C. elegans Enabled exhibits novel interactions with N-WASP, Abl, and cell-cell junctions.

Ena/VASP proteins are associated with cell-cell junctions in cultured mammalian cells [1] and Drosophila epithelia [2, 3], but they have only been extensively studied at the leading edges of migratory fibroblasts, where they modulate the protrusion of the leading edge [4]. They act by regulating actin-filament geometry, antagonizing the effects of actin-capping protein [5]. Embryos lacking the C. elegans Ena/VASP, UNC-34, display subtle defects in the leading edges of migrating epidermal cells but undergo normal epidermal morphogenesis. In contrast, embryos lacking both UNC-34 and the C. elegans N-WASP homolog have severe defects in epidermal morphogenesis, suggesting that they have parallel roles in coordinating cell behavior. GFP-tagged UNC-34 localizes to the leading edges of migrating epidermal cells, becoming redistributed to new junctions that form during epidermal-sheet sealing. Consistent with this, UNC-34 contributes to the formation of cadherin-based junctions. The junctional localization of UNC-34 is independent of proteins involved in Ena/VASP localization in other experimental systems; instead, junctional distribution depends upon the junctional protein AJM-1. We also show that Abelson tyrosine kinase, a major regulator of Enabled in Drosophila, is not required for UNC-34/Ena function in epithelia. Instead, our data suggest that Abelson kinase acts in parallel to UNC-34/Ena, antagonizing its function.


Supplemental Results and Discussion
UNC-34 Aids CeSCAR/WAVE in Initiating Cell Migration Several barbed-end modulators act via the Arp2/3 complex, which nucleates new actin filaments via the formation of side branches from existing filaments [S1]. Reducing the function of various members of the Arp2/3 complex results in embryonic lethality in C. elegans well before the initiation of morphogenesis [S2]. However, in order to perform its role as an actin nucleator, Arp2/3 must be bound and activated by one of several families of molecules, allowing for the localized activation of Arp2/3 in specific places and at specific times during development.
The SCAR/WAVE family of molecules is one such family of Arp2/3 activators [S3]. SCAR/WAVE forms part of a polypeptide complex that responds to activated Racs, or other membrane-associated signals, which in turn activates Arp2/3, thereby promoting actin-filament nucleation [S4-S7]. The C. elegans genome contains one predicted member of this family, wve-1 [S8]. wve-1(RNAi) results in embryonic lethality (75.6%, n = 1018), with affected embryos displaying severe defects in embryogenesis ( Figure S1B). wve-1(RNAi) does not affect the formation of the epidermis, but cells fail to perform the migrations necessary for proper dorsal intercalation and enclosure. Embryos ultimately arrest without completing morphogenesis with a terminal phenotype identical to Gex (gut on the exterior) mutants [S9].

Ena/VASP and WASP/WIP Act in Parallel during Ventral Enclosure
The WASP family of molecules constitutes another important group of Arp2/3 activators [S11]. This activation can occur by WASP binding to active Cdc42 and/or PIP 2 at the membrane, and it often leads to the formation of Figure S1. Initiation of Epidermal Migration Requires WVE-1 and UNC-34 Nomarski time-lapse images. Ventral is up, and anterior is to the left. (A) Wild-type enclosure. Ventral cells migrate around the lateral surfaces (arrowheads) of the embryo to meet at the ventral midline (arrow). (B) wve-1(RNAi) embryos in which epidermal cells fail to migrate and never enclose (asterisk). (C) unc-34(gm104); wve-1(RNAi) embryos also fail during morphogenesis, with epidermal cells becoming slightly rounded and never migrating to enclose the animal (asterisk). (D) ced-10(n3417) embryos phenocopy unc-34(gm104); wve-1(RNAi) embryos, including enclosure defects (asterisk). The scale bar represents 10 mm. filopodia [S12, S13]. In our hands, the removal of the function of the sole C. elegans N-WASP, wsp-1, via RNAi resulted in embryos that were largely phenotypically normal, with a low level of lethality (15.6%, n = 596); dead embryos displayed phenotypes consistent with disruptions in cell division (data not shown). However, as previously reported [S8], the reduction of WSP-1 in an unc-34 null background yielded 100% lethality (n = 125); affected embryos displayed substantially disrupted morphogenesis (see Figure 1).
Because these defects are not observed in either single mutant, these results show that during morphogenesis, unc-34 and wsp-1 act in a genetically redundant manner. In order to support the genetic specificity of this interaction, we examined the role of the C. elegans WIP (WASP-interacting protein) during enclosure. The precise molecular role of WIP is unclear (reviewed in [S14]). It is capable of binding WASP family members [S15] and often appears to be important for WASP function [S16, S17], but it also participates in actin dynamics independent from WASP. C. elegans has a single WIP homolog, wip-1 [S18]. In wild-type animals, wip-1(RNAi) results in low levels of embryonic lethality similar to wsp-1 (RNAi), and double RNAi of wsp-1 and wip-1 does not enhance this lethality. unc-34(gm104); wip-1(RNAi) and unc-34(gm104); wsp-1(RNAi) animals are phenotypically indistinguishable (Figures S2A and S2B;cf. Figures 1B and 1B 0 ), strongly suggesting that wip-1 and wsp-1 act together during enclosure and that, in this context, WIP acts as a positive regulator of WASP function. One caveat to these experiments is that the RNAi knockdown for wip-1 might have been incomplete. Future experiments involving wsp-1 and wip-1 null mutants would definitively confirm these results.
Overall Actin Morphology Is Grossly Normal in unc-34(gm104);wsp-1(RNAi) Embryos The presumptive roles of UNC-34 and WSP-1 involve the modulation of the actin cytoskeleton. Actin is known to be generally required for ventral enclosure [S19], so we performed phalloidin staining to investigate the gross structure of actin in unc-34;wsp-1 embryos ( Figure S3). Phalloidin-stained unc-34(gm104);wsp-1(RNAi) embryos resemble wild-type embryos. Although leadingedge protrusions are less readily apparent, actin is enriched at the leading free edges and cell-cell borders of ventral cells that fail to migrate, and the bundling of cellular actin into nascent circumferential cables appears to initiate normally ( Figure S3B).

Models for the Genetic Interaction between unc-34 and wsp-1
Our results regarding the genetic interaction between unc-34/Ena and other components involved in leadingedge migration, collated with those of others, is shown in Figure S5A. The synergistic genetic interaction between unc-34 and wsp-1 is consistent with several models at the molecular level ( Figure S5B). WSP-1 is known to activate the C. elegans Arp2/3 complex [S20] to promote actin-filament nucleation via side branching. UNC-34 presumably acts in a manner similar to other members of the Ena/VASP family because unc-34 mutants display phenotypes [S8] that appear to be similar to neuronal phenotypes associated with the loss of Ena/VASP family proteins in other systems (reviewed in [S21]). One model of filopodium formation involves actin branches in the dendritic network becoming privileged via the presence of filament tip complexes that include Ena/VASP proteins [S22], which allow pre-existing filaments to elongate and ultimately become bundled into filopodia. So that redundancy between UNC-34 and WSP-1 can be integrated into this model, in C. elegans, most of the dendritic network could be generated via WVE-1-mediated branching. In addition, a local microenvironment with elevated WSP-1 activity-possibly because of active Cdc42 or another signal from the membrane-combined with UNC-34/tip complex activity could lead to a more elaborate protrusion structure.  (B) unc-34(gm104);wsp-1(RNAi) embryos show grossly normal actin structure, including enriched actin at leading edge (arrowheads), small protrusions (arrow), and cellcell borders, as well as initial aggregation of bundled actin that will ultimately become circumferential bundles. The scale bar represents 10 mm.

S2
In such a growing-end model ( Figure S5B, left), a certain critical threshold level of available barbed ends must be achieved through the activity of both UNC-34 and WSP-1, but each molecule is dispensable on its own. The loss of both molecules simultaneously is clearly catastrophic for protrusion formation.
An alternative, but not mutually exclusive, de novonucleation model ( Figure S5B, right) offers a slightly different explanation for our results. Recent work in Dictyostelium has shown an interaction between the formin dDia2 and an Ena/VASP family protein during filopodium formation [S23]. If UNC-34 acts as a scaffolding molecule linking a formin (or other nucleator) to the existing cytoskeleton, it becomes easier to understand the genetic interaction of wsp-1 and unc-34. In this case, WSP-1 provides one method of de novo filament nucleation, and UNC-34 provides another by recruiting a formin, with the two methods being functionally redundant in this context. The examination of the fine structure of the actin cytoskeleton in unc-34; wsp-1 embryos, or their equivalents in another system, would no doubt provide insight into the molecular nature of this genetic interaction.
So that germline mosaics of hmr-1(zu389) and hmp-1(zu278) could be obtained, individual adult hermaphrodites derived from transgenic lines carrying extrachromosomal arrays containing wild-type hmr-1 or hmp-1 genes, respectively, were picked to separate plates, and their progeny were examined for 100% embryonic arrest. Animals with this phenotype were placed on fresh plates without food for approximately 16 hr, and their broods were processed for immunostaining.

Generation of unc-34::gfp Transgenic Lines
The HMR-1A promoter (P HMR-1A ) was fused to the unc-34 gene via PCR fusion, and the resultant amplicon was fused to the gfp gene and unc-54 3 0 untranslated region (UTR) via a second polymerase chain reaction (PCR) fusion reaction. A list of the primer sequences used to generate the fusion amplicons is available on request to J.P. The P HMR-1A ::unc-34::gfp fusion amplicon was coinjected along with the rol-6(su1006) marker gene into unc-34(gm104) hmp-1(fe4)/ mIs10 hermaphrodites, and three independent lines carrying both transgenes as an extrachromosomal array were established. All three gave identical patterns of expression.
To express unc-34::gfp under the control of the dlg-1 promoter, the upstream region of dlg-1 was amplified from wild-type DNA and cloned into a modified version of pPD95.75 (provided by A. Fire, Stanford University), which can be used as a destination vector in the Gateway cloning system (Invitrogen). The primers used to amplify the dlg-1 upstream region were dlg1PL, 5 0 -CGAAGCTTCA CAGTTTACCAAACTAGTC-3 0 , and dlg1PR, 5 0 -CGGCATGCGCTTCC TTCCTTCGGTG-3 0 . The unc-34 cDNA was then cloned into this vector via recombination. This construct produced a fusion of the GFP gene to the C terminus of UNC-34. The resultant vector, pPE#JP101 was injected into unc-34(gm104) hmp-1(fe4)/mIs10 hermaphrodites.

RNAi
Double stranded RNA-mediated interference was performed as previously described via either injection [S30] or feeding [S31] methods. Ambion MegaScript T7, T3, and SP6 kits were used for the preparation of injection samples. Plasmid pBG1 (described in [S8]) is a partial wsp-1 cDNA cloned into Fire vector kit plasmid L4440 for feeding RNAi. R144.4 RNA was made with partial cDNA yk433c4, from the laboratory of Yuji Kohara, as a template. wve-1 RNA was made with a genomic fragment amplified with primers 5 0 -AAGTTAAGCT GAAGCCGAGAGCCTGCGG-3 0 and 5 0 -GAACATCTTCGGAGAGATT TATCACGACG-3 0 as the template.

Microscopy
4D Nomarski microscopy was performed as previously described [S32]. In brief, gravid adults were bisected in M9 solution so that embryos could be collected. Embryos were transferred by mouth

S3
pipette to a 5% agar pad on a glass slide, submerged in M9, and sealed with a glass coverslip for filming. Embryos were filmed on a Nikon Eclipse E600 or Optiphot-2 upright microscope equipped with DIC optics and Ludl shutter controllers. Dage-MTI analog video cameras collected the data onto Macintosh G3 computers with Scion AG-5 frame grabbers. NIH Image software with custom macros was used for the compression of 3D time-lapse data into 4D QuickTime movies. Two-photon excitation microscopy was performed as previously described [S33] with a custom setup for the collection of images into Bio-Rad software, which were then compressed into 3D projected time-lapse movies with NIH Image and custom macros. Spinning-disc confocal microscopy on live GFP specimens was performed with a Yokogawa CSU10 scanhead attached to a Nikon Eclipse E600 microscope. Data were collected with a Hamamatsu ORCA-ER CCD camera wth Perkin Elmer Ultraview software on a Pentium 4 PC. Fixed specimens were viewed either with spinning-disc confocal microscopy or laser-scanning confocal microscopy with a Bio-Rad MRC1024. All live specimens were filmed at 20 C.

Quantitative Protrusion Analysis
Embryos expressing dlg-1D7::gfp [S29] were filmed with spinningdisc confocal microscopy. Twenty focal planes were captured from the ventral-most plane moving 9.5 mm into the specimen in 0.5 mm intervals. Such Z stacks were taken every 50 s throughout morphogenesis. Stacks were projected into a single image and compressed into AVI movies. These movies were then viewed on a Macintosh G4 computer with ImageJ software. Movies were expanded to 200% zoom so that analysis could be facilitated. The perimeter of the LCPZ was measured at specified time points; this measurement was divided by the width of the cell bodies so that slight differences in viewing angle and differences in cell width in the mutants could be internally controlled for ( Figure S4A). These data were collected over five consecutive time points (at 50 s intervals) per embryo: the time point at which the leading cell protrusions first touch their contralateral partners and the four preceding time points for the wild-type and the last five time points of productive migration toward the ventral midline for mutant embryos. For leading cell analysis, the perimeter of the protrusions on one side was traced Figure S5. Models for UNC-34 and WSP-1 Action during Epithelial Migration (A) Summary of genetic interactions during epidermal cell migration. The initiation of cell migration is largely dependent on WVE-1E activity, with partial redundancy of UNC-34, which might act downstream of CED-10/Rac (indicated by the question mark). In addition to WVE-1, leading cell migration requires UNC-34 or WSP-1/WIP activity for the proper formation of protrusions. (B) Molecular models of UNC-34/WSP-1 synergy. In a growing-ends model, a threshold level of available barbed ends is required for filament elongation, leading to protrusion. These ends are made available by either WSP-1-mediated de novo nucleation or UNC-34-mediated barbed-end protection. Either local activity alone is sufficient to generate protrusion. In a de novo-nucleation model, a threshold level of newly created barbed ends is required. WSP-1 performs this activity by activation of the Arp2/3 complex in the same way as the previous model. UNC-34 acts primarily as a scaffolding molecule, locally recruiting a formin to nucleate new barbed ends. Again, either WSP-1 or UNC-34 local activity alone is sufficient to generate proper LCPZ morphology.
in ImageJ with a Wacom Graphire3 input tablet, and the path length of the perimeter was measured. The width of the bodies of the leading cells was then measured with a segmented straight line tool, and the perimeter was divided by the width to yield an internally normalized metric. This process was repeated on the other side and again at each of the other time points in the analyzed sample. The same process was repeated for the pocket cells. Data were compared between genotypes for leading cells or pocket cells with a Wilcoxon rank sum test with a = 0.05. A nonparametric test was chosen largely because of concerns that the population values might not be normally distributed and probably did not have equal variances; blunted mutant protrusions probably have lower variance values than do the highly dynamic wild-type protrusions. The resulting Z statistic value for leading cells was 27.085, and a similar test on pocket cells between wild-type and unc-34(gm104) yielded Z = 22.723, which are both well below the cutoff value for a two-tailed test with an a = 0.05.

Antibody Staining
UNC-34 staining was performed with a modified Finney-Ruvkun [S34] protocol with 1:100 diluted polyclonal anti-UNC-34 and overnight room temperature incubations. Phalloidin staining was performed in a manner similar to previous descriptions [S19] but with fixations and incubations performed on a poly-L-lysine coated slide. Texas Red phalloidin (Molecular Probes) was applied with an overnight 4 C incubation.