The EFF-1A Cytoplasmic Domain Influences Hypodermal Cell Fusions in C. elegans But Is Not Dependent on 14-3-3 Proteins

Background Regulatory and biophysical mechanisms of cell-cell fusion are largely unknown despite the fundamental requirement for fused cells in eukaryotic development. Only two cellular fusogens that are not of clear recent viral origin have been identified to date, both in nematodes. One of these, EFF-1, is necessary for most cell fusions in Caenorhabditis elegans. Unregulated EFF-1 expression causes lethality due to ectopic fusion between cells not developmentally programmed to fuse, highlighting the necessity of tight fusogen regulation for proper development. Identifying factors that regulate EFF-1 and its paralog AFF-1 could lead to discovery of molecular mechanisms that control cell fusion upstream of the action of a membrane fusogen. Bioinformatic analysis of the EFF-1A isoform’s predicted cytoplasmic domain (endodomain) previously revealed two motifs that have high probabilities of interacting with 14-3-3 proteins when phosphorylated. Mutation of predicted phosphorylation sites within these motifs caused measurable loss of eff-1 gene function in cell fusion in vivo. Moreover, a human 14-3-3 isoform bound to EFF-1::GFP in vitro. We hypothesized that the two 14-3-3 proteins in C. elegans, PAR-5 and FTT-2, may regulate either localization or fusion-inducing activity of EFF-1. Methodology/Principal Findings Timing of fusion events was slightly but significantly delayed in animals unable to produce full-length EFF-1A. Yet, mutagenesis and live imaging showed that phosphoserines in putative 14-3-3 binding sites are not essential for EFF-1::GFP accumulation at the membrane contact between fusion partner cells. Moreover, although the EFF-1A endodomain was required for normal rates of eff-1-dependent epidermal cell fusions, reduced levels of FTT-2 and PAR-5 did not visibly affect the function of wild-type EFF-1 in the hypodermis. Conclusions/Significance Deletion of the EFF-1A endodomain noticeably affects the timing of hypodermal cell fusions in vivo. However, prohibiting phosphorylation of candidate 14-3-3-binding sites does not impact localization of the fusogen. Hypodermal membrane fusion activity persists when 14-3-3 expression levels are reduced.

Despite its prevalence and significance, understanding the molecular, biochemical, and biophysical mechanisms of developmental cell fusions is in its infancy [3,5,7,26]. The molecular mechanisms of virus-cell membrane fusion and intracellular membrane fusion are much better understood than cell-cell fusion [26][27][28][29], and that knowledge has already yielded crucial medicines that work via direct modulation of fusogenic reactions during infection [30][31][32]. Detailed understanding of the mechanism of cell-cell fusion in animals could be crucial for the development of many possible technologies and therapies, including tissue engineering and regeneration, cancer immunotherapy, and design of therapies for fusion abnormalities affecting muscle, bone, and reproduction [3].
The nematode Caenorhabditis elegans (C. elegans) is an ideal organism in which to study cell fusion because of its nearly invariant sequence of embryonic and postembryonic fusion events, wherein over one-third of somatic cells fuse to develop multiple syncytial cells in major tissues such as the vulva, pharynx and hypodermis (epidermis) [33][34][35]. Until recent years, no molecule was known to act as an intercellular fusion protein during the developmentally programmed formation of these multinucleated tissues. It is now known that two paralogous "FF" proteins found in C. elegans, EFF-1 and AFF-1 (epithelial-fusion-failure and anchor cellfusion-failure), are necessary and sufficient to fuse complementary sets of neighboring cells into multinucleated structures [36][37][38][39][40][41][42].
Significantly, EFF-1::GFP (which has detectable, but only partial, Eff rescuing activity) becomes concentrated at the borders of fusion-fated cells in the developing embryonic hypodermis immediately before fusion occurs [38]. Discrete localization to fusion competent cell-cell borders has also been observed in SF9 and S2R+ insect cells, which fuse when induced to express nematode EFF-1 [41,43]. While polarization and localization of cellular fusogens is important for the proper patterning of tissues, the molecular mechanism behind this tightly regulated selection remains unknown. The importance of such control, however, has been demonstrated in vivo, as unregulated expression of EFF-1 or AFF-1 leads to a lethal phenotype, in which promiscuous and destructive cell-cell fusion occurs throughout and between tissues [37,38,40].
Many questions remain unanswered regarding how EFF-1 is regulated. How is EFF-1 signaled to specifically localize to the membrane border between fusion-fated cells prior to fusions? What constrains EFF-1 from accumulating at the membrane contact with a cell neighbor that is not a fusion partner? Once localized to the membrane interface, how is EFF-1 triggered to actuate the fusion between membrane bilayers? Identifying factors that interact with EFF-1 to control its activity and localization patterns will contribute to a better understanding of the general mechanism of cell fusion mediated by FF proteins, and fusogens to be discovered in other systems.
Sequence analysis of FF proteins shows FF members in other nematode species in addition to a small number of arthropods and a ctenophore, chordate, and protist [27,40,42,44]. Recent data shows structural homology between EFF-1's ectodomain and class II viral fusion proteins; however, the mechanism by which EFF-1 fuses membranes is different from class II viral fusogen mechanisms [45,46]. Therefore, we focused our studies on the FF protein domains' structure and function. EFF-1 has four alternatively spliced isoforms, of which only two, EFF-1A and EFF-1B, have a transmembrane anchor. EFF-1A and EFF-1B share similar extracellular and transmembrane domains; however, EFF-1B's C-terminal cytoplasmic domain is shorter than EFF-1A's and varies greatly in its sequence [36]. Three lines of evidence indicate that EFF-1A functions more potently as a developmental fusogen than does EFF-1B, and that much of this increased potency in encoded in the EFF-1A C-terminus. First, EFF-1A cDNA rescues eff-1(oj55) mutant defects in C. elegans, while EFF-1B cannot rescue these mutants [38]. Second, Podbilewicz et al. showed that both EFF-1A and EFF-1B localize to the cell surface and fuse transfected SF9 insect cells in culture [41]. However, a large amount of EFF-1B cDNA, five times higher than that used for EFF-1A, was required in order to achieve similar cell surface expression levels and multinucleation. These results suggest that much more EFF-1B mRNA must be translated for an equivalent concentration of fusogen to be activated and function at the plasma membrane. Third, Sapir et al. reported that replacing AFF-1's native transmembrane domain and C-terminal cytoplasmic domain with those of EFF-1A increased cell surface expression and multinucleation in SF9 cells [40]. These results all support the hypothesis that EFF-1's cytoplasmic domain has a functional role in contributing to efficient and potent cell fusions, possibly by increasing surface expression, maintaining EFF-1 membrane stability, or triggering EFF-1 to actuate fusion. Interestingly, cytosolic C-terminal truncation of some viral fusogens greatly reduces or eliminates fusogenic activity, and cell surface expression [47][48][49][50]. Thus, intracellular sequences appear to be important to the function of a broad range of unrelated fusogenic proteins.
Many of the known actions of 14-3-3 family members upon their binding partners correspond to notable aspects of EFF-1 function in vivo. 14-3-3 proteins are often referred to as "scaffolds" that bring proteins into proximity with each other by virtue of their rigidity and propensity to form dimers [55,58]. Multiple studies have demonstrated a role for 14-3-3s in asymmetrically restricting cell polarity determinants in C. elegans zygotes, mammalian epithelia, and Drosophila oocytes and epithelial cells [59][60][61][62][63]. Evidence also implicates 14-3-3 proteins in facilitating forward transport of plasma membrane proteins through the secretory pathway [56,64,65]. 14-3-3 proteins have been shown to mask membrane protein translocation signals, such that release from 14-3-3-binding enhances membrane protein expression. Conversely, 14-3-3 binding to a cell surface protein could mask ER retention signals, thereby releasing the protein from the ER. 14-3-3 dimers are also modeled to initiate transport from the ER through the secretory pathway by simultaneously binding both to a membrane-bound protein and either to forward transport accessory protein complexes or to proteins that inactivate ER retention machinery. Based on these various models of action by 14-3-3s, we hypothesized that 14-3-3s could regulate EFF-1 delivery, accumulation, or activation at the fusion-fated cell surface. Subsequent experiments suggested that 14-3-3 proteins might be required for timely embryonic hypodermal cell fusions in C. elegans. First, mutation of the candidate 14-3-3-binding motifs in EFF-1A caused loss of the Eff-phenotype-rescuing activity of an eff-1 transgene tested in vivo. Second, human 14-3-3η bound to EFF-1A::GFP when both proteins were co-expressed in mammalian cells [51]. Based on these findings, we hypothesized that 14-3-3 proteins specifically regulate EFF-1-mediated cell fusion events in vivo, perhaps by governing EFF-1's translocation from internal organelles to the cell surface, by anchoring and/or clustering EFF-1A at the plasma membrane border programmed to fuse, or by effecting an activating change in the conformation of EFF-1A that drives membrane fusion. EFF-1's ectodomain shows structural homology to class II viral fusion proteins and a unique cell-cell fusion mechanism from class II viral fusogens for which the trigger of EFF-1 mediated cell fusion is still unknown [45,46]. Intracellular factors, such as 14-3-3, that are predicted to bind to EFF-1's cytoplasmic domain could act as the "trigger" for fusion initiation.
In this paper, we further investigate the role of the only known C. elegans 14-3-3 proteins, PAR-5 and FTT-2 [66], in controlling EFF-1's spatiotemporal function and localization. We confirm, through analysis of endogenous mutations, that the C-terminal tail of EFF-1A is required for the precise timing of cell fusions in the embryo. However, we find that normal expression of 14-3-3 proteins is not essential for EFF-1-induced hypodermal cell fusions. Furthermore, the candidate 14-3-3-binding sites within EFF-1A are not required for timely localization of EFF-1A::GFP to fusion-competent hypodermal cell contacts. These new results combine with previously published findings to indicate that potentiation of EFF-1 function and localization in the hypodermis by these putative C-terminal phospho-motifs does not require interaction with normal levels of 14-3-3 proteins. However, fusion activity is noticeably enhanced by presence of the EFF-1A C-terminal cytoplasmic tail.

RNAi
A par-5 RNAi feeding construct was generously provided by Sieu Sylvia Lee's lab [72] and was transformed into the E. coli feeding strain HT115 (DE3) [73]. The transformed bacteria were streaked from a frozen stock onto a Luria Broth (LB)-Amp agar plate overnight at 37°C for 16 hours. One colony of par-5 RNAi was grown in 5 ml LB-Amp media overnight at 37°C for 16 hours. Sixty-millimeter NGM plates supplemented with 2 mM IPTG were seeded with 200 μl of par-5 RNAi overnight culture, a sufficient quantity to sustain multiple generations of propagating C. elegans, and the plated bacteria were allowed to induce siRNA expression for 24 hours at room temperature. Five L2/L3-stage FC275 larvae (P0) were washed twice in a 200 μL drop of M9 buffer then transferred to the induced par-5 RNAi NGM plate. Larvae were cultured at 20°C and monitored for hatching and growth of their offspring (F1) to the L3/L4 stage before the F1-L3/L4 larvae were mounted for imaging (see below).

Monitoring Cell-Cell Fusion
Embryos from SU93, FC75, HC396, FC280, FC80, FC275, FC276, and FC279 were isolated and mounted as above. A spinning disk confocal microscope (PlanFluor 40X, NA 1.30 objective, Perkin Elmer/Prairie Ultraview RS5 laser launch system, Yokogawa CSU 21 microlens scanhead, Nikon Eclipse TE2000-E inverted microscope, Hamamatsu ORCA-AG cooled CCD camera) fitted with a Perfect Focus System (Nikon) was controlled by MetaMorph software (Molecular Devices) to create 4-dimensional (4D) renditions of cell fusions during embryonic development. YFP signal was excited with a 514 nm laser or GFP signal was excited with a 488 nm laser. Stacks of confocal optical sections spaced 1μm apart through the whole embryo were collected every 2.5 minutes for 12 hours. Maximum intensity Z-projections of the confocal stacks over the 12-hour time period were rendered using MetaMorph or ImageJ [74]. (For HC396 and FC80, images were taken without the Perfect Focus System.) Larvae from FC275 (par-5 RNAi), HC396 and FC280 were transferred to a 200 μL drop of M9 buffer (22mM KH 2 PO 4 , 42mM Na 2 HPO 4 , 86mM NaCl) containing 0.1 M levamisole to paralyze the worms for imaging purposes. Larvae were mounted between a slide and coverslip in M9 buffer with 0.1 M levamisole, 1% methyl cellulose, 0.09% 20-μm polystyrene beads (Polysciences, Inc., #18329) and imaged using spinning disk confocal microscopy (Plan Fluor 20X, NA 0.50 objective). YFP signal was excited with a 514 nm laser. Confocal stacks of optical sections spaced 1μm apart through the whole larva were collected. Maximum intensity Z-projections from sections through the hypodermis were rendered using ImageJ [74].

EFF-1's Cytoplasmic Domain Is Required for Normal Cell Fusion Activity in vivo
We have briefly discussed and characterized the eff-1(zz1) mutant elsewhere [51]. This mutant has a premature Trp>Stop nonsense mutation at residue 587 that eliminates all but two residues of EFF-1A's cytoplasmic domain (S1 Fig). Nonsense mediated decay (NMD) analysis suggests that the loss of function in eff-1(zz1) animals is due to defects within the expressed protein, rather than message instability brought on by the nonsense mutation (see S1 Supporting Information). To more fully understand the contribution of the missing domain to normal function, we extended our analysis of the eff-1(zz1) phenotype. All eff-1(zz1) mutant hatchlings exhibited abnormal tail whip morphology visible using a dissecting microscope. This phenotype is also manifest in every other known eff-1 loss-of-function allele. In order to more precisely appraise the severity of this molecular defect, we compared other morphological measures of the Eff phenotype among a collection of alleles. We found that eff-1(zz1) animals have a rather normal-looking body morphology, in striking contrast to strong loss-of-function eff-1 mutants (e.g. zz8, zz10, and ku433 alleles), which are Dumpy. Quantitative comparison of adult (96 hr) body length also showed an intermediate defect in eff-1(zz1) when compared to both mild and null alleles of eff-1 (Fig 1A). These phenotypic characteristics suggest that eff-1 (zz1) mutants have sufficient functional EFF-1 activity to execute cell fusions required for most of normal morphogenesis, but that some fusions are sufficiently delayed to reveal fully penetrant defects in the larval tail whip and adult body length. Alternatively, other non-fusionrelated events might be affected.
To gain further insight into the defects induced by this mutation, we directly monitored the timing of embryonic fusion events in the hypodermis of eff-1(zz1) mutant embryos by observing AJM-1::GFP at intercellular junctions via high-resolution imaging. By the 1.5-to 2-fold stage of normal embryonic development, 17 dorsal hypodermal cells (comprising 16 membrane contacts) normally complete fusion as they begin forming the hyp7 syncytium [33,75]. In strong eff-1 mutant embryos, none of these cell junctions disappear (refer to Fig 4 for AJM-1::GFP control images). However, in eff-1(zz1) embryos, we did observe disappearance of some hypodermal cell junctions-a phenomenon associated with completion of cell fusion eventswithin this normal timeframe (Fig 1B, S1 Movie). We more precisely quantified the timing of these fusions relative to other contemporaneous developmental events. To compare the activity of EFF-1 expressed from eff-1(zz1) against the activity from the wild-type gene (Fig 4, t = 450), we counted the number of fused dorsal hyp7 junctions before the first embryonic movement at the 1.5-fold stage. This test revealed a significant difference between eff-1(zz1) embryos and wild-type embryos in the number of fusion events completed by this stage (p<0.001 in independent t-test, Fig 1C).

-Expressing Cells
Despite evidence that the identified motifs are required for the full fusogenic function of EFF-1 and an observation that a 14-3-3 protein can bind EFF-1 only in the presence of these motifs, there have been no data reported assessing whether C. elegans 14-3-3 proteins, FTT-2 and PAR-5, are needed for embryonic fusion events. We therefore examined 14-3-3 loss-of-function mutants for absence or aberrations of cell fusions in EFF-1-expressing hypodermal cells. Cell fusions were monitored during two major steps of cell fusion: 1) fusion pore formation indicated by diffusive cytoplasmic content mixing using the elt-3p::yfp reporter [70], a diffuse cytoplasmic, hypodermis specific reporter characterized here for the first time as a cell-cell fusion reporter (Fig 3 and S4 Movie) or lbp-1p::gfp, which shows similar hypodermis specific expression and cytoplasmic diffusion as elt-3p::yfp [36,51,76], and 2) widening of the fusion aperture seen by displacement and disappearance of intercellular junctions between fused cells using the AJM-1:: GFP marker, a sub-adherens junction marker previously established as a cell-cell fusion reporter (Fig 4) [36,69,75]. These hallmark processes of cell fusion, cytoplasmic diffusion and disappearance of intercellular junctions, are delayed or absent in eff-1 fusion mutants (Fig 3 lower panel  and Fig 4 inset respectively). Transgenic 14-3-3 loss-of-function embryos expressing these cell fusion reporters were imaged throughout embryonic development and analyzed for defects in cell-cell fusions responsible for formation of the large hypodermal syncytium, hyp7.
First, we observed that functionally null ftt-2 (n4426Δ) mutants showed no disruptions in the reproducible timing, position, or orientation of cell fusions, when compared to wild-type embryos and worms [33,35,69]. Cytoplasmic mixing (Fig 5A and S5 Movie) and recession and dissolution of intercellular junctions (Fig 5B and S6 Movie) occurred between all dorsal fusioncompetent cells within the normal time-span of development. No ectopic fusions were seen.
Next, we assessed the requirement for PAR-5 in embryonic cell fusions. As previously reported, par-5(it55) animals exhibit the strongest expressivity compared to other characterized alleles and the overwhelming majority of par-5(it55) embryos fail to complete morphogenesis; however, cellular differentiation of multiple cell types does occur [63]. In our hands, homozygous par-5(it55) mutant hermaphrodites produced embryos that were embryonic lethal and characterized by failed morphogenesis. We assessed whether cell fusions occur in the resulting disordered tissues by use of the transgene lbp-1p::gfp, which is highly expressed in a subset of fusogenic cells of the hypodermal lineage. This reporter has previously been used to study hypodermal fusions by observing cytoplasmic diffusion of GFP between fusion-fated cells [36,51,76]. In par-5 mutant embryos, we observed several cell-cell fusions taking place, even within aberrantly formed embryos. Fig 6 (S7 Movie) shows sets of neighboring bright and dark cells rapidly exchange fluorescent cytoplasm, a characteristic trait of cell fusions in these assays [36,38,51]. The fusion-competent cells seen in Fig 6 are most likely differentiated dorsal hypodermal cells (misshapen and poorly organized), because they display the strong fluorescence intensity of lbp-1p::gfp reporter expression that is typical of this cell type [76]. These results indicate that cell fusion is possible after loss of par-5, even while other aspects of hypodermal morphogenesis are severely affected.

EFF-1 Induced Cell Fusions Persist After Reduction of Both 14-3-3 Paralogs
FTT-2 and PAR-5 are both expressed in the developing embryo [63,66]. Previous studies have shown that ftt-2 mRNA expression more than doubles upon specific knockdown of par-5 RNA, suggesting a possible compensation mechanism between par-5 and ftt-2 [72]. These data   and the high sequence identity shared by PAR-5 and FTT-2 proteins (86.2%, [77]) puts forward the possibility that PAR-5 and FTT-2 could functionally compensate for one another in the single-mutant knockouts previously described. Consequently, a loss-of-fusion phenotype would not be seen in the single-mutant knockouts if either protein could regulate EFF-1 function. Accordingly, we reduced both PAR-5 and FTT-2 levels in developing embryos to examine the effect of loss of both species of 14-3-3 protein upon cell fusion.
In light of the diverse roles that 14-3-3s play in various cellular functions of eukaryotic organisms, it could be intractably lethal to completely eliminate all 14-3-3 proteins during C. elegans development. However, RNA interference in C. elegans allows genes to be knockeddown at different stages of development with adjustable expressivity. We therefore conducted double-knockdown experiments by inducing par-5-specific RNAi in ftt-2 null hermaphrodites, using established dsRNA feeding techniques to induce systemic RNAi [78]. Initially, par-5 RNAi was fed to ftt-2(n4426Δ) larvae beginning at the L1 stage; however, this resulted in potent sterility and gonad defects as the treated animals reached adulthood, a severe par-5 mutant phenotype previously reported [63]. We subsequently adjusted the larval stages at which par-5 RNAi was fed to ftt-2 null larvae, to determine the developmental time period at which sufficient levels of maternal PAR-5 were expressed to allow for early gonad and oocyte development and the production of fertilized embryos. We found that feeding ftt-2 null larvae with par-5 dsRNA starting at the late-L2 to early-L3 stage (L2/L3) allowed for fertility in the treated larvae and avoided early stage maternal-effect embryonic lethality of their progeny. In this system, EFF-1 Cytoplasmic Domain and Cell-Cell Fusion any hatching offspring from L2/L3 par-5-RNAi fed parents continued, themselves, to receive par-5 RNAi through feeding during larval development post-hatching.
A strong par-5 knockout phenotype was observed in these "escaping" progeny, as they showed gonad defects and sterility in adulthood. This par-5 RNAi maternal effect phenotype corresponds with previous characterization of "escaping" progeny after RNAi [63]. Despite their fully expressed defects in gonadogenesis, however, these hatching offspring of ftt-2-null par-5-RNAi animals displayed no visible defects in hyp7 cell fusion, as seen in the pattern of elt-3p::yfp fluorescence, during their larval growth (Fig 7). Fig 7B shows a double row of hyp7 syncytial nuclei, resulting from embryonic and post-embryonic fusions, in a double-knockdown L4-stage larva. This pattern appears unperturbed when compared to a wild-type larva (Fig 7A) and contrasted with an eff-1 null larva (Fig 7C), each also at the L4 stage. These results indicate that the four waves of larval hyp7 fusions occur correctly in the absence of ftt-2 expression and with par-5 expression as low as it can practically be suppressed. We concluded that these dozens of EFF-1-dependent hypodermal cell fusion events do not require normal levels of 14-3-3 proteins.

Discussion
It is poorly understood how EFF-1 is regulated to accumulate and become actively fusogenic at the apical border of cells preparing for fusion. In vivo mosaic analyses by Podbilewicz et al. [41] and us (S2 Fig) show that fusion occurs only when both neighboring cells express EFF-1. This mutual dependence upon EFF-1 activity in adjacent fusion-fated cells invites a model of homotypic interaction between EFF-1 molecules at the interface between cell fusion partners [27,38,41,46]. As mentioned, recent structural studies of EFF-1's ectodomain show homology to class II viral fusion proteins but with a unique mechanism from class II viral fusogens [45,46]. The current study explored possible interactions in cis, between EFF-1's endodomain and other intracellular proteins that might regulate its function and localization. Our previous results [51] had offered seemingly strong evidence suggesting that 14-3-3 adaptor molecules bind to predicted 14-3-3 binding motifs in EFF-1's cytoplasmic domain and thereby potentiate cell-fusion activity. From those preliminary results and the multiple cellular roles in which 14-3-3s engage, we modeled several hypotheses of how 14-3-3 could regulate EFF-1 at the level of localization, oligomerization, or the activation of fusogenicity.
Despite the earlier evidence that stimulated these hypotheses, however, the current study reveals that 14-3-3s are not key regulators, either spatially or temporally, of hypodermal cell fusion directed by EFF-1. We have shown here that EFF-1A::GFP is correctly translocated and retained at hypodermal cell membranes programmed to fuse, even when the possibility of 14-3-3 binding has been abrogated. Likewise, the EFF-1 cell fusion machinery is not affected by the loss of 14-3-3 proteins, as evidenced by the persistence of hypodermal cell-cell fusions in embryogenesis and larval development when PAR-5 and FTT-2 have been reduced or eliminated. However, we cannot exclude the possibility that EFF-1A may be regulated by 14-3-3s in other syncytial tissues or as part of distinct functions or mechanisms. The range of tissues and timing at which EFF-1 is thought to trigger membrane fusion is quite diverse [2,36,39,79,80]. We have not surveyed all of these instances for fidelity in either the rate or accuracy of fusion events. Nonetheless, in the visibly prominent cell fusions that we studied here, loss of potential interaction motifs from EFF-1 or loss of potential interacting proteins PAR-5 and FTT-2 did not induce any measurable deficit in protein localization or fusogenic activity of EFF-1, respectively.

EFF-1's Cytoplasmic Domain and Its Effect on Fusogenicity
It appears that the C-terminal cytoplasmic tail of the EFF-1A splice variant, harboring the putative 14-3-3-binding consensus motifs, is required for only some actions of the eff-1 gene during development. The quantitative analysis from our high-resolution imaging of cell fusions in the embryo suggest that visible defects in the form of eff-1(zz1) larvae may result from slight delays in fusions that are critically important to formation of the tail-whip structure. It was recently shown that AFF-1, but not EFF-1, is necessary for tail-spike cell fusion [81,82]. Therefore, the actions of EFF-1 function must be necessary for other events required for normal tailwhip formation, perhaps hyp10 cell fusion, considering the highly penetrant tail-whip defects seen in eff-1(zz1) and all other currently characterized eff-1 mutants. Our new data combine with previous descriptions of abnormal tail-whip morphogenesis to suggest that the highly Larval hypodermal cell fusions occur normally in 14-3-3 double mutants. Double-knockdown mutant phenotype was generated using par-5-specific RNAi on ftt-2(n4426Δ) null mutant animals. Cells fated to fuse into the hyp7 syncytium of L4 larvae are labeled with elt-3p::yfp in three different genotypes: wild-type (A), 14-3-3 double knockdown (B), and eff-1(zz10) null mutant (C). In panels A and B, fields of syncytial hyp7 cytoplasm and nuclei display even and continuous distribution of YFP (arrows), as seen in 100% of observed larvae (n = 12 and 4 respectively). In the eff-1 null larva in panel C, arrows indicate labeled hypodermal cells that have failed to fuse with hyp7 (arrowheads), as seen in in 100% of observed larvae (n = 6). Scalebar = 10 μm.
doi:10.1371/journal.pone.0146874.g007 EFF-1 Cytoplasmic Domain and Cell-Cell Fusion penetrant larval tail-whip defects seen in eff-1(zz1) are likely due to a rather short delay in cell fusions. It seems, therefore, that the morphogenesis of some structures, such as the tail, resulting from EFF-1 function are more exacting in their need for temporally precise fusion events than are other structures in the body plan. Alternatively, it is possible that other non-tail tissues simply suffer less-visible or less-functionally obvious defects when EFF-1-dependent fusions are slightly delayed by a minor decrease in EFF-1 activity. It may also be true that the EFF-1A isoform is uniquely required for tail-whip morphogenesis, the only context in which the functions of EFF-1A and -1B have been compared directly in vivo [38]. These possibilities remain to be tested.
We hypothesize that EFF-1(zz1)'s decreased potency is a result of decreased protein stability, impairment of cell surface expression, or fusogen activation. Reduced viral fusogen oligomerization and reduced interaction of viral fusogens with accessory proteins at the membrane have both been reported in viral fusion-protein mutants with truncated cytoplasmic tails [49,83]. We can exclude the possibility that EFF-1A's cytoplasmic tail is essential to the core membrane fusion mechanism, because cell fusions in eff-1(zz1) do occur, albeit at a slightly delayed rate. But the possibility remains that the EFF-1A C-terminus offers a fusion-enhancing function, as has been reported for some viral fusogens. For example, truncation of the simian parainfluenza virus 5 (SV5) fusion protein's cytoplasmic domain impedes fusion pore enlargement and the endodomain of reovirus FAST fusogens is important for the membrane fusion mechanism and syncytiogenesis [50,84]. Future studies will be needed to determine whether the fusogenic activity per EFF-1 molecule is actually affected by this or any mutation. Currently, our data do not discriminate between changes in stability, activity, or localization.

Comparing Analyses of EFF-1 Function in vivo and in Exogenous Assay Systems
This report describes techniques that more precisely measure fusogen function potency in worm strains carrying modified versions of eff-1. Fine tail whip structure remains the most sensitive bioassay for the detection of a slight reduction in eff-1 function, as it is disrupted in all alleles that we have studied. The strong penetrance of tail defects produces an essentially binary signal. Defects are seen in the tail whip of all offspring with reduced or absent EFF-1 activity versus a normal tail whip phenotype in all offspring of wild-type EFF-1 animals. Alleles with severe loss of molecular integrity present with defects in body length and morphology. Analysis of a range of mutant alleles using these assays allows us to rank-order molecular defects by the degree of decreased body length (Fig 1A). In contrast, the embryonic hyp7 fusion-timing assay -used here to measure deficits in the eff-1(zz1) mutant (Fig 1B and 1C)-should allow for quantification of subtle differences in fusogenic activity among weakly hypomorphic or hypermorphic eff-1 alleles. Fine tail whip structure, body length, and hyp7 fusion-timing could be used to assess the activity of transgenic EFF-1 variants expressed during mutant-rescue assays but the strength of EFF-1 activity produced from endogenous and transgene loci has yet to be quantitatively reconciled for eff-1. Until reconciled, direct evaluative comparisons between mutations in loco and in vitro are not possible.
Concordance between heterologous organism-and cell-based assays, in vivo assays, and predictive algorithms varies. We found in our analysis of eff-1(zz1) that the loss of the full EFF-1A C-terminal tail produces a measurable deficit in the fusogenic activity of eff-1 gene products in vivo. These conclusions agree with data previously published using transfected EFF-1A and EFF-1B cDNAs to induce fusion of cultured insect cells [41]. Thus, some structure/function analyses carried out in an alternative experimental system can correctly predict the behavior of the molecule within the cells of developing nematode tissues. In contrast, we saw no appreciable impact on EFF-1 localization or fusogenic function when we examined mutations specifically disrupting a predicted interaction between EFF-1 and endogenous 14-3-3 proteins. In this case, our observations do not harmonize with previous evidence or predicted function for a physical binding interaction between EFF-1A and a human 14-3-3 protein. Apparently, in this case, the extrapolations from in vitro data are not supported in vivo. Alternatively, the impact of the loss of a 14-3-3/EFF-1 interaction, if one exists in vivo, must be slight or difficult to visualize in the tissues we examined.

Post-Translational Modification of the Cytoplasmic Domain of EFF-1
Our previous results showed that alanine-substitution of serines S632, S634, and S654 reduces the activity of EFF-1 (expressed from a transgene) and blocks the ability to rescue endogenous mutations [51]. While we have not directly shown that these serines are phosphorylated in vivo, our previous in vitro results showed that a phospho-specific protein, human 14-3-3, will only bind EFF-1A::GFP if these three serine residues have not been replaced with non-phosphorylatable alanines. In addition, threonine-substitution of these serines, to retain a phosphorylatable residue at each position, conserves the cell-fusion activity of an eff-1 transgene expressed in the nematode [51]. The NetPhos 2.0 phosphorylation prediction algorithm scores S632 and S634 as highly likely to be phosphorylated, and S654 as not as likely [85]. Interestingly, site-directed deletion of S654 (653-655) from EFF-1A has no detrimental effect on a transgene's ability to rescue eff-1(oj55) mutants (data not shown). For reasons we cannot explain, it seems that mutation of the S654 to alanine proves to be more detrimental than this triplet deletion. The strong prediction of S632 and S634 phosphorylation gives reason to believe that these serines on the EFF-1A C-terminus may become phosphorylated in C. elegans in vivo.
Using a prediction tool (KinasePhos 2.0) for phosphorylation sites and the kinases that act on such sites, we found that multiple kinases are predicted to act on S632, S634, and S654 [86]. Testing whether EFF-1-dependent cell fusion is sensitive to the activity of C. elegans kinase homologues may be an interesting course of investigation as control of EFF-1 fusogenicity by phosphorylation would be a novel form of fusogen regulation. Likewise, generation of phosphomimetic mutations (aspartate or glutamate) at S632 and S634 might reveal phosphorylation effects on EFF-1 function that are detectable by one of the in vivo assays established in this study. To our knowledge, phosphorylation has been shown to only indirectly affect virus-cell fusion and virally induced syncytium formation [87,88]. In any hypothesis, modification of EFF-1A's cytoplasmic tail can only modulate its activity to generate efficiently timed cell fusions, since we have shown here that delayed EFF-1-dependent fusions can still occur in the absence of its C-terminus.  [41], and therefore strongly support the model that EFF-1 acts homotypically, required by both cells for fusion to occur. Strain Construction: FC196: N2 (Bristol) C. elegans hermaphrodites were transformed by microinjection of pSur5Rc and pJE8 (wild-type eff-1) to generate extrachromosomal array zzEx78. pSur5Rc, a gift from Morgan Tucker and Min Han at the University of Colorado, includes the DsRed2 coding region (Clonetech) ligated via KpnI/EcoRI subcloning downstream of a 3.6 kb sur-5 promoter, originally derived from pTG96.2 [89]. Worms with red nuclear fluorescence were selected from the progeny following injection and were crossed to N2 males. FC204: FC196 (zzEx78 [eff-1+; pSur5Rc]) males were mated with FC75 (eff-1(zz10) II; jcIs1 IV) hermaphrodites. A strain carrying zzEx78 and homozygous for both eff-1(zz10), and jcIs1 was identified by observing that all worms not carrying zzEx78 exhibited 100% fusion-defective phenotypes (homozygous eff-1(zz10)), while progeny carrying zzEx78 were rescued for larval tail-whip defects and disappearance of AJM-1::GFP junctions in the hypodermis. Imaging: Loss of the extrachromosomal array zzEx78 expressing eff-1+and DsRed2 during embryonic cell division results in a mosaic pattern of red fluorescence. Mosaic animals were identified in which eff-1-rescued cells expressing DsRed2 lay adjacent to non-red (eff-1 null) cells. Larvae were paralyzed with 1M sodium azide and confocal image stacks were acquired on either a Perkin Elmer Ultraview RS5 or a Zeiss LSM 510 Meta confocal scanning microscope. Laser excitation used was at 488nm for GFP excitation and either at 568nm or at 543nm for DsRed2. GFP and DsRed2 channels were separated using linear unmixing software (Zeiss). Confocal z-stacks were converted to TIFF format and rendered as projections using Image J software [74]. (TIF) S1 Movie. Animals expressing EFF-1A with a C-terminal truncation have delayed embryonic cell fusions. Maximum-intensity projection of an eff-1(zz1) embryo expressing an adherens junction marker (AJM-1::GFP) imaged by 4-dimensional confocal microscopy. Arrows denote fused junctions and arrowheads indicate unfused cell borders, with intact junctions still observed before the embryo begins muscular movement. Anterior is left, dorsal is up. Time shown is approximate age since fertilization. Scalebar = 10 μm. Early cytoplasmic fluorescence seen in gut-fated cells (no longer visible at the time of adherens junctions phenotyping) is expressed from the mIs12 transgene, which was included in the background in which we screened for the zz1 mutation and is tightly linked to eff-1 on chromosome II. S3 Movie. EFF-1(S632/634/654A)::GFP accumulation at a fusion-fated cell border on the dorsal embryo surface. Time-lapse maximum-intensity projection of the dorsal surface of the embryo. Arrow indicates EFF-1(S632/634/654A)::GFP accumulation at a cell contact. Onemicron-spaced image stacks were captured every 2.5 minutes using widefield microscopy, and maximum intensity Z-projections of the dorsal surface were rendered. In 100% of the mutant embryos (n = 4), the same pattern of junctional localization is seen as for wild-type EFF-1::GFP [38]. Scalebar = 10 μm.