The Diaphanous-related Formin FHOD1 Associates with ROCK1 and Promotes Src-dependent Plasma Membrane Blebbing*

Diaphanous-related formins (DRFs) mediate GTPase-triggered actin rearrangements to regulate central cellular processes, such as cell motility and cytokinesis. The DRF FHOD1 interacts with the Rho-GTPase Rac1 and mediates formation of actin stress fibers in its deregulated form; the physiologically relevant activities and molecular mechanisms of endogenous FHOD1, however, are still unknown. Here we report that FHOD1 physically associates via the N-terminal part of its FH2 domain with the central domain of ROCK1. Although FHOD1 does not affect the kinase activity of ROCK1, the DRF is an efficient substrate for phosphorylation by ROCK1. Co-expression of FHOD1 and ROCK1 results in the generation of nonapoptotic plasma membrane (PM) blebs, to which the DRF is efficiently recruited. Blebbing induced by FHOD1 and ROCK1 depends on F-actin integrity, the Rho-ROCK cascade, and Src activity and is reminiscent of the recently described PM blebs triggered by expression of Src homology 4 (SH4) domain PM targeting signals. Consistently, endogenous FHOD1 is required in SH4 domain expressing cells for efficient PM blebbing and rounded cell morphology in two-dimensional cultures or three-dimensional matrices, respectively. Efficient association of FHOD1 with ROCK1, as well as recruitment of the DRF to blebs, depends on Src activity, suggesting that the functional interaction between both proteins is regulated by Src. These results define a role for endogenous FHOD1 in SH4 domain-induced blebbing and suggest that its activity is regulated by ROCK1 in a Src-dependent manner.


Diaphanous-related formins (DRFs) mediate GTPase-triggered actin rearrangements to regulate central cellular processes, such as cell motility and cytokinesis. The DRF FHOD1
interacts with the Rho-GTPase Rac1 and mediates formation of actin stress fibers in its deregulated form; the physiologically relevant activities and molecular mechanisms of endogenous FHOD1, however, are still unknown. Here we report that FHOD1 physically associates via the N-terminal part of its FH2 domain with the central domain of ROCK1. Although FHOD1 does not affect the kinase activity of ROCK1, the DRF is an efficient substrate for phosphorylation by ROCK1. Co-expression of FHOD1 and ROCK1 results in the generation of nonapoptotic plasma membrane (PM) blebs, to which the DRF is efficiently recruited. Blebbing induced by FHOD1 and ROCK1 depends on F-actin integrity, the Rho-ROCK cascade, and Src activity and is reminiscent of the recently described PM blebs triggered by expression of Src homology 4 (SH4) domain PM targeting signals. Consistently, endogenous FHOD1 is required in SH4 domain expressing cells for efficient PM blebbing and rounded cell morphology in two-dimensional cultures or three-dimensional matrices, respectively. Efficient association of FHOD1 with ROCK1, as well as recruitment of the DRF to blebs, depends on Src activity, suggesting that the functional interaction between both proteins is regulated by Src. These results define a role for endogenous FHOD1 in SH4 domain-induced blebbing and suggest that its activity is regulated by ROCK1 in a Src-dependent manner.
In response to intra-and extracellular cues, remodeling of the submembranous cytoskeleton constantly adjusts the plasma membrane (PM) 2 of eukaryotic cells. These cytoskeletal reorganizations are primarily controlled by small Rho-GTPases and their downstream signaling cascades, resulting in distinct types of invaginations or protrusions, depending on the specific set of GTPases and effectors involved. In addition to well described PM protrusions, such as lamellipodia and filopodia (1), under certain conditions, cells display on their surface highly dynamic rounded structures referred to as PM blebs (2). PM blebbing results from local destabilization of the cortical actin meshwork that causes expansion of the PM due to the osmotic pressure of the cell interior. Following this expansion phase, blebs typically briefly remain static before local actin polymerization and actin-myosin contraction events are thought to guide retraction of the bleb (3)(4)(5)(6)(7). PM blebbing has long been observed as an early event in apoptotic and necrotic processes (8 -10). More recently, nonapoptotic PM blebs were identified to play roles in distinct cellular processes, such as cytokine release, cytokinesis, embryonic stem cell motility, or cancer cell invasion (11)(12)(13)(14). Although PM blebbing seems to follow the common overall scheme of expansion and retraction, mechanistic differences exist between distinct types of nonapoptotic PM blebs, in particular in regard to the stimulus that initiates blebbing. Typically, blebbing is induced in a three-dimensional environment by yet to be identified stimuli and is thought to facilitate directed cell movement of, for example, tumor or germ cells (11, 14 -16). More amenable for molecular analysis, several model systems have been described in which cell blebbing is efficiently observed under two-dimensional cell culture conditions. Deficiency in the actin-binding protein filamin A (5,6) or the tumor suppressor p53 (17) as well as expression of the Dia-interacting protein DIP (18) or an effector loop mutant of active Rac1 (19) can cause efficient PM blebbing. We recently reported that expression of SH4 membrane-targeting domains, corresponding to an 18-aa short peptide with N-terminal acylation that mediates PM targeting of, for example, Src kinase family members, also induces PM blebbing in two-dimensional as well as three-dimensional cultures. In addition, SH4 domain-induced blebbing correlates with accelerated cell invasion and rounded cell morphology in three-dimensional matrices (20). Virtually all types of blebs analyzed so far share the dependence on the small GTPase Rho and its effector kinase ROCK (2). SH4 domain blebbing additionally involves activity of the protooncogene Src. How precisely actin destabilization and subsequent polymerization is achieved, however, has not been addressed for individual types of blebbing.
Of the four classes of actin nucleators described so far (Arp2/3 complex, Spire proteins, cordon-bleu, and formins), the last have been implicated in PM blebbing. Formins are modular proteins containing characteristic formin homology (FH) domains 1 and 2 that mediate binding to profilin as well as actin nucleation and bundling, respectively, and a less well defined FH3 domain. Actin nucleation mediated by formins results in unbranched filaments and occurs at the barbed end, at which the formin persists for protection from capping proteins (21). A subgroup of formin proteins, the Diaphanous-related formins (DRFs), contain an additional GTPase binding site and function in Rho-GTPase-regulated pathways (see Refs. 21 and 22 for a review). DRFs are autoinhibited by an intramolecular interaction between their C-terminal Diaphanous autoinhibitory domain (DAD) and the N terminus (23). Activation of DRFs includes release of their autoinhibition, achieved by binding to the GTPase and/or other cellular interaction partners (24 -26). The mammalian DRF family comprises the Dia, Daam, FRL, and FHOD formins that are distinct in their GTPase specificity, tissue distribution, and biological function. Regarding PM blebbing, Dia1 (an interaction partner and regulator of Rho) and mDia2 (an interaction partner of Cdc42, RhoA, and Rif) were shown to be critical for bleb formation of MDA-MB-435 cancer cells in three-dimensional matrices and DIP-expressing fibroblasts in two-dimensional ones, respectively (15,18). The role of, for example, Rac-interacting DRFs, such as FHOD1, in PM blebbing has not yet been addressed.
FHOD1 is currently one of the least well understood DRFs on the molecular and cellular level. Although interacting with Rac1, a constitutively active FHOD1 variant induces formation of F-actin stress fibers in a Rho-ROCK-dependent manner and coordinates F-actin and microtubule networks (27)(28)(29)(30)(31)(32). Expression of active Rac1 recruits FHOD1 to the PM but seems insufficient for its activation (29). FHOD1 shares typical features with other DRF family members, such as overall domain organization, multimerization, and autoinhibition (32)(33)(34), but its GTPase interaction and FH3 domains are not well characterized. By which mechanism autoinhibition of FHOD1 is released and which biological activities endogenous full-length FHOD1 exerts is not entirely clear; however, recent results imply that phosphorylation of the FHOD1 DAD by ROCK facilitates activation of the DRF and stress fiber formation in epithelial cells (35,36).
Here we describe that FHOD1 directly interacts with ROCK1 and find that both proteins synergize to promote PM blebbing. ROCK1/FHOD1-induced blebs displayed characteristics similar to the recently described blebbing induced by SH4 domains, including the requirement for Src activity. Interestingly, both endogenous FHOD1 and ROCK1 were required for efficient SH4 domain-induced blebbing. Since both physical and functional interactions between FHOD1 and ROCK1 required Src activity, these results imply that Src acts as key regulator of the functional interplay between FHOD1 and ROCK1 in SH4 domain-induced PM blebbing.
Constructs-The cDNA library from human peripheral blood mononuclear cells used in the yeast two-hybrid screen and cloned in the pGAD3S2X vector has been already described (38). DNA constructs for expression in yeast of the wild type (WT) and deleted forms (⌬C, 1-863, 1-807, and 1-611) of FHOD1 fused to LexA as well as mammalian expression constructs for HA-tagged FHOD1 and FHOD1-⌬C variants have been described previously (33). Plasmids for expression of Myctagged ROCK1 and ROCK1-⌬3 were gifts from Shuh Narumiya (Kyoto University Faculty of Medicine). The expression constructs for fusion proteins of wild-type or activated GTPases as well as for N18-HASPB-GFP and N18-Yes-WT-GFP were described earlier (20,29). The C3-GFP plasmid as well as expression plasmids for Myc-tagged GTPases were kindly FHOD1 in Membrane Blebbing provided by Stefan Offermanns (University of Heidelberg). Small interfering RNA oligonucleotides were purchased at MWG Biotech. Target sequences were FHOD1 (5Ј-UACCA-GAGCUACAUCCUUAUU-3Ј) and nonspecific control (5Ј-AGGUAGUGUAAUCGCCUUGUU-3Ј).
In Vitro Kinase Assay-Immunoprecipitation (IP) of Myctagged ROCK1 was performed as described previously (8). Briefly, 250 l of whole cell lysate derived from a 10-cm dish, which was grown with HeLa cells at 90% confluence were pretreated with 25 l of Protein G-Sepharose beads (50% suspension) for 30 min. IP was performed by the addition of 7 l of mouse anti-c-Myc antibody which was coupled to 30 l of Protein G-Sepharose beads (50% suspension). For the kinase reaction, beads were suspended in 30 l of kinase assay buffer (39), containing 4 Ci of [␥-32 P]ATP, 10 mM ATP, 3 or 5.6 g of the substrates MLC (Sigma) or FHOD1, respectively, and either 5.5 l of solvent or ROCK inhibitor (Y-27632; final concentration 10 M) were added for specificity control. Following 30 min of shaking at 30°C, the kinase reaction was stopped, subjected to separation by SDS-PAGE, and transferred by Western blotting to a polyvinylidene fluoride membrane, which was finally incubated for 24 h with a photosensitive film.
RNAi-For targeted knockdown of gene expression, doublestranded RNA oligonucleotides were transfected into target cells by lipofection. For transfection, 3.0 ϫ 10 5 HeLa cells, 1.5 ϫ 10 5 CHO cells, or 2 ϫ 10 5 SYF plus c-Src cells were seeded in a 6-well format. On the following day, cells were washed twice with phosphate-buffered saline before Opti-MEM I (Invitrogen) was added as indicated. For transfection with Oligofectamine (HeLa cells; Invitrogen) or Lipofectamine 2000 (CHO and SYF ϩ c-Src cells; Invitrogen), the transfection reagent and double-stranded RNA were separately resolved in Opti-MEM I, mixed, incubated for 10 min at room temperature, combined, and incubated again for 15 min at room temperature. Finally, the mixture was carefully added to the cells dropwise. After 3 h of incubation in the incubator, either 2-fold amounts of medium were added to the transfection solution (HeLa cells) or transfection solution was removed first before medium was added (CHO and SYF ϩ c-Src cells). On the following day, cells were split and seeded on coverslips in a 6-well format. 48 h after the first transfection, cells were once more transfected as described above. Further transfections with plasmid DNA were subsequently performed, or expression of SH4 domains was induced by the addition of 1 g/ml doxycycline. On the following day, cells were harvested for microscopic analysis and Western blot analysis of knockdown efficiency.
Drug Treatment-HeLa, SYF Ϫ/Ϫ , or SYF ϩ c-Src cells expressing ROCK1 either alone or in combination with FHOD1, N18-Yes-GFP, or N18-HASPB-GFP were synchronized in PM bleb formation. Therefore, membrane blebbing was efficiently abrogated by treatment with Y-27632 (SYF cells, Induction of Apoptosis-For induction of apoptosis, HeLa cells grown on coverslips were treated for 3 h with a combination of 50 g/ml recombinant human tumor necrosis factor ␣ and 5 g/ml cycloheximide. Apoptotic HeLa cells were determined by an Annexin V-mediated surface-specific stain of externalized phosphatidylserine. Therefore, living cells were carefully washed with AB buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl 2 ) and incubated with staining solution consisting of AB buffer containing fluorescently labeled Annexin V-Alexa-Fluor-568 at a dilution of 1:50 (v/v) for 15 min at room temperature. Subsequently, cells were fixed and stained for F-actin.
Co-immunoprecipitation-To analyze FHOD1-ROCK1 association by co-immunoprecipitation, 2 ϫ 10 6 COS-7 cells were co-transfected with 10 g of the indicated plasmid DNA using the electroporation method. PP1, PP2, or PP3 was added to transfected cells for 4 h with PP1 and PP3 in normal conditions or overnight with PP2 at 37°C in serum-starved conditions (Dulbecco's modified Eagle's medium supplemented with 0.1% bovine serum albumin and 10 mM HEPES, pH 7.5). 48 h after transfection, cells were lysed in assay buffer (25 mM Tris-HCl (pH 8), 2 mM EDTA, and 150 mM NaCl) containing 1% Triton for 30 min at 4°C. The cleared lysate was precipitated with an anti-Myc antibody in the presence of protein G-Sepharose and incubated overnight at 4°C. The immunoprecipitates were then analyzed by Western blotting.
Yeast Two-hybrid Analysis-All of the procedures used to analyze interactions in the two-hybrid system were performed in the L40 yeast reporter strain as previously described in detail (41). The screening of the peripheral blood mononuclear cell cDNA library was performed using the WT LexA-FHOD1 fusion as bait. Plasmid DNA was rescued from positive clones and tested for specificity by retransformation into L40 with WT FHOD1 or the Ras extraneous target as described previously (41).
Invasion Assay-For invasion assays, CHO-N18-HASPB-GFP cells were treated with control or FHOD1-specific RNAi, and SH4 domain expression was induced by the addition of doxycycline as described above. Transwell inserts (Greiner Bio-One) were coated with 50 l of growth factorreduced Matrigel (BD Biosciences). 8,500 cells were seeded and allowed to adhere before the lower chamber was filled with 1 ml of minimum Eagle's medium containing 0.5% fetal bovine serum and the upper chamber was filled with 200 l of medium. Invasion assays were stopped after 24 h, and cells were fixed with 8% paraformaldehyde and visualized using rhodamine-phalloidin and 4Ј,6-diamidino-2-phenylindole. Invasion assays were analyzed by taking z intervals of 5 m with a ϫ20 objective using confocal microscopy (Leica TCS SP2). For quantification, the amount of invaded cells in each optical section with a distance of more than 10 m from the membrane from nine randomly chosen fields was counted. For imaging cell morphology of invaded cells, single confocal sections were taken with a ϫ20 objective. OCTOBER

RESULTS
ROCK1 and FHOD1 Physically Interact-In a search for FHOD1-interacting proteins, we performed a yeast two-hybrid screen using full-length FHOD1 as a bait and a human peripheral blood mononuclear cell cDNA library. One of the positive clones that interacted with FHOD1 was identified as a fragment encompassing aa 368 -1357 of ROCK1, as indicated by growth on medium lacking histidine as well as activation of the ␤-galactosidase reporter gene (Fig. 1, A and B). The specificity of this interaction was validated by the lack of interaction of FHOD1 or ROCK1 with the Ras and Raf negative controls, respectively (Fig. 1B). The analysis of various FHOD1 fragments indicated that the constitutively active FHOD1⌬C (which lacks aa 1011-1164 and thus its autoregulation domain) can also interact with ROCK1 and mapped the ROCK1 interaction to the N-terminal region of the FHOD1 FH2 domain (aa 612-807) (Fig. 1C). Coimmunoprecipitation experiments from COS-7 cells expressing epitope-tagged ROCK1 and FHOD1 confirmed the specific association of the proteins (Fig. 1D). WT FHOD1 as well as FHOD1⌬C also associated efficiently with the constitutively active ROCK1⌬3 (aa 1-727) variant, indicating that aa 368 -727 of ROCK1 are required for the association with FHOD1. Association with ROCK1 was also detected for FHOD1 variants lacking the entire FH1 domain (⌬FH1, lacking aa 570 -610) or the core region of the FH2 domain in the C-terminal part (⌬FH2, lacking aa 807-866) (data not shown). Together, FHOD1 physically associates with ROCK1 in yeast and mammalian cells, and this interaction is governed by the N-terminal half of the FHOD1 FH2 domain and the central core region of ROCK1.
FHOD1 and ROCK1 Synergize for Efficient PM Blebbing-To address potential functional consequences of the interaction of FHOD1 with ROCK1, both proteins were co-expressed in HeLa cells (Fig. 2). 24 h post-transfection with expression constructs for HA-tagged FHOD1 and Myc-tagged ROCK1, cells were fixed, stained for F-actin and the overexpressed proteins, and then analyzed by microscopy. Expectedly (8,9), cells expressing ROCK1 alone typically (61.6 Ϯ 6.8% of all cells) displayed large blebs emanating from the PM, with ROCK1 being enriched in the limiting membrane of the blebs as well as in the bleb lumen. Such ROCK1-induced PM blebs revealed an enrichment of F-actin at the basis as well as at the outer rim (Fig. 2, A (bd) and B). Similar structures were rarely observed in cells expressing FHOD1 alone (Fig. 2, e-h) that displayed the typical cytoplasmic localization of the DRF and normal F-actin organization or in vector-transfected control cells (13.3 Ϯ 1.6% and 9.2 Ϯ 3.0% of all cells, respectively). Co-expression of FHOD1 with ROCK1, however, markedly changed cell morphology, resulting in the generation of a large number of relatively small PM blebs. On average, most of the cells showed more than 12 blebs/cell (approximately 77% compared with 39% of solely ROCK1-expressing cells; Fig. 2D), and the size of individual PM blebs was relatively reduced (ϳ75% smaller than 3.5 m in diameter, as compared with 49% in ROCK1-expressing cells; Fig. 2C). Importantly, virtually all cells (92.8 Ϯ 3.7%) that were positive for ROCK1 and FHOD1 displayed PM blebbing, demonstrating a significant increase in PM blebbing efficacy when compared with the expression of ROCK1 alone (Fig. 2B). Although the subcellular localization of ROCK1 in blebbing cells remained unchanged upon co-expression with FHOD1, the DRF was efficiently recruited to the bleb membrane and lumen in the presence of ROCK1. Blebs generated in the presence of both ROCK1 and FHOD1 were also more enriched in F-actin than ROCK1 blebs ( Fig. 2A, i). PM blebbing induced by ROCK1 or by ROCK1 and FHOD1 did not result in rapid cell death, exposure of phosphatidylserine, or nuclear condensation and was thus unrelated to apoptosis (Fig. S1). We conclude that FHOD1 and ROCK1 synergize to efficiently generate nonapoptotic PM blebs.
FHOD1/ROCK1 PM Blebbing Depends on Rho Signaling, F-actin Integrity, and Src Activity-To obtain a more detailed picture of the signaling pathway that controls FHOD1/ROCK1stimulated PM blebbing, we first analyzed the involvement of individual Rho-GTPases (Fig. 3). FHOD1 was co-expressed together with ROCK1 and a GFP fusion protein of either dominant negative Rac1 (Rac1-N17), Cdc42 (Cdc42-N17), the Clostridium botulinum C3 transferase (for inactivation of Rho), or a GFP control. Expression of dominant negative Rac1 or dominant negative Cdc42 had no inhibitory effect on PM blebbing but rather caused an increase in blebbing. In contrast, inhibition of Rho efficiently prevented PM blebbing induced by ROCK1 alone or by ROCK1 and FHOD1 (Fig. 3, A and B). Consistently, inhibition of ROCK activity with Y-27632 potently abolished PM blebbing induced by ROCK1 alone or in combination with FHOD1 (Fig. 3C). Both types of blebs also displayed comparable sensitivity to F-actin disruption, with latrunculin B being more efficient than cytochalasin D. In contrast, disruption of microtubules by nocodazole had no effect. However, the use of further pharmacological inhibitors revealed statistically significant differences between blebs induced by ROCK1 alone or in conjunction with FHOD1; blebbistatin, an inhibitor of the myosin II motor protein, reduced blebbing in both cases. However, ROCK1 blebs were markedly more sensitive to the treatment than ROCK1/FHOD1 blebs. Finally, the Src inhibitor PP1 had virtually no effect on ROCK1 blebs but reduced PM blebbing of cells co-expressing ROCK1  OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 and FHOD1 to levels achieved by ROCK1 only. This suggests that PM blebs generated in the presence of FHOD1 and ROCK1 represent protrusions governed by a machinery including Src that is dispensable for the generation of blebs by ROCK1 alone.

FHOD1 in Membrane Blebbing
Active Forms of ROCK1 and Src Localize in FHOD1/ROCK1induced PM Blebs-We next asked whether ROCK1 and Src act locally during bleb formation and if they are active in PM blebs induced by expression of ROCK1 alone or in combination with FHOD1. To localize the activity of ROCK1, the staining pattern of endogenous myosin light chain (MLC), a major substrate of ROCK, was investigated. Overall, MLC was detected at the PM and along stress fibers in control cells (vector) and was found enriched in bleb membrane and lumen in ROCK1-and ROCK1/FHOD1-expressing cells (Fig. 4A). This distribution was indistinguishable from that of the phosphorylated, active pMLC (Fig.  4B). Together and consistent with the inhibitory effects of Y-27632, these results suggest that ROCK1 acts locally at both types of blebs to provide contractility. The analysis of Src distribution revealed a pattern reminiscent of microtubules and a significant localization of the kinase to ROCK1 as well as ROCK1/ FHOD1 PM blebs (Fig. 4C). The use of a phosphospecific antibody to detect active Src revealed a marked accumulation of phospho-Src at the PM of control cells and a moderate presence at ROCK1 PM blebs. In line with the observed sensitivity to the Src inhibitor PP1 (Fig. 3C), active Src was significantly more enriched in PM blebs induced by coexpression of ROCK1 and FHOD1 (Fig. 4D). These results suggested that efficient PM blebbing induced by ROCK1 and FHOD1 involves the localized activity of ROCK1 as well as of Src. Based on this requirement for Rho-ROCK, F-actin, and Src, FHOD1/ROCK1 blebs appear strikingly related to recently described blebs induced by overexpression of SH4 membrane-targeting domains (20).
The Association between ROCK1 and FHOD1 Is Regulated by Src Activity-Since FHOD1 and ROCK1 functionally synergized for efficient PM blebbing in a Src-dependent manner, we next addressed whether Src is involved in the regulation of the physical association between FHOD1 and ROCK1 and performed co-precipitation experiments in the presence of increasing amounts of the Src inhibitor PP1 (Fig. 5A, left). Of note, treatment with PP1, which interfered with FHOD1/ ROCK1-mediated PM blebbing (Fig. 3C) substantially reduced the association of FHOD1 with WT or ⌬3 ROCK1 proteins without affecting FHOD1 or ROCK1 expression levels. Similar interference with the FHOD1-ROCK1 interaction was also obtained with the structurally unrelated Src inhibitor PP2 (Fig. 5A, right). No disruption of the association between FHOD1 and ROCK1 was observed upon treatment with the inactive structural homologue PP3 (Fig. 5B). Of note, also endogenous FHOD1 protein was found to associate with ROCK1 in a Srcdependent manner (Fig. 5C). These results were confirmed by localization analyses in intact cells; although FHOD1 was effi-ciently recruited to the PM and to blebs in the presence of ROCK1, interference with blebbing by inhibition of Src activity abrogated PM localization in the presence of the kinase (Fig. 5D). The localization of ROCK1 as well as PM blebbing induced by expression of ROCK1 alone, however, were unchanged upon inhibition of Src activity (Fig.  5, D and E). We conclude that Src activity is essential for the efficient association between FHOD1 and ROCK1 in mammalian cells and governs the recruitment of FHOD1 to the PM in the presence of ROCK1.
FHOD1 Does Not Affect Overall ROCK1 Activity but Is a ROCK1 Substrate-To address if overall ROCK1 activity was affected by coexpression of FHOD1, we scored its ability to phosphorylate a MLC substrate in vitro following immunoisolation from transfected cells (Fig.  6A). Although negative controls (lanes 1 and 6) expectedly did not yield phosphorylation of MLC, active ROCK1⌬3 generated high levels of pMLC (lane 5). WT ROCK1 (lane 2) resulted in lower but significant pMLC levels that were reduced to background in the presence of the specific ROCK1 inhibitor Y-27632 (lane 4). Co-expression of FHOD1 with ROCK1 (lane 3) had only minor effects and did not result in elevated ROCK activity, ruling out that FHOD1 functionally synergizes with ROCK1 via general up-regulation of its kinase activity. To test whether FHOD1 could instead be subject of regulation via phosphorylation by ROCK1, we repeated the kinase reaction using recombinant purified FHOD1 as substrate for immunoprecipitated ROCK1 (Fig. 6B). Indeed, potent phosphorylation of full-length FHOD1 as well as of a truncated FHOD1 fragment were observed when FHOD1 protein was added to immunoprecipitates containing active ROCK⌬3 (lane 4). This reaction was fully inhibited by the addition of Y-27632 (lane 5) and much more pronounced than nonspecific background phosphorylation observed in the absence of ROCK1 (lane 2). Immunoisolated WT ROCK1 only displayed weak activity in this assay, resulting in a slight  OCTOBER  domain-induced blebbing in cells lacking Src expression. To this end, SYF cells that lack expression of Src, Yes, and Fyn (SYF Ϫ/Ϫ ) or SYF cells with reconstituted Src expression (SYF ϩ c-Src) were used. Expectedly, expression of the Yes SH4 domain fused to GFP (N18-Yes-GFP) caused pronounced PM blebbing in more than 60% of the transfected cells in SYF ϩ c-Src cells, whereas only background blebbing activity was observed in SYF Ϫ/Ϫ cells lacking Src (Fig. 7, A and B). SH4 domain-induced blebbing in SYF ϩ c-Src cells was suppressed by PP1 in a dose-dependent manner. Similar results were obtained in SYF cells co-expressing ROCK1/FHOD1 (data not shown).

FHOD1 in Membrane Blebbing
These results now allowed us to test whether endogenous FHOD1 is involved in Src-dependent PM blebbing induced by SH4 domains. To this end, FHOD1 expression was efficiently reduced by specific RNAi-mediated knockdown in SYF ϩ c-Src cells (Fig. 7E, top). Importantly, PM blebbing was significantly reduced albeit not abrogated from 64.5 Ϯ 1.4% in the control to 39.3 Ϯ 7.1% upon reduction of FHOD1 expression (Fig. 7, C and  D). Similar results were obtained in N18-HASPB-GFP-expressing CHO cells; although the SH4 domain induced efficient PM blebbing in cells treated with control RNAi (79.6 Ϯ 4.3%), reduction of FHOD1 expression significantly decreased the number of blebbing cells (42.9 Ϯ 16.8%) (Fig. 7, C-E, middle). Such ϳ2-fold reduction in blebbing efficiency was also observed under knockdown conditions where FHOD1 was undetectable by Western blotting (data not shown), suggesting that FHOD1 is critically involved in but is not essential for SH4 domain-induced PM blebbing.
To analyze the specificity of FHOD1 for SH4 domain-induced blebs, we next tested the involvement of endogenous FHOD1 in various types of PM blebbing by RNAi in HeLa cells (Fig. 7E, bottom). In line with the nonapoptotic nature of ROCK1/FHOD1 blebs, FHOD1 knockdown had no effect on the morphology (data not shown) and frequency of apoptotic blebs induced by tumor necrosis factor ␣/cycloheximide treatment (Fig. 7F). Similarly, FHOD1 expression levels were not limiting for PM blebbing induced by overexpression of ROCK1 alone (Fig. 7G), confirming that PM blebs induced by ROCK1 or ROCK1/FHOD1 require distinct machineries. Since FHOD1 physically interacts with Rac1 and a specific Rac1 effector loop FIGURE 5. The interaction of FHOD1 with ROCK1 requires Src kinase activity. A, co-immunoprecipitation of FHOD1 by ROCK1. COS-7 cells expressing HA-tagged FHOD1-WT and the indicated Myc-tagged ROCK1 constructs were treated with the indicated amounts of Src kinase inhibitor PP1 (left) or PP2 (right) for 4 h in medium supplemented with 5% fetal calf serum (PP1) or overnight in medium supplemented with 0.1% fetal calf serum (PP2). Lysates were subjected to IP with an anti-Myc antibody, separated by SDS-PAGE, and analyzed by Western blotting with either anti-Myc or anti-HA antibodies. B, co-immunoprecipitation of FHOD1 by ROCK1 as in A but in presence of the inactive PP2 structural homologue PP3. C, Src-dependent association of endogenous FHOD1 with ROCK. Myc-tagged WT ROCK1 was immunoisolated from COS-7 cells treated with the indicated concentrations of PP1, and the immunoprecipitates were analyzed for the presence of endogenous FHOD1. D and E, confocal micrographs of HeLa cells co-expressing ROCK1 and FHOD1 (D) or expressing ROCK1 alone (E). Cells were treated either with solvent or 10 M PP1 for 3 h, fixed, and stained for ROCK1 and FHOD1. Shown are representative cells for co-expression of Myc-ROCK1 and HA-FHOD1 (PM blebs in the control, no PM blebs after PP1 treatment) or ROCK1 alone (PM blebs at both conditions). See Fig. 3C for quantification of these phenotypes. Scale bars, 10 m. FIGURE 6. FHOD1 fails to activate ROCK1 but represents a ROCK1 substrate. A, in vitro kinase assay of immunoprecipitated Myc-tagged ROCK1 transiently expressed alone or in combination with HA-tagged FHOD1 in HeLa cells. The kinase reaction was started by the addition of the immunoprecipitates to kinase activation buffer containing 4 Ci/30 l [␥-32 P]ATP and 1 g/30 l MLC as a substrate together with either solvent or Y-27632 (10 M). After 30 min at 30°C, the kinase reaction was terminated by addition of 6ϫ SDS sample buffer, and proteins were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane that was exposed to a photosensitive film for 24 h. Depicted are the signals obtained from the radioactively labeled MLC, the amount of overall MLC stained with Amido Black dye, the input levels of Myc-ROCK1 and HA-FHOD1, and tubulin levels as loading control. B, in vitro kinase assay as in A but using purified recombinant FHOD1 as a substrate. WB, Western blot. OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 mutant (L61A37) has been reported to induce PM blebbing (19,27,29), we also tested if FHOD1 knockdown affects PM blebbing induced by Rac1-L61A37. Although this Rac1 variant efficiently induced large numbers of relatively small nonapoptotic PM blebs, FHOD1 RNAi caused only a slight and statistically nonsignificant reduction of blebbing (Fig. 7H). Together, these results define FHOD1 as a DRF specifically involved in PM blebbing induced by SH4 domains.

FHOD1 in Membrane Blebbing
We next sought to address whether the support of SH4 domain-induced PM blebbing by FHOD1 had functional consequences. PM blebbing can facilitate invasion of tumor cells that use a rounded, amoeboid mode of motility in three-dimensional environments. This mode of invasion is in contrast to the mesenchymal type of cell movement used by other tumor cells, where more elongated cells achieve motility by proteolytic degradation of the surrounding matrix (42). Since SH4 domain-expressing cells adopt a rounded cell morphology in a three-dimensional environment and invade into three-dimensional Matrigel (20), we analyzed cell invasion and morphology of these cells treated with control or FHOD1-specific RNAi (Fig. 8). Surprisingly, we noted in three independent experiments that efficient reduction of FHOD1 expression significantly enhanced invasion of SH4 domain-expressing cells into Matrigel (Fig. 8, A and B). Importantly, enhanced cell invasion upon FHOD1 knockdown coincided with a relative decrease in cells that displayed rounded cell morphologies (Fig. 8C). Instead, the majority of the cells adopted an elongated cell shape with pronounced differences in the length of the long and short axis of the cell. Such elongated cells were 2-5 times more frequent upon FHOD1 knockdown as compared with RNAi control-treated cells. We conclude that FHOD1 is involved in the regulation of cell morphology and invasion efficiency of SH4 domain-expressing cells in three-dimensional matrices.

DISCUSSION
We report here the physical interaction of the DRF FHOD1 with the Rho effector kinase ROCK1. Initially identified as direct interaction partners by a yeast two-hybrid screen, both proteins were found to associate in intact cells, and FHOD1 was recruited to the PM in the presence of the kinase. At the functional level, co-expression of both proteins resulted in efficient PM blebbing in a Rho-ROCK-, F-actin-, and Src-dependent manner. These characteristics were consistent with those of SH4 domain-triggered PM blebbing, and RNAi knockdown analysis revealed that endogenous FHOD1 is specifically involved in PM blebbing mediated by these membrane-targeting signals. Importantly, the requirement for Src activity in FHOD1-mediated PM blebbing was reflected in the need of Src for the efficient physical association of FHOD1 with ROCK1 and the PM recruitment of the DRF by the kinase.
The data presented provide important clues regarding the molecular mechanism by which FHOD1 and ROCK1 synergize to induce PM blebbing. FHOD1 did not influence the subcellular localization of ROCK1 and did not impact on its kinase activity. These effects are therefore clearly distinct from a recently described mechanism where the ability of ROCK1 to mediate membrane blebbing is governed by PDK1, a kinase that replaces the negative regulator RhoE on ROCK1, thereby ensuring the PM localization of ROCK (43). Rather, our data are consistent with a regulation of FHOD1 activity via its interaction with ROCK1. One important aspect of this regulation is related to the recruitment of FHOD1 to the PM in the presence of ROCK1. Our results do not allow us to conclude whether this translocation is a direct consequence of the physical interaction of FHOD1 with ROCK1. Although we were able to roughly map the interaction to the N-terminal region of the FH2 domain of FHOD1, we failed to generate a stable FHOD1 mutant that does not interact with ROCK1 but retains other biological activities (data not shown), precluding direct assessment of this question. However, the fact that FHOD1 can be potently phosphorylated by ROCK1 in vitro suggests that the regulation of FHOD1 activity may be direct. Similar to other DRFs, FHOD1 is autoinhibited by an intramolecular interaction between the C-terminal DAD and the FH3 domain (23,34). Although this autoinhibition can be efficiently released by the interaction with a GTPase (e.g. in the case of Dia1 and Rho-GTP (25)), expression of an active variant of the FHOD1-interacting GTPase Rac1 does not result in full activation of the DRF in cells (29). Other stimuli must therefore be required for activation of FHOD1. Conceivably, phosphorylation of residues present in the protein surfaces involved in the autoinhibitory interaction may facilitate autoinhibition release. Indeed, the FHOD1 C terminus is known to be subject to phosphorylation (36), and results presented here consistently demonstrate that FHOD1 is a ROCK1 substrate in vitro. These results are entirely consistent with work published during preparation of this manuscript revealing that in intact mammalian cells, phosphorylation of three serine residues in the FHOD1 DAD by ROCK1 causes activation of the DRF (35). Together, these studies imply that FHOD1, after associating with ROCK1 at the PM, is subsequently activated by phosphorylation and then synergizes with the kinase in cytoskeletal remodeling. This mechanism can lead to formation of actin stress fibers in thrombin-treated endothelial cells (35) or PM blebbing (this study). The functional outcome of ROCK1-mediated activation of FHOD1 probably depends on the cell system and upstream trigger that initiates actin remodeling and on the subcellular localization at which the FHOD1-ROCK1 interaction occurs. In line with this scenario, constitutive active FHOD1⌬C associates with ROCK1 but does not facilitate PM blebbing (data not shown), presumably due to its localization to cytoplasmic stress fibers.
Another important aspect of our study is the involvement of endogenous Src activity in the functional and physical interaction between FHOD1 and ROCK1. Although the precise molecular mechanism remains unclear, our results are most compatible with a scenario where Src-mediated phosphorylation events facilitate the physical interaction between FHOD1 and ROCK1. Either in the context of a tripartite complex or via a stepwise process where the association with Src facilitates a subsequent interaction of FHOD1 with ROCK1, this governs sustained PM recruitment of the active DRF and thus PM blebbing. This would explain why PM blebbing induced by ROCK1 alone, which occurs via an FHOD1-independent mechanism, is not regulated by Src. Src has been reported to regulate cytoskeletal remodeling by various formins, including FHOD1 (44 -47). Our results therefore underscore the general role of Src for DRF activity and suggest that Src may directly regulate protein interactions of DRFs.
It is important to note that PM blebbing was reduced but not abolished in the RNAi experiments despite strong reduction of FHOD1 protein levels and that, once reduced below an assumed critical concentration, further decrease in FHOD1 expression by more efficient knockdown did not result in further reduction of PM blebbing. We cannot exclude the possibility that such residual blebbing reflects the activity of a cellular FHOD1 pool refractory to RNAi by our experimental approach. More likely, even if FHOD1 is necessary for optimal PM blebbing, some other factors are sufficient to maintain appreciable blebbing in its absence. Whether this involves the closely related FHOD2, other DRF proteins, or additional actin regulators is currently being investigated. However, recent studies revealed that specific DRFs are involved in various types of PM blebbing; human Dia1 is critical for blebbing and motility of MDA-MB-435 cancer cells in three-dimensional environments (15), and mammalian Dia2 mediates blebbing upon overexpression of DIP (18). Thus, DRFs emerge as important regulators of PM blebbing, with individual family members apparently facilitating distinct types of membrane blebbing. This specificity may reflect the particular trigger for PM blebbing in individual cases, suggesting that GTPase specificity of a given DRF may determine in which kind of PM dynamization event it will be involved. An important open question is what exactly the DRFs contribute to facilitate the blebbing process. Although this remains formally to be demonstrated for FHOD1, the enrichment of F-actin and the DRF at the bleb base and lumen, respectively, indicates that its involvement in blebbing may reflect nucleation of new actin filaments, which may synergize with ROCK-mediated contractility for bleb retraction. Since actin nucleation is a conserved feature of DRFs, this may apply to many different types of blebbing. The involvement of individual DRFs in select types of blebbing indicates that, depending on the individual DRF involved, blebbing could also be facilitated by less well conserved DRF activities, such as bundling of actin filaments (48) or overall activation of Rho signaling (15).
Another important issue is the physiological relevance of FHOD1-mediated PM blebbing. Since SH4 domain expression promotes PM blebbing and cell invasion in Matrigel, we hypothesized that FHOD1 knockdown would diminish invasion. Instead, an increase in invasion efficacy was observed that correlated with the adoption of a more elongated rather than rounded cell morphology. These changes in cell morphology were presumably facilitated by the weakening of PM-cortex interactions due to expression of the SH4 domain and by reduced contraction in FHOD1 knockdown cells. Similar changes in cell morphology of invading cells have already been reported, albeit in the opposite direction; tumor cells employing an elongated invasion mode readily adopt rounded cell morphology and blebbing motility when matrix degradation is inhibited (49). In analogy, our results may indicate that the reduction of FHOD1 expression not only caused morphology alteration but triggered the cells to switch to an elongated mesenchymal mode of motility. These results emphasize DRFs as versatile regulators of cell morphology and invasiveness and warrant a detailed analysis of the specific role of DRFs for the invasive potential of individual types of cancer cells.