Requirements for F-BAR Proteins TOCA-1 and TOCA-2 in Actin Dynamics and Membrane Trafficking during Caenorhabditis elegans Oocyte Growth and Embryonic Epidermal Morphogenesis

The TOCA family of F-BAR–containing proteins bind to and remodel lipid bilayers via their conserved F-BAR domains, and regulate actin dynamics via their N-Wasp binding SH3 domains. Thus, these proteins are predicted to play a pivotal role in coordinating membrane traffic with actin dynamics during cell migration and tissue morphogenesis. By combining genetic analysis in Caenorhabditis elegans with cellular biochemical experiments in mammalian cells, we showed that: i) loss of CeTOCA proteins reduced the efficiency of Clathrin-mediated endocytosis (CME) in oocytes. Genetic interference with CeTOCAs interacting proteins WSP-1 and WVE-1, and other components of the WVE-1 complex, produced a similar effect. Oocyte endocytosis defects correlated well with reduced egg production in these mutants. ii) CeTOCA proteins localize to cell–cell junctions and are required for proper embryonic morphogenesis, to position hypodermal cells and to organize junctional actin and the junction-associated protein AJM-1. iii) Double mutant analysis indicated that the toca genes act in the same pathway as the nematode homologue of N-WASP/WASP, wsp-1. Furthermore, mammalian TOCA-1 and C. elegans CeTOCAs physically associated with N-WASP and WSP-1 directly, or WAVE2 indirectly via ABI-1. Thus, we propose that TOCA proteins control tissues morphogenesis by coordinating Clathrin-dependent membrane trafficking with WAVE and N-WASP–dependent actin-dynamics.


Introduction
The coordination and functional cooperation between endocytic trafficking of membranes and membrane proteins with actinbased motility is required for the correct execution of many cellular phenotypes. These include directional cell migration, tissue morphogenesis, cell-fate determination, and the establishment of cell polarity in epithelial and in neuronal cells. Consistently, endocytic trafficking and actin-based motility and dynamics are intimately linked. Results obtained in several species have established that endocytosis and trafficking events rely on propelling forces generated by actin treadmilling [1]. Consistent with these results, an increasing number of actin binding and regulatory proteins have been shown to participate in a variety of internalization and trafficking processes, ultimately controlling the signaling response of cells to extracellular stimuli. In addition, genetic and cellular biochemical evidence has revealed how cycles of endocytosis and recycling of plasma membranes and plasma membrane proteins are essential to promote the spatial restriction of signaling [2][3][4][5].
However our understanding of the molecular circuitry involved in these processes is still in its early stages. Proteins that sit at the crossroads of membrane remodeling and actin dynamics are predicted to play a prominent role in these processes, simultaneously binding regulators of actin dynamics and sensing or inducing membrane curvature. A prototypical example of this kind of protein is the BAR (Bin, Amphiphysin, Rvs) domain superfamily of proteins-including ''classical'' BAR domains, F-BAR (FCH-BAR or EFC Extended-FCH) and I-BAR (Inverse-BAR) domains, which have emerged as important players in membrane-remodeling processes [6]. Members of the superfamily are recruited from the cytoplasm to trigger the formation of plasma-membrane extensions, invaginations, tubular organelles, and transport intermediates, including endocytic vesicles [7][8][9][10][11][12]. Much of what is known about the structure-function relationships of the BAR superfamily has been obtained from crystallographic and in vitro biochemical studies, showing that members of the family are elongated dimers formed by the antiparallel association of a-helical coiled coils which can deform liposome into tubules of different diameters [13][14][15][16][17][18][19].
A defining feature of a subfamily of F-BAR-containing proteins that includes three mammalian members, TOCA-1 (Transducer of Cdc42 dependent actin assembly), CIP4 (Cdc42 interacting protein 4) and FBP17 (Formin binding protein 17) (hereafter referred to as the TOCA family), is to possess additional protein:protein interaction domains that enables them to function as signal transducers, physically bridging membrane trafficking with signalling that controls actin dynamics. TOCA-1 and CIP4, through their imperfect HR1/CRIB-like (Cdc42 and Rac interacting binding motif) domain act as direct downstream effectors of the small GTPase Cdc42 [20][21][22]. In addition all members of the TOCA family bind, through their conserved C-terminal SH3 domain, to either prototypical endocytic proteins such as dynamin [6,8,12], or to actin nucleation promoting factors (NPFs) such as N-WASP and WASP [12,[22][23][24]. In this latter case, the association of TOCA-1 with the inhibited WASP-WIP complex has been shown to be critical for the activation of Arp2/3-mediated actin polymerization induced by Cdc42 [18,22,24].
Despite this wealth of structural and biochemical observations, the functional and cellular roles of the TOCA family proteins have remained largely elusive. Consistent with their biochemical properties, concomitant downregulation of TOCA-1 and FBP17 in vivo resulted in a relative slight inhibition of Transferrin receptor internalization [6,12,18]. Whereas the ectopic expression of FBP17 and CIP4, but not of TOCA-1, caused the appearance of membrane tubules, whose accumulation was enhanced by inhibition of dynamin or actin dynamics [6,8,12]. Recently, by somatic gene disruption in dorsal epithelial cells, the only Drosophila TOCA-family was demonstrated to mediate E-cadherin endocytosis in conjunction with the Cdc42/Par6/aPKC polarity complex [25]. However the precise molecular details in this pathway remain to be elucidated.
Here, by combining genetic approaches in the nematode C. elegans with cellular biochemical analysis in mammalian cells we have identified a requirement of the TOCA subfamily of proteins in WASP and WAVE-dependent pathways controlling actin dynamics and membrane trafficking. Specifically we find that CeTOCA-1 and CeTOCA-2 are important for the regulation of Clathrin-mediated endocytic processes during oocyte growth, and in the control of epithelial morphogenesis in developing embryos. Remarkably, mammalian TOCA-1, like C. elegans CeTOCA-2, associates with ABI-1, a key member of the WAVE complex. Furthermore mammalian TOCA-1 localizes at tight junction in epithelial cells, suggesting that this function may be conserved.

Results
The C. elegans TOCA family comprises two genes products, CeTOCA-1 and CeTOCA-2 In C. elegans two distinct genes display a significant level of overall similarity to the three mammalian members of the TOCA family: CeTOCA-1 and CeTOCA-2. CeTOCA-1 and CeTOCA-2 contain, like their mammalian counterparts, a predicted N-terminal extended FCH or F-BAR domain, a central Cdc42-binding HR1 region, and a C-terminal SH3 domain ( Figure S1A). The secondary structure prediction of the N-terminus of the C. elegans TOCA-1 protein is in good agreement with the one described for the human F-BAR domain of FBP17 [18] ( Figure S1A). Thus we built a structural atomic model of the F-BAR domains of C. elegans CeTOCA-1. With an estimated precision of 100% (E-value 2.6 e 230 ), the model predicted this domain to fold into a nearly flat zeppelin shape. This analysis also showed full conservation of all key cationic residues required for membrane lipid bending ( Figure 1A), suggesting that the biochemical functions of this domain in CeTOCA-1 is equivalent to that of its mammalian homologues. Consistent with this prediction, the ectopic expression into mammalian cells of GFP-tagged CeTOCA-1 and CeTOCA-2 induced the formation of tubular/vesicular-like structures in nearly 100% of cells expressing the transgene ( Figure 1B), like their mammalian homologues [6]. Next, we tested whether the SH3 domain of CeTOCA-1 is also functional and able to associate with one of the known mammalian ligands, N-WASP or C. elegans WSP-1. Indeed, immobilized GST-SH3 domains of CeTOCA-1 and CeTOCA-2 bound to mammalian N-WASP and CeWSP-1 ( Figure 1C and Figure S1F), CeTOCA-2 SH3 was less efficient than the SH3 domains of human TOCA-1 or CeTOCA-1. Thus, C. elegans TOCAs appear to possess some of the same biochemical features of their mammalian orthologues.
To define the functional roles of CeTOCAs in the context of an intact animal model, we generated and analyzed various single and double toca deletion mutants in C. elegans (described in detail in Figure S1B). All mutations resulted in the elimination of the proteins as shown by immunoblotting with antibodies against the entire C-terminal coiled-coil and SH3 domains of CeTOCA-1 and CeTOCA-2 ( Figure 1D). Analysis of CeTOCA-1 and CeTOCA-2 expression by immunofluorescence in developing embryos revealed that the proteins are ubiquitously expressed at various stages ( Figure 1E and 1F and Figure S1C and S1D and data not

Author Summary
Cells continuously remodel their shape especially during cell migration, differentiation, and tissues morphogenesis. This occurs through the dynamic reorganization of their plasma membrane and actin cytoskeleton: two processes that must therefore be intimately linked and coordinated. Molecules that sit at the crossroads of membrane remodeling and actin dynamics are predicted to play a pivotal role in coordinating these processes. The TOCA family of proteins represents a case in point. These proteins bind to and deform membranes during processes such as membrane trafficking. They also control actin dynamics through their interactions with actin remodeling factors, such as WASP and WAVEs. Here, we characterize the functional role of TOCA proteins in a model organism, the nematode Caenorhabditis elegans. We established that toca genes regulate Clathrin-mediated membrane trafficking during oocyte growth. We further discovered that these proteins play an important role in epithelial morphogenesis in developing embryos, and in egg production in adult nematodes. Moreover, the TOCA interacting proteins WASP/WSP-1 and WAVE/WVE-1, as well as other components of the WVE-1 complex, appear to be involved in TOCA-dependent processes. Thus, we propose that TOCA proteins control tissue morphogenesis by coordinating Clathrin-dependent membrane trafficking with WAVE and N-WASP-dependent actin-dynamics. Figure 1. The C. elegans TOCA family protein comprises two genes products, toca-1 and toca-2, which display conserved and functional F-BAR and SH3 domains. (A) Model of the predicted tertiary structure of the CeTOCA-1 F-BAR domain. The model was obtained with Phyre software (Protein Homology/analogy Recognition Engine). Residues highlighted in red, corresponding to the ones marked by red asterisks in the sequence alignment shown in Figure S1A, are conserved and involved in phospholipids binding. (B) CeTOCAs induces tubular-vesicular structures in vivo. Mouse embryo fibroblasts were transfected with the C. elegans CeTOCA-1 or 2-fused to GFP. Cells were fixed, stained with DAPI to detect cell nuclei (blue), or processed for epifluorescence. The formation of elongated tubular and vesicular-like structures is shown at higher magnification in the insets. Bar, 10 mm. (C) The SH3 domain of C. elegans TOCA-1 is functional. In vitro translated [S35]-M-labeled C. elegans WSP-1 (upper panel, WSP-1) or total cellular lysates (1 mg) of HeLa cells were incubated with equal amounts (10 mg) of the SH3 domain of either mammalian or C. elegans CeTOCA-1 or CeTOCA-2 fused to GST or GST, as a control. Bound proteins and an aliquot of total cell lysates (50 mg) or of the in vitro translated WSP-1 (1/200 of the material used in the in vitro binding experiment) were immunoblotted with the antibodies indicated on the right or exposed to autoradiography to detect in vitro translated WSP-1. The SH3 domain of CeTOCA-1 and CeTOCA-2 migrates slower since both constructs include residues extending 59, which were required for producing a soluble protein. (D,E) Expression levels of CeTOCA-1 and CeTOCA-2 in Wt and mutant worms. (D) Total cellular lysates of the indicated WT and toca-1 (left panels) or WT and toca-2 (right panels) mutant adult worms were immunoblotted with antibodies raised against the Cterminal regions of either C. elegans CeTOCA-1 or CeTOCA-2, and against Actin. Black arrows point to TOCAs proteins, white arrows indicate unspecific bands. (E) C. elegans embryos of Wt or toca-1(tm2056) and toca-2(ng11) mutants, to show the specificity of the antibodies, were fixed and immunostained with anti-CeTOCA-1 or CeTOCA-2 as indicated (right) or processed for differential interference contrast microscopy (DIC) (left). Bar, 10 mm. (F) Germline expression of CeTOCA-1 and CeTOCA-2. Full-length CeTOCA-1 C-DNA and full-length genomic CeTOCA-2 were fused to GFP and the transgenes expression was driven by the germ-line specific promoter pie-1. Constructs were then bombarded in their respective mutant background (toca-1(tm2056) and toca-2(ng11)). Images show expression of both CeTOCA-1 (upper left) and CeTOCA-2 (upper right) in oocytes and in developing embryos (bottom left and right). Vesicular-like structures in the oocytes are shown at higher magnification in the insets. White arrows point to the plasma membrane in oocytes and white arrowheads to the vesicles. Yellow arrow points to the rachis. Bar  shown). Both proteins labeled intracellular structures and appeared enriched along cell-cell junctions. This was particularly pronounced for CeTOCA-1, which colocalized with the junctional protein AJM-1 ( Figure 1E and Figure S1C and Figure S2A), while TOCA-2 appears to be more diffuse in the cytoplasm or cytoplasmic structures, suggesting that these proteins may function in the formation or maintenance of adhesive structures. We further confirmed the intracellular localization of CeTOCA proteins by analyzing the expression of CeTOCA-1::GFP and CeTOCA-2::GFP transgenes in their respective mutant strains lacking the endogenous protein. Attempts to use the endogenous promoters to drive the transgenes failed. We therefore employed the commonly used pie-1 promoter, to drive maternal protein expression in the germline, and early embryos [26]. This was particularly relevant because of the endocytic defect of oocytes in animals lacking CeTOCA proteins ( Figure 2). In oocytes, both gene products are present at the plasma membrane and in vesicular-like structures, consistent with their predicted role in membrane trafficking ( Figure 1F). Importantly, we observed a similar staining pattern of endogenous CeTOCA-1 in dissected gonads ( Figure S2B) (the anti-CeTOCA-2 antibody was not sufficiently efficient to recognize specifically the endogenous protein in this organ and adult worms). Finally, CeTOCA-2, but not CeTOCA-1 localizes to the partial membranes of the rachis ( Figure 1F and Figure S1D), the central core of cytoplasm that connects developing oocytes in the syncytial gonad.
Efficient Clathrin-mediated endocytosis (CME) of the yolk protein YP170 by oocytes requires CeTOCA proteins Lipids and proteins derived from yolk are thought to provide essential nutrients required for rapid embryo development. Accordingly, adult C. elegans hermaphrodites synthesize massive quantities of yolk particles in their intestines and secrete them basolaterally into the pseudocoelomatic space (body cavity), from which they are taken up into growing oocytes via receptormediated endocytosis [27][28][29] (see also Figure 2A). Disruption of CME impedes the internalization of yolk protein YP170, causing characteristic aberrant accumulation of aggregated yolk in the pseudocoelomatic space [29]. Since some of the mammalian TOCA family members are involved in CME [7,12] we tested whether this was the case also in the nematode.
Furthermore we found that the amount and distribution of YP170 in the oocytes was reduced in toca double mutant, as one would expect from defective internalization. As the growing oocytes move toward the spermatheca, where fertilization takes place, they progressively accumulate YP170, which in Wt distributes in a gradient that generally encompasses three or more oocytes [29][30]. Conversely, in endocytosis defective mutants, YP170 is either not detectable, or present only in the last most proximal oocyte ( Figure 2E). toca-1(tm2056);toca-2(ng11) strain displays a significant reduction in the number of oocytes positive for YP170, with YP170::GFP only detectable in the single most proximal oocyte, in more than 50% of mutant worms ( Figure 2E and and Figure S3B). Additionally, the amount of YP170::GFP, as determined by measuring total fluorescence of the worm with a similar number of YP170-positive oocytes, was significantly reduced ( Figure S3C). It is relevant to point out that when we compared worms at a similar stage and displaying a similar number of oocytes in the gonads, most wild type worms had three YP170 positive oocytes (.80% of cases), while most toca-1(tm2056);toca-2(ng11) worms had only the most proximal of the oocyte positive for YP170 (.85% of cases), with striking accumulation of YP170 in the pseudocoelomatic cavity ( Figure S3B).
Finally, we found that yolk receptor was correctly localized, and actually slightly, but significantly enriched at the plasma membrane, in the toca-1(tm2056);toca-2(ng11) double mutant strain ( Figure 2F and Figure S4), indicating that the accumulation of YP170 was not due to poor expression or mistargeting of yolk receptors during secretion. Thus, CeTOCA proteins are essential for the efficient endocytosis of yolk protein during oogenesis, suggesting that their primary role is to control CME. The specific role of CeTOCA protein in CME is further demonstrated by the observation that fluid-phase endocytic processes, such as the internalization of ssGFP into coelomocytes was not altered (not shown).

WSP-1 and the members of the WVE-1 complex are required for yolk uptake
The putative role of CeTOCAs at the crossroads of membrane trafficking and actin dynamics predicts that this endocytic function may involve TOCA-binding actin regulators. We thus tested whether the two major pathways, mediated by WSP-1 and WVE-1, affect endocytosis. Nematode WVE-1 (also called GEX-1) is associated also in nematode with GEX-2, and GEX-3 [27][28], the homologues of mammalian WAVE2, PIR121, NAP-1, and ABI-1 respectively. These proteins together with HSPC300 form a complex [31], which is conserved across different species and is required to activate Arp2/3-dependent actin polymerization [32,33]. To this end, we analyzed YP170::GFP endocytosis in mutant strains or after RNAi approaches. The RNAi approach was required to assess the role of WVE-1 and its interactors in adult germlines, since complete loss of these proteins leads to embryonic lethality. The accumulation of YP170 in the pseudocoelomatic space was clearly detected in 10% of partial loss of function mutant abi-1(ok640), in 20% of abi-1(RNAi) animals, and in 20% of wsp-1(gm324) mutant animals [34] ( Figure 3A). We observed a similar degree of inhibition of endocytosis after cdc-42(RNAi), while ablation of chc-1/Clathrin completely blocked YP170 accumulation by oocytes ( Figure 3B) [35].These results suggest that the WSP-1 pathway contributes to optimal endocytosis in the nematode, as it does in mammals ( Figure 3A).
Notably, the mutant abi-1(ok640) carries a deletion of the exons coding for the C-terminal SH3 domain, which mediates activation of N-WASP in mammals [36]. The N-terminal region of ABI-1, instead, which is essential for the assembly and the stability of the WAVE complex [37][38][39], is predicted to remain unaltered in the ok640 mutant, suggesting that WVE-1 function might not be disrupted. Consistent with this idea, the level of WVE-1 protein in abi-1(ok640) mutant was similar to Wt controls ( Figure 3A). WVE-1 levels were significantly reduced, as expected, upon gex-3(RNAi) ( Figure 3B), another component of the WAVE complex necessary for its stability [40]. Additionally, abi-1(RNAi) which is known to destabilize the WAVE2 complex [40], was as effective in inhibiting YP170 internalization as RNAi of wve-1, gex-2, or gex-3, indicating that the WVE-1 complex contributes to endocytosis.
Collectively, these observations support the notion that WSP-1 and, more surprisingly, the WVE-1 complex, which in mammalian Quantification of Wt and mutant worms accumulating YP170::tdimer2 in the body cavity. Data represent the percentage of worms accumulating YP170 into the body cavity and are expressed as the mean6s.e.m (n = 100) of at least three independent experiments. One asterisk indicates that a significant difference between Wt and mutants was detected, two asterisks a significant difference between mutants and rescue strains (P,0.0001 two-tailed t-test). (E) toca-1;toca-2 double mutant display reduced endocytosis of YP170 into oocytes. Left, examples of the distribution of YP170::GFP in the growing oocytes of Wt worms. Three gonads categories (one, two, or three green oocytes) are shown depending on the accumulation of YP170 into the oocytes. The most represented categories are two (46%) and three (48%) GFP-positive oocytes. Right, the distribution of gonad-categories with respect to the indicated genotype is expressed as percentage of the total (n = 100, in three independent experiments) (see also Figure S3B and S3C for quantification). Oocytes proximal to the spermatheca are numbered as 21. Bar, 10 mm. (F) The YP170 RME-2 receptor cellular distribution and expression is not altered in the toca-1;toca-2 double mutant. Confocal images of synchronized WT and toca-1;toca-2 double mutant worms expressing RME-2::GFP. Images were acquired with Axiovert 200 M microscope using MetaMorph and deconvoluted by AutoDeblur. Surface and middle sections are shown. Note that in the toca-1;toca-2 double mutant, RME-2 is correctly localized at the membrane, like in the WT. We quantified the levels of RME-2 fluorescent signals and show that they are increased on the surface, but significantly reduced in the cell cytoplasms (see Figure S4 for details of the quantification methods). Bar, 10 mm. doi:10.1371/journal.pgen.1000675.g002 Figure 3. Genetic or functional interference of WSP-1 and of the member of the WVE-1 complex causes accumulation of YP170 into the body cavity. (A) abi-1(ok640) or abi-1(RNAi) and wsp-1(gm324) mutants accumulate YP170 into the body cavity. Localization of YP170::GFP in synchronized young adult Wt, abi-1(ok640), abi-1(RNAi), and wsp-1(gm324) mutant worms. Bar, 100 mm. Bottom graph, quantification of Wt and mutants worms accumulating YP170::GFP in the body cavity. Data represent the percentage of worms accumulating YP170 into the body cavity and are expressed as the mean6s.e.m (n = 100) of at least three independent experiments. P,0.0001 two-tailed t-test. Bottom panels, lysates of adult Wt, wve-1(zu469), abi-1(ok640) mutants worms or worms fed with control or wve-1 specific RNAi-expressing bacteria were immunoblotted with the antibodies indicated on the right. Actin was used as loading control. Note, that in the mutant abi-1(ok640), which retains the binding surfaces to WVE-1, the level of this latter protein is similar to Wt. The asterisk indicates unspecific bands that were occasionally observed depending on the lysis procedure. (B) RNAi-mediated interference of gex-2, gex-3, or wve-1 causes accumulation of YP170 in the body cavity. Localization into the intestine, body cavity, and oocytes of YP170::GFP in synchronized young adult worm fed with bacteria expressing the indicated RNAi. cdc-42 and chc-1 (Clathrin heavy chain-1) RNAi-treated animals were used as control for defective endocytosis of YP170 [35]. Bar, is 100 mm. Graph, quantification of control and RNAi-treated worms accumulating YP170::GFP in the body cavity. Data represent the percentage of worms accumulating YP170 into the body cavity and are expressed as the mean6s.e.m (n = 100) of at least three to four independent experiments. P,0.0001 two-tailed t-test. Bottom panel, lysates of control (ctr) and gex-3 RNAi-treated and of Wt and wve-1(zu469) mutant worms were immunoblotted with the antibodies indicated on the right. Actin was used as loading control. (C) RNAi-mediated interference of wve-1 or gex-3 in wsp-1(gm324) mutant causes an increased accumulation of YP170 in the body cavity. Localization into the intestine, body cavity, and oocytes of YP170::GFP in synchronized young adult worm fed with bacteria expressing the control and the wve-1 or gex-3 RNAi in Wt and wsp-1(gm324). In all images, arrows indicate the YP170::GFP accumulated into the body cavity. Graph, quantification of control and RNAi-treated worms accumulating YP170::GFP in the body cavity. Data represent the percentage of worms accumulating YP170 into the body cavity and are expressed as the mean6s.e.m (n = 100) of at least three to four independent experiments. P,0.0001 two-tailed t-test. doi:10.1371/journal.pgen.1000675.g003 cells has never been implicated in CME, are both concomitantly required for efficient internalization of YP170 in C. elegans oocytes. In keeping with this notion, RNAi mediated interference of wve-1 or gex-3 ( Figure 3C) significantly worsened YP170 accumulation into the body cavity of the wsp-1(gm324) strain, indicating that the two NPFs act redundantly in this process ( Figure 3C).
CeTOCA-dependent impairment of yolk uptake is accompanied by a reduction in egg production Reduced internalization of vitellogenin frequently leads to impaired oocyte production resulting in a reduced total number of eggs laid [29]. Accordingly, while mutation of toca-1 alleles gave only a slight, but not statistically significant, reduction of the total number of eggs laid, toca-2(ng11) and toca-2(tm2088) mutant animals laid only 50-60% of the eggs laid by Wt strains and egg production in toca-1(tm2056);toca-2(ng11) double mutants was about 20% of Wt ( Figure 4A). Germ-line specific expression of CeTOCA-2::GFP in the toca-2(ng11) mutant rescued the defect in egg production ( Figure 4A), demonstrating the cell autonomy of the requirement for CeTOCA-2.
In wve-1/gex-1, gex-2 and gex-3 mutants, dorsal intercalation and migration of ventral and lateral cells are disrupted, resulting in extrusion of intestinal cells concomitant with failure to properly enclose the embryo [41]. We observed a similar set of morphological alterations in toca-1(tm2056);toca-2(ng11) double mutant. Time-lapse analysis of developing toca double mutant revealed that virtually all dying embryos are defective in morphogenesis ( Figure 5B and Video S1). In all these cases gut cells appeared to differentiate normally ( Figure 5B). Using the DLG-1::GFP transgene to label the apical junctions [42] we further monitored morphogenesis in live embryos and detected Gex-like morphogenetic defects, including failure of epidermal enclosure (full Gex-phenotype, leading to extrusion of gut to the exterior), 1-fold arrest and 2-fold arrest ( Figure 5C, arrow, and Video S1), just as it has been described for loss of the WVE-1/ WAVE protein complex. Mutations in WAVE/SCAR components lead to an expansion of the apical lumen of the intestine [40]. Using the DLG-1::GFP transgene, we measured the width of the intestinal lumen and found that wve-1 mutant and the toca-1(tm3334);toca-2(ng11) double mutant display a similar increase in the width of the intestinal lumen ( Figure 5C). This Gex-like intestinal morphogenesis defect was further confirmed by staining embryos at the same stage of development with anti-IFB-2 antibody, MH33 ( Figure 5D and Figure S5A and S5B), which recognizes intermediates filaments forming the terminal web beneath the microvilli of intestinal cells. WAVE complex mutants and the toca double mutant show an expanded MH33 region ( Figure 5D and Figure S5B). Together these results strengthen the notion that the embryonic lethality of toca-1(tm2056);toca-2(ng11) double or toca-2(ng11) mutants is due to a Gex epithelial morphogenesis phenotype [43].
We further examined the effect of toca mutations on apical junctions, focusing on the key junctional protein AJM-1 (also know as JAM-1) [44][45]. In the toca-1;toca-2 double mutant, AJM-1::GFP was localized at cell-cell junctions, as in the Wt strain, but hypodermal cells failed to intercalate dorsally and to correctly migrate ventrally, similar to gex-2 and gex-3 mutants strains [40] ( Figure 6A and Figure S5C). Finally, like DLG-1, AJM-1::GFP also marks the apical borders of cells lining the intestinal lumen, which is enlarged in toca-1;toca-2 double mutants, similar to defects previously reported for gex mutants [41]. Furthermore, quantification of AJM-1::GFP at junction (along the cell perimeter) in embryos at similar stage of development revealed a significant 1.5-fold higher fluorescent signal of AJM-1 in toca-1;toca-2 mutant than in Wt strain ( Figure 6A, right graph, and Figure S6), while its overall levels were unchanged as determined by immunoblotting of total worm lysates (not shown). These data suggest an altered cellular trafficking that may lead to increase junctional accumulation of the protein.
We caution, however, that junctional AJM-1 levels are an indirect measurement of the amounts on the cell surface of transmembrane proteins, reflecting possible trafficking impairment. Collectively, these results indicate the requirement of CeTOCA proteins in the earliest event of epidermal morphogenesis due to a Gex phenotype and suggest a role in regulation of junctional protein traffic.

Mammalian TOCA-1 physically associates with N-WASP and the ABI-1/WAVE2 complex
The interactions in the nematode between CeTOCAs and WSP-1 are consistent with the demonstrated biochemical and cellular role of mammalian TOCA-1 and FPB17, which directly interact with WASP/N-WASP and promote WASP/N-WASP activation at the plasma membrane [22,24], thus presumably promoting actin dynamics during CME [6,12]. Consistent with this idea, we found that mammalian TOCA-1 binds to N-WASP and to WIP, an N-WASP interacting regulatory protein, -through the TOCA-1 SH3 domain ( Figure 1C and Figure S7A and S7B).
The relationship between CeTOCA proteins and the WVE-1/ WAVE complex in the nematode is, instead, more complex, suggesting unexpected levels of molecular interactions and regulation. To overcome the lack of reliable antibodies and the intrinsic limitations of cellular biochemistry of C. elegans, we set out to define whether any physical or functional interaction between these proteins is conserved in mammals. To this end, we utilized the SH3 domain of TOCA-1 to search for novel interactors, employing phage display libraries of human polypeptides [47]. We found ABI-1 among the most represented interactors (not shown). We validated this interaction by showing that endogenous and ectopically expressed TOCA-1 and ABI-1 coimmunoprecipitated ( Figure 7A and 7B), and interacted in in vitro binding experiments using immobilized SH3 domain of TOCA-1 ( Figure S7C). A similar interaction was also detected between ABI-1 and FBP17 or CIP4 ( Figure S7D), indicating that all family members have the capacity to form a complex with ABI-1. Notably, we could recover endogenous TOCA-1, but not CIP4 or FBP17 (not shown), in ABI-1 immunoprecipitates ( Figure 7B), suggesting that TOCA-1 is the most likely physiological relevant F-BAR containing interaction partner of ABI-1. In keeping with this notion, we found that in vitro translated C. elegans ABI-1 readily interacted with the SH3 domain of CeTOCA-2 ( Figure 7C), indicating that this interaction is conserved in the nematode.
The quantitatively more relevant binding partner of ABI-1 in mammalian cells is WAVE2 [37]. Consistent with this, we also detected WAVE2 in TOCA-1 and ABI-1 immunoprecipitates ( Figure 7B) and in in vitro binding experiments using TOCA-1 SH3 domain ( Figure S7C). The extent of this interaction was lower than the one between TOCA-1 and ABI-1 ( Figure 7A and Figure S7C). These results suggest that ABI-1 may serve as a bridge between TOCA-1 and WAVE2. Accordingly, the amount of WAVE2 in TOCA-1 immunoprecipitates was significantly increased when ABI-1 was over-expressed ( Figure 7D).
TOCA-1, like CIP4, was originally identified as an effector of the small GTPase Cdc42. We thus sought to assess whether ABI-1 could be linked directly to Cdc42 in the presence of TOCA-1. The binding of TOCA-1 to purified, GTP-loaded Cdc42 was specific and readily detectable ( Figure 7E), as previously reported [22]. More importantly, ABI-1 interacted with GTP-loaded Cdc42 only in the presence of TOCA-1 ( Figure 7F). Under these conditions, WAVE2 was also recovered on immobilized and activated Cdc42 (not shown), suggesting that a signaling complex connecting Cdc42-TOCA-1 and ABI-1/WAVE-2 may form. Thus, mammalian TOCA-1 can associate with the WAVE complex through ABI-1, in addition to N-WASP, recapitulating the interactions observed in the nematode.

Discussion
In this manuscript, we provide evidence that a primary role of the CeTOCA family proteins is to control endocytic processes with important implications for the regulation of actin-dependent epithelial morphogenesis and migration. toca genes genetically interact with wve-1, acting in the same pathway as wsp-1, most likely downstream of cdc-42. This, combined with the observation that TOCA family members can form complexes with N-WASP or WAVE2 through ABI-1, indicates that these proteins coordinate N-WASP and WAVE-dependent actin dynamics with membrane trafficking.
TOCAs, WVE-1, and WSP-1 are essential for optimal Clathrin-dependent endocytosis The concomitant genetic disruption of the two C. elegans toca homologues results in a fully penetrant endocytic defect in oocytes. This defect in yolk uptake likely accounts also for the significant reduction in the number of eggs laid by double toca-1;toca-2 mutant worms. The finding, however, that CeTOCA-2, but not CeTOCA-1, is enriched at the cortical surface of rachis suggests that this protein may have a specific role beyond vitellogenin endocytosis, possibly by directly affecting oocyte structure and organization. Recently, proteins that display a cortical rachis enrichment, like that of CeTOCA-2, such as Anillin-2 (ANI-2) [48] and Flightless-1 (FLI-1) [49], have been shown to be required for germline morphogenesis and oocyte growth. Both of these proteins are actin-binding factors, pointing to the importance of actin dynamics and architectural organization in this process. In keeping with this observation, one possible, albeit speculative, function for CeTOCA-2 would be to provide a link between actin and membranes during germline morphogenesis. We note however that we did not observe obvious abnormalities in germline structure (e.g. the distribution of nuclei in the syncytium and the size of the oocytes appear relatively normal in toca-2(ng11) strain (not shown). More experiments will be needed to assess whether toca-2 is a germline morphogenesis gene and its precise role in the process.
In YP170 internalization, CeTOCA-1 and CeTOCA-2 act in a redundant fashion. However, in this process, as in embryo development and oocyte growth, CeTOCA-2 display a more prominent role than CeTOCA-1, likely reflecting a differential intracellular localization and/or different binding affinity for their common partners, namely ABI-1 or N-WASP. In these latter cases, the relative low affinities of the interactions observed suggest that CeTOCAs may modulate the function of WSP-1 and ABI-1/ WVE-1, but are not obligatory partners of either protein complexes.
Notably, the concomitant disruption of toca-1 and toca-2 reduced or delayed, but did not abrogate, YP170 entry into oocytes, indicating that these proteins are critical, but not essential for endocytosis of YP170. This is not unexpected given the complexity of CME, where .50 accessory proteins [50,51], including membrane binding and bending BAR-domain-containing molecules [6,15], have been shown to aid the core machinery of internalization in a cell context and often cargo-dependent manner [52]. The C. elegans genome contains additional F-BAR, N-BAR and SH3-containing proteins. These proteins may, thus, potentially function in a partially redundant fashion with the tocas ( Figure S8).
The membrane tubulation activity of the F-BAR domain coupled with the ability to bind key actin regulators is predicted to enable TOCA proteins to assist and promote the initial events leading to the internalization of plasma membrane proteins. The genetic evidence provided in this study in part confirms this proposition [7,12,18,19]. We found that: i) the two major actin NPFs WSP-1, and more surprisingly WVE-1 (and its binding partners), are important for optimal internalization of YP170; ii) CeTOCAs genetically interact with WVE-1, while acting in the same pathway of WSP-1. The existence of the latter pathway is consistent with the demonstrated role of mammalian TOCA-1 in mediating the biochemical activation of WASP downstream of Cdc42 [22]. According to this model, the direct and concomitant interaction of TOCA-1 and Cdc42 with WASP relieves the autoinhibited state of the latter protein leading to Arp2/3mediated actin polymerization [22]. A similar mechanisms is likely functional in YP170 internalization, as indicated by the fact that RNAi-interference of Cdc42 phenocopies [35] the removal of WSP-1 and CeTOCAs ( Figure S8).
The pathways emanating from Cdc42 and regulating endocytosis are, however, more complex that the one depicted here. Indeed, the Cdc42-dependent polarity effector complex, PAR-3/ PAR-6/atypicalPKC was also shown to mediate the endocytosis of YP170 in C. elegans [35]. Whereas, in Drosophila, Cdc42, PAR-6 and atypical PKC, but not PAR-3, have been recently demonstrated to be required for E-cadherin endocytosis, working in concert with CIP4 [25], WASP and Dynamin [53]. The full Cdc42, Par6, Par3, and aPKC complex affect the trafficking of Crumbs and apically applied FM 4-64 in Drosophila during neuroectoderm development [54]. While the hierarchical organization and precise mechanisms through which these proteins function remains unclear, these data and our findings suggest a role of a CDC42/PAR-3/atypicalPKC/TOCA/WASP axis in the regulation of early endocytic steps of multiple cargos.
An additional level of complexity is revealed by the requirement for the WAVE/WVE-1 in YP170 endocytosis, suggesting a previously unsuspected regulatory role for the WAVE complex in a bona fide Clathrin-dependent internalization process. In mammals, however, no co-localization of this protein complex with Clathrin was ever detected at the plasma membrane, suggesting that it does not directly control internalization [36,55]. Assuming that this is also the case in the nematode, the most likely role of the WAVE/ WVE-1 complex is in later steps of the endocytic route, or to regulate Clathrin-independent pathways.
Finally, it is worth noting that toca mutants display more severe endocytic impairment than the combined genetic/functional interference of wsp-1 and/or wve-1 or Cdc42. This might simply reflect an incomplete down-regulation of wve-1 and its associated components by RNAi. Alternatively, toca genes, by analogy to the role exerted in embryonic development, may act upstream of or in parallel to these NPFs, exerting a modulatory role on their activities, and further regulating additional actin-independent pathways. In favor of this latter possibility, the SH3 domains of all the mammalian TOCA paralogues have been shown to associate with Dynamin [6], a GTPases that is essential for vesicle scission during multiple internalization mechanisms.
One possible scenario that emerges from, and may account for these and previous published observations, is that CeTOCA-1 and CeTOCA-2 may be part of a network of intertwined pathways acting downstream or in parallel to Cdc42 ( Figure S8). CeTOCA-1 and CeTOCA-2 may directly connect Cdc42 to WSP-1 and/or Dynamin. Concomitantly, a WVE-1 complex cascade may contribute to the generation of actin dynamics-based forces required for optimal endocytosis. TOCA proteins might drive the correct localization of the WVE-1 complex, rather than modulating its biochemical activity. This possibility is suggested by the interaction between CeTOCA-2 and ABI-1, which is conserved in mammals and nematodes. All of these pathways, including the Cdc42-PAR-6-PAR-3-aPKC complex, may converge, coordinating actin and membrane dynamics during the internalization of different cargos, such as YP170-RME-2 and, presumably, junctional complexes ( Figure S8).

How do TOCAs control epidermal morphogenesis?
Our results indicate that CeTOCA proteins are localized at cellcell junctions, preferentially at the apical/lateral side, where junctional complexes are also located. A similar localization is also conserved in mammalian epithelial cells. Additionally, at these sites, removal of CeTOCAs impairs the organization of cortical filamentous actin and enhances accumulation of the junctional protein AJM-1. Since the primary function of CeTOCA proteins in C. elegans appears to regulate endocytosis, it is reasonable to assume that CeTOCAs may similarly control the internalization or trafficking of key cell-cell junction proteins. Indeed, intracellular trafficking of junctional proteins is emerging as a critical device to ensure dynamic remodeling of epithelial cell-cell adhesion molecules [56][57][58][59]. This process may become particularly relevant when epithelial cells migrate, such as during morphogenesis in embryonic development, or in wound closure [60]. Similarly, disruption of CeTOCA proteins may impair the proper internalization of junctional complexes, resulting in altered morphogenetic movements of hypodermal cells. It is of note that defective endocytosis of E-cadherin, a key component of adherens junctions, in the Drosophila notum epithelium or dorsal thorax has been recently linked to the only Drosophila TOCA family member, CIP4 [25]. In this system, Drosophila CIP4 appears to somehow connect the Cdc42-dependent polarity subcomplex, PAR-6/atypicalPKC with WASP and Dynamin, ultimately regulating vesicle scission and internalization of E-cadherin [25,53]. A similar situation may also occur in hypodermal cells in C. elegans. In this case, CeTOCA proteins may directly regulate the activity of WSP-1/WASP, downstream of CDC-42, while cooperating with WVE-1, ultimately mediating the generation of actin dynamics-dependent forces needed for the correct remodeling of junctional complexes during cell migration. Future studies exploiting the mammalian epithelial cell model systems will be required to address these hypotheses. Nevertheless our data suggest that TOCA proteins sit at a critical crossroad of actin and membrane dynamics in the regulation of the intracellular trafficking during epithelial morphogenesis.
The monoclonal anti-WAVE2 antibody was generated against the C-terminal portion of human protein (VCA domain, aa 422-498) produced as a GST fusion protein.
All genetic experiments and phenotypic analysis in C. elegans and cell biochemistry in mammalians are described in Text S1. Figure S1 TOCA genes and proteins. (A) Multiple sequence alignment of TOCA family members from various species (Homo sapiens, Hs; Mus musculus, Mm; Xenopus tropicalis, Xt; Caenorhabditis elegans, Ce). Protein sequences were aligned using the ClustalW program. Manual adjustments were introduced on the basis of secondary structure information, and the picture was produced using Jalview. The secondary structure of the F-BAR domain of the human FBP17 (black) and the predicted one of the C. elegans TOCA-1 (coloured as in Figure 1A) are reported at the bottom of the alignment. Asterisks indicate residues of FBP17 involved in phospholipids binding. (B) Genomic organization of toca-1 and toca-2 genes and of the available deletion alleles. Schematic representation of C. elegans toca-1 and toca-2 intron/exon organization (intron = lines; exon = black boxes). The locus position of the putative F-BAR, HR1, and SH3 domains is indicated on top. Bar, 1 Kb. The deletion of the various tocas mutant worms utilized is also indicated. The toca-1(tm2056) is a deletion encompassing the exon that precedes the one coding for the SH3 domain; toca-1(tm3334), harbours a deletion extending from exon 3 to 4 over the F-BAR domain. Both these deletions result in an out-of-frame shift of the remaining gene products, which cannot be detected by immunoblotting ( Figure 1D), indicating that the mutations lead to destabilization of the entire mRNA. toca-2(tm2088) is a short deletion of exons 1-4, also causing an out-of-frame shift; finally toca-2(ng11), which was generated by TMP/UV mutagenesis, is a large deletion encompassing almost the entire locus (from exon 4 to 9). To obtain double toca mutants, we crossed either one of the two strains carrying the toca-1 mutated alleles with toca-2(ng11). (C) TOCA-1 and TOCA-2 localization in developing embryos. C. elegans embryos were fixed and immuno-stained with anti-CeTOCA-1 or CeTOCA-2 antibodies as indicated (right) or processed for differential interference contrast microscopy (DIC)(left). Bar, 10 mm.  Figure S4 RME-2 levels in toca-1;toca-2 oocytes. RME-2, the yolk receptor, is correctly localized and enriched at the plasma membrane. RME-2::GFP fluorescent intensities (arbitrary units, A.U.) along selected (distance, pixel) areas and lines were quantified by ImageJ software (see Materials and Methods). Different areas from at least 20 Wt and toca-1;toca-2 animals were analyzed. The images in red represent a typical example of Wt and toca-1;toca-2 animals and were obtained by applying a threshold algorithm (ImageJ) to equalize and remove background staining and evidence pixel intensities values above threshold, which correspond to surface RME-2 signals. This procedure permits us to appreciate that the levels of cortical RME-2 are higher in toca-1;toca-2 animals with respect to Wt. Graph, the GFP intensity along the junctions (upper) and the average intensity of cytoplasmic RME-2 per area (bottom) is plotted for Wt and toca-1;toca-2. double mutant worm displays altered intestinal morphology during embryo development. Wt and the indicated mutant worms were fixed and stained with anti-MH33 antibody or DAPI to detect the intestinal cells and cell nuclei, respectively. Embryos die at 1.5 fold stage, just before elongation starts; DAPI shows that Wt and mutants have a similar number of nuclei, indicating a similar developmental stage. Bar, 10 mm. The percentage of gut-defective embryos of the various genotypes, quantified as described in Materials and Methods, is shown in the bottom graph. Please note that in the case of gex-3(zu196) we used a balanced heterozygous strain OX169 gex-3(zu196)/DnT1 in which only 25% of the progeny is homozygous for gex-3(zu196) according to Mendelian distribution. Nearly 100% of these homozygous gex-3(zu196) embryos display the morphogenetic intestinal defect as previously reported (Soto et al., 2002). Data are the mean6s.e.m. (n = 100) of at least three independent experiments. P,0.0001, two-tailed t-test is indicated by an asterisk. (B) Intestinal morphology of Wt and mutant embryo at different stages of development. Wt and toca-1;toca-2 mutant worms were fixed and stained with anti-MH33 antibody. All dying (,12% of total embryos, Figure 5A) toca-1;toca-2 embryos are arrested at 1.5 fold stage, just before elongation starts (left); A significant fraction of toca-1;toca-2 embryos display altered intestinal morphology with enlargement of the intestinal lumen at the 2 fold stage with respect to Wt. Of note, at this stage it is easy to appreciate that MH33 display an apical distribution as previously reported (Patel et al., 2008) (right). Bar, 10 mm. (C) toca-1;toca-2 Gex (Gut on the exterior) embryos display an altered epidermal cell morphology typical of gex mutants. Left, lateral and ventral view of Wt and toca-1;toca-2 expressing AJM-1::GFP. Right, a scheme of the morphogenetic defects caused by loss of toca-1 and toca-2. Loss of tocas leads to the distinctive Gex (Gut on the exterior) phenotype due to failures in cell movement and cell shape changes. Lateral view: gex embryos fail to initiate epidermal ventral movements. By 400 min after first cleavage, Wt embryos initiate circumferential constrictions to squeeze the embryo into a worm. gex embryos undergo constriction that leads the epidermis to collapse inwardly. The internal organs (pharynx and intestine) end morphogenesis exposed on the ventral surface. Ventral view: in gex embryos epidermal cells fails to correctly migrate to the ventral side as in the Wt. Arrows indicate epidermal cells. Bar, 10 mm. Found at: doi:10.1371/journal.pgen.1000675.s005 (2.88 MB TIF) Figure S6 The surface levels of AJM-1 of toca-1;toca-2 mutant embryos are slightly higher than wt embryos. Quantification of AJM-1 along the cell perimeter of hypodermal cells in WT and toca-1;toca2 mutant embryos. Embryos at the two-cell stage were kept at 22u for 5 hours before fixation between 360 and 400 min, as indicated, when the ''bean shape stage'' was reached. Embryos were stained with anti-actin (not shown). The second and third raw images were used to determine the levels of F-actin at junctions as described in Figure 6B, or processed for epifluorescence or DIC. Images were captured with a Leica Microsystems confocal microscope using the HCX PL APO CS 63.061.40 OIL objective lens, and objective zoom (3.496). Exposure time and gain setting were fixed as follows: (Dapi PMT1 (Photo Multiplier Tube) = 500, GFP-PMT2 = 583, Cy3-PMT3 = 550, PMT Trans (HV) = 298). Identical settings were used for all samples so that direct comparison of the signal intensities among the images of embryos of different genetic backgrounds was possible. The ImageJ threshold alogorithm (red channels) were then applied to eliminate cytoplasmic and background signals by placing ''Regions of Interest'' (ROI) over areas outside the junctional contour. We manually determined the cell perimeters and calculated mean intensities and perimeter length. The mean intensitiy values were then multiplied by the perimeter of each cell to obtain the total intensity along the cell perimeter, as exemplified in the boxed cells on the bottom raw (the line around the cells is meant to outline the chosen cell, not the actual perimeter). Examples of the values obtained for the outlined (white circles) cells are shown. We repeated the procedures to obtain statistically meaningful data that are expressed as mean6s.e.m (n.20 from at least 4 to 5 independent embryos of each genotype). The values of the total intensities along the cell perimeters were plotted as described in Figure 6A. Bar is 10 mm. The results indicate that there is a slight, but significant enrichment of AJM-1 at the junction. This may suggest an altered trafficking of junctional proteins. However, AJM-1 is not a transmembrane protein in C. elegans, but rather associates with apical junctional molecules and thus presumably reflects the cell surface levels of such transmembrane proteins. P,0.0002, two-tailed t-test is indicated by asterisks. , may directly promote localized actin dynamics during early steps of Clathrin-mediated endocytosis (CME). The polarity complex PAR-3/PAR-6 whose activity is required for endocytic and recycling events downstream of Cdc42 (Balklava et al., 2007) may define an alternative branch of the pathway that may also converge in controlling the TOCA-1/2/WASP(WSP-1) axis. An unexpected contribution of the WAVE(WVE-1) axis in this process is evidenced by; i) the increased accumulation of YP170 after interference with WAVE (WVE-1) complex components; ii) the genetic interactions of these latter genes with toca-1/2; iii) the biochemical link between TOCA-1/2 and ABI1, which is conserved also in mammals. The WAVE (WVE-1) complex may function in later endocytic steps of CME since, at least, in mammals it does not localize to Clathrin-coats at the plasma Membrane (Benesch et al., 2005). The precise relation of the WAVE (WVE-1) complex with CDC42/TOCA-1/2 is unclear at present. However, TOCA-2 appears dominant with respect to TOCA-1. TOCA1/2 may also directly associate to Dynamin (Itoh et al., 2005; Tsujita et al., 2006), whose pinching activity is critical to promote vesicle scission. Dynamin-and actin-dependent activities may work in concert with TOCA-1/2 to promote tubule scission. Other F-BAR containing proteins, such as Nostrin may add further layers of complexity to this network, which may coordinate membrane tubulation and curvature sensing with the activity (WASP/WSP-1) ( Video S1 toca-1;toca-2 double mutants show a Gex phenotype. Time-lapse observation of embryogenesis with Nomarski microscopy of Wt and toca-1;toca-2 double mutant worms. Movie recording started approximately 300 min after first cleavage. Frames (451) were taken every 1 min for a total of 7 1/2 hours.