Actin is required for specific processes during early embryogenesis

To ask if different processes in the early embryo that depend on an intact actin cytoskeleton are differentially sensitive to the amount of available actin, we performed a time-course analysis of early embryogenesis after injection of dsRNA targeting all actin genes using act-1(RNAi) (see Materials and Methods, Figure 1, and Additional File 1: Supplementary Figure 1). To verify and quantify the depletion of actin, we used an anti-actin antibody to detect the amount of protein in the gonad over time. We observed that protein levels were reduced by about half at approximately 6 hours post-injection, declined sharply between 7–8 h, and finally tapered off (Figure 1A and 1B). We will refer to the embryos produced by treated animals as actin(RNAi) embryos. Analysis of actin(RNAi) embryos at distinct time points revealed a phenotypic series analogous to what might be seen if we had an allelic series for a family of redundantly functioning genes (Figure 1B and 1C).

The wild-type (WT) embryo undergoes a series of canonical events that are easily observable using time-lapse DIC microscopy (Figure 1C, Additional File 4: Movie 1). We review them here and compare them with events that occur in actin(RNAi) embryos (Figure 1C, Additional Files 5, 6, 7, 8 Movies 2–5). After the completion of meiosis and the extrusion of two polar bodies, approximately 25 minutes after fertilization, the maternal and paternal pronuclei form at anterior and posterior ends, respectively. Concurrently, the cortex of the embryo undergoes intense contractions that cease at the posterior end of the embryo when the paternal pronucleus closely associates with it. These contractions culminate in a deep invagination near the middle of the embryo referred to as the pseudocleavage furrow. At this time, the maternal pronucleus migrates toward the paternal pronucleus. The pseudocleavage furrow relaxes as the pronuclei meet in the posterior half of the embryo. The pronuclei do not immediately fuse upon meeting, but move together toward the middle and rotate, such that the division axis is established perpendicular to the long axis of the embryo before nuclear envelope breakdown. The spindle is displaced at anaphase to the posterior of the embryo, giving rise to two daughter cells of different size. These cells have distinct fates giving rise to different parts of the adult worm. The distinct identities of these cells are characterized by the way they divide in the second round of mitosis: they undergo mitosis at different times, with the larger anterior AB cell dividing before the smaller posterior P1 cell, and their division axes are perpendicular to each other (Figure 1C, Figure 2A, Additional File 4: Movie 1).

The effect of actin(RNAi) on these canonical early embryonic events depends on the time elapsed between injection of dsRNA and initiation of embryogenesis, due to a progressive depletion of actin over time in the treated animal (Figure 1). Eventually, actin(RNAi) leads to a drastic reduction of fertilized embryos, approximately 12–15 hours after injection; in contrast, control animals produce fertilized embryos for at least twice as long (see Additional File 2: Supplementary Figure 2; [18]). These results are consistent with a recent analysis showing that the lateral plasma membrane collapses in gonads cultured with high concentrations of actomyosin inhibitors [36], which could lead to disruption of oocyte integrity and fertilization. Thus, the existence of any embryos indicates that oocyte integrity is maintained by residual actin activity, and therefore represents a hypomorphic condition. Prior to this study, the only hypomorphic phenotypes reported in actin(RNAi) embryos, as part of a large scale study, were failure to extrude polar bodies, loss of spindle displacement, and loss of cytokinesis [18]. By selecting embryos for analysis from RNAi-treated animals in a carefully timed way, we were able to observe a reproducible series of defects that we could calibrate for each batch of dsRNA (see Materials and Methods). We grouped embryos showing similar ranges of progressively more severe defects into four phenotypic classes (Classes I-IV; Figures 1C and 2). Specific phenotypes appearing together within a class thus represent processes that require similar levels of actin in the cell. The range of phenotypes observed in the four phenotypic classes is summarized in Additional File 3: Supplementary Table 1.

Class I defects (Figures 1C (n = 11) and 2B (n = 5), Additional File 5: Movie 2) highlight the processes that are affected by the mildest reduction in available actin, and thus require the highest levels of actin. In the first cell cycle, cortical ruffling and pseudocleavage were absent. Spindle movements in P0 appeared normal and the first cell division was asymmetric, whereas the AB and P1 cells divided in the correct orientation but with abnormally synchronous timing. Thus, Class I defects appear to separate two processes, spindle orientation and timing of the second division, which are thought to be coupled and to depend on proper polarity establishment in the one-cell stage. In Class II embryos (Figures 1C (n = 12) and 2C (n = 5), Additional File 6: Movie 3), the paternal pronucleus failed to maintain its close association with the posterior cortex, causing the pronuclei to meet more centrally. After nuclear fusion, the spindle failed to move toward the posterior prior to cytokinesis, giving rise to more symmetric daughter cells than in WT. Finally, the second mitotic cell divisions were both synchronous and oriented parallel to each other, transverse to the AP axis (Figure 2C). Class III embryos (Figures 1C (n = 12) and 2D (n = 4), Additional File 7: Movie 4) began to show cytokinesis defects, as well as disorganized furrows during subsequent cellular divisions that gave rise to polynucleated cells. Nuclear reformation occurred close to the contractile ring, an indication of reduced elongation of the central spindle during anaphase B. Severely depleted Class IV embryos (Figures 1C (n = 14) and 2E (n = 4), Additional File 8: Movie 5) failed both to extrude both polar bodies and to execute cytokinesis, and contained small karyomeres. Interestingly, even among the most strongly affected embryos, the rotation of the spindle in P0 occurred normally.

The reformation of the nuclei near the cell division remnant and the presence of karyomeres in actin(RNAi) embryos suggested that they might also experience chromosome segregation defects. To further explore the effect of actin depletion on chromosome segregation, we analyzed the same time points in a line expressing a GFP-histone fusion (GFP::HIS) [37], which reveals chromosomal movements and dynamics. In addition to confirming the orientation of mitotic chromosomes and cell cycle timings, we also observed that anaphase figures showed partially decondensed chromosomes, lagging chromosomes, and chromosome bridges in Class III and IV embryos (Figure 2D and 2E). These data indicate that severe actin depletion interferes with the proper resolution of chromosomal pairing and/or movement of chromosomes to the spindle poles.

Cortical F-actin undergoes three phases of morphological transitions in the one-cell embryo

To visualize actin dynamics in vivo in the early embryo we fused GFP with the D. melanogaster Moesin F-actin binding domain [38]. This GFP::MOE marker has been shown to faithfully report the distribution of microfilaments in both D. melanogaster [39] and C. elegans [38]. All the structures that we observe with our GFP::MOE line are therefore F-actin-rich structures decorated with GFP molecules. Previous analyses using phalloidin-stained fixed specimens [19, 20] or in vivo visualization of NMY-2::GFP dynamics [23] during the first cell cycle of embryogenesis have shown that actomyosin forms a cortical lattice that appears to contract asymmetrically to the anterior half of the embryo [23]. GFP::MOE embryos similarly show a dynamic meshwork of cortical actin fibers and dense focal accumulations, as well as numerous small punctate structures. These structures all become enriched to the anterior with similar kinetics as NMY-2::GFP. However a persistent collection of small actin-rich puncta remains at the extreme posterior throughout the first cell cycle. The asymmetric redistribution of microfilaments does not appear to result from collapse of the meshwork under tension; rather, the dynamic formation and dissipation of F-actin structures that we observe in the anterior indicates that microfilaments are in rapid flux. This suggests that the capacity for filament assembly becomes segregated to the anterior domain while assembly is reduced in the posterior. At pronuclear meeting, the fibrous actin meshwork revealed by GFP::MOE transforms into a collection of discrete puncta in the anterior half of the embryo that persists until just prior to contractile ring formation, at which point cortical F-actin rapidly dissipates (Figure 3, Additional File 9: Movie 6).

Thus, there appear to be three phases involving a major reorganization of the cortical actin cytoskeleton that are coordinated with the cell cycle in the one-cell embryo: first, the asymmetric redistribution of F-actin structures to the anterior (we refer to this as "Phase I"); second, concomitant with pronuclear meeting and centration, a shift in the overall morphology of predominant cortical F-actin structures from a dynamic meshwork to discrete puncta ("Phase II"); and finally, a rapid clearance of cortical F-actin in anticipation of cytokinesis ("Phase III").

The proper anterior redistribution of microfilaments depends on PAR proteins

The diversity of actin-rich cortical structures and regulated transitions in their morphology led us to investigate the role of F-actin regulators on their dynamics. Regulators of the actin cytoskeleton have been shown to influence the localization of PAR proteins in C. elegans and other organisms [6, 24]. Severe depletion of CDC-42, which binds PAR-6 [40], leads to sterility and osmotic sensitivity, and the actin cytoskeleton fails to assemble on the cortical surface (data not shown) [24, 41, 42]. Less severe depletion of CDC-42 does not affect the initial polarization of F-actin to the anterior during Phase I, but by pronuclear meeting actin becomes hyper-asymmetrically concentrated in the anterior (n = 6; Figure 3, Additional File 13: Movie 10), similar to par-1(RNAi) embryos. Unlike par-1(RNAi), however, this hyper-asymmetry persists throughout Phase II, and the domain of F-actin-rich puncta at the extreme posterior appears to be expanded relative to WT during Phase II. Overall, the association of F-actin structures with the cortex appears weaker than in WT, and the contractile ring is very faint. Therefore, CDC-42 is needed to maintain the initial asymmetric distribution of F-actin and for a strong association of F-actin with the cortical surface.

Actin-binding proteins and nucleators of actin affect the morphology of cortical microfilaments

Profilins are actin-binding proteins that also interact with other regulators of the actin cytoskeleton and can affect actin dynamics in a variety of ways depending on the molecular context (reviewed in [43]). C. elegans contains three profilins with similar biochemical properties that all behave as classical nonvertebrate profilins [44]. Only one of these, PFN-1, is essential for embryogenesis, where it is required for cytokinesis and polarity in the one-cell embryo and is also required for the proper localization of F-actin and NMY-2 to the cortex [16]. To investigate profilin's role in polarity and its effect on cortical microfilaments in more detail, we analyzed embryos depleted of PFN-1 by time-lapse microscopy. We found that pfn-1(RNAi) prevented the appearance of an actin-rich meshwork at the cortex, while actin-rich puncta appeared much earlier and moved more freely around the cortex than in WT (n = 6; Figure 3, Additional File 14: Movie 11). These puncta began to coalesce into large actin-rich clumps between pronuclear meeting and centration (when the Phase II transition normally occurs in WT). Cytokinesis never occurred, although some residual contractile activity could be observed on the cortex during mitosis (Additional File 14: Movie 11 shows a striking ring-like formation of cortical actin-rich structures that appears to contract into a single aggregate). These observations indicate that the early embryo contains two genetically separable populations of cortical actin-based structures, a fibrous meshwork and distinct punctate accumulations, and that profilin appears to be required exclusively for the assembly of filaments that form the meshwork. Profilin is also required to correctly position and coordinate contractile activity required for cytokinesis with the mitotic spindle, as has been observed previously in C. elegans and other systems [45].

The Arp (actin-related protein) 2/3 complex promotes rapid actin polymerization in response to various signals and results in the formation of branched filaments, which allow for cytoskeletal reorganization in the cell [5, 46–49]. In C. elegans, depletion of either ARX-1 (Arp2) or ARX-2 (Arp3) leads to embryonic lethality; while no early embryonic phenotypes have been reported, ventral closure and postembryonic defects have been observed [50, 51]. Disruption of the Arp2/3 complex by depletion of ARX-1 prevented the formation of an F-actin meshwork at the cortex; a few transient F-actin structures were observed, but no foci formed upon pronuclear meeting (n = 4; Figure 3, Additional File 16: Movie 13). Despite the lack of F-actin enrichment in the anterior, the cell was able to undergo an asymmetric cell division, indicating that other factors required for the establishment of polarity and the boundary between anterior and posterior cortical domains were unaffected by depletion of ARX-1. F-actin was properly recruited to the contractile ring and cytokinesis occurred as in wild type.

The C. elegansembryo contains highly dynamic actin-rich "comets"

In the course of characterizing actin dynamics in the GFP::MOE strain, we were struck by the presence of "comet tails" that moved rapidly through the cytoplasm and near the cortex in the one-cell embryo. Actin-rich comets have been described in association with both endocytic vesicles [5, 26–28, 30] and intracellular pathogens such as Listeria, Shigella, and Rickettsia (reviewed in [5, 31–33]). In C. elegans, we noticed that the comets have no consistent direction – often appearing as curvy or curly structures – and that they vary in length, speed and number throughout the first cell cycle (Figure 4, Additional File 17: Movie 14). We also observed comets in developing and maturing oocytes and in later stages of embryogenesis (data not shown). The observation of similar unknown structures has been reported in fixed samples [19], but their dynamic nature has not previously been described. To confirm that actin comet tails were not artifacts of the GFP::MOE transgenic line, we visualized F-actin in fixed WT and GFP::MOE embryos with rhodamine-conjugated phalloidin (Rh-phalloidin). Rh-phalloidin clearly labeled comet tails in both lines, and co-localized with GFP antibodies in the GFP::MOE line (Figure 4B). DAPI did not stain the heads of the comet tails, dispelling the notion that they are associated with an intracellular bacterial pathogen (Figure 4B).

To begin to elucidate the potential role of actin comets, we characterized their behavior. Since these comets are more numerous during the first cell cycle, we quantified the number of comets present during this time and observed that the number of comets peaks around prometaphase (Figure 4C). Additionally, during the period from pronuclear meeting to prometaphase, the comets appear to coalesce around the pronuclei and sometimes to associate with the centrosomes (Additional File 11: Movie 8). This suggests that an accumulation of any cargo propelled by the comets may ultimately be required at some location near or defined by the nuclei and/or mitotic spindle. We measured speeds of actin comets in C. elegans and found that they travel at an average of 0.22 ± 0.10 μm/s (Figure 4D), similar to the mean speed and range of endocytic vesicles in transfected rat basophilic leukemia (RBL) cells (0.24 ± 0.10 μm/s) [27]. Although we observed that some comets were thicker than others, we found no correlation between their widths and speed (Figure 4E). We note that the trajectories and speeds of actin comets are quite distinct from those observed for an early endosome marker in the early embryo [34]; thus we believe they represent a different population of cellular structures.

We next asked if the actin comets were dependent on known regulators of F-actin assembly such as CDC-42, ARX-1 (Arp2), and PFN-1, which are required for the proper formation of actin comets in other systems. The Arp2/3 complex nucleates microfilaments, and CDC-42 promotes Arp2/3 activity by stimulating nucleation promoting factors (NPFs) that interact with Arp2/3 (reviewed in [5, 31–33]). Depletion by RNAi showed that all three actin regulators are required to achieve the maximal number of comets observed at prometaphase in WT (Figure 4C). Depletion of PFN-1, which promotes polarized filament growth and increases the rate of movement of comets in other systems [31, 32], caused the most dramatic reduction in the basal number of comets, which rose only slightly at prometaphase (Figure 4C). We cannot distinguish between the possibilities that further reduction of these proteins would lead to complete loss of comets, or that other cellular factors may be able to promote a basal level of filament nucleation in these comets. In contrast, depletion of PAR-6 (data not shown) or NMY-2 (Figure 4C) had no significant effect on either basal or maximal comet numbers, although in nmy-2(RNAi) embryos comets persisted well beyond telophase, particularly when the contractile ring was weak or failed to form (Figure 4C and data not shown).

We noticed no qualitative change in the speed or length of the comet tails in arx-1(RNAi), cdc-42(RNAi), pfn-1(RNAi), or nmy-2(RNAi) embryos (data not shown). We conclude that actin-rich comets in the C. elegans one-cell embryo are influenced by a cell cycle-dependent process that is coordinated with the activity of actin cytoskeletal regulators and that their speed is consistent with a potential role in endocytosis, but further characterization is needed to determine their precise role during development.

Actin-dependent processes in the early embryo are differentially affected by depletion of actins by RNAi

Since actin is required for many diverse processes in the cell, deciphering its specific contribution to different cellular processes during development is challenging. By performing a time-course depletion of multiple actin isoforms simultaneously, we have described four classes of increasingly severe phenotypic responses in the early embryo. We have shown that among the processes most sensitive to actin depletion are cortical ruffling, pseudocleavage, and establishment of polarity. Progressively more severe depletion of actins led to reduced elongation of the central spindle, failure to displace the spindle toward the posterior, aberrant coalescence of telophase chromosomes, failure to extrude polar bodies, cytokinesis failure, and the appearance of chromatin bridges during mitosis. Eventually, sterility ensues, indicating a requirement for actin in oogenesis [36] and potentially in fertilization. We never observed defects in centrosome duplication or separation, or in rotation of the pronuclei in the one-cell embryo, indicating that microtubule-dependent processes are largely unperturbed by defects in the actin cytoskeleton. However, defects in spindle positioning and elongation during anaphase suggest that microtubule interactions with the cortex are perturbed by disruption of the actin cytoskeleton. These analyses allow us to differentiate processes that have a strict requirement for actin (cortical contractility, pseudocleavage, and polarity) from those that can take place in the presence of a lower level of actin (spindle displacement and elongation, polar body extrusion, cytokinesis, and fertilization) or those that appear to be completely independent of actin at these stages (nuclear migration, spindle rotation, and centrosomal duplication and separation).

Microfilament dynamics during early embryogenesis

We have constructed a GFP::MOE line that enables observation of the dynamic nature of the actin cytoskeleton during early development in C. elegans. Using this GFP::MOE line, we have examined the dynamics of actin during the first cell division at high resolution and describe in detail the dynamic process in which F-actin becomes asymmetrically distributed to the anterior half of the embryo during the first mitotic cell division. Our observations clearly reveal three distinct phases of cortical F-actin dynamics during the first cell cycle: I) an anterior redistribution of a dynamic meshwork of fibers and foci corresponding to the period of cortical flow, II) a subsequent morphological transition to punctate structures in the anterior, and III) a global clearing from the cortex prior to assembly of the cytokinetic furrow. This final clearing of cortical F-actin in anticipation of cytokinesis (Phase III) has not been previously described.

Studies using either fixed specimens stained with phalloidin derivatives [19, 20] or in vivo imaging with NMY-2::GFP [23] or GFP::MOE [24] to visualize the actomyosin network have described an asymmetric distribution of both F-actin and myosin to the anterior during the first cell cycle. We also observe that F-actin becomes asymmetrically enriched in the anterior half of the embryo (Phase I), concomitant with the anterior PAR proteins and NMY-2) [10, 23, 24]. However, in contrast to the NMY-2 lattice at the cortex, which visibly contracts toward the anterior, the cortical F-actin meshwork appears to contain dynamic fibers that continuously form and disappear, which become progressively restricted to the anterior half of the embryo (Figure 3, Additional File 9: Movie 6). As discussed below, this anterior redistribution depends on actin regulators and PAR proteins.

At pronuclear meeting, the morphology of actin structures transforms from a fibrous meshwork into a collection of distinct punctate structures (Phase II). This transition coincides with the change of morphology of cortical NMY-2::GFP structures from large dynamic aggregates to a field of small puncta [23], and a transition in the appearance of anterior PAR-6 from a punctate to smooth morphology [24]. This morphological transition represents a second phase of actin reorganization.

Establishment of polarity

During the establishment of cortical polarity, the observation that filaments in the anterior domain apparently undergo rapid turnover suggests that some factor(s) required for filament nucleation and/or elongation becomes restricted to the anterior, such that filaments are rapidly lost in the posterior during the period of cortical flow. This observation is consistent with the finding that at the onset of polarity ECT-2, a putative guanine exchange factor and activator of RHO-1, becomes excluded from the posterior cortex in the vicinity of the sperm pronucleus, and this is dependent on an intact centrosome [24]. Both ECT-2 and RHO-1 are required for the localization of actin to the cortex and for polarization of the embryo [24]. Interestingly, the introduction of sperm-associated RhoGAP CYK-4 (which inactivates RHO-1), is also required for polarity, but cyk-4(RNAi) embryos retain cortical contractility [25]. Thus while contractile myosin activity is important for establishing polarity, other factors regulated by Rho signaling must also play a central role in this process. Taken together, these results suggest that microfilaments in the cortical meshwork are normally in flux, that these filaments rapidly disassemble in the posterior after onset of polarity due to the depletion or inactivation of factors capable of sustaining filament nucleation and dynamics, and that these factors are regulated downstream of RHO-1.

In other systems, Rho family proteins regulate both of the primary actin nucleation mechanisms in the cell, the Arp2/3 complex and Formin Homology (FH) proteins [2]. Arp2/3 appears to be dispensable for polarity in the C. elegans embryo [16] but is necessary for the formation of the actomyosin meshwork at the cortex (Figure 3). Formins recruit profilin to rapidly extend the free ends of uncapped actin filaments (reviewed in [52, 53]). Our observations show that depletion of C. elegans PFN-1 leads to a marked reduction in the number of actin comets in the cytoplasm (Figure 4C), prevents the formation of the fibrous meshwork observed in WT, and ultimately leads to loss of polarity and cytokinesis failure. pfn-1(RNAi) embryos initially display small punctate structures at the cortex that increase in number and begin to aggregate as the cell cycle progresses. The reduction of the actin meshwork but the retention of prominent actin-rich puncta upon depletion of PFN-1 suggests that the populations of microfilaments in these two types of structures are distinct, and that PFN-1 contributes primarily to the elongation or turnover of filaments in the meshwork. The existence of different subpopulations of cortical microfilaments would be consistent with work in other systems showing that different types of actin structures are nucleated and organized by different combinations of actin-binding proteins [40, 54–56].

Our observations of highly dynamic actin fibers in WT and their loss in pfn-1(RNAi) embryos, combined with data demonstrating that polarization depends on RHO-1 activity, suggest a mechanism for breaking the actomyosin symmetry in response to the polarization cue. Like PFN-1, the FH protein CYK-1 is required for the maintenance of microfilaments on the cortex, and PFN-1 binds to CYK-1 [16]. Profilin retains its cortical localization even when microfilaments are eliminated in the embryo by treatment with Latrunculin A [16]. Together these data suggest the possibility that filaments comprising the actin meshwork in the C. elegans early embryo are nucleated and rapidly extended primarily by CYK-1 working together with PFN-1, and that this activity is regulated by RHO-1 and its effectors. Down-regulation of RHO-1 activity in the posterior would lead to a localized loss of this nucleation/elongation activity and consequent rapid disassembly of actin filaments. This localized absence of filament integrity in the posterior would in turn lead to cortical flow, mediated by the migration of cortical contractile activity toward the anterior as the uniform meshwork maintaining balance in contractile forces is lost. Additional factors could also contribute to this mechanism, as Rho family proteins in other systems are known to also regulate contractility and actin dynamics through myosin light chain kinase and cofilin [2]. Further investigation will be required to precisely define the mechanisms through which microfilament dynamics are regulated in the early embryo.

PAR proteins and the cortical actin cytoskeleton

As previously described, the anterior PAR proteins are required for the proper localization and maintenance of F-actin in the anterior [20]. We also found that PAR-6 is needed for the proper segregation of F-actin to the anterior domain. We observed an initial weak clearing of actin in the posterior cortex that was lost by the time of pronuclear centration and rotation, at which time actin was distributed uniformly throughout the cortex. par-2(RNAi) resulted in a similar effect on the cortical actin cytoskeleton, including an initial asymmetry that was then lost. These results suggest that the PAR system is not required for breaking actin symmetry, but is required for reinforcement and maintenance of actin asymmetry, and that the anterior PAR complex and actin cytoskeleton are mutually dependent for the maintenance of polarity in the one-cell embryo. The PARs and the actin cytoskeleton may thus respond to the same initial cues that eventually lead to the establishment and maintenance of distinct anterior and posterior domains. This is consistent with previous reports providing evidence for positive feedback between the PAR system and the actomyosin cytoskeleton during the establishment and maintenance phases [23, 24].

Our results suggest that PAR-1 and CDC-42 are needed to establish the proper boundary of F-actin distribution at pronuclear meeting, as depletion of these proteins causes a hyper-asymmetric enrichment of F-actin to the anterior. Loss of CDC-42 also allows some F-actin to begin to accumulate at the posterior pole by the time of pronuclear centration and rotation. Overall we observed fewer foci, and cortical fibers appeared fainter, in embryos depleted of CDC-42 as compared with WT, indicating that regulation of filament dynamics by CDC-42 contributes to general actin cytoskeletal organization in the early embryo. CDC-42 is required for the maximal cortical accumulation and maintenance of the PAR-6 domain [57]; thus a reduced level of cortical PAR-6 in cdc-42(RNAi) embryos could contribute to the observed reduction in the extent of the anterior F-actin domain. Severe depletions of CDC-42 caused embryos to become osmotically sensitive ~36 hours after injection and worms eventually became sterile, potentially revealing cross-talk between signaling pathways controlling the actin cytoskeleton and other cortical activities such as egg shell formation.

Actin comets and their potential role in the cell

In this paper we describe the presence and dynamics of highly motile actin comets that have not been previously recognized in C. elegans. Similar comet tails have been observed in association with the movement of different types of endocytic vesicles and intracellular pathogens, which exploit the cell's cytoskeletal machinery to propel them through the cytoplasm and into neighboring cells (reviewed in [5, 31–33]). The Arp2/3 complex has been implicated in the internalization of membranous organelles in Dictyostelium, budding yeast, Xenopus extracts and eggs, and mammalian cells. In vivo studies and in vitro reconstitution experiments have also established that the Arp2/3 complex is one of the primary cellular factors driving the propulsion of actin-rich tails. The Arp2/3 complex nucleates new actin filaments and cross-links newly formed actin filaments in a branched pattern, and is regulated by Rho family proteins and a variety of nucleation promoting factors (NPFs) (reviewed in [53]). Organelles and bacterial pathogens initiate filament assembly by stimulating Arp2/3 activity through a combination of such factors, and use polymerization at the rear surface to generate the propulsive force for intracellular movement.

To begin to address what the role of such comets could be in C. elegans, we characterized the dynamics and behavior of the actin comets in the first embryonic cell cycle, where they are most easily observed. We found that the speed at which actin comets move in the C. elegans early embryo (0.22 ± 0.10 μm/s) is very close to reported speeds for actin comet tails associated with endocytic vesicles in mammalian cells (0.24 ± 0.10 μm/s) [27]. We observed no correlation between the speed of the comets and their width (thickness), suggesting that the rate of movement is not limited by physical factors such as resistance to movement due to viscosity of the cytoplasm, but may be determined by molecular factors such as actin filament nucleation or polymerization rates. In support of this idea, the motility of Listeria occurs at the rate of actin polymerization and depends on the availability of renewable ATP and molecular tail components required for propulsion (such as actin and profilin) [58, 59].

We found that the number of actin comets in the one-cell embryo is cell-cycle-dependent, peaking around prometaphase. Consistent with observations in reconstituted systems [31], NMY-2 appears to be dispensable for the formation and dynamics of comets in C. elegans. However partial depletion of ARX-1 (the Arp2 subunit of the Arp2/3 complex), CDC-42 (a GTPase that stimulates assembly of actin filaments nucleated by Arp2/3), or PFN-1 (which promotes polarized filament growth), led to a significant reduction in the peak number of actin comets. Similarly, in yeast, a temperature-sensitive mutation in the Arp2p subunit of the Arp2/3 complex – which is known to cause defects in endosome internalization but does not affect actin cables or trafficking of internalized endosomes [60] – led to a reduction in the number, but not the velocity, of retrograde actin patch movements away from the cell cortex [5]. Because we cannot completely deplete these proteins in the embryo by RNAi (more severe depletions lead to sterility and osmotic defects), it is possible that further depletion would lead to more severe effects. However, our results are consistent with the idea that Arp2/3 activity stimulates the nucleation of actin comets in C. elegans embryos. Studies performed in vitro have shown that profilins affect the rate of movement of actin tails but not their numbers [31, 32]. In contrast, we noticed no significant change in the velocity of comets in PFN-1 depleted embryos, but the number of comets was drastically reduced at all stages of the cell cycle. Further work will thus be needed to understand the mechanisms by which actin filament dynamics are controlled in these structures and how this translates into propulsive force.

We do not yet understand the biological role fulfilled by actin comets in C. elegans. While the nature of their cargo is still unknown, they display similar morphological and dynamic characteristics with Arp2/3-dependent comet tails associated with endocytic vesicles in other metazoan cells. It thus seems likely that actin comets in the C. elegans embryo represent the movement of some type of endocytic vesicles. Based on their numbers and behavior, they appear to be distinct from early endosomes recently described in the C. elegans embryo, which are much more numerous on the cortex and move internally directly toward centrosomes at a rate consistent with dynein-mediated transport along microtubules [34]. The observations that few comets appear to be present at any one time, and that their numbers are coordinated with the cell cycle, suggests a highly specialized function. For example, actin comets might mediate the regulated internalization of specific cell-surface factors [61]. The precise roles that actin comets play in the one-cell embryo and whether they have additional roles in other cellular and developmental processes remain open questions.

C. elegansstrains and GFP fusions

General methods for culturing and handling were carried out as previously described [62]. All experiments were carried out at 25°C unless otherwise indicated. The strains and genetic markers used were as follows: wild-type Bristol strain (N2) and nnIs [unc-119(+) pie-1 promoter::gfp::Dm-moesin^437–578(amino acids 437–578 of D. melanogaster Moesin)] (GFP::MOE) [38].

RNAi

RNAi was performed by injection or soaking [63]. Clones were obtained from the genome-wide feeding library [50], verified by restriction digests, and PCR amplified with T7 primer. In some cases, gene-specific primers containing partial T7 polymerase promoter sequences were first used to amplify a portion of each gene from genomic DNA, followed by a second round of PCR with full T7 primer. PCR products were used as templates to make dsRNA using the T7 Ribomax kit (Promega). Injections and soaking were then performed as described [63, 64]. Worms were analyzed 18–24 hours after injection except for cdc-42(RNAi) and arx-1(RNAi) embryos, which were imaged 30–36 hours after injection. We controlled for the efficacy and variability of RNAi by scoring for lethality and comparing phenotypes manifested in our treated embryos with published embryonic phenotypes of strong genetic alleles, where such information was available.

For phenotypic series of actin(RNAi), each batch of dsRNA was calibrated for reproducibility of phenotypes and the same batch of dsRNA was used to quantify the amount of actin protein in the gonad over time. For antibody staining, groups of 7–10 worms injected within 15 minute intervals were fixed and stained at 6 h, 7 h, 8 h, 9 h, 10 h, 11 h, or 12 h after injection, and fluorescence intensity measured by Openlab 3.1.7 software. Series of timed gonad fixations were repeated 3 times for each batch of dsRNA, with similar trends in the levels of actin protein detected over time (Figure 1A and 1B show data from one representative series). Class I phenotypes were observed as early as 6–8 hours after injection depending on the batch of dsRNA used. Subsequent classes occurred as follows: Class II (1 hour after Class I), Class III (1.5–2 hours after Class I) and Class IV (2–3 hours after Class I). We assume that the variability in timing was due to variations in concentration of dsRNA in different preparations (which were not quantified).

Live imaging and analysis of GFP fluorescence

For embryo observation, gravid worms were isolated in egg buffer and dissected on coverslips, which were then inverted in 2% or 4% agarose pads that were sealed with Vaseline. Observation and recordings for DIC and fixed fluorescent images were performed using a Leica DMLA microscope using a 100× (1.3 NA) objective with either an ORCA-ER (Hammamatsu C4742–95) digital camera. Fluorescent confocal images were obtained using the Spinning Disk Confocal system (Improvision) with an Electron Multiplier CCD camera (Hammamatsu C9100), a Ludl motorized XY stage, and an Improvision Piezo Focus Drive. The GFP::MOE strain was imaged at the cortex with a 63× (1.46 NA) objective by taking a 7-step (1 μm) Z-stack at the cortex every 10 s, at 75% laser power, 80 ms exposure, and a sensitivity of 169. Images were processed to a 3D-max projection using Volocity software and Adobe Photoshop 7.0 (Adobe Systems). Actin comet speed, length, and number were determined by taking an 8-step (1 μm) Z-stack through the top half of the embryo using the CARV confocal imager (BD Biosciences) or the Spinning Disk Confocal system (Improvision) and the Electron Multiplier CCD camera (Hamamatsu C9100); the acquisition system was controlled by OpenLab 3.1.7 or Volocity 4.1 (Improvision). Comets that stayed in one focal plane for 3 or more frames were used for measurement of speed and length. To quantify the number of comets present during different stages of the cell cycle, a 7 μm Z-stack with a 1 μm step was taken and the number of observed comets was recorded during five intervals of 3–4 minutes each between the following six timepoints: ~3.5 minutes before pseudocleavage, pseudocleavage, prometaphase, metaphase, completion of cytokinesis, and ~3.5 minutes after cytokinesis.

Immunocytochemistry

For fixation and visualization of the F-actin cytoskeleton in the early embryo we used the protocol described in [23]. Gonad dissections, fixation and immunohistochemistry were carried out as described [65] for 6 timepoints after injection with act-1 dsRNA. Antibodies and dilutions were as follows: 1:1000 rabbit anti-GFP primary antibody (Abcam) or 1:200 monoclonal mouse anti-Actin Clone C4 (MP Biomedicals) with 1:1000 fluorescein isothiocyanate (FITC)-conjugated secondary anti-rabbit or anti-mouse IgG antibodies (Jackson Immuno Research). Rhodamine-conjugated phalloidin (1 unit/200 μl; Molecular Probes) was included with the secondary antibody to label F-actin specifically. Slides were washed in PBS and mounted in Vectashield (Vector Labs Inc. #H-1000).