Differential Effects of Cytohesins 2 and 3 on β1 Integrin Recycling*

ADP-ribosylation actor 6 (ARF6) regulates the endocytosis and recycling of a variety of proteins and also promotes peripheral actin rearrangements and cell motility. ARF6 is activated by a large number of guanine nucleotide exchange factors, which likely regulate ARF6 at different locations and during different processes. In this study we investigate the roles of the cytohesin ADP-ribosylation factor (ARF)-guanine nucleotide exchange factors during the recycling of integrin β1. Intriguingly, we find that knockdown and overexpression of ARNO/cytohesin 2 and GRP1/cytohesin 3 have opposing effects on cell adhesion and spreading on fibronectin and on cell migration. We find that ARNO/cytohesin 2 is required for integrin β1 recycling, whereas GRP1/cytohesin 3 is dispensable for this process. This is the first demonstration of unique roles for these proteins.

ADP-ribosylation factors 1-6 (ARF1-6) 2 are small GTPases of the Ras superfamily and regulators of vesicular transport. There are six known ARF isoforms divided into three classes. Class I ARFs (ARF1-3) regulate the assembly of vesicle coat complexes in the secretary pathways and at endosomes. Class II ARFs (ARF4 and -5) are still poorly understood. The only member of class III, ARF6, is localized at the cell periphery and regulates trafficking between the plasma membrane and endosomal system (1). ARF6 also regulates assembly of the cytoskeleton and cell motility (2)(3)(4)(5).
ARF6 regulates the endocytosis of ␤1 integrin and a number of other membrane proteins. These proteins are internalized via clathrin-independent endocytosis. They then join with clathrin dependent cargo in the early endosome where they are sorted either for degradation in the lysosome or for recycling to the plasma membrane (6).
In epithelial tissues cells are attached to each other by cellcell adhesions and further anchored by cell-substratum adhesions (14,15). These cells become migratory during developmental morphogenesis, wound healing, and cancer metastasis and when motile, extensively remodel their adhesions (14). ARF6 regulates adhesion remodeling by controlling the endocytosis and recycling of adhesion proteins (16,17). Adhesion remodeling is one mechanism to facilitate cell migration (18).
Integrins are cell adhesion proteins that interact with matrix proteins such as fibronectin. Cell surface levels of integrins are determined by a balance between endocytosis and recycling. Integrin trafficking may promote cell migration by internalizing integrins at the trailing edge of the cell and recycling them out to the leading edge (19,20). Integrin ␤1 is a component of the major receptors of fibronectin. ARF6 has been shown to regulate both the internalization and recycling of integrin ␤1 (17,21). Recent studies demonstrated that Brag2 activates ARF6 to regulate integrin ␤1 endocytosis, whereas an unknown GEF activates ARF6 to regulate integrin ␤1 recycling (22). The established roles for cytohesins and ARF6 in motility suggest that cytohesins could activate ARF6 to regulate integrin ␤1 recycling during cell migration.
In this study we test this hypothesis and demonstrate that cytohesin function is required for integrin ␤1 recycling. Furthermore, we found that cytohesin 2/ARNO, but not cytohesin 3/GRP1, is required in integrin ␤1 recycling. We also show that ARNO and GRP1 have opposing effects on cell adhesion, spreading, and migration. ARNO knockdown inhibits cell migration, whereas GRP1 knockdown increases cell migration. To our knowledge this is the first demonstration that these two similar proteins, ARNO and GRP1, have unique functions. These findings outline not only the endocytic recycling pathway of integrin ␤1 but also demonstrate that ARF6 activation is under extensive regulatory control by multiple GEFs during cell migration.
siRNA and Plasmids-siRNA duplexes against the sequence GCAAUGGGCAGGAAGAAGU targeting human and canine ARNO and siRNA duplexes against the sequence GGAGAAG-GCCUAAAUAAGA targeting human and canine GRP1 were obtained from Dharmacon. Control siRNAs against firefly luciferase or scrambled sequences (siControl 1) were also obtained from Dharmacon. The siRNAs were transfected with Lipofectamine RNAiMax for HeLa and MCF-7 cells and with Lipofectamine 2000 for T23 MDCK cells using the manufacturer's suggested protocol for reverse transfection (Invitrogen). Experiments were carried out 48 h after transfection with siRNAs. Myc-tagged ARNO and FLAG-tagged GRP1 were obtained from James Casanova. Transient transfections were performed using Lipofectamine 2000 or the Neon transfection system (Invitrogen). Experiments were carried out 24 h after transient transfection.
Adhesion Assay-HeLa cells were transfected with Myctagged ARNO or FLAG-tagged GRP1 or with siRNAs targeting ARNO or GRP1. After 24 h of expression or 48 h of knockdown, confluent cells were nonenzymatically harvested and dissociated into single cells by treatment with 1 mM EGTA, 4 mM EDTA in phosphate-buffered saline for 20 min at 37°C. Harvested cells were then resuspended in serum-free medium. 5 ϫ 10 4 cells in 100 l were plated in 96-well plates in wells coated with different concentrations of fibronectin. The cells were allowed to attach for 1 h and then fixed with 4% formaldehyde. The fixed cells were stained with 0.1% crystal violet. The dye was extracted into 1% sodium deoxycholate and quantitated by measuring the optical density at 590 nm.
Analyzing Integrin ␤1 Recycling by Microscopy-Integrin ␤1-specific antibody TS2/16 was bound to the surface of serum-starved cells at 4°C, and then the cells were warmed to make the cells internalize the antibody-bound integrin. Residual integrin antibodies remaining on the cell surface were removed with an acid wash (0.5% acetic acid, 0.5 M NaCl, pH 3.0) on ice, and then 20% FBS was applied to initiate integrin ␤1 recycling at 37°C (17). Duplicate samples were fixed at 0 and 5 min after the initiation of integrin ␤1 recycling. One sample was permeabilized by 0.1% Triton to observe total antibody bound integrin, whereas the other sample was left unpermeabilized to observe surface integrin ␤1. Both samples were stained with AlexaFluor 546-conjugated goat anti-mouse secondary antibody. Cells were observed and photographed using a Zeiss AxioPlan microscope equipped with a Spot RT3 camera (Diagnostic Instruments). During each experiment standardized microscope and camera settings were used, and all images were recorded using identical exposure times. Scale bars were added using ImageJ.
Biochemical Integrin ␤1 Recycling Assay-Antibody binding, integrin internalization, and integrin recycling were performed as described above except for fixation of the samples. Instead, recycled integrin antibodies at the surface were removed by a second acid wash after 0 and 5 min of recycling. Then, the cells were lysed, and the lysate was incubated with protein G-Sepharose to isolate internal antibody bound integrin ␤1 (17). Precipitated integrin ␤1 was analyzed by Western blot for integrin ␤1.
Surface Integrin ␤1 Labeling Assay-Integrin ␤1-specific antibody TS2/16 was bound to the cell surface at 4°C, and then the cells were washed with serum-free media twice. The cells were lysed, and the lysate was incubated with protein G-Sepharose to isolate surface antibody-bound integrin ␤1 (17). Precipitated integrin ␤1 was analyzed by Western blot for integrin ␤1.
Migration Assay-Cell migration was assessed using Oris migration chambers (Oris TM Universal Cell Migration Assembly kit; Platypus Technologies, Madison, WI). T23 cells were transfected with siRNAs targeting ARNO or GRP1 for 2 days. Cells (1 ϫ 10 5 ) were then seeded into each test well and incubated at 37°C in a 5% CO 2 humidified chamber to permit cell attachment. After 24 h, all well inserts were removed, and the medium was refreshed. The cells were incubated for 24 h to permit cell migration and then fixed with 4% formaldehyde. The fixed cells were stained with 0.1% crystal violet. Images of each chamber were taken, and cell migration was quantitated by measuring the remaining cell-free surface area using ImageJ (23).
Spreading Assay-MCF-7 cells were transfected with siRNA targeting ARNO or GRP1. The cells were incubated for 2 days and then nonenzymatically harvested and dissociated into single cells as described above for the adhesion assay. Harvested cells were then resuspended in serum-free medium and replated on coverslips coated with 10 g/ml fibronectin. The cells were allowed to spread for 30 min at 30% confluency and then fixed with 4% formaldehyde. The samples were stained with rhodamine-conjugated phalloidin, and images of each spreading were taken. Cell spreading was quantitated by measuring cell surface area with ImageJ (23).
Flow Cytometry-Cell surface integrin levels were measured by flow cytometry. HeLa cells were transfected with GFP, GFP and ARNO, or GFP and GRP1. Cells were detached by using 1 mM EGTA, 4 mM EDTA in phosphate-buffered saline for 10 min on ice. Harvested cells were washed with phosphate-buffered saline and fixed in cold 4% paraformaldehyde for 15 min at 4°C while rotating. Cells were stained with AlexaFluor 647conjugated anti-integrin ␤1 (TS2/16) antibody for 1 h at room temperature. After staining, the cell suspension was filtered through a 37-m mesh to remove multicell aggregates. The median level of AlexaFluor 647 fluorescence in GFP-positive cells was determined by flow cytometry on an FC500 benchtop cytometer.
Reverse Transcription-PCR-Total RNA was isolated using the RNeasy kit (Qiagen). Custom primers to amplify bases 156 -505 of human GRP1, bases 272-571 of canine GRP1, bases 548 -872 of human ARNO, bases 532-872 of canine ARNO, and ReadyMade primers to amplify glyceraldehyde-3-phosphate dehydrogenase were obtained from Integrated DNA Technologies. Reverse transcription-PCR was performed with 0.5 g of total RNA as template using the Qiagen One-Step reverse transcription PCR kit.
Deconvolution Microscopy-MCF-7 cells were co-transfected with FLAG-GRP1 and Myc-ARNO. The cells were spread on fibronectin for 30 min as described above for the spreading assay. The cells were then fixed and stained with M2 anti-FLAG conjugated to fluorescein isothiocyanate and biotinylated 9e10 anti-Myc followed by staining with AlexaFluor 546-conjugated streptavidin. The cells were imaged with an Olympus IX 81 microscope and a Hamamatsu Orca camera. Z-stacks with 0.3-m spacing were obtained. Out of focus light was removed using the Constrained Iterative Deconvolution algorithm in Slidebook 5.0 (Intelligent Imaging Innovations, Denver, CO). Merged images through the Z and x axis were created using a maximum projection, and scale bars were added using Slidebook.

Cytohesins Are Critical for Cell Adhesion to Fibronectin-In-
tegrin ␤1 is a component of the major fibronectin receptors. Therefore, if cytohesin function is required for integrin ␤1 recycling, then adhesion to fibronectin will require a cytohesion function. We used SecinH3, a recently identified cytohesin inhibitor, to determine whether cytohesin function is required for adhesion to fibronectin (24). The importance of cytohesin function on integrin-mediated cell to substratum adhesion was first examined by plating HeLa cells on different concentrations of fibronectin in the presence or absence of 15 M SecinH3. We found that cell adhesion was significantly decreased in SecinH3-treated cells (Fig. 1A). We next investigated which cytohesin member is the major GEF acting in this process. We reasoned that cytohesin 2/ARNO and cytohesin 3/GRP1 are attractive candidates because only ARNO and GRP1 are ubiquitously expressed, whereas cytohesins 1 and 4 are primarily leukocyte-specific (25). We, therefore, transfected Myc-tagged ARNO or FLAG-tagged GRP1 into HeLa cells and expected that overexpression of the relevant cytohesin should enhance the adhesion in the same assay. Interestingly, cell adhesion to fibronectin was enhanced in ARNO-transfected cells but repressed in GRP1-transfected cells (Fig. 1B). This unexpected inhibition may indicate that the extra GRP1 acts in a dominant negative fashion by either competing with endogenous ARNO or by sequestering necessary downstream components.
We confirmed the differential effects of ARNO and GRP1 on cell adhesion by reducing the expression of each GEF using siRNAs. Transfection of HeLa, MCF-7, or MDCK cells with siRNAs targeting ARNO or GRP1 reduces the expression level of mRNA for these proteins (supplemental Fig. S1). Cell adhesion was inhibited in cells with reduced ARNO expression as we expected. Strikingly, we also observed that adhesion was enhanced in cells with reduced GRP1 expression (Fig. 1C). This observation may indicate either that components sequestered by GRP1 are released and are, therefore, more available to ARNO or that ARNO expression is increased to compensate for GRP1 knockdown. Either way, these data suggest that there is more surface integrin ␤1 integrin or increase integrin activity in the cells with reduced GRP1 expression.
Cells with enhanced surface levels or activity of integrin ␤1 should not only adhere more tightly to fibronectin but should also spread more rapidly when plated on fibronectin. Conversely, cells with reduced levels or activity of surface integrin should show impaired spreading on fibronectin. Therefore, we confirmed the results shown in Fig. 1, A-C, by investigating the spreading of MCF7 cells treated with siRNAs to reduce expression of ARNO or GRP1. Consistent with the results of the adhesion experiments, we found that spreading was impaired in cells transfected with siRNAs targeting ARNO and enhanced in cells transfected with siRNAs targeting GRP1 (Fig. 1D). These experiments also confirm that ARNO and GRP1 have opposing effects on adhesion in multiple cell types.
Enhanced adhesion to fibronectin can be due either to increased levels of surface integrin or to increased activity of the integrins present on the surface (inside-out signaling). Previous work demonstrated that ARF6 regulates the endocytosis and recycling of ␤1 integrin, whereas cytohesin 1 has also been implicated in inside-out signaling (17,26). Therefore, ARNO could enhance adhesion through either pathway. To distinguish between the effects on integrin trafficking and effects on integrin activity, we directly measured the surface levels of ␤1 integrin.
Cell surface levels of ␤1 integrin in cells transfected with siRNAs targeting ARNO or GRP1 were determined by labeling of the cell surface integrin with antibody to integrin ␤1 followed by precipitation of the antibody-bound protein. Surface integrin levels were reduced in the ARNO knockdown MCF-7 cells (Fig. 2, A and B). Next, we performed the same assay in HeLa cells overexpressing Myc-ARNO or FLAG-GRP1. Surface integrin levels were elevated in the ARNO-expressing cells and reduced in the GRP1-expressing cells (Fig. 2, C and D). We confirmed that HeLa cells overexpressing Myc-ARNO or FLAG-GRP1 have altered levels of surface integrin using flow cytometry. These cells were fixed and stained with labeled antibody directed against ␤1 integrin. Cell surface integrin levels measured by flow cytometry were also enhanced in cells overexpressing ARNO and reduced in cells overexpressing GRP1 (Fig. 2E). These data suggest that ARNO modulates adhesion by regulating cell surface levels of integrin and not by altering integrin activity. Given the enhanced adhesion seen in cells treated with GRP1 siRNA (Fig. 1), we expected to see elevated levels of surface integrin in these cells. However, we did not observe a significant change in the surface levels of ␤1 integrin in these cells. This assay may not be sensitive enough to see a modest change in cell surface integrin. Alternatively, these data may suggest that GRP1 has an inhibitory effect on inside-out signaling. Given the reduced cell-surface ␤1 integrin levels seen in cells overexpressing FLAG-GRP1, we favor the former explanation.
␤1 Integrin Recycling Requires Cytohesins-Surface levels of adhesion proteins are balanced by internalization and recycling. Cell surface levels of integrins are increased during cell migration (27). Both integrin ␤1 internalization and recycling are under the control of ARF6 (17,21). Internalization of integrin ␤1 requires activation of ARF6 by the GEF Brag2 (22). However, the GEF that activates ARF6 during the recycling of integrin ␤1 remains unclear. Figs. 1 and 2 demonstrate that altering cytohesin expression modulates cell surface levels of ␤1 integrin. Therefore, we directly tested whether cytohesins are required for integrin recycling. First, an antibody to integrin ␤1 was bound to cell surface integrins. The cells were then incu-bated to allow internalization of the antibody-bound proteins. Any antibody remaining on the cell surface was removed, and the cells were triggered to recycle internalized antibody-bound proteins to the cell surface. The amount of recycled antibody on the cell surface was determined by fluorescent staining. Alternatively, immunoprecipitation was used to determine the levels of antibody-bound proteins retained within the cells (17). We found that integrin ␤1 recycling in MCF-7 cells follows a time course similar to that previously seen in HeLa cells (supplemental Fig. S2) (17). Most of the integrins were recycled back to the cell surface within 5 min of treatment with 20% FBS and were subsequently re-internalized within 15 min. (supplemental Fig. S2). We directly tested the hypothesis that cytohesins are required in integrin ␤1 recycling. Consistent with adhesion results, integrin ␤1 recycling was reduced when cytohesin func- tion was impaired by treatment of the cells with SecinH3. Most of integrin ␤1 was still inside SecinH3-treated cells after 5 min of stimulation (Fig. 3A), and almost none of previously internalized integrin ␤1 was on the cell surface after the stimulation with 20% FBS, in contrast to the untreated cells (Fig. 3B).
Only ARNO Is Required for Integrin ␤1 Recycling-Figs. 1 and 2 suggest that ARNO but not GRP1 is required for integrin ␤1 recycling. Therefore, we measured ␤1 integrin recycling in MCF-7 cells transfected with siRNAs targeting ARNO or GRP1. Integrin ␤1 recycling was significantly inhibited in cells transfected with an siRNA targeting ARNO (Fig. 4). Cells transfected with an siRNA directed against GRP1, on the other hand, showed no impairment of integrin ␤1 recycling at a 5-min time point. Most of integrin ␤1 was recycled back to the cell surface in the control and GRP1 siRNA-treated cells, but ARNO-siRNA-treated cells still had not recycled internal integrin ␤1 after 5 min of stimulation with 20% FBS (Fig. 4, A and C). Fig. 1C suggests that the GRP1 knockdown should have increased recycling, but we did not observe this effect when recycling was directly tested. The reason may be differences in sensitivity of the two assays or the different cell types used in the two experiments.

ARNO and GRP1 Knockdown Have Opposing Effects on Cell Migration-Because integrin-mediated adhesion has been
shown to promote cell migration, we tested the effect of cytohesin knockdown on migration. Migration was measured by using the Oris migration chamber. In this system cells are grown around a silicon plug. The plug is then removed, and the cells migrate into the area vacated by the plug. We seeded MDCK cells transfected with siRNAs targeting ARNO or GRP1 into the chambers. After an overnight recovery, the stopper was pulled out to initiate the cell migration, and the cells were allowed to migrate into the open area overnight. As expected, cell migration was reduced in cells treated with siRNA targeting ARNO but enhanced in cells transfected with siRNA directed against GRP1 (Fig. 5). Our data all suggest that ARNO/cytohesin 2 regulates integrin ␤1 recycling and that GRP1/cytohesin 3 does not regulate this process.
Localization of Cytohesins in Spreading Cells-One explanation for the fact that cells express multiple GEFs is that the different GEFs act at different subcellular locations. Therefore, we investigated the localization of cytohesins 2 and 3 in cells spreading on fibronectin. MCF-7 cells expressing either FLAGtagged GRP1 or Myc-tagged ARNO were spread on fibronectin for 30 min then fixed, and the cytohesin localization was determined by immunofluorescence. Cytohesin 2/ARNO was found at both the edges and throughout the interior of spreading cells. Cytohesin 3/GRP1, on the other hand, was almost exclusively located at the very peripheral edge of spreading MCF-7 cells (Fig. 6A). We confirmed the differential localization of cytohesins 2 and 3 using deconvolution microscopy of MCF-7 cells co-expressing both cytohesins. The deconvolved images confirm that in spreading cells cytohesin 3 is exclusively located at the cell periphery, whereas cytohesin 2 is present at the periph-  Integrin ␤1 recycling requires cytohesins. A, integrin ␤1 is retained intracellularly in cells treated with SecinH3. Cell surface integrins were labeled with antibody to ␤1 integrin, and the integrin was internalized as described under "Experimental Procedures." Recycling was initiated in the presence or absence of 15 M SecinH3. Internal integrin was isolated as described under "Experimental Procedures" and was quantitated by Western blot. B, surface and total integrin ␤1 was visualized by immunofluorescence after the integrin ␤1 recycling assay for the cells without or with SecinH3 treatment as described in A. Bar, 25 m.  ery and at more interior regions (Fig. 6B). Furthermore projection along the z axis of the deconvolved stack shows that cytohesin 3 is restricted to the most basal level of the cell, whereas cytohesin 2 is also present in more apical regions (Fig. 6C).

DISCUSSION
Cellular migration is a well coordinated process that requires altered cell shape and polarity (20). Recent studies have shown that small GTPases of the ARF family are involved in these processes (28). Furthermore, it has been demonstrated that ARF6 regulates endocytosis and recycling of adhesion proteins, which is one mechanism that can facilitate cell migration (17,21). In this study we demonstrated that integrin ␤1 recycling is regulated by cytohesin 2/ARNO. We also found that two similar ARF6 GEFs, cytohesin 2/ARNO and cytohesin 3/GRP1, have different functions in cell adhesion, spreading, and migration. ARNO knockdown reduces adhesion, spreading, and migration, whereas GRP1 knockdown enhances adhesion, spreading, and migration. Additionally we found that ARNO but not GRP1 is required for ␤1 integrin recycling.
Integrin ␤1 Recycling Pathway-Integrins are continuously cycling by endocytosis and reinsertion into the plasma membrane (29,30). This trafficking mechanism regulates cell migra-tion. Integrin is redistributed from the trailing edge to the leading edge of migrating cells (31). ARF6 is known to regulate both integrin ␤1 endocytosis and recycling (17,21).
The itinerary taken by ARF6-dependent cargo has been elucidated. These cargo proteins, such as ␤1 integrin, are internalized by clathrin-independent endocytosis. They then enter the early endosome where they meet up with cargo that was internalized in a clathrin-dependent manner. Recycling cargo then travel to the Rab11 positive endosomal recycling compartment. ARF6-dependent cargo is then returned to the plasma membrane via tubular recycling endosomes that align along microtubules. Both the internalization and recycling steps of this pathway require ARF6 activity (6).
Despite the existence of 15 ARF GEFs, there are only 6 ARFs expressed in mammalian cells (10,32). This suggests that ARF activation is regulated by different GEFs in different signaling pathways. The 15 Sec7 family members are divided into 5 classes: GBF/BIG, cytohesins, EFA6, BRAGs, and Fbox (9).
Previous research showed that Brag2 is localized to the plasma membrane and activates ARF6 to regulate integrin ␤1 endocytosis (22). It also has been demonstrated that ACAP1, a GTPase-activating protein for ARF6, is involved in recycling of this protein (21). However, the ARF6 GEF involved in integrin ␤1 recycling remained unknown. Cytohesin 2/ARNO has been shown to stimulate epithelial cell migration, and cytohesin 3/GRP1 induces actin cytoskeleton rearrangement (5,33). In addition, recycling of adhesion proteins is one of the mechanisms to regulate cell migration. Therefore, we tested cytohesins as promising candidates and identified ARNO as the ARF GEF required for integrin ␤1 recycling.
Different Functions of ARNO and GRP1 in Cell Adhesion, Spreading, and Migration-The cytohesin family has four isoforms: cytohesin 1, cytohesin 2/ARNO, cytohesin 3/GRP1, and cytohesin 4. They share similar sizes, extensively conserved sequences, and a common domain structure: coiled coil, sec7, PH, and basic domain (9). Among the members, ARNO and GRP1 are ubiquitously expressed, and they are very closely related (80% identity) (25). Previous research demonstrated that knockdown of either ARNO or GRP1 has similar effects on insulin signaling (24). Here, we demonstrated different functions of ARNO and GRP1 in cell adhesion, spreading, and migration. ARNO knockdown reduced cell adhesion, spreading, and migration, whereas GRP1 knockdown had opposing effects on these processes. To our knowledge, this is the first demonstration of differential actions of ARNO and GRP1.
In addition to the different actions of ARNO and GRP1 in integrin ␤1 recycling, we find that these proteins are uniquely localized in spreading cells. Cytohesin 3/GRP1 is exclusively located in the most peripheral and basal regions of spreading MCF-7 cells. This localization suggests that GRP1 may regulate the protrusive actin rearrangements that occur in these locations as the cells spread. Cytohesin 2/ARNO is co-localized with cytohesin 3/GRP1 in these protrusive regions. In addition, cytohesin 2/ARNO is located throughout the interior of the cell. It is likely that some of this more centrally located ARNO promotes the recycling of integrin ␤1 from endosomal structures to the cell surface. In the merged image the cytohesin is pseudocolored green, and the phalloidin is pseudocolored red. Bar, 10 m. B, MCF-7 cells were cotransfected with FLAG-cytohesin 3 and Myc-cytohesin 2, spread on fibronectin, and fixed as above. The cells were then stained with fluorescein isothiocyanate-conjugated M2 anti-FLAG and biotinylated 9e10 anti-Myc followed by staining with AlexaFluor 546-conjugated streptavidin. A z-stack through the entire depth of the cell was collected and analyzed by deconvolution as described under "Experimental Procedures." A maximum projection along the z axis is shown. In the merged image cytohesin 3 is pseudocolored green, and cytohesin 2 is pseudocolored red. Bar, 10 m. C, a maximum projection through the x axis of the deconvolved stack in B is shown in order to visualize the location of cytohesins 2 and 3 in the z dimension. Bar, 10 m.
Although ARNO and GRP1 are 80% identical overall, their coiled-coil domains are only 50% identical. Therefore, one explanation of the different subcellular locations of ARNO and GRP1 is that they may interact with different partners via their coiled-coil domains. Several scaffolding proteins have been identified that bind to the coiled-coil domain of ARNO and/or GRP1. GRASP/Tamalin and CYBR/CASP/CyTip can bind both ARNO and GRP1 (34 -36). IPCEF interacts with both ARNO and GRP1 by two-hybrid assay, although only the interaction with ARNO was confirmed with recombinant proteins (37). GRSP1 interacts with GRP1, and it is not known whether this protein can also bind to ARNO (38). It has been shown that these proteins co-localize with cytohesins and modify their ARF6-activating activity (37). Therefore, endogenous ARNO and GRP1 may be bound via their coiled-coil domains to different scaffolding proteins and thereby are recruited to different places.
A possible additional explanation for the different actions of ARNO and GRP1 is the presence of a protein kinase C site in the polybasic domain of ARNO that is absent in GRP1 (3). The cytohesin polybasic domain is an intramolecular autoinhibitory domain (39). The polybasic domain and the linker region between the Sec7 and PH domains act as a pseudosubstrate to occlude the active site. This autoinhibition is relieved either by binding of active ARF6 to the PH domain or by phosphorylation of polybasic domain. These alterations disrupt the pseudosubstrate interactions and allow the catalytic Sec7 domain to interact with and activate ARFs (39). Protein kinase C activation in ARNO-expressing HeLa cells produces extensive ruffling and actin rearrangement (3). Integrin ␤1 recycling may depend upon an activation of protein kinase C that GRP1 cannot respond to. Thus, future investigations into the mechanisms underlying the differing actions of ARNO and GRP1 during integrin ␤1 recycling will likely provide a better understanding of how migration is regulated by ARF. Such studies will also help to illuminate why ARF6 is subject to regulation by a large number of GEF proteins.