Cell division^*

The C. elegans embryo is a powerful model system for studying the mechanics of metazoan cell division. Its primary advantage is that the syncytial gonad makes it possible to use RNA interference (RNAi) to generate oocytes whose cytoplasm is reproducibly (>95%) depleted of targeted essential gene products. Introduction of dsRNA rapidly catalyzes the destruction of the corresponding mRNA in many different systems. However, depletion of pre-existing protein is generally a slow process that depends on the half-life of the targeted protein. In contrast, in the C. elegans gonad, the protein present when the dsRNA is introduced is depleted by the continual packaging of maternal cytoplasm into oocytes (Figure 1). Since depletion relies on the rate of embryo production instead of protein half-life, the kinetics tend to be similar for different targets. By 36-48 hours after introduction of the dsRNA, newly formed oocytes are typically >95% depleted of the target protein.

Several additional advantages contribute to the usefulness of the C. elegans embryo as a model system. Of particular importance is their rapid and highly stereotypical mitotic divisions; the time between the onset of DNA condensation and the completion of furrow ingression during cytokinesis is approximately 14 minutes. The invariant nature of the first few divisions also facilitates the development of methods to assess the consequences of molecular perturbations. Quantitative methods have already been developed to monitor pronuclear migration (Figure 4; see also Albertson, 1984; O'Connell, 2000), cortical flows (for examples see Cheeks et al., 2004; Hird and White, 1993; Munro et al., 2004), chromosome segregation and spindle elongation during anaphase (for examples see Cheeseman et al., 2004; Grill et al., 2001; Labbe et al., 2004), and the asymmetric positioning of the spindle within the embryo (for examples see Colombo et al., 2003; Labbe et al., 2004; Tsou et al., 2002). Assay development has been accelerated by the emergence of microparticle bombardment mediated transformation (Praitis et al., 2001) and the availability of vectors containing regulatory sequences that direct efficient germline expression (Strome et al., 2001), which together have led to the generation of a large number of strains expressing fluorescent fusion proteins in the early embryo. Analysis of the mechanical consequences of depleting essential cell division proteins is also facilitated by the relatively weak DNA damage (Brauchle et al., 2003) and spindle checkpoints (Encalada et al., 2004), which allow the embryo to proceed through the cell cycle despite dramatic defects in nuclear structure, spindle assembly, chromosome segregation and centrosome function.

Genetic and RNAi-based approaches have identified a large number of loci important for cell division. Mutants in proteins required for cell division have been uncovered in screens of collections of nonconditional maternal effect and temperature sensitive mutations that result in embryonic lethality (for some examples see Encalada et al., 2000; Golden et al., 2000; Gönczy et al., 1999; Kemphues et al., 1988; O'Connell et al., 1998). The ability to reproducibly deplete oocytes of target proteins by RNAi, which can be performed by feeding, soaking or injection of hermaphrodites (see Reverse genetics), has also led to a series of genome-wide screens that identified a set of ~2100 genes required for embryonic viability (Fernandez et al., 2005; Fraser et al., 2000; Gönczy et al., 2000; Kamath et al., 2003; Maeda et al., 2001; Piano et al., 2000; Rual et al., 2004; Simmer et al., 2003; Sõnnichsen et al., 2005). Filming of embryos depleted of each of these 2100 gene products using differential interference contrast (DIC) microscopy has defined a set of 660 genes whose inhibition results in detectable defects during the first two cell divisions (Gönczy et al., 2000; Piano et al., 2000; Sõnnichsen et al., 2005; Zipperlen et al., 2001). Roughly half of these genes are specifically required for cell division processes such as chromosome segregation or cytokinesis, whereas the other half contributes to cell maintenance, via roles in processes such as translation and mitochondrial function (Sõnnichsen et al., 2005). The embryonic lethality resulting from RNAi of some of the 1440 genes for which no DIC defect was observed may be due to cell division defects that remain undetected, either due to incomplete penetrance of the RNAi or failure to score the resulting defects by DIC (for example subtle defects in chromosome segregation are very difficult to detect using this assay). Alternatively, depletion of many of these gene products may cause embryonic lethality due to developmental defects that preclude hatching.

Due to its accessibility to RNAi-based molecular perturbation, the first embryonic division that ensues following fertilization has been the most intensively studied. In this section, we provide a brief timeline for the events between fertilization and the completion of the first cytokinesis (outlined schematically in Figure 2; for reviews see Cowan and Hyman, 2004; Pelletier et al., 2004; Schneider and Bowerman, 2003).

Prior to fertilization, C. elegans oocytes are arrested in meiotic prophase with nuclei containing two copies of the diploid genome packaged into recombined bivalent chromosomes. The two rounds of meiotic chromosome segregation that generate the haploid oocyte pronucleus are completed in the zygote after the oocytes are fertilized. During each meiotic division, chromosome segregation is accomplished by a small acentriolar meiotic spindle that forms in the embryo anterior. During anaphase of meiosis I and again in meiosis II, the meiotic spindle associates with the cortex in an end-on fashion, and a highly asymmetric cytokinesis-like event extrudes a polar body (Figure 2; Albertson and Thomson, 1993; Clark-Maguire and Mains, 1994; Yang et al., 2003). In addition to the haploid pronucleus, the sperm brings a pair of centrioles into the oocyte, which lacks centrioles due to their degradation during oogenesis. As meiosis completes, the haploid oocyte and sperm-derived pronuclei, located at opposite ends of the embryo increase in size, becoming visible by DIC microscopy. After entering the oocyte, the sperm-derived centriole pair recruits pericentriolar material and acquires the ability to nucleate microtubules (O'Connell, 2000; Pelletier et al., 2004). Subsequently, the two sperm-derived centrioles separate, forming two centrosomes positioned on either side of the paternal pronucleus. Coincident with chromosome condensation during mitotic prophase, the pronuclei migrate towards each other. After the pronuclei meet, the nuclear-centrosome complex moves to the center of the embryo and rotates to align with the long axis of the embryo (Albertson, 1984; Hyman and White, 1987). The miotitc spindle begins to move towards the embryo posterior during metaphase (Labbe et al., 2004; Oegema et al., 2001), and asymmetric elongation during anaphase contributes to its posterior displacement (Albertson, 1984; Grill et al., 2001). Since the cleavage furrow bisects the mitotic spindle, this displacement results in an asymmetric first cleavage (For more on the mechanisms that generate this asymmetry see Asymmetric cell division and axis formation in the embryo).

The C. elegans nuclear envelope is structurally similar to that of vertebrates, consisting of two concentric membranes (outer and inner) enclosing a lumenal space, nuclear pore complexes that mediate bidirectional transport between the cytoplasm and the nucleus, and an underlying laminar network (Figure 3; Cohen et al., 2002). The molecular composition of the C. elegans inner nuclear membrane/lamina also resembles that in vertebrates (Table 1). C. elegans expresses a single B-type lamin (LMN-1; Liu et al., 2000; Riemer et al., 1993), that forms a meshwork of intermediate filaments beneath the inner nuclear membrane (reviewed in Gruenbaum et al., 2005). The C. elegans inner nuclear membrane/lamina also contains three proteins, Ce-emerin, CeMAN-1, and CeLem2, that contain a LEM domain, a defining 40 amino acid motif shared by a family of nuclear envelope proteins (Gruenbaum et al., 2002; Lee et al., 2000; Lin et al., 2000; Liu et al., 2003). LEM family proteins all bind to lamins (reviewed in Lee and Wilson, 2004) and to the small inner nuclear membrane associated protein BAF (Segura-Totten and Wilson, 2004; Zheng et al., 2000). Depletion of LMN-1, Ce-BAF, or simultaneous depletion of the LEM family proteins Ce-Emerin and Ce-MAN-1, results in a similar spectrum of defects in nuclear structure, chromosome condensation, and chromosome segregation (Liu et al., 2000; Liu et al., 2003; Margalit et al., 2005; Zheng et al., 2000).

The dynamics of nuclear envelope disassembly and reassembly during the first mitotic division of the C. elegans embryo are illustrated in Figure 3. LMN-1 leaves the nuclear envelope during prometaphase (Lee et al., 2000; Liu et al., 2000). In contrast, inner nuclear membranes containing Ce-emerin and Ce-MAN-1 remain largely intact and surround the mitotic spindle everywhere except near spindle poles during metaphase and early anaphase, disassembling fully only during mid to late anaphase (Figure 3; Lee et al., 2000; Lee et al., 2002; V. Galy, P. Askjaer and I.W. Mattaj, personal communication). As the remnants of the old nuclear envelopes disperse, the formation of new nuclear envelopes around the segregated chromatin is detected beginning about 1 minute after anaphase onset (Figure 3; V. Galy, P. Askjaer and I.W. Mattaj, personal communication).

The molecular composition of the nuclear pore complexes (NPCs) is also similar to that in vertebrates. C. elegans orthologs of at least one component of each vertebrate NPC sub-complex have been identified (Galy et al., 2003; Kuznetsov et al., 2002). 17 genes encoding 19 nucleoporins are essential for embryonic viability. Depletion of 14 of these proteins results in defects in nuclear morphology and, in some cases, to reduced nuclear size consistent with a defect in nucleo-cytoplasmic transport (Galy et al., 2003). Nuclear envelope assembly also requires the small GTPase, Ran and the nuclear transport receptor importin-γ (IMB-1; Askjaer et al., 2002; Bamba et al., 2002; Walther et al., 2003). High concentrations of RanGTP in the vicinity of chromatin are thought to promote the dissociation of importin-γ from nucleoporins to trigger NPC assembly on the chromatin surface.

The site of sperm entry defines the embryo posterior (Goldstein and Hird, 1996). As the pronuclei become visible by DIC, the sperm-derived pronucleus and its associated centrosome(s) sit on the posterior cortex. The oocyte-derived pronucleus forms following two rounds of meiotic chromosome segregation, typically in the embryo anterior. The two pronuclei migrate towards each other coincident with chromosome condensation during the first mitotic prophase. Pronuclear migration consists of movement of the oocyte pronucleus towards the sperm pronucleus and movement of the sperm pronucleus away from the cortex towards the embryo center (Albertson, 1984; O'Connell et al., 2000). Initially, the oocyte pronucleus moves ~ 12 μm towards the posterior at a slow rate (~ 3.5 μm/min). As it approaches the sperm pronucleus, the oocyte pronucleus accelerates, moving an additional 10 μm at ~5-10 times its initial rate (Figure 4; Albertson, 1984; O'Connell et al., 2000). The sperm pronucleus begins its migration later than its female counterpart and travels at a slow rate of ~ 3.5 μm/min till it meets the oocyte pronucleus near the embryo center (~7 μm).

Rapid movement of the oocyte pronucleus towards the sperm pronucleus and pronuclear meeting, both require an intact microtubule cytoskeleton (Strome and Wood, 1983). Two nuclear envelope proteins, ZYG-12 and SUN-1 recruit dynein to pronuclei, and are required for centrosomes to maintain their association with nuclei (Table 1; Malone et al., 2003). The centrosomes separate around the sperm pronucleus in a dynein dependent manner (Gönczy et al., 1999). As the pronuclei move towards each other, dynein on the oocyte pronucleus is thought to come into contact with microtubules emanating from the centrosomal asters associated with the sperm pronucleus. The fast phase of pronuclear migration and pronuclear meeting is thus mediated by nuclear envelope associated dynein pulling on the two centrosomal microtubule asters (Cowan and Hyman, 2004; Gönczy et al., 1999; Hamill et al., 2002; Malone et al., 2003; O'Connell et al., 2000; Schmidt et al., 2005).

In addition to the migration of the female pronucleus towards the male pronucleus, the migration and centration of the male pronucleus within the embryo has also been analyzed to distinguish between two possible models: (1) a “pushing mechanism,” in which the male pronucleus is pushed away from the cortex by the polymerization of astral microtubules and, (2) a “pulling mechanism” in which the male pronucleus is pulled by minus-end-directed motors anchored throughout the cytoplasm. Comparisons between simulations and actual migration indicate that the second “pulling” model is the primary mechanism (Kimura and Onami, 2005). Although microtubule-based mechanisms for pronuclear movement are the best characterized, some slow pronuclear movement is still observed in embryos in which the microtubule cytoskeleton is compromised, or which lack centrosomes or fail to recruit dynein to nuclei (Cowan and Hyman, 2004; Gönczy et al., 1999; Hamill et al., 2002; Malone et al., 2003; O'Connell et al., 2000; Schmidt et al., 2005). Consistent with these observations, correlative evidence suggests that cortical flows may also contribute to the slow phase of pronuclear migration (Hird and White, 1993).

Centrosomes consist of a single centriole or centriole pair surrounded by pericentriolar material that nucleates and anchors microtubules. Like the centrioles in Drosophila embryos (Callaini and Riparbelli, 1990), C. elegans centrioles are composed of nine singlet microtubules symmetrically positioned around a central tube (Figure 5B; Albertson, 1984; Kirkham et al., 2003; O'Connell et al., 2001; Wolf et al., 1978). Each cylindrical centriole is approximately 200– 250 nm in length and 175 nm in diameter. In contrast, vertebrate centrioles typically have nine triplet microtubules (Marshall, 2001; Preble et al., 2000). Consistent with this structural difference, the C. elegans genome lacks homologs of delta- and epsilon-tubulin, two tubulin family members required for the formation of triplet microtubules (Dutcher, 2003).

During fertilization, the amoeboid sperm brings a pair of centrioles into the egg. After entering the egg, the centrioles recruit pericentriolar material and separate around the sperm-derived nucleus, forming two centrosomal microtubule asters (Figure 5; O'Connell, 2000; Pelletier et al., 2004). After they separate, a new daughter centriole begins to form adjacent to each sperm centriole. The centrosomes recruit additional pericentriolar material as the embryo enters mitosis in a process called centrosome maturation, increasing about 5 fold in size and nucleating capacity by metaphase (Hannak et al., 2001). Coincident with the recruitment of additional PCM, the daughter centrioles elongate, reaching full length by metaphase. In late anaphase/telophase the centriole pairs split. Concurrent with cytokinesis, the mitotic PCM disassembles, releasing two small centrosomes into each daughter cell.

Pericentriolar material is thought to consist of a proteinacious matrix, called the “centromatrix”, that recruits other PCM components (Bornens, 2002; Palazzo et al., 2000). Three centrosomal proteins, SPD-5, SPD-2 and AIR-1 (Aurora-A kinase), are required for the assembly of pericentriolar material (see Table 2 for a list of C. elegans centrosome/centriole components). SPD-5 is a coiled-coil protein thought to be a component of the centromatrix. In spd-5 embryos, no PCM forms around the centrioles, and no centrosomal microtubule asters are observed (Hamill et al., 2002). Severe defects in PCM assembly are also observed when SPD-2 or the aurora A kinase homolog AIR-1 are inhibited (Hannak et al., 2001; Kemp et al., 2004; O'Connell et al., 2000; Pelletier et al., 2004; Schumacher et al., 1998). In these embryos, centrosomal microtubule asters form, but are very small, and additional PCM fails to accumulate around the centrioles during mitotic entry.

Several PCM proteins are required for the activity rather than the assembly of the PCM. The microtubule nucleating activity of the pericentriolar material requires the centrosomal tubulin isoform gamma-tubulin, and two gamma-tubulin associated proteins, CeGrip-1 and CeGrip-2. Embryos depleted of any of these proteins fail to form centrosomal microtubule asters during interphase (Hannak et al., 2002). As depleted embryos enter mitosis, relatively robust microtubule asters form around the centrosomes, suggesting that a gamma-tubulin independent pathway contributes to their assembly (Strome et al., 2001). However, chill and rewarm experiments reveal that the rate of centrosomal microtubule nucleation is highly compromised in gamma-tubulin depleted embryos and spindle assembly fails (Hannak et al., 2002; Strome et al., 2001). ZYG-9 and TAC-1 are two conserved PCM proteins that associate to form a complex that promotes microtubule growth. Embryos depleted of either protein exhibit defects in pronuclear migration, presumably because centrosomal microtubules are too short to capture the oocyte pronucleus, and small spindles form around the sperm chromatin in the embryo posterior (Bellanger and Gönczy, 2003; Le Bot et al., 2003; Matthews et al., 1998; Srayko et al., 2003).

Seven C. elegans proteins have been shown to contribute to new centriole assembly. Four of these, SAS-4 (Kirkham et al., 2003; Leidel and Gönczy, 2003), SAS-5 (Dammermann et al., 2004; Delattre et al., 2004; Schmutz and Spang, 2005), SAS-6 (Dammermann et al., 2004; Leidel et al., 2005), and the atypical protein kinase ZYG-1 (O'Connell et al., 2001), localize to centrioles and are specifically required for centriole formation. Depletion of any of these proteins by RNAi results in specific failure of centrosome duplication. In this characteristic phenotype (O'Connell et al., 2001), the two sperm centrioles separate after fertilization (as in wild-type) and organize the two centrosomes that form the poles of an apparently normal spindle during the first mitotic division. However, since daughter centrioles fail to form adjacent to each of the sperm centrioles, only one centriole instead of the normal two are released into each daughter cell when the centrosomes break down in telophase. Since each daughter cell contains only one centriole, it can form only one centrosome, and monopolar spindles are observed in both cells at the two-cell stage. SPD-2 is a bi-functional protein that localizes to centrioles as well as the PCM and is required for centriole assembly in addition to its role in PCM recruitment (Kemp et al., 2004; Pelletier et al., 2004). Embryos depleted of two pericentriolar material proteins, SPD-5 and gamma-tubulin, also exhibit severe defects in centriole assembly, leading to the idea that the PCM contributes to centriole assembly by recruiting gamma-tubulin, which may promote the assembly of the microtubules that make up the centriolar cylinders, as well as those nucleated by the PCM (Dammermann et al., 2004).

The formation of mitotic chromosomes begins when cohesin is loaded onto chromosomes and establishes cohesion between the duplicated chromosomes (sister chromatids) during DNA replication in S phase. Proteolytic cleavage of one of the cohesin subunits later in the cell cycle is thought to release the linkage between the sister chromatids to allow their segregation during anaphase (Haering and Nasmyth, 2003). The components of the C. elegans cohesin complex have been identified and depletion has revealed roles in mitotic chromosome segregation and the pairing of homologous chromosomes during meiosis (Table 3; Chan et al., 2003; Mito et al., 2003; Pasierbek et al., 2003).

Two additional protein complexes, condensin and the chromosomal passenger complex, also have critical roles in the formation and segregation of mitotic chromosomes. In contrast to vertebrates, which have two complexes, condensins I and II, that mediate spatially distinct aspects of condensation (Hirano, 2004), C. elegans has only condensin II (Table 3, Hagstrom et al., 2002; Ono et al., 2003). Chromosome condensation is delayed in condensin depleted embryos, but DNA compaction is ultimately achieved, indicating the existence of condensin-independent mechanisms that can compact mitotic chromatin. However, sister chromatid pairs formed in the absence of condensin are structurally defective and cannot be efficiently separated from each other by the mitotic spindle (Hagstrom et al., 2002; Kaitna et al., 2002; Chan et al., 2004). Concurrent with, but largely independent of condensation, kinetochores assemble to create chromosomal attachment sites for spindle microtubules (described in greater detail below). Like condensin, the chromosomal passenger protein complex (including the aurora B kinase, AIR-2, BIR-1, ICP-1 and CSC-1) is recruited to mitotic chromosomes as they form and is required for their proper segregation, (Schumacher et al., 1998; Severson et al., 2000; Kaitna, 2000; Oegema, 2001; Kaitna et al., 2002; Rogers 2002; Romano 2004; Hsu et al., 2000). Although it does not appear to be required for kinetochore assembly in C. elegans, the passenger complex plays important roles in chromosome structure and likely functions to correct aberrant kinetochore-microtubule attachments (reviewed in Vagnarelli and Earnshaw, 2004).

Eukaryotes can be divided into two groups based on the architecture of their mitotic chromosomes. Monocentric organisms assemble kinetochores on a single localized chromosomal site defined by the presence of dedicated centromeric chromatin. In contrast, holocentric organisms, including C. elegans, assemble diffuse kinetochores along the entire poleward face of each sister chromatid (Figure 6A). Holocentric chromosome architecture is present in widely divergent and highly successful metazoan lineages (including nematodes, hemipteran insects, and lower plants) that constitute a large part of the earth’s biomass, and C. elegans has emerged as an important model system for studying holocentric chromosome architecture (reviewed in Dernburg, 2001; Maddox et al., 2004).

Despite differences in the extent of the chromosomal length occupied by the kinetochore, the molecular composition of kinetochores in C. elegans and monocentric organisms is very similar (Table 4). Importantly, mitotic kinetochores in both monocentric and holocentric organisms assemble on a base of specialized centromeric chromatin defined by the presence of nucleosomes containing the histone H3 variant CENP-A (Buchwitz et al., 1999; Sullivan, 2001). Depletion of the C. elegans homolog of CENP-A leads to a characteristic “kinetochore-null” phenotype in which chromosomes fail to distribute over the spindle equator and to segregate (Oegema et al., 2001). This defect results from the failure to assemble kinetochores that can interact with spindle microtubules. To date only 4 proteins have been identified whose depletion gives a kinetochore-null defect: CENP-A^HCP-3, CENP-C^HCP-4, KNL-3 and KNL-1, which can be placed in a linear assembly hierarchy (Figure 6B) with CENP-A^HCP-3 at the top (Moore and Roth, 2001; Oegema et al., 2001; Desai et al., 2003; Cheeseman et al., 2004). KNL-3 and KNL-1 are components of a 10-protein complex that is critical to forming the outer domains of the kinetochore that interact with spindle microtubules (Figure 6B; Cheeseman et al., 2004; Desai et al., 2003). A number of other kinetochore proteins have been identified whose depletion results in less severe chromosome segregation defects (for examples see Howe et al., 2001; Moore et al., 1999; for a detailed summary see Table 4). The reproducible phenotypes obtained following protein depletion have been useful in grouping proteins that function together in the context of the kinetochore (Figure 6B; Cheeseman et al., 2004).