Intramolecular Interactions between the Dbl Homology (DH) Domain and the Carboxyl-terminal region of Myosin II-interacting Guanine Nucleotide Exchange Factor (MyoGEF) Act as an Autoinhibitory Mechanism for the Regulation of MyoGEF Functions*

Background: MyoGEF is implicated in regulating cytokinesis and breast cancer cell invasion. Results: Binding of the MyoGEF carboxyl-terminal region to its DH domain suppresses breast cancer cell invasion and interferes with cytokinesis. Conclusion: Intramolecular interactions of MyoGEF act as an autoinhibitory mechanism to regulate MyoGEF functions. Significance: Autoinhibitory intramolecular interactions of MyoGEF serve as a control point to regulate cytokinesis and breast cancer cell invasion. We have reported previously that nonmuscle myosin II-interacting guanine nucleotide exchange factor (MyoGEF) plays an important role in the regulation of cell migration and cytokinesis. Like many other guanine nucleotide exchange factors (GEFs), MyoGEF contains a Dbl homology (DH) domain and a pleckstrin homology domain. In this study, we provide evidence demonstrating that intramolecular interactions between the DH domain (residues 162–351) and the carboxyl-terminal region (501–790) of MyoGEF can inhibit MyoGEF functions. In vitro and in vivo pulldown assays showed that the carboxyl-terminal region (residues 501–790) of MyoGEF could interact with the DH domain but not with the pleckstrin homology domain. Expression of a MyoGEF carboxyl-terminal fragment (residues 501–790) decreased RhoA activation and suppressed actin filament formation in MDA-MB-231 breast cancer cells. Additionally, Matrigel invasion assays showed that exogenous expression of the MyoGEF carboxyl-terminal region decreased the invasion activity of MDA-MB-231 cells. Moreover, coimmunoprecipitation assays showed that phosphorylation of the MyoGEF carboxyl-terminal region by aurora B kinase interfered with the intramolecular interactions of MyoGEF. Furthermore, expression of the MyoGEF carboxyl-terminal region interfered with RhoA localization during cytokinesis and led to an increase in multinucleation. Together, our findings suggest that binding of the carboxyl-terminal region of MyoGEF to its DH domain acts as an autoinhibitory mechanism for the regulation of MyoGEF activation.


We have reported previously that nonmuscle myosin II-interacting guanine nucleotide exchange factor (MyoGEF) plays an important role in the regulation of cell migration and cytokinesis. Like many other guanine nucleotide exchange factors (GEFs), MyoGEF contains a Dbl homology (DH) domain and a pleckstrin homology domain. In this study, we provide evidence demonstrating that intramolecular interactions between the DH domain (residues 162-351) and the carboxyl-terminal region (501-790) of MyoGEF can inhibit MyoGEF functions. In vitro and in vivo pulldown assays showed that the carboxyl-terminal region (residues 501-790) of MyoGEF could interact with the DH domain but not with the pleckstrin homology domain. Expression of a MyoGEF carboxylterminal fragment (residues 501-790) decreased RhoA activation and suppressed actin filament formation in MDA-MB-231 breast cancer cells. Additionally, Matrigel invasion assays showed that exogenous expression of the MyoGEF carboxyl-terminal region decreased the invasion activity of MDA-MB-231 cells. Moreover, coimmunoprecipitation assays showed that phosphorylation of the MyoGEF carboxyl-terminal region by aurora B kinase interfered with the intramolecular interactions of MyoGEF. Furthermore, expression of the MyoGEF carboxyl-terminal region interfered with RhoA localization during cytokinesis and led to an increase in multinucleation. Together, our findings suggest that binding of the carboxyl-terminal region of MyoGEF to its DH domain acts as an autoinhibitory mechanism for the regulation of MyoGEF activation.
There are approximately 80 guanine nucleotide exchange factors (GEFs) 2 in humans. They belong to a family of signaling proteins that play a critical role in the regulation of a variety of physiological processes, such as cell migration, cell cycle progression, membrane trafficking, angiogenesis, morphogenesis, apoptosis, and gene expression (1)(2)(3). GEFs generally execute their biological functions through activation of overlapping downstream Rho GTPase proteins such as RhoA, Rac1, and Cdc42 (2,4). Rho GTPase proteins exist in active, GTP-bound forms or in inactive, GDP-bound forms. GEFs activate Rho GTPase proteins by enhancing the exchange of bound GDP for GTP (5). Dysfunctional regulation of Rho GTPase signaling has been associated with numerous human diseases and disorders (6). Both clinical and animal studies have demonstrated that Rho GTPase signaling plays an important role in promoting tumor progression and metastasis (7).
It is generally thought that the invasion of tumor cells into the surrounding tissues is a prerequisite for tumor metastasis (8). Therefore, the roles of Rho GTPase proteins in the regulation of cell migration and invasion have been investigated intensively (9). Activation of Rho GTPase proteins leads to the remodeling of cytoskeleton organization that plays a central role in the regulation of cell migration and invasion (10). Accordingly, it has been demonstrated that RhoA and RhoC are key promoters of breast cancer invasion and metastasis (11,12). Consistent with the fact that numerous RhoGEFs exist in humans (1), multiple RhoGEFs, such as leukemia-associated RhoGEF (LARG), PDZ-RhoGEF, and myosin II-interacting GEF (MyoGEF), are implicated in promoting breast cancer cell invasion through activation of RhoA and/or RhoC (13)(14)(15).
Rho GTPase signaling also plays a central role in the regulation of cytokinesis. The localization and activation of the Rho GTPase protein RhoA at the cleavage furrow are essential for the assembly of the actomyosin contractile ring during cytoki-nesis (16 -18). RhoA can activate Rho kinase, which, in turn, inhibits myosin phosphatase and directly phosphorylates myosin light chain, hence increasing myosin contractile activity (19,20). It is generally thought that Ect2, a GEF for RhoA, is a key activator of RhoA at the cleavage furrow during cytokinesis (18,(21)(22)(23). Nonetheless, several other RhoGEFs, such as GEF-H1, LARG, and MyoGEF, have also been implicated in regulating cytokinesis (24 -26). Furthermore, a line of evidence has demonstrated that two mitotic kinases, aurora B and polo-like kinase (Plk1), play a central role in mitotic progression and cytokinesis (27,28). Aurora B and Plk1 can activate Ect2 and MyoGEF during cytokinesis and localize both proteins to the central spindle (29 -32).
GEFs invariably contain at least a Dbl homology (DH) domain and a pleckstrin homology (PH) domain (1). The DH domain is the catalytic unit of the GEFs. It can catalyze the dissociation of GDP from the inactive, GDP-bound forms of Rho GTPase proteins, thus stimulating the exchange of GDP for GTP and promoting the formation of the active, GTPbound forms of GTPase proteins (33,34). The PH domain is adjacent and carboxyl-terminal to the DH domain. The tandem DH and PH domains often form a functional unit to catalyze the guanine nucleotide exchange activity (34 -36). In addition, the PH domain can bind to phospholipids and participate in protein-protein interactions, thus targeting GEFs to proper subcellular locations such as the cell membrane and cytoskeleton (34). Activation of GEFs can be achieved through various modes of regulation, such as protein-protein interactions, associations with the membrane and cytoskeleton, and relief of autoinhibitory intramolecular interactions (1,34,(37)(38)(39). It has been demonstrated that intramolecular interactions between the DH domain and the amino-or carboxyl-terminal regions of GEFs can inhibit the catalytic activity of the DH domain (40 -42). Various molecular events, such as protein-protein interactions and posttranslational modifications, can disrupt such intramolecular interactions and result in the activation of GEFs (40,41). Intermolecular interactions have also been implicated in the regulation of GEFs. However, unlike intramolecular interactions, intermolecular interactions of GEFs can be either activatory or inhibitory. For instance, oligomerization of Dbl through the DH domain is required for the oncogenic activity of Dbl (43). On the other hand, hetero-and homo-oligomerization of PDZ-RhoGEF, LARG, and p115RhoGEF through their carboxyl-terminal regions lead to the inhibition of the respective GEFs (44,45).
We have reported previously that MyoGEF plays an important role in the regulation of cytokinesis and breast cancer cell invasion (15,24,31,32,46,47). MyoGEF is highly expressed in invasive breast cancer cells and tissues (15). Depletion of Myo-GEF decreases RhoA/RhoC activation and suppresses breast cancer cell invasion (15). We also found that phosphorylation of MyoGEF at Thr-544 by aurora B creates a docking site for Plk1, thus leading to the binding of Plk1 to MyoGEF (31). In turn, Plk1 phosphorylates MyoGEF at Thr-574 (32). Such sequential phosphorylation of MyoGEF by aurora B and Plk1 localizes MyoGEF to the central spindle and leads to MyoGEF activation during cytokinesis. In this study, we provide evidence demonstrating that intramolecular interactions between the DH domain and the carboxyl-terminal region of MyoGEF could serve as an autoinhibitory mechanism for the regulation of MyoGEF functions.

EXPERIMENTAL PROCEDURES
Cell Culture-MDA-MB-231, U2OS, and HeLa cells were purchased from the ATCC. MDA-MB-231 cells were grown in Leibovitz's L-15 medium supplemented with 10% fetal bovine serum. U2OS and HeLa cells were grown in DMEM supplemented with 10% fetal bovine serum. Lipofectamine was used for transfection with plasmids according to the instructions of the manufacturer (Invitrogen). For immunofluorescence analysis, the transfected cells were trypsinized 24 h after transfection and grown on fibronectin-coated coverglasses for an additional 3 h.
Expression and Purification of Recombinant Polypeptides-A bacterial expression system was used to express GST-and Histagged recombinant polypeptides. BL21 bacterial cells expressing GST-tagged recombinant polypeptides were resuspended in PBS and homogenized by sonication. Triton X-100 was added to a final concentration of 1%. The bacterial lysates were then incubated by shaking at 23°C for 1 h. A glutathione-conjugated agarose (Sigma) column was used to purify GST-tagged recombinant polypeptides. The proteins were eluted with 100 mM Tris-HCl (pH 8.0) and 10 mM glutathione and dialyzed against 50 mM Tris-HCl (pH 7.5) and 50 mM NaCl. BL21 bacterial cells expressing His-tagged recombinant polypeptides were resuspended in 50 mM sodium phosphate (pH 8.0) and 300 mM NaCl and homogenized by sonication. A nickel-nitrilotriacetic acid-agarose (Qiagen) column was used to purify Histagged recombinant polypeptides. The proteins were eluted with 50 mM sodium phosphate (pH 8.0), 300 mM NaCl, and 250 mM imidazole and dialyzed against 50 mM Tris-HCl (pH 7.5) and 50 mM NaCl.
GST Pulldown Assays-GST pulldown assays were done as described previously (47,50). Briefly, the immobilized GST-MyoGEF polypeptides were incubated with in vitro-translated Myc-MyoGEF fragments or transfected cell lysates overnight at 4°C. After washing four times with binding buffer (50 mM Tris-HCl (pH 7.4), 100 mM NaCl, 0.05% Triton-X-100, 10% glycerol, 0.2 mM EDTA, and 1 mM DTT), the beads were resuspended in SDS loading buffer to elute the bound proteins. In vitro-trans-lated Myc-MyoGEF fragments were synthesized using the TNT SP6 quick-coupled transcription/translation system (Promega, Madison, WI) according to the instructions of the manufacturer.
Coimmunoprecipitation Assays-5 g of purified His-or GST-MyoGEF fragments was mixed with the in vitro-translated MyoGEF fragments in BC100 buffer (20 mM HEPES (pH 7.9), 100 mM NaCl, 10% glycerol, 0.01% Nonidet P-40, and 1 mM DTT). The mixtures were precleared with protein A/G-agarose beads. The precleared mixtures were incubated with agaroseconjugated anti-Myc antibody overnight at 4°C. After three washes with BC100 buffer, the bound proteins were eluted with SDS loading buffer. To examine the effect of aurora B-and Plk1-mediated phosphorylation on intramolecular interactions of MyoGEF, 5 g of purified GST-tagged MyoGEF polypeptides was incubated with buffer alone or with 1 g of recombinant His-aurora B or His-Plk1 (Invitrogen) in kinase buffer (5 mM MOPS (pH 7.2), 2.5 mM ␤-glycerophosphate, 1 mM EGTA, 4 mM MgCl 2 , 0.05 mM DTT, and 250 M ATP). The reaction mixtures (50 l) were incubated at 30°C for 30 min. The treated GST-MyoGEF polypeptides were then used for coimmunoprecipitation assays.
Quantification of Actin Filaments-To assess the effect of the MyoGEF carboxyl-terminal region on actin filament formation, MDA-MB-231 cells transfected with the mCherry-MyoGEF-501-790 plasmid were stained with phalloidin, and the fluorescence intensities in transfected cells were quantitated using NIH ImageJ software (51). Briefly, the freehand selection tool was used to mark the cells of interest in the same field. Regions next to cells were also selected and served as a background. The corrected total cell fluorescence was calculated on the basis of the integrated density from transfected and surrounding untransfected cells as well as background readings: corrected total cell fluorescence ϭ integrated density Ϫ (area of selected cell ϫ mean fluorescence of background readings). A more than 2-fold decrease in the corrected total cell fluorescence in transfected cells compared with that in surrounding untransfected cells was used as the threshold for an indicator of decreased actin filament formation.
To assess the effect of the MyoGEF carboxyl-terminal region on actin filament formation induced by the MyoGEF aminoterminal region, we used NIH ImageJ software to count the actin filaments in transfected cells. Briefly, the line plot profile tool was used to examine the elongated actin stress fibers. Each peak in the line plot profile image represents a stress fiber. It should be noted that the lines were drawn perpendicular to the parallel actin bundles. Also, the areas containing large actin aggregates were avoided.
Matrigel Invasion Assays-Transfected MDA-MB-231 cells were trypsinized, and ϳ1 ϫ 10 5 cells (in Leibovitz's L-15 medium containing 3% of BSA) were seeded on the upper wells of Biocoat Matrigel chambers (Corning). The lower wells were filled with Leibovitz's L-15 medium containing 10% FBS. The transfected cells then underwent chemoattraction across the Matrigel and filter (pore size, 8 m) to the lower surface of the transwells for 22 h. The nonmigrating cells on the upper chambers were removed by a cotton swab. A fluorescence microscope was used to image the migrated cells on the lower surface of the membrane at five different and random fields at ϫ10 magnification. Data were collected from three independent experiments, each done in triplicate. Migrated cells were counted and mean differences (Ϯ S.E.) between groups were analyzed using Student's t test.

RESULTS
The Carboxyl-terminal Region of MyoGEF Binds to the DH Domain-It has been shown that intramolecular interactions between the DH domains and the amino-or carboxyl-terminal regions of GEFs can act as an autoinhibitory mechanism to regulate the activation of numerous GEFs (1,34). Like many other GEFs, MyoGEF contains a DH domain and a PH domain at the amino-terminal region of the molecule (Fig. 1A). Therefore, we asked whether an autoinhibitory intramolecular interaction exists for the regulation of MyoGEF activation. To this end, we used in vitro pulldown assays to examine the interaction between the amino-and carboxyl-terminal regions of MyoGEF. GST pulldown assays showed that a GST-MyoGEF carboxyl-terminal fragment (GST-MyoGEF-501-790) coprecipitated with the in vitro translated Myc-tagged amino-terminal fragment (Myc-MyoGEF-1-515) ( Fig. 1C; compare lane 2 with lane 3). These results suggested that the carboxyl-terminal region of MyoGEF could bind to the amino-terminal region of the molecule. As shown in Fig. 1A, the amino-terminal region (residues 1-515) of MyoGEF contains a DH domain (residues 162-351) and a PH domain (residues 363-515). We then asked whether the carboxyl-terminal region of MyoGEF could bind to its DH and/or PH domains. Immunoprecipitation assays showed that Myc-DH (but not Myc-PH) coprecipitated with His-MyoGEF-501-790 ( To determine the specificity of the interaction between the MyoGEF carboxyl-terminal region and the DH domain, we examined whether the MyoGEF carboxyl-terminal region could bind to other RhoGEFs such as Ect2 (21). MDA-MB-231 cells transfected with a plasmid encoding Myc-MyoGEF-501-790 were subjected to coimmunoprecipitation assays with anti-Myc antibody. As shown in Fig. 2B, endogenous MyoGEF coprecipitated with Myc-MyoGEF-501-790 (Fig. 2B, top panel). However, endogenous Ect2 (a closely related Rho-GEF) did not coprecipitate with Myc-MyoGEF-501-790 (Fig. 2B, bottom panel). These results suggested that the MyoGEF carboxyl-terminal region could specifically bind to its DH domain.
Exogenous Expression of the MyoGEF Carboxyl-terminal Region Suppresses Actin Filament Formation-GST pulldown and immunoprecipitation assays showed that the carboxyl-terminal region of MyoGEF could bind to the DH domain (Figs. 1 and 2). These findings led us to speculate that the intramolecular interaction between the carboxyl-terminal region and the DH domain could suppress the activation of MyoGEF. We have shown previously that MyoGEF is highly expressed in MDA-MB-231 breast cancer cells and that it can promote the cell invasion activity through activation of RhoA and RhoC (15). Furthermore, exogenous expression of MyoGEF induces actin bundling (15). Therefore, we asked whether exogenous expression of the MyoGEF carboxyl-terminal region in MDA-MB-231 cells interfered with actin filament formation. MDA-MB-231 breast cancer cells were transfected with a plasmid encoding mCherry-MyoGEF-501-790. 24 h after transfection, the transfected cells were trypsinized and cultured on fibronectincoated coverglasses for 3 h. The transfected cell were then fixed with paraformaldehyde and stained for actin filaments. As shown in Fig. 3B, a-c, exogenous expression of mCherry-MyoGEF-501-790 in MDA-MB-231 cells decreased the formation of actin filaments (green, n ϭ 82/97 cells). These results suggested that the carboxyl-terminal region of MyoGEF (residues 501-790) could inhibit MyoGEF activation through interactions with the DH domain.
Our previous findings have shown that the carboxyl-terminal end of MyoGEF contains a PDZ-binding motif that can interact with the PDZ domain-containing protein GIPC1 (46). Therefore, it is possible that exogenous expression of MyoGEF-501-790, which contains the carboxyl-terminal PDZ-binding motif, could interfere with actin organization by disrupting the GIPC1-MyoGEF interaction. To test this possibility, we examined whether exogenous expression of a carboxyl-terminal fragment lacking the carboxyl-terminal PDZ-binding motif (MyoGEF-501-780-⌬SEV) interfered with actin filament formation. We have found previously that MyoGEF polypeptides that lack the carboxyl-terminal three residues (SEV) do not bind to GIPC1 (46). As shown in Fig. 3B, d-f, exogenous expression of MyoGEF-501-790-⌬SEV also decreased actin filament formation (n ϭ 59/101 cells), suggesting that exogenous expression of MyoGEF carboxyl-terminal fragments could interfere with actin filament formation without disrupting GIPC1-MyoGEF interactions. However, we do not yet know whether binding of GIPC1 to the carboxyl-terminal end of MyoGEF has an impact on the intramolecular interactions of MyoGEF. It is of note that the impact of MyoGEF-501-790 on actin filament formation (n ϭ 82/97 cells, 84%) was much greater than that of MyoGEF-501-790-⌬SEV (n ϭ 59/101 cells, 58%).
We next examined the ability of amino-terminally truncated MyoGEF fragments to induce actin bundling in transfected MDA-MB-231 cells. The following MyoGEF fragments were examined (Fig. 4A): MyoGEF-46 -790 (lacking the amino-terminal 45 amino acid residues), MyoGEF-154 -790 (lacking the amino-terminal 153 amino acid residues), and MyoGEF-154 -540 (containing the tandem DH-PH domain only). Our results showed that MyoGEF-46 -790 could induce actin bundling and localize to the actin bundles (Fig. 4C, a-c, arrowheads). However, neither MyoGEF-154 -790 nor MyoGEF-154 -540 could induce actin bundling (Fig. 4C, d-i). These results suggested that the tandem DH-PH domain (residues 154 -515) alone was not sufficient to induce actin bundling in transfected cells. Currently, we do not know why the presence of the amino-terminal region (residues 46 -153) was required for the DH-PH domain of MyoGEF to induce actin bundling and to localize to actin bundles. Although the PH domain is often required for the DH domain to function in numerous GEFs (35,36), it has also been shown that the PH domain can bind to the DH domain and inhibit the DH domain (53). It is unlikely that the PH domain of MyoGEF could inhibit the function of the DH domain because exogenous expression of the MyoGEF fragment 1-354 (containing the amino-terminal region and the DH domain) did not induce actin bundling in transfected cells (data not shown). Also, exogenous expression of a MyoGEF fragment (residues 361-515, containing the PH domain) did not affect actin filament formation (Fig. 3B, j-l).

Exogenous Expression of the MyoGEF Carboxyl-terminal Region Inhibits the Invasion Activity of MDA-MB-231 Breast
Cancer Cells-We have found previously that MyoGEF can promote the invasion activity of MDA-MB-231 breast cancer cells through activation of RhoA and RhoC (15). In this study, we further showed that the MyoGEF carboxyl-terminal region could bind to the DH domain and inhibit MyoGEF-induced actin bundling (Figs. 1, 2, and 5). Our results also showed that exogenous expression of a carboxyl-terminally truncated Myo-GEF fragment (MyoGEF-1-540) induced stress fiber formation (Figs. 4B and 5). Therefore, we asked whether the intramolecular interaction between the carboxyl-terminal region of Myo-GEF and its DH domain played a role in the regulation of cell invasion activity. MDA-MB-231 cells were transfected with either the mCherry empty vector or a plasmid encoding mCherry-MyoGEF-501-790. Matrix gel invasion assays were used to examine the invasion activity of the transfected cells. The transfected cells were allowed to invade Matrigel for 22 h. A fluorescence microscope was used to examine the fluorescence-positive cells that successfully migrated to the underside of the Matrigels. As shown in Fig. 6, exogenous expression of mCherry-MyoGEF-501-790 significantly decreased the invasion activity of MDA-MB-231 cells (Fig. 6, A and B). Our results suggested that the carboxyl-terminal region of MyoGEF could bind to the DH domain of MyoGEF and inhibit MyoGEF activation, thus decreasing cell invasion activity.
We next asked whether exogenous expression of the Myo-GEF carboxyl-terminal region in U2OS cells interfered with cytokinesis. U2OS cells were used for this experiment because this cell line expresses endogenous MyoGEF and because the background levels of multinucleation in U2OS cells are much lower than those in MDA-MB-231 cells (data not shown). U2OS cells were transfected with the mCherry empty vector or a plasmid encoding mCherry-MyoGEF-501-790. Forty-eight hours after transfection, the cells were stained for nuclei. As shown in Fig. 7, C and D, we observed multinucleation in U2OS cells expressing mCherry-MyoGEF-501-790 (Fig. 7C, d-f, arrowheads) but not in cells expressing mCherry alone (Fig. 7C,  compare a-c with d-f). Fig. 7A shows that in vitro phosphorylation of the MyoGEF carboxyl-terminal region by aurora B could inhibit the intramolecular interaction between the The transfected cells were subjected to Rhotekin pulldown assays as described previously (24). D, quantitation of the results of Rhotekin pulldown assays from three independent experiments. *, p Ͻ 0.05. amino-and carboxyl-terminal regions of MyoGEF. Therefore, we also asked whether exogenous expression of the phosphomimetic MyoGEF carboxyl-terminal fragment (MyoGEF-501-790-T544D) had an impact on cytokinesis. We found that exogenous expression of MyoGEF-501-790-T544D did not cause multinucleation in U2OS cells (Fig. 7C, g-i), suggesting that phosphorylation of MyoGEF at Thr-544 by aurora B could relieve the autoinhibitory intramolecular interactions of MyoGEF.
It is well known that localization and activation of RhoA at the cleavage furrow are essential for the completion of cytokinesis (16 -18). We next asked whether exogenous expression of the MyoGEF carboxyl-terminal fragment MyoGEF-501-790 affected RhoA activation and localization at the cleavage furrow. Immunofluorescence staining of TCA-fixed cells with anti-RhoA antibody has been used to monitor the localization of activated RhoA at the cleavage furrow (54,55). The cortical staining of RhoA is generally considered as evidence for the activated forms of the protein. U2OS cells transfected with a plasmid encoding mCherry-MyoGEF-501-790 were fixed with TCA and then subjected to immunofluorescence analysis with antibodies specific for RhoA and MyoGEF. The MyoGEF antibody is a peptide antibody that can specifically recognize the carboxyl-terminal end of MyoGEF (24). Therefore, this Myo-GEF peptide antibody was used to detect the expression of mCherry-MyoGEF-501-790 in TCA-fixed U2OS cells. In cells that did not express mCherry-MyoGEF-501-790, RhoA concentrated at the cleavage furrow (Fig. 8, b, d, f, and h, arrowheads). In contrast, RhoA was diffusely localized to the cleavage furrow (Fig. 8, j and l, arrowheads) and the pole regions (Fig. 8,  j and l, arrows) during anaphase in cells expressing mCherry-MyoGEF-501-790 (Fig. 8, compare b and d with j and l). Fur- thermore, we found that RhoA did not concentrate to the midbody at late stages of cytokinesis (Fig. 8, n and p, arrowheads,  compare f and h with n and p). Our results suggested that exogenous expression of the MyoGEF carboxyl-terminal region interfered with the localization of RhoA at the cleavage furrow. These observations are consistent with our previous findings that RNAi-mediated depletion of MyoGEF can disrupt RhoA localization at the cleavage furrow (47).
The T544D Phosphomimetic Carboxyl-terminal Fragment Does Not Affect Actin Filament Formation-Our results showed that expression of the phosphomimetic mutant MyoGEF-501-790-T544D did not cause multinucleation in transfected U2OS cells (Fig. 7C). This finding was also in agreement with our observations that phosphorylation of the MyoGEF carboxylterminal region by aurora B could suppress the intramolecular interactions of MyoGEF (Fig. 7, A and B). We next asked whether phosphomimetic mutation of the MyoGEF carboxylterminal region interfered with actin filament formation in interphase cells. We found that exogenous expression of the phosphomimetic mutant MyoGEF-501-790-T544D did not affect actin filament formation (Fig. 9; compare a-c with d-f), further confirming that phosphorylation of MyoGEF at Thr-544 by aurora B could inhibit the intramolecular interactions of MyoGEF. It has been shown that aurora B is also implicated in the regulation of cell spreading and cell migration (56,57). However, treatment of transfected MDA-MB-231 cells with an inhibitor specific for aurora B did not interfere with actin bundling induced by exogenously expressed MyoGEF (data not shown), suggesting that, during interphase, phosphorylation of MyoGEF by aurora B was not likely to be a major pathway to activate MyoGEF.
MyoGEF Forms Homodimers/Oligomers-Our results demonstrated that the MyoGEF carboxyl-terminal region could bind to its DH domain and suppress MyoGEF activation. It is conceivable that the interaction between the DH domain and the carboxyl-terminal region of MyoGEF can occur either intramolecularly or intermolecularly. Therefore, we asked whether MyoGEF could form homodimers/oligomers. As shown in Fig. 10A, in vitro pulldown assays demonstrated that GFP-MyoGEF could coprecipitate with Myc-MyoGEF. Furthermore, coimmunoprecipitation assays also showed that endogenous MyoGEF coprecipitated with Myc-MyoGEF from transfected MDA-MB-231 cells (Fig. 10B). These results suggested that MyoGEF could form homodimers/oligomers. Next we asked whether the interaction between the MyoGEF carboxyl-terminal region and the DH domain was required for MyoGEF oligomerization. In vitro pulldown assays demonstrated that aurora B-mediated phosphorylation of the Myo-GEF carboxyl-terminal region and the T544E phosphomimetic mutation interfered with the interaction between the MyoGEF carboxyl-terminal region and the DH domain (Fig. 7, A and B). Therefore, we asked whether the T544E phosphomimetic mutation interfered with the formation of MyoGEF dimers/ oligomers. Fig. 10C shows that the T544E phosphomimetic mutation did not interfere with the intermolecular interactions of MyoGEF. Therefore, our findings indicated that MyoGEF could form homodimers/oligomers independently of the interaction between the MyoGEF carboxyl-terminal region and the DH domain. However, it is still not clear which domains mediate MyoGEF dimerization/oligomerization.

DISCUSSION
In this study, we demonstrated that the intramolecular interaction between the DH domain and the carboxyl-terminal region of MyoGEF serves as an autoinhibitory mechanism to regulate MyoGEF activation. In vitro and in vivo pulldown assays show that the carboxyl-terminal region of MyoGEF could bind to its DH domain. We also demonstrated that exogenous expression of the MyoGEF carboxyl- terminal region interferes with actin filament formation, inhibits breast cancer cell invasion activity, and leads to the formation of multinucleated cells. Our findings suggest that the autoinhibitory intramolecular interactions of MyoGEF serve as a control point to regulate cytokinesis and breast cancer cell invasion.  Autoinhibitory Regulation of MyoGEF Functions-Our previous findings have shown that MyoGEF is implicated in the regulation of cytokinesis and breast cancer cell invasion (15,24). In this study, we found that the intramolecular interaction between the DH domain and the carboxyl-terminal region of MyoGEF could inhibit MyoGEF functions. Our findings demonstrate that exogenous expression of the MyoGEF carboxylterminal region interferes with MyoGEF functions, leading to cytokinesis defects and decreasing breast cancer cell invasion. It has been shown that the carboxyl-terminal region of MyoGEF can interact with centrosome/spindle-associated protein (CSPP), ezrin, and GIPC1 (46,47,58). Aurora B and Plk1 can phosphorylate the carboxyl-terminal region of MyoGEF (31,32). Therefore, exogenous expression of the MyoGEF carboxylterminal region in transfected cells could potentially cause cell phenotypes through interference with protein-protein interactions and phosphorylation involving the carboxyl-terminal region of MyoGEF. However, exogenous expression of the T544D phosphomimetic MyoGEF carboxyl-terminal fragment did not interfere with actin filament formation in interphase cells (Fig. 9) and did not cause multinucleation (Fig. 7C). Furthermore, phosphorylation of the MyoGEF carboxyl-terminal region by aurora B decreased the intramolecular interactions of MyoGEF. Moreover, the MyoGEF carboxyl-terminal region did not bind to a closely related RhoGEF Ect2 (Fig. 2B). Therefore, the dominant negative effects of the MyoGEF carboxyl-terminal region are likely dependent on its interaction with the DH domain, leading to the inhibition of MyoGEF functions.
Intramolecular interactions have been shown to serve as an autoinhibitory mechanism for the regulation of GEFs (34). It is conceivable that domains or regions participating in an intramolecular interaction are also used for intermolecular interactions. Our results demonstrate that the MyoGEF carboxyl-terminal region can bind to its DH domain and suppress MyoGEF activation, raising the possibility that intermolecular interactions of MyoGEF can occur between the carboxyl-terminal region and the DH domain. However, we found that phosphomimetic mutation at Thr-544 interfered with the interaction between the carboxyl-terminal region and the DH domain (Fig.  7B). In contrast, phosphomimetic mutation at Thr-544 did not disrupt the formation of MyoGEF dimers/oligomers (Fig. 10C). Therefore, our findings suggest that the interaction between the MyoGEF carboxyl-terminal region and its DH domain can occur intramolecularly. However, we cannot completely rule out the possibility that the interaction between the carboxylterminal region and the DH domain of MyoGEF also occurs intermolecularly.
Our results show that the carboxyl-terminal region of Myo-GEF could bind to its DH domain, thus interfering with Myo-GEF activation (Figs. 3 and 5). However, exogenous expression of full-length MyoGEF also induced actin bundling in transfected cells compared with that of carboxyl-terminally truncated MyoGEF fragments (Fig. 4B), suggesting that an unknown mechanism exists to activate full-length MyoGEF in the transfected MDA-MB-231 cells (see below).
Regulation of Breast Cancer Cell Invasion by MyoGEF-We have reported previously that MyoGEF can activate RhoA/ RhoC in MDA-MB-231 breast cancer cells (15). Furthermore, MyoGEF-induced actin bundling can be inhibited by a dominant negative mutant of RhoA but not by a dominant negative mutant of Rac1 (15). Therefore, our findings suggest that RhoA and RhoC are implicated in MyoGEF-induced actin bundling. Cell migration involves lamellipodium formation and membrane ruffles at the front as well as retraction of the rear part of migrating cells (59). Actin polymerization at the front of migrating cells is critical for lamellipodium formation and membrane ruffles, whereas myosin-based contractility is required for retraction of the rear part of migrating cells (59 -62). It is generally thought that Rac1 is restricted to the front of migrating cells and is responsible for lamellipodium formation and membrane ruffles, whereas RhoA is localized to the rear of migrating cells and is responsible for tail retraction (62)(63)(64)(65). This appears to be at odds with our observations that exogenous expression of MyoGEF resulted in the formation of actin bundles that are often associated with membrane ruffles (Fig.  4B), which are a typical function of Rac1 (66). However, another line of research also demonstrates that RhoA/RhoC can be activated in the front of migrating cells (67,68), consistent with findings that RhoA/RhoC can induce membrane ruffles (69 -71). In support of this concept, a study has shown that S100A4 and Rhotekin (a downstream effector of RhoA) can form a complex with the activated forms of RhoA, thus converting an activated RhoA from a stress fiber inducer to a membrane ruffle inducer (72).
It has been shown that the PH domain of GEFs can associate with actin filaments and phospholipids, thus targeting GEFs to the actin cytoskeleton and/or the cell membrane (1,34). We have found previously that the PH domain of MyoGEF can interact with nonmuscle myosin II (15,24). The PDZ domain of GIPC1 can bind to the carboxyl-terminal PDZ-binding motif of MyoGEF, and the GIPC1-MyoGEF interaction is implicated in promoting breast cancer cell invasion (46). The ezrin-radixinmoesin proteins can link the plasma membrane to actin filaments and play a critical role in organizing the cell cortex and promoting breast tumor metastasis (73,74). It has been shown that ezrin can bind to the carboxyl-terminal region (residues 579 -790) of MyoGEF/PLEKHG6 (58). In addition, ezrin is localized to membrane ruffles and required for membrane ruffling (75)(76)(77). In this study, we found that full-length MyoGEF could induce and localize to membrane ruffle-associated actin bundles (Fig. 4B). In contrast, carboxyl-terminally truncated MyoGEF with deletion of 190 -250 amino acid residues from the carboxyl-terminal end of MyoGEF predominantly localized to stress fibers (Fig. 4B). Therefore, we propose that binding of proteins such as ezrin and GIPC1 to the carboxyl-terminal region of MyoGEF may not only interfere with the autoinhibitory intramolecular interactions, leading to the activation of MyoGEF, but also recruit MyoGEF to the cell membrane, where MyoGEF activates RhoA and induces actin bundling and membrane ruffling, thus promoting cell invasion (Fig. 10D). Accordingly, carboxyl-terminally truncated MyoGEF mutants lacking the binding sites for ezrin and GIPC1 predominantly localize to the stress fiber (Fig. 4B). It would be interesting to know whether binding of ezrin or GIPC1 to the carboxyl-terminal region of MyoGEF has an impact on the intramolecular interactions of MyoGEF. DECEMBER 5, 2014 • VOLUME 289 • NUMBER 49

JOURNAL OF BIOLOGICAL CHEMISTRY 34045
Regulation of Cytokinesis by MyoGEF-We have reported previously that MyoGEF is implicated in promoting cytokinesis (24,47). Aurora B and Plk1 are two mitotic kinases that play critical roles in the regulation of cytokinesis (27,28). Aurora B and Plk1 can promote MyoGEF activation and localization at the central spindle during cytokinesis by phosphorylating Myo-GEF at Thr-544 and Thr-574, respectively (31,32). Our findings in this study show that phosphorylation of the MyoGEF carboxyl-terminal region by aurora B relieves the intramolecular interaction between the DH domain and the carboxyl-terminal region of MyoGEF (Figs. 7 and 10D). These findings are consistent with our previous observations that aurora B-mediated phosphorylation of MyoGEF at Thr-544 leads to the activation of MyoGEF during cytokinesis. Our previous studies have also demonstrated that phosphorylation of MyoGEF at Thr-544 creates a Plk1 docking site and promotes the binding of Plk1 to MyoGEF (31). In turn, Plk1 phosphorylates MyoGEF at Thr-574, leading to the activation and localization of MyoGEF at the central spindle during cytokinesis (32). However, it is still not clear how Plk1-mediated phosphorylation of MyoGEF at Thr-574 is implicated in the regulation of MyoGEF localization and activation during cytokinesis. One possibility is that Plk1-mediated phosphorylation of MyoGEF promotes the binding of signaling molecules such as CSPP to the carboxyl-terminal region of MyoGEF, thus suppressing the intramolecular interactions of MyoGEF (Fig. 10D). In support of this speculation, we have found previously that CSPP can bind to the carboxylterminal region of MyoGEF in a phosphorylation-dependent manner and that the CSPP-MyoGEF interaction plays a role in promoting MyoGEF localization to the central spindle (47). In addition, phosphorylation of the MyoGEF carboxyl-terminal region by Plk1 promotes the binding of CSPP to MyoGEF. 3 In summary, we propose that phosphorylation and protein-protein interactions involving the carboxyl-terminal regions of MyoGEF decrease the autoinhibitory intramolecular interactions of MyoGEF, thus leading to the activation and localization of MyoGEF at the central spindle during cytokinesis (Fig. 10D). It would be interesting to know whether binding of CSPP to the carboxyl-terminal region of MyoGEF has an impact on the intramolecular interactions of MyoGEF.