ZYG-9, TAC-1 and ZYG-8 together ensure correct microtubule function throughout the cell cycle of C. elegans embryos

Regulation of microtubule behavior is crucial for correct cell division. Microtubule dynamics is governed by the intrinsic properties of the polymer and microtubule-associated proteins (MAPs). Whereas much is known about the mechanisms by which MAPs influence microtubule dynamics in vitro, comparatively little is known about how they function in vivo.

The early embryo of Caenorhabditis elegans is an attractive model system to investigate microtubule-dependent processes within live cells, owing notably to the exquisite temporal and spatial resolution that can be achieved with time-lapse differential interference contrast (DIC) microscopy (reviewed in Oegema and Hyman, 2005). Distinct microtubule-dependent processes occur during the first cell cycle, including migration of the male and female pronuclei during interphase, as well as spindle assembly during mitosis. Importantly, forward genetic and functional genomic studies have led to the systematic identification of genes required for correct microtubule-dependent processes, including genes encoding MAPs.

Correct growth of microtubules in eukaryotic cells generally relies on a XMAP215-related MAP (reviewed in Kinoshita et al., 2002). Originally identified as a growth-promoting factor in Xenopus egg extracts (Gard and Kirschner, 1987), XMAP215 and related proteins are thought to favor polymerization of microtubules, although XMAP215 itself and its S. cerevisiae homologue Stu2p can also promote microtubule depolymerization (Shirasu-Hiza et al., 2003; van Breugel et al., 2003). In C. elegans embryos, the XMAP215 homologue ZYG-9 is present in the cytoplasm and enriched at centrosomes throughout the cell cycle (Matthews et al., 1998). Embryos depleted of ZYG-9 exhibit slower rates of microtubule growth and, consequently, have shorter astral microtubules (Srayko et al., 2005). As a result, several microtubule-dependent processes fail in the one-cell stage (Kemphues et al., 1986). For instance, male and female pronuclear migration are defective during interphase and, as a consequence, the spindle assembles in an aberrant position during mitosis. Thus, ZYG-9 is required for microtubule growth throughout the cell cycle but the consequences of inactivating it solely during mitosis have not been addressed.

In several organisms, XMAP215 family members associate with a protein of the transforming and acidic coiled-coil (TACC) family (reviewed in Gergely, 2002). In Drosophila, TACC is required for the efficient recruitment of the XMAP215 protein Msps to centrosomes (Cullen and Ohkura, 2001; Lee et al., 2001). Similarly, TACC3 helps recruit the XMAP215 protein ch-TOG to spindle microtubules in human cells (Gergely et al., 2003). In C. elegans, the only TACC family member TAC-1 localizes in the cytoplasm and is enriched at centrosomes, like ZYG-9 (Bellanger and Gönczy, 2003; Le Bot et al., 2003; Srayko et al., 2003). Moreover, ZYG-9 and TAC-1 form a complex and stabilize each other in vivo: in either zyg-9(RNAi) or tac-1(RNAi) embryos, both ZYG-9 and TAC-1 protein levels are severely diminished (Bellanger and Gönczy, 2003). Therefore, in the absence of mutant alleles to analyze the contribution of each protein separately, it is not clear whether either ZYG-9 or TAC-1 is limiting for promoting microtubule growth in vivo.

Another MAP acting in the one-cell stage C. elegans embryo is ZYG-8, which harbors a doublecortin domain and a kinase domain (Gönczy et al., 2001). ZYG-8 is enriched at microtubule asters throughout the cell cycle and is present on the spindle during mitosis. The evolutionarily conserved doublecortin domain binds microtubules and promotes their stability or polymerization (reviewed in Horesh et al., 1999). In humans, the doublecortin gene DCX is mutated in patients with a neuronal migration disorder thought to result from defective microtubule-dependent processes (des Portes et al., 1998; Gleeson et al., 1998). Furthermore, the mammalian orthologue of ZYG-8, known as DCLK (doublecortin-like kinase), regulates spindle assembly and ensures the correct fate of cortical neural progenitors (Shu et al., 2006). In C. elegans, zyg-8-mutant embryos undergo a sequence of events comparable with that in the wild type during interphase. During mitosis, however, they exhibit a defect in anaphase spindle positioning that results in aberrant cell division (Gönczy et al., 2001). An analogous phenotype is observed after treatment of wild-type embryos with nocodazole during mitosis (Gönczy et al., 2001), indicating that ZYG-8 acts during that stage of the cell cycle to promote microtubule growth, a conclusion supported by measurements of growth rates in metaphase zyg-8(RNAi) embryos (Srayko et al., 2005). However, whether zyg-8 also acts during interphase is not apparent from this analysis, nor that of null-mutant alleles (Gönczy et al., 2001). The absence of a requirement during interphase might reflect the fact that ZYG-8 acts strictly during mitosis. Alternatively, ZYG-8 might act throughout the cell cycle but become essential only during mitosis when the requirement for growth promoting factors may be particularly high.

Here, we report the identification of new zyg-9 and tac-1 alleles that allowed us to address several questions regarding the respective contribution of ZYG-9 and TAC-1 to microtubule-dependent processes in vivo. Moreover, we were able to achieve rapid conditional inactivation of the ZYG-9-TAC-1 complex and, thus, test its requirement specifically during mitosis. In addition, using synthetic gene inactivation, we demonstrate that ZYG-9-TAC-1 and ZYG-8 act in a partially redundant manner throughout the cell cycle of one-cell stage C. elegans embryos.

We identified eight conditional-recessive parental-effect embryonic-lethal mutant strains with phenotypes at the restrictive temperature that are indistinguishable from those observed when zyg-9 or tac-1 is inactivated (Materials and Methods; Table 1). We genetically mapped the mutation in five mutant strains to the center of chromosome II, where zyg-9 is located, and in three mutant strains to the right arm of chromosome II, where tac-1 is located (Materials and Methods). The five strains of the first group fail to complement each other and zyg-9(b244), demonstrating that they correspond to zyg-9-mutant alleles. The three strains of the second group also fail to complement each other.

We conducted western blot analysis of embryonic extracts from each mutant strain to estimate the overall levels of TAC-1 or ZYG-9 protein (see supplementary material Fig. S1). These extracts were prepared from embryos at all developmental stages and, thus, might only partially reflect the impact of the mutations on TAC-1 and ZYG-9 in the early embryo. Therefore, we systematically examined the distribution of ZYG-9 and TAC-1 in early embryos by immunofluorescence analysis in the eight mutant strains at the restrictive temperature. Representative specimens are shown in Figs 1 and 2, which also report quantification of the cytoplasmic and centrosomal protein levels observed in each case. In addition, we performed sequence analysis of the eight mutant strains.

Sequencing of the zyg-9 gene in the first complementation group revealed the following point mutations (Fig. 1A and supplementary material Fig. S2A): D189G in zyg-9(or623), G203E in zyg-9(or628), G488E in zyg-9(or593), I862K in zyg-9(or635) and E1317K in zyg-9(or634). The mutations in zyg-9(or623), zyg-9(or628), zyg-9(or593) and zyg-9(or635) lie in three so-called TOG domains that are found in all members of the XMAP215 family and are likely to mediate tubulin binding (supplementary material Fig. S2A) (Al-Bassam et al., 2007). TOG domains form paddle-like structures composed of six HEAT (huntingtin, elongation factor 3, PP2A regulatory subunit, TOR1) repeats (HR1 to HR6) (Al-Bassam et al., 2007). HEAT repeats form rod-like helical structures and mediate protein-protein interactions (Andrade and Bork, 1995), suggesting that these four zyg-9 alleles encode proteins defective in binding tubulin and/or another partner protein. We found that centrosomal ZYG-9 is decreased in two of these alleles, zyg-9(or628) and zyg-9(or593), despite cytoplasmic levels being comparable to those of the wild type (Fig. 1F,1H). It is tempting to speculate that the alteration in the TOG domains 1 and 2 in these two alleles compromises interaction with a centrosomal anchor. Although the nature of such a putative TOG-domain-interacting centrosomal anchor remains to be determined, it is unlikely to be TAC-1 because a fragment of ZYG-9 lacking the two first TOG domains can still interact with TAC-1 (Bellanger and Gönczy, 2003).

The I862K mutation in zyg-9(or635) changes an amino acid that is invariant and located within a conserved stretch of amino-acids in all metazoan XMAP215 family members (see supplementary material Fig. S2A). We found that, in this allele, ZYG-9 is also significantly decreased at centrosomes and, to a lesser extent, in the cytoplasm (Fig. 1J). As I862 is also part of a weak TAC-1-binding region (TBR, Fig. 1A) (Bellanger and Gönczy, 2003), it is possible that the I862K mutation alters formation of the TAC-1-ZYG-9 complex. To test this hypothesis, we performed an in vitro binding assay using the TBR of ZYG-9 I862K and GST-TAC-1. As shown in Fig. 3, we found that the I862K mutation causes a sizeable reduction in the ability of the TBR of ZYG-9 to bind TAC-1. Accordingly, we observed a slight reduction of TAC-1 levels in zyg-9(or635) early embryos (Fig. 1K), as anticipated if the interaction between TAC-1 and ZYG-9 is less efficient in vivo.

Although the E1317K mutation in zyg-9(or634) represents a drastic electrostatic change, this region of the protein is quite divergent among XMAP215 family members; an arginine occupies this position in vertebrates, demonstrating that a positively charged amino-acid can be accommodated at this site (see supplementary material Fig. S2A). Nonetheless, in C. elegans, this mutation results in a severe decrease of ZYG-9 levels, both at centrosomes and in the cytoplasm (Fig. 1L). E1317 is located within a domain that undergoes robust interaction with TAC-1 (Bellanger and Gönczy, 2003), raising the possibility that the E1317K mutation disrupts TAC-1 binding. Accordingly, we found that the E1317K mutation causes a significant reduction in the binding of the ZYG-9 TBR to TAC-1 (Fig. 3). Compatible with this view, TAC-1 levels are also markedly decreased in zyg-9(or634) embryos, both at centrosomes and in the cytoplasm (Fig. 2M). Overall, the analysis of these newly identified zyg-9 alleles uncovers new centrosomal targeting and protein-stabilizing functions for the ZYG-9 TOG domains, as well as for the two residues I862 and E1317 that are also crucial for efficient binding to TAC-1.

Sequencing the tac-1 gene in the second complementation group revealed the following point mutations (Fig. 2A and supplementary material Fig. S2B): L229F in both tac-1(or369) and tac-1(or402) (hence considered collectively as tac-1(or369) in the entire manuscript) and M58I in tac-1(or455). These strains identify the first mutant alleles of tac-1. The residues mutated in these alleles are not invariant across evolution. Nevertheless, they are part of the TACC domain (Fig. 2A and supplementary material Fig. S2B), whose function is likely to be hampered by the mutations. We found that centrosomal TAC-1 is diminished approximately twofold in tac-1(or369) embryos and approximately threefold in tac-1(or455) embryos compared with the wild type (Fig. 2E,G). By contrast, cytoplasmic TAC-1 levels in tac-1(or369) embryos are comparable to those in the wild type and slightly diminished in tac-1(or455) embryos (Fig. 2E,G). We addressed whether these two mutations impair the interaction between TAC-1 and ZYG-9. We found that, whereas the binding of GST-TAC-1 M58I to the ZYG-9 TBR is not impaired, that of GST-TAC-1 L229F is severely diminished compared with the wild type (Fig. 3). Overall, we conclude that the mutations M58I and L229F both alter the ability of TAC-1 to be targeted or retained at centrosomes. Furthermore, we conclude that, M58I has an additional, slightly deleterious effect on protein stability and L229 is crucial for ZYG-9 binding.

Inactivation of either ZYG-9 or TAC-1 using RNA interference (RNAi) compromises the entire ZYG-9-TAC-1 complex, a fact that has precluded an analysis of the relative contribution of each protein separately. For instance, it is not known whether each protein can be present at centrosomes without the other, nor whether either ZYG-9 or TAC-1 is limiting for microtubule-dependent processes in one-cell-stage embryos. We reasoned that an examination of ZYG-9 and TAC-1 distribution in a selection of our newly identified mutant alleles might shed light on these questions.

First, centrosomal TAC-1 is not reduced in zyg-9(or628) embryos, despite centrosomal ZYG-9 being approximately twofold lower than in the wild type (Fig. 1G,F). Therefore, centrosomal ZYG-9 is not limiting for the presence of TAC-1 at centrosomes. Second, despite being present at relatively normal levels in the cytoplasm, centrosomal ZYG-9 is severely diminished in tac-1(or369) embryos (Fig. 2D), in which centrosomal but not cytoplasmic TAC-1 is diminished (Fig. 2E). This indicates that centrosomal TAC-1 is limiting for ZYG-9 enrichment at centrosomes. Since the interaction between TAC-1 L229F encoded by tac-1(or369) and the ZYG-9 TBR is significantly reduced in vitro (see Fig. 3), it is tempting to speculate that TAC-1 recruits ZYG-9 to centrosomes in wild-type embryos. Third, cytoplasmic levels of ZYG-9 in tac-1(or369) embryos are comparable to those in wild-type embryos, even though centrosomal ZYG-9 is diminished (Fig. 2D). Since these mutant embryos exhibit a canonical zyg-9-tac-1 phenotype, we conclude that normal levels of ZYG-9 in the cytoplasm are not sufficient for correct microtubule-dependent processes. An analysis of ZYG-9 distribution in embryos, in which tac-1 has been inactivated only partially by RNAi indicated that normal levels of ZYG-9 at centrosomes are not sufficient either (Bellanger and Gönczy, 2003). Together, these findings establish that ZYG-9 is limiting both in the cytoplasm and at centrosomes in one-cell-stage embryos. Fourth, we found that tac-1(or455) embryos maintained at the permissive temperature exhibit a significant diminution of TAC-1 at centrosomes and to a lesser extent in the cytoplasm (Fig. 2I), but no detectable change in ZYG-9 distribution (Fig. 2H). These embryos undergo correct microtubule-dependent processes and are fully viable (Table 1). Therefore, in contrast to ZYG-9, TAC-1 is not limiting for microtubule-dependent processes in one-cell-stage embryos. Furthermore, TAC-1 levels and distribution in zyg-9(or628) embryos are comparable to those in wild type (Fig. 1G), indicating that TAC-1 is not sufficient either for these processes.

We next decided to take advantage of the conditional nature of these alleles to specifically investigate the function of ZYG-9-TAC-1 during mitosis. The requirement of ZYG-9-TAC-1 during interphase has prevented an analysis of the specific requirement after this stage using RNAi or other non-conditional methods of gene inactivation. We reasoned that the new temperature-sensitive alleles might provide a means of compromising ZYG-9-TAC-1 selectively during mitosis.

Using time-lapse DIC microscopy, examination of embryos from the new mutant alleles raised at ∼15°C and ∼26°C indicated that short exposure to the restrictive temperature results in the canonical zyg-9-tac-1 phenotype in the case of zyg-9(or634) and tac-1(or455) (data not shown). We used zyg-9(or634) to inactivate zyg-9 function during mitosis (Fig. 4C). Animals were raised, dissected and initially observed on the microscope stage at ∼15°C (Materials and Methods). Towards the end of interphase, when the two joined pronuclei moved to the cell center, the temperature was shifted to ∼26°C and the observation continued. We found that zyg-9(or634) embryos treated in this manner assemble a spindle in the cell center and aligned along the anterior-posterior (A-P) axis - like in the wild type (compare Fig. 4C1 with Fig. 4A1). Strikingly, however, the spindle is displaced in an exaggerated manner towards the lateral and posterior cortex during anaphase (compare Fig. 4C2 with 4A2). These phenotypes are comparable to those observed in embryos with compromised zyg-8 function or after nocodazole treatment during mitosis (Fig. 4B) (Gönczy et al., 2001). As shown in Fig. 4D, we observed a similar sequence of events towards the end of interphase when tac-1(or455) embryos were shifted to ∼26°C. As anticipated from the phenotypic similarity, we found that the microtubule network in temperature-shifted zyg-9(or634) and tac-1(or455) embryos resembled that of embryos with compromised zyg-8 function (supplementary material Fig. S3). We conclude that ZYG-9-TAC-1 is required during mitosis for correct anaphase spindle positioning in one-cell-stage C. elegans embryos, something that could not be tested prior to this work due to the earlier requirements of ZYG-9-TAC-1 during the cell cycle.

Our analysis indicates that ZYG-9-TAC-1 functions during mitosis, simultaneously with ZYG-8. Therefore, we set out to investigate whether these two microtubule-growth-promoting factors genetically interact during mitosis, as well as other stages of the cell cycle. To this end, we sensitized one-cell-stage embryos to the lack of growth-promoting factors throughout the cell cycle by compromising zyg-9-tac-1 function, and addressed whether the additional inactivation of zyg-8 results in synthetic phenotypes (Fig. 5, Table 2).

In wild-type embryos (Fig. 5A), the female pronucleus forms in the presumptive anterior. The male pronucleus, together with the two centrosomes, are in close proximity of the presumptive posterior cortex (Fig. 5A1). Thereafter, the female pronucleus moves towards the centrosomes, whereas the centrosomes and associated male pronucleus move away more slowly from the posterior cortex. After the two pronuclei have met in the posterior half of the embryo, centration and rotation of the two pronuclei and associated centrosomes ensues, which results in them moving ∼20 μm away from the posterior cortex; the spindle thus sets up in the cell center and along the A-P axis (Fig. 5A2). During anaphase, the spindle elongates asymmetrically and becomes positioned slightly towards the posterior (Fig. 5A3,A5), resulting in an unequal cleavage (Fig. 5A4).

In tac-1(RNAi) embryos (Fig. 5B), there is sometimes more than one female pronucleus, indicative of a partially penetrant defect during the female meiotic divisions (Table 2) (Le Bot et al., 2003; Srayko et al., 2003). Apart from this, the sequence of events does not differ from that in wild type until the onset of pronuclear migration (Fig. 5B1). Thereafter, the male pronucleus and associated centrosomes move away slightly (on average 2.9 μm) from the posterior cortex, but further movement of male or female pronucleus does not occur. Moreover, centration and rotation of centrosomes and pronuclei do not take place. As a result, the spindle assembles in the cell posterior and elongates slightly, orthogonal to the A-P axis (Fig. 5B2,B3,B5). A cleavage furrow ingresses from the posterior cortex, bisecting the orthogonally placed spindle (Fig. 5B4).

The phenotype of zyg-8(or484)-mutant embryos (Fig. 5C) is analogous to that of null alleles of zyg-8, such as zyg-8(t1650) (Gönczy et al., 2001). Apart from some embryos having more than one female pronucleus (Table 2) (Gönczy et al., 2001), the sequence of events does not differ significantly from that in wild type prior to mitosis (Fig. 5C1,C2). During anaphase, however, the spindle is displaced in an exaggerated manner towards the lateral and posterior cortex, where it often reorientates itself orthogonally to the A-P axis (Fig. 5C3,C5). Thereafter, a cleavage furrow, usually ingressing from the posterior cortex, bisects the spindle (Fig. 5C4).

Strikingly, we found that inactivating tac-1 using RNAi in a zyg-8(or484) mutant results in markedly more severe cell-division defects than those observed in either tac-1(RNAi) embryos or zyg-8(or484) embryos (Fig. 5D). First, 82% of doubly compromised embryos exhibit more than one female pronucleus (Fig. 5D, left), compared with 8% in tac-1(RNAi) embryos and 26% in zyg-8(or484)-mutant embryos (Table 2). Second, the male pronucleus and associated centrosomes in doubly compromised embryos barely move away (on average 1.7 μm) from the posterior cortex during interphase (Fig. 5D1). Third, the spindle is shorter to start with and does not elongate substantially during mitosis (Fig. 5D2,D3,D5); probably, as a consequence, the cleavage furrow invariably fails to ingress in doubly compromised embryos (Fig. 5D4). Analogous synthetic phenotypes were observed by performing tac-1(RNAi) in a zyg-8(t1650) mutant or zyg-9(RNAi) in a zyg-8(or484) mutant (data not shown).

We also addressed whether increasing the levels of TAC-1 can compensate for a diminution of zyg-8 function. Accordingly, we found that the presence of transgenic GFP-TAC-1 (Bellanger and Gönczy, 2003), in addition to that of endogenous TAC-1, improves the viability of zyg-8(or484)-mutant embryos raised at a semi-permissive temperature (20.0% in zyg-8(or484) GFP-TAC-1, n=370; 8.6% in zyg-8(or484), n=236; Materials and Methods). Overall, these findings establish that ZYG-9-TAC-1 and ZYG-8 are required, in a partially redundant manner, for microtubule-dependent processes during meiosis, as well as during interphase and mitosis of one-cell-stage embryos.

In an attempt to understand better the relationship between ZYG-9-TAC-1 and ZYG-8 at a molecular level, we investigated the physical interactions between these three proteins. We initially identified TAC-1 during a yeast two-hybrid screen with ZYG-8 as a bait (Bellanger and Gönczy, 2003). Previous work focused on the analysis of ZYG-9 and TAC-1 (Bellanger and Gönczy, 2003; Le Bot et al., 2003; Srayko et al., 2003), and the interaction between ZYG-8 and TAC-1 has not been investigated to date. Because TAC-1 and ZYG-9 stabilize each other, we first addressed whether the presence of TAC-1 is similarly needed for the stability of ZYG-8, and vice versa. Western-blot analysis indicates that the levels of ZYG-8 are not significantly reduced in tac-1(RNAi) embryonic extracts (see supplementary material Fig. S4A), nor are they reduced in the eight new tac-1- and zyg-9-mutant strains (data not shown). Reciprocally, the levels of TAC-1 are not diminished when zyg-8 is inactivated (see supplementary material Fig. S4A). Furthermore, the subcellular distribution of ZYG-8 is not changed in tac-1(or455) (supplementary material Fig. S4C, compare with S4B) and that of TAC-1 is not altered in zyg-8-mutant embryos (supplementary material Fig. S4E, compare with S4D). Thus, in contrast to TAC-1 and ZYG-9, TAC-1 and ZYG-8 do not appear to be mutually dependent for achieving correct protein levels or subcellular distribution.

We next tested whether the interaction observed between ZYG-8 and TAC-1 by using the yeast two-hybrid assay reflects a direct association between the two proteins. As shown in Fig. 6A, we found that in vitro translated full-length ZYG-8 binds TAC-1 in a pull-down assay. Fragments of ZYG-8 that lack the doublecortin domain either fail to interact in a detectable manner (fragment 1-207) or only interact weakly (fragment 477-802) with TAC-1 (Fig. 6B). Conversely, a fragment comprising amino acids 206-423 that contains the doublecortin domain, mediates robust interaction with TAC-1 (Fig. 6B). Together, these findings indicate that the doublecortin domain of ZYG-8 is necessary and sufficient to directly interact with TAC-1. By analyzing the protein encoded by a mutant allele of zyg-8, we obtained evidence that the ability of the doublecortin domain of ZYG-8 to interact with TAC-1 is separate from its ability to bind microtubules. The G219E mutation in zyg-8(or484) alters a residue in the first β-strand of the doublecortin domain that is invariant in all proteins that share this domain (Gönczy et al., 2001; Kim et al., 2003). We found that, in contrast to wild-type ZYG-8 (see supplementary material Fig. S5A) (Gönczy et al., 2001), ZYG-8 G219E does not colocalize with microtubules in COS7 cells (see supplementary material Fig. S5B). Nevertheless, the doublecortin domain of this variant of ZYG-8 interacts with TAC-1 in yeast two-hybrid (see supplementary material Fig. S5C), indicating that association with TAC-1 does not require binding to microtubules.

We next sought to address whether ZYG-8 associates with TAC-1 in vivo by performing reciprocal co-immunoprecipitation experiments using C. elegans embryonic extracts. As shown in Fig. 6C, we found that TAC-1 antibodies co-immunoprecipitate trace amounts of ZYG-8. By contrast, ZYG-9 antibodies do not co-immunoprecipitate detectable amounts of ZYG-8. Together, these findings indicate that, although most of TAC-1 is associated with ZYG-9 (Bellanger and Gönczy, 2003) a minor fraction of TAC-1 is associated with ZYG-8, perhaps because their interaction is restricted in time or space. Furthermore, these findings indicate that TAC-1 and ZYG-8 can form a complex in vivo that does not contain ZYG-9.

Microtubule-dependent processes are crucial for cell division and are notably modulated by MAPs, which can act in a spatially and temporally restricted manner. In this study, we investigated in C. elegans embryos the contribution of the XMAP215 homologue ZYG-9 and of its associated partner TAC-1, as well as that of the doublecortin-domain-containing MAP ZYG-8.

We report the identification of five new conditional alleles of ZYG-9, as well as the first three mutant alleles of TAC-1 in C. elegans, which also represent the first temperature-sensitive mutant alleles of a TACC protein in any organism. The positions of the mutations in these newly identified alleles underscore the importance of protein-protein interactions for correct ZYG-9 and TAC-1 function. Indeed, the seven distinct molecular lesions all target domains that mediate protein-protein interactions: in the case of zyg-9 alleles, TOG domains or domains interacting with TAC-1 and, in the case of tac-1 alleles, the TACC domain. That the TACC domain is mutated in the three tac-1 alleles might not be significant, given that this domain accounts for ∼80% of the entire protein. By contrast, we noticed that three out of five zyg-9 mutations lie in the fifth HEAT repeat of TOG domains 1 or 2, a finding unlikely to be owing to chance, given that together the three HR5 regions of ZYG-9 represent only ∼8% of the entire protein (see supplementary material Fig. S2). Moreover, the two other zyg-9 mutations target residues located in domains known to interact with TAC-1 (Bellanger and Gönczy, 2003) and, indeed, cause a severe reduction in the ability to bind TAC-1 in vitro. Overall, it is tempting to speculate that TOG-domain-mediated interactions between ZYG-9 and some unidentified partner, as well as TOG-domain-independent interaction between ZYG-9 and TAC-1 play crucial roles during the first embryonic division of the nematode.

ZYG-9 and TAC-1 form a complex and are mutually dependent for its stability (Bellanger and Gönczy, 2003; Le Bot et al., 2003; Srayko et al., 2003). Therefore, prior analysis did not distinguish between ZYG-9 and TAC-1 being crucial for stimulating growth of microtubules, although the growth-promoting activity ascribed to XMAP215 family members made ZYG-9 a likely candidate. Moreover, it was not clear whether this complex acts solely at centrosomes, where it is enriched, or also in the cytoplasm. Our analysis of the new zyg-9- and tac-1-mutant alleles sheds light on these questions in several ways. First, by analyzing zyg-9(or628) embryos, we found that the reduction of centrosomal ZYG-9 does not result in a similar depletion of centrosomal TAC-1, raising the possibility that TAC-1 can localize to centrosomes - at the least to some extent - independently of ZYG-9. Second, by analyzing tac-1(or369) embryos, we found that cytoplasmic ZYG-9 is indistinguishable from the wild type. Nevertheless, tac-1(or369) embryos exhibit a strong phenotype, indicating that normal levels of cytoplasmic ZYG-9, by themselves, are not sufficient for function. Since normal levels of centrosomal ZYG-9 are not sufficient either (Bellanger and Gönczy, 2003), ZYG-9 levels appear to be critical both in the cytoplasm and at centrosomes to ensure correct microtubule-dependent processes in one-cell-stage embryos. Third, analyzing tac-1(or455) embryos at the permissive temperature, we found that a significant reduction of TAC-1 levels at centrosomes does not interfere with microtubule-dependent processes. Moreover, by analyzing zyg-9(or628) embryos, we found that normal levels and distribution of TAC-1 are also not sufficient for microtubule-dependent processes. Taken together, these findings establish that ZYG-9, but not TAC-1, is limiting both in the cytoplasm and at centrosomes for promoting microtubule growth, whereas TAC-1 is important for maintaining correct ZYG-9 levels and contributing to its centrosomal enrichment.

TACC proteins appear to exert slightly distinct functions in other organisms. Thus, although the Xenopus TACC protein Maskin interacts with XMAP215, immunodepletion and add-back experiments in egg extracts indicate that Maskin can promote microtubule assembly without XMAP215 (O'Brien et al., 2005; Peset et al., 2005). In Drosophila, it has been proposed that TACC functions primarily to recruit the XMAP215 homologue Msp to centrosomes (Lee et al., 2001). TACC proteins have been implicated in tumor progression in human cells (reviewed in Raff, 2002), and it will be interesting to use the newly identified temperature-sensitive tac-1 alleles to test whether TAC-1 is required for cell division in later stages of C. elegans development. Suggestively, tac-1(or455)-mutant animals raised at the restrictive temperature throughout development become uncoordinated (our unpublished observations), compatible with a later requirement for cell division. Therefore, the new tac-1 alleles described in this study represent an important resource for future investigation of the role of TACC proteins during development of metazoan organisms.

TACC-domain-containing proteins and doublecortin-domain-containing proteins are evolutionarily conserved and we found that TAC-1 physically interacts with ZYG-8, both in vitro and in vivo. The biological significance of this physical interaction remains to be clarified. We found that ZYG-8 is not required for correct levels or distribution of TAC-1, as expected from the fact that null-alleles of zyg-8 do not exhibit the earlier phenotypes associated with loss of zyg-9-tac-1 function. Conversely, TAC-1 is not required for correct levels or distribution of ZYG-8, even though this could have been envisaged given that zyg-8 mutants exhibit defects later than embryos lacking zyg-9-tac-1. Therefore, ZYG-8 and TAC-1 do not mutually stabilize each other, in contrast to ZYG-9 and TAC-1. Nonetheless, it remains possible that ZYG-8 somehow promotes the function of ZYG-9-TAC-1 without affecting protein levels or distribution. For instance, the kinase domain of ZYG-8 could phosphorylate ZYG-9 or TAC-1, the latter being more likely given that ZYG-8 and TAC-1 are part of a complex that does not contain ZYG-9. In this manner, ZYG-8 could modulate the activity of the ZYG-9-TAC-1 complex. There is precedent for phosphorylation modulating XMAP215 and TACC proteins. Thus, phosphorylation of XMAP215 by CDK1 dampens its growth promoting activity in vitro (Vasquez et al., 1999), whereas phosphorylation by Aurora-A activates TACC at centrosomes in Drosophila (Barros et al., 2005), and is similarly required for the recruitment of Maskin to the spindle poles in Xenopus egg extracts (Peset et al., 2005). By analogy, one could speculate that the kinase domain of ZYG-8 similarly activates TAC-1.

A plausible alternative to ZYG-8 modulating ZYG-9-TAC-1 activity is that the growth-promoting activities of ZYG-9-TAC-1 and that of ZYG-8, perhaps associated with TAC-1, independently ensure efficient microtubule-dependent processes throughout the cell cycle. Compatible with this view, we found that, compromising ZYG-9-TAC-1 in addition to ZYG-8 yields more severe phenotypes than compromising either MAP alone. Doubly compromised embryos exhibit enhanced meiotic defects, more severe pronuclear migration defects during interphase, and synthetic defects in spindle elongation and cytokinesis. Therefore, ZYG-9-TAC-1 and ZYG-8 act in a partially redundant manner during meiosis, interphase and mitosis to ensure correct microtubule-dependent processes. To our knowledge, such functional redundancy between these MAPs has not, in any organism, been documented before.

Whereas ZYG-9-TAC-1 has previously been known to promote growth of microtubules early in the cell cycle (Matthews et al., 1998; Bellanger and Gönczy, 2003; Le Bot et al., 2003; Srayko et al., 2003; Srayko et al., 2005), our findings uncover that ZYG-8 functions also during interphase and not only during mitosis as previously suspected. The interphase requirement is masked by the presence of ZYG-9-TAC-1, and becomes apparent only when ZYG-9-TAC-1 and ZYG-8 are compromised simultaneously. Because anaphase in early C. elegans embryos is accompanied by extensive growth of astral microtubules, MAPs - such as ZYG-8 - that contribute to a lesser extent to microtubule growth throughout the cell cycle may become limiting only during anaphase. By contrast, MAPs - such as ZYG-9-TAC-1 - that contribute to a greater extent are already limiting during interphase. A combination of different MAPs contribute to the overall behavior of microtubules in other cases. For instance, a balance of the growth-promoting XMAP215 and of the microtubule-destabilizing XKCMI determines microtubule dynamics in Xenopus egg extracts (Tournebize et al., 2000; Kinoshita et al., 2001). In C. elegans embryos, our findings establish that ZYG-9-TAC-1 and ZYG-8, two growth-promoting MAPs, ensure together the correct microtubule-dependent processes throughout the cell cycle.

C. elegans strains were maintained according to standard methods (Brenner, 1974). zyg-8-mutant alleles have been described previously (Gönczy et al., 2001). Temperature-sensitive mutant alleles were generated in a lin-2(e1309) background (Encalada et al., 2000). For linkage group mapping, tac-1(ts); him-8(e1498) IV and zyg-9(ts); him-8(e1498) IV males were mated with MT3751 [dpy-5(e61) I; rol-6(e187) II; unc-32(e189) III], MT464 [unc-5(e53) IV; dpy-11(e224) V; lon-2(e678) X] hermaphrodites. All eight mutant alleles showed linkage to chromosome II. All three tac-1 alleles mapped to the right arm of LGII, distal to lin-7 at map position 22.9 (tac-1 is located at ∼23.46), based on three-factor mapping with strains of the genotype tac-1(ts)/rol-1(e91) lin-7(e1413) II. Further mapping of tac-1-mutant alleles was conducted by standard three-factor crosses and SNP mapping (Brenner, 1974; Wicks et al., 2001). Sequencing of mutant alleles was performed on a pool of three independently PCR-amplified genomic DNA using the Beckman Quickstart Dye Terminator Kit and a Beckman CEQ8000 instrument, at the University of Oregon DNA Sequencing Facility.

To determine the lethality of embryos produced by homozygous tac-1 and zyg-9 mutants at 15°C, individual worms were allowed to lay eggs overnight followed by removal of the adults. Embryo lethality was determined 24 hours after removal of the adults by counting the number of unhatched embryos and larvae using a stereomicroscope. The same procedure was used at 26°C, except that L4 larvae raised at 15°C were shifted to the restrictive temperature of 26°C overnight, and then transferred to fresh plates and allowed to lay eggs overnight.

Unless specified otherwise, homozygous mutant animals of temperature sensitive alleles were shifted to the restrictive temperature of ∼25°C for >12 hours prior to analysis for immunofluorescence or time-lapse microscopy. To compare embryonic lethality conferred by zyg-8(or484) to that conferred by zyg-8(or484) expressing GFP-TAC-1 (Bellanger and Gönczy, 2003), embryos were collected at room temperature for 7-15 hours and the population of surviving adults determined was 2 days later. For simplicity, we refer throughout the manuscript to `mutant embryos' when mentioning the progeny derived from homozygous mutant animals.

Bacterial strains for RNAi by feeding of zyg-9 and of tac-1 were generated essentially as described (Kamath et al., 2001). L4 hermaphrodites were placed onto feeding plates and incubated typically for >30 hours at 20°C before analysis. While feeding zyg-8(or484) animals, plates and control plates were transferred to 25°C 6-15 hours prior to analysis.

For most experiments, embryos were analyzed by time-lapse DIC microscopy at 23±1°C, capturing one image every 5 seconds (Gönczy et al., 1999). Movies corresponding to the embryos displayed in Figs 4 and 5 are available upon request. To ensure conditional inactivation of tac-1 or zyg-9 following a temperature shift, tac-1(or455) and zyg-9(or634) animals were kept at ∼15°C until the time of dissection. Embryos were prepared and mounted on an agarose pad containing a thermal probe to monitor the temperature of the specimen during the experiment. The preparation was placed on the microscope stage, where the temperature was kept at 15-18°C using a custom-made fan coupled to a cooling unit. When one-cell-stage embryos were undergoing centration/rotation, the temperature was set to ∼26°C, which was reached within <90 seconds.

Methanol fixation and staining of embryos for indirect immunofluorescence followed established protocols (Gönczy et al., 1999), with the following exceptions. In some cases (Fig. 2H,I), animals grown at 15°C were dissected and fixed in the cold-room to prevent heat-induced protein inactivation. Such a brief (<3 minutes) stay in the cold room did not alter the microtubule network in a significant manner in the wild type (data not shown). For analysis of mutant embryos following a brief temperature-shift (supplementary material Fig. S3), animals were placed within an Eppendorf tube into a PCR machine at 26°C for 7 minutes prior to fixation. For staining with ZYG-8 antibodies, fixation in methanol of ∼1 minute was necessary to achieve robust staining of the spindle and spindle poles.

The following primary antibodies were used: 1:200 mouse anti-α-tubulin (DM1A, Sigma), 1:500 rabbit anti-TAC-1 (Bellanger and Gönczy, 2003), 1:5000 rabbit anti-ZYG-9 (Gönczy et al., 2001), 1:200 rabbit anti-ZYG-8 (Gönczy et al., 2001), 1:3000 rabbit anti-SPD-5 (Hamill et al., 2003). Secondary antibodies were 1:1000 goat anti-mouse conjugated to Alexa-Fluor-488 (Molecular Probes) and 1:1000 goat anti-rabbit conjugated to Alexa-Fluor-543. Slides were counterstained with ∼1 μg/ml Hoechst-dye 33258 (Sigma) to detect DNA. Unless indicated otherwise, optical slices (∼1 μm thick) were collected on an LSM510 Zeiss confocal microscope and processed using Adobe Photoshop, maintaining the relative intensities among embryos.

Quantification of ZYG-9 and TAC-1 following indirect immunofluorescence was performed on a wide-field microscope with a 63× lens using Metamorph software (Universal Imaging), analyzing 400×300 pixel images of embryos during late prometaphase (occasional embryos), during metaphase (some embryos) or anaphase (most embryos) of the first or second cell cycle, following background removal. To determine the relative levels of ZYG-9 or TAC-1 (Figs 1, 2), the average pixel intensity of the ZYG-9 or TAC-1 signal in the cytoplasm was compared with the average pixel intensity of the α-tubulin signal in two ∼150-pixel circles encompassing the middle of the asters, which served as an internal control for staining efficiency. To determine the ratio between the levels of ZYG-9 or TAC-1 at centrosomes and in the cytoplasm, the average pixel intensity of the ZYG-9 or TAC-1 signal in two ∼50-pixel circles centered on the centrosomes was compared with the average pixel intensity of signal in the cytoplasm. Although some pre-anaphase embryos were included to increase sample size, restricting the quantification strictly to anaphase embryos did not alter the results in a significant manner (data not shown).

To generate GST-TAC-1 M58I and L229F, as well as the two ZYG-9 mutations I862K and E1317K, we mutated (site-directed mutagenesis strategy from Stratagene) pGex-TAC-1 (Bellanger et al., 2003) and a vector containing the TAC-1-binding region (TBR) of ZYG-9 (aa 654-1415) (Bellanger et al., 2003). Recombinant GST and GST-TAC-1 proteins were produced in E. coli, purified, dialyzed against PBS and snap-frozen. The ZYG-9 TBR, the full-length ZYG-8 or fragments thereof, as well as the first GEF domain of human Trio, were translated in vitro using the TNT T7 kit (Promega) in the presence of [³⁵S]methionine (Amersham-Pharmacia), as described (Bellanger et al., 2000). Translation efficiency was verified by SDS-PAGE and autoradiography. Of each labeled protein 1-10 μl were then incubated for 2 hours (ZYG-9) or overnight (ZYG-8) at 4°C with 5 μg of GST or GST-TAC-1 in 200 μl binding buffer (Hepes 50 mM pH 7.5, NaCl 150 mM, MgCl₂ 5 mM, DTT 0.1 mM and Tween-20 0.01%). The GST proteins and associated radioactive products were purified using glutathione beads, and the retained material was separated by SDS-PAGE prior to autoradiography.

Western blot analysis was performed using nitrocellulose membranes (Protran, Schleicher&Schuell). Primary antibodies used were diluted in PBS with 5% milk and 0.1% Tween-20. Secondary HRP-coupled anti-mouse and anti-rabbit antibodies were diluted in PBS and used at 1:10,000. The ECL reaction was performed as recommended by the manufacturer (Amersham Pharmacia).

Co-immunoprecipitations were performed as described (Bellanger and Gönczy, 2003). Briefly, worm embryonic extract was prepared in IP buffer (Hepes 100 mM, NaCl 750 mM, MgCl₂ 10 mM, EDTA 50 mM, DTT 1 mM, protease inhibitors) by grinding snap-frozen embryos in a ceramic bowl. The extract was pre-cleared on proteinG-sepharose beads prior to addition of antibodies (5 μg/ml final concentration) and incubation for 2 hours at 4°C, followed by addition of proteinG-sepharose and further incubation for 1 hour at 4°C. The beads were washed once with IP buffer and twice with PBS before western blot analysis.

We are grateful to Karine Baumer, Sylvie Fromont and Nga N'Guyen for technical help, to Virginie Hachet for providing some wild-type recordings, as well as to Anne Debant, Katayoun Afshar and Marie Delattre for critical reading of the manuscript. J.-M.B. was supported by an EMBO long-term post-doctoral fellowship (ALTF 497-2001), J.B.P. by an NIH training grant. In addition, this work was supported by grants R01GM49869 from the NIH (to B.B.) and 3100A0-102087 from the Swiss National Science Foundation (to P.G.).