␤Arr1 ϩ/ϩ MEFs (Fig. 5A).
We examined whether activation of Ras is required for dephosphorylation and activation of cofilin. To this end, we overexpressed the empty vector, FLAG-RasGRF2 (which activates both Ras and Rac) or FLAG-⌬Cdc25-RasGRF2 (which activates Rac only), and determined cofilin phosphorylation. As shown in Fig. 5B, both FLAG-RasGRF2 and FLAG-⌬Cdc25-RasGRF2 were expressed to similar levels. In cells with overexpression of either protein, the level of cofilin phosphorylation was lower than that in control empty vector-transfected cells (Fig. 5B). To ascertain that the overexpressed ⌬Cdc25-RasGRF2 is functional, we performed pulldown assays to measure active Rac. Western blot results indicated that overexpression of GFP-⌬Cdc25-RasGRF2 increased Rac⅐GTP levels by an average 3-fold (Fig. 5C). These results suggest that elevated Rac activity is sufficient to activate cofilin in ␤Arr1 Ϫ/Ϫ MEFs.
We asked if RasGRF2 is necessary for cofilin activation in ␤Arr1 Ϫ/Ϫ MEFs. Knockdown of endogenous RasGRF2 expression in ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs was achieved using siRNA and confirmed at the gene level by RT-PCR (Fig. 5D) and the protein level by Western blot (Fig. 5E). We next examined whether RasGRF2 knockdown would affect Rac activity using pulldown assays. Suppression of RasGRF2 expression lowered active Rac levels in ␤Arr1 Ϫ/Ϫ MEFs by about 2-fold. We then examined whether suppression of RasGRF2 expression affected cofilin phosphorylation. Western blot results indicated that the knockdown of RasGRF2 did not impact total cofilin expression in ␤Arr1 ϩ/ϩ MEFs (Fig. 5, F and G). However, the knockdown of RasGRF2 significantly increased cofilin phosphorylation levels in ␤Arr1 Ϫ/Ϫ MEFs (Fig. 5, F and G), suggesting that RasGRF2 is a required component for cofilin activation following depletion of ␤Arr1.

␤Arrestin1 Regulates Protrusion and Cell Motility
To provide direct evidence that up-regulated RasGRF2 expression is critical for the morphological properties of ␤Arr1 MEFs, we examined the effect of RasGRF2 knockdown. Control siRNA-transfected ␤Arr1 Ϫ/Ϫ MEFs appeared round with multiple protrusions (Fig. 5H, panels a and b), whereas ␤Arr1 Ϫ/Ϫ MEFs with RasGRF2 knockdown became elongated with fewer protrusions (Fig. 5H, panels c and d). These data support that Rac and cofilin regulate cell morphology of ␤Arr1 Ϫ/Ϫ MEFs.
We performed chromatin immunoprecipitation (ChIP) to test whether ␤Arr1 bound to the promoter region of the rasgrf2 gene. The binding of ␤Arr1 to the p21 gene promoter has been reported (23), and we included the p21 promoter as a positive control (Fig. 6A). ␤Arr1 bound to the rasgrf2 promoter region, and the ChIP signal was much weaker in ␤Arr1 Ϫ/Ϫ MEFs compared with ␤Arr1 ϩ/ϩ MEFs (Fig. 6A). The rasgrf2 promoter was reported to be heavily methylated in human cancer cells (24), and we examined whether promoter methylation contributed to the differential expression of the rasgrf2 gene. We treated ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs with the demethylating agent decitabine (i.e. 5-aza-2Ј-deoxycytidine) and determined the rasgrf2 mRNA levels using real time PCR. Basal rasgrf2 mRNA levels were higher in ␤Arr1 Ϫ/Ϫ than in ␤Arr1 ϩ/ϩ MEFs (Fig.  6B). Treatment with decitabine increased rasgrf2 mRNA levels by about 400-fold (over corresponding untreated samples) in  both ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs (Fig. 6B). These results suggest that ␤Arr1 regulates rasgrf2 gene expression, at least in part, through promoter DNA methylation.
␤Arr1 Regulates RCC Cell Migration and Invasion-Actin polymerization is crucial for formation of polarized protrusions that are required for the directional cell migration and invasion. We examined the effect of ␤Arr1 knockdown on the migration of several RCC cell lines using Transwells. The results revealed that knockdown of ␤Arr1 dramatically decreased directional migration of Caki-1 (Fig. 7A) and RCC7 (data not shown) cells. As migration constitutes a critical step of cell invasion, we examined the effects of ␤Arr1 knockdown on the cell invasion through a Matrigel matrix. Suppression of ␤Arr1 expression with siRNA significantly reduced invasion of RCC7 (Fig. 7B) A B ␤Arrestin1 Regulates Protrusion and Cell Motility and SN12C (Fig. 7C) cells. A similar effect was observed using ␤Arr1 knock-out MEFs, i.e. ␤Arr1 ϩ/ϩ MEFs invaded through Matrigel more efficiently than ␤Arr1 Ϫ/Ϫ MEFs (Fig. 7D). Next, we examined whether higher Rac activity contributes to the reduced cell migration following depletion of ␤Arr1 expression. Inhibition of Rac activity with NSC23766 did not affect migration of ␤Arr1 ϩ/ϩ MEFs (Fig. 7E). Distinctly, the inhibition of Rac activity increased the migration of ␤Arr1 Ϫ/Ϫ MEFs (Fig. 7E). These results suggest that deregulated high Rac activity contributes to the lowered rate of directional migration . ␤Arr1 regulates cell migration and invasion through Rac. A-C, ␤Arr1 was required for RCC cell migration and invasion. Transient knockdown of ␤Arr1 expression was performed using siRNA in Caki-1, RCC7, and SN12C cells and verified by Western blot, using actin as a loading control (lower panels below the bar graphs). Caki-1 cell migration was examined using Boyden chamber (A) and RCC7 (B), and SN12C (C) cell invasion was examined using Matrigel-coated filters. *, p Ͻ 0.05 versus siCON. D, deletion of ␤Arr1 inhibited MEF invasion. ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs were used for the cell invasion, and expression of ␤Arr1 was examined by Western blot using actin as a loading control (lower panels). *, p Ͻ 0.05 versus ␤Arr1 ϩ/ϩ MEFs. E, inhibition of Rac activity increases migration rate of ␤Arr1 Ϫ/Ϫ MEFs. ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs were treated with NSC23766 (NSC) (50 M) for 48 h, and migration toward a gradient of FBS was examined using Transwells. *, p Ͻ 0.05 versus untreated samples. F, knockdown of rasGRF2 increases migration rate of ␤Arr1 Ϫ/Ϫ MEFs. ␤Arr1 ϩ/ϩ and ␤Arr1 Ϫ/Ϫ MEFs were transfected with control or RasGRF2 siRNA, and cell migration was examined using Transwells. *, p Ͻ 0.05 versus control. G, schematic diagram of ␤Arr1-mediated inhibition of rasgrf2 gene expression. RasGrf2, a dual Ras/Rac guanine nucleotide exchange factor, activates Rac to promote actin polymerization through dephosphorylation and activation of cofilin. This leads to the formation of multiple protrusions and increased circularity of cells with a consequent decrease of directional cell migration. In the nucleus, ␤Arr1 inhibits the gene expression of rasgrf2 through, at least in part, enhanced methylation of the promoter region. Suppression of ␤Arr1 expression results in higher RasGRF2 and Rac activity.
of ␤Arr1 Ϫ/Ϫ MEFs. As our results suggest that higher Rac activity in ␤Arr1 Ϫ/Ϫ MEFs resulted from the up-regulated expression of RasGRF2, we examined possible effects of RasGRF2 knockdown. Suppression of RasGRF2 expression with siRNA did not impact migration of ␤Arr1 ϩ/ϩ MEFs (Fig. 7F) but increased the migration rate of ␤Arr1 Ϫ/Ϫ MEFs (Fig. 7F), consistent with the idea that increased expression of RasGRF2 leads to deregulated Rac activity that, in turn, contributes to the lowered cell migration.

DISCUSSION
␤Arrs control GPCR signaling through receptor desensitization and down-regulation, as well as perturbation of the second messenger level (25). In addition, ␤Arrs function as scaffolds to transduce signals from GPCRs, and the best described examples include activation of c-Src and ERK (6). Because of the absence of nuclear export signal, ␤Arr1 is also present in the nucleus (26), where it can regulate gene transcription. Indeed, ␤Arr1 was shown to increase the transcription of IL-1␤ (27), p27, and c-fos (28) genes. ␤Arr1 was also shown to negatively regulate transcription of ␥-interferon-responsive genes (29). Here, we observed inhibition of rasgrf2 gene expression by ␤Arr1; depletion of ␤Arr1 resulted in the up-regulation of ras-grf2 mRNA and protein levels. The results also show that ␤Arr1 binds to the promoter region of rasgrf2 gene, and treatment with the demethylating agent decitabine significantly increased rasgrf2 gene expression. Together, these results suggest that ␤Arr1 regulates cell morphology and migration through RasGRF2-related mechanisms.
The rasgrf2 gene was reported to be aberrantly methylated in the CpG island at the 5Ј region in human pancreatic cancer cells, and treatment with decitabine restored rasgrf2 gene expression (24). Similarly, reduced rasgrf2 gene or protein expression was observed in mammary carcinoma tissues (30), in non-small cell lung cancers (31), and in benign colorectal adenoma as well as invasive colon carcinomas (32). Restoration of RasGRF2 expression resulted in a 56% reduction of colony formation of the colorectal cancer cell line HCT-116 (32), and deletion of RasGRF2 accelerated development of lymphoblastic lymphoma-like tumors in mice (33), suggesting a tumor-suppressive role for RasGRF2. A recent study showed that ␤Arr1 functions as an E3 ligase adaptor in the nucleus to mediate p53 degradation and accumulation of DNA damage in response to chronic stress (23). Hence, nuclear ␤Arr1 can either suppress the expression or facilitate the degradation of tumor suppressors, which may contribute to human tumor initiation and progression.
RasGRF2 promotes exchange activities on both Ras and Rac GTPases (34); activation of either pathway can have profound effects on actin remodeling and cancer cell migration (35). A recent study showed that RasGRF2 regulates cancer cell invasion through direct binding to Cdc42 thereby sequestering it from association with cognate GEFs, and knockdown of RasGRF2 increased rounded invasion of melanoma cells (36). Here, we observed activation of Rac by RasGRF2 following depletion of ␤Arr1. It is not clear at present how RasGRF2 activates Rac in MEFs and RCC cell lines yet inhibits Cdc42 activation in melanoma cells (36).
Rac is known to promote actin polymerization at cell protrusions through the activation of, among others, the actin-severing protein cofilin (37) whose activity is tightly controlled by phosphorylation (38). Phosphorylation of serine 3 by LIM kinase inhibits cofilin binding to actin, and dephosphorylation of cofilin by slingshot or chronophin activates it (38). Our results show that depletion of ␤Arr1 results in cofilin dephosphorylation, which is a direct effect of up-regulated expression of RasGRF2 and consequent activation of Rac. This effect can be mimicked by overexpression of RasGRF2 or constitutively active Rac and, concordantly, the knockdown of RasGRF2 in ␤Arr1 Ϫ/Ϫ MEFs reversed this effect. As shown in Fig. 7G, we propose a model where ␤Arr1 suppresses rasgrf2 gene expression through, at least in part, enhanced methylation of the promoter region. Knock-out of ␤Arr1 results in higher RasGRF2 and Rac activity. Treatment with the demethylation agent decitabine increased rasgrf2 gene expression in ␤Arr1 knockout cells. Decitabine also increased rasgrf2 expression level in wild-type cells, implying additional factors are involved in the methylation of rasgrf2 gene promoter. It can be envisioned that activated Rac leads to activation of slingshot family phosphatase to dephosphorylate and activate cofilin, as has been reported for keratinocyte migration (39). It is also worth mentioning that Rac was reported to inactivate cofilin through LIM kinase-mediated phosphorylation (40,41). Therefore, additional regulatory mechanisms are likely involved in the finetuning of cofilin activity.
␤Arr2 was reported to function as a scaffold for cofilin and the phosphatases slingshot and chronophin in HEK293, leukocytes, and breast cancer cells (42)(43)(44). This multiprotein complex formation facilitated cofilin dephosphorylation, actin polymerization, and protrusion formation. Depletion of ␤Arr2 resulted in cell migration defects (43). Here, we showed that ␤Arr1 regulates rasgrf2 expression at the transcription level with a consequent effect on Rac and cofilin activities and cell migration. Therefore, ␤Arr1 and ␤Arr2 may function coordinately at the transcriptional and post-translational levels to regulate cofilin activity, actin remodeling, and cell migration.
In summary, our results show that ␤Arr1 suppressed rasgrf2 gene expression presumably through promoter hypermethylation. Depletion of ␤Arr1 resulted in up-regulated expression of RasGRF2 with consequent activation of Rac and cofilin and concomitant formation of multipolar protrusions and cell rounding, leading to reduced cell migration and invasion. Inhibitory effects on cell migration and invasion may contribute to the putative tumor-suppressive role for RasGRF2, and expression or function of nuclear ␤Arr1 may be targeted to restore expression of RasGRF2 for human cancer therapy.