Xenopus NEDD1 is required for microtubule organization in Xenopus egg extracts

The centrosome serves as the major microtubule-nucleating and -organizing center in animal cells. It is composed of hundreds of proteins. The molecular details of how centrosomal proteins contribute to centrotome function are only beginning to emerge. Members of the neuron-precursor-cell-expressed developmentally downregulated protein 1 (NEDD1) family of conserved proteins have recently been implicated in recruiting γ-tubulin and its associated proteins, which together make up the γ-tubulin ring complex (γTuRC), to the centrosome. Human NEDD1 and its Drosophila ortholog Dgp71WD are WD-repeat proteins that interact with the γTuRC. Experimental knockdown of human NEDD1 was recently shown to result in loss of γ-tubulin from the centrosome. By contrast, however, Dgp71WD knockdown has no effect on targeting the γTuRC to the centrosome in flies. Using Xenopus egg extracts, we show that Xenopus NEDD1 is mostly dispensable for targeting γ-tubulin to centrosomes, but that microtubule organization is disrupted in NEDD1-depleted extracts. We show that NEDD1 exists in a complex that is distinct from the γTuRC, suggesting that NEDD1 may not be a bona fide subunit of the Xenopus γTuRC. We propose that the main function of NEDD1 in Xenopus is in microtubule organization.


Introduction
Microtubules serve important roles in cellular organization both during interphase -when they help to position cellular organelles and, facilitate directional transport of proteins and vesicles -and during cell division -when they provide the framework for the segregation of the chromosomes. The overall arrangement of microtubules in a cell -and their dynamic behavior -is regulated by microtubule-stabilizing and -destabilizing activities as well as microtubule-crosslinking and -motor proteins. In most cell types microtubules are nucleated and organized by microtubuleorganizing centers (MTOCs). The major MTOC in animal cells is the centrosome. Although proteomic approaches have contributed valuable information to our knowledge of the components of the centrosome, a detailed understanding of how these components interact to allow the centrosome to function is still missing.
Centrosome function requires a ring-shaped protein complex that contains ␥-tubulin, a specialized member of the tubulin superfamily (Zheng et al., 1995;Moritz et al., 1995). In metazoa, ␥-tubulin associates with at least six additional proteins that are highly conserved across phylogenetically distant species (Wiese and Zheng, 2006). One of these, Dgp71WD (also known as Grip71), is a tryptophan-aspartate (WD)-repeat-containing protein of ~71 kDa first identified as part of the Drosophila ␥-tubulin ring complex ␥(TuRC) (Gunawardane et al., 2003). Dgp71WD was later shown to probably be a peripheral subunit of the ␥TuRC because its depletion does not disrupt the complex (Vérollet et al., 2006). Subsequent work identified a human ortholog of Dgp71WD as a member of the neuron-precursor-cell-expressed developmentally downregulated protein 1 (NEDD1) family of conserved proteins.
Like Dgp71WD, human NEDD1 is a centrosomal protein that interacts with the ␥TuRC but is not required for the integrity of the complex, suggesting that the peripheral association of the WDcontaining protein with the ␥TuRC is evolutionarily conserved (Lüders et al., 2006;Haren et al., 2006).
Despite these similarities, important functional differences between Dgp71WD and human NEDD1 were revealed by experiments performed in tissue culture cells. For example, knockdown of human NEDD1 by RNA interference (RNAi) resulted in the reduction of the amount of ␥-tubulin associated with the centrosome (Lüders et al., 2006;Haren et al., 2006), which suggests that NEDD1 is required for ␥TuRC recruitment to the centrosome in human cells. Surprisingly, however, knockdown of Dgp71WD had no effect on the amount of ␥-tubulin associated with the centrosome in Drosophila tissue culture cells, suggesting that Dgp71WD is dispensable for ␥TuRC recruitment (Vérollet et al., 2006).
To gain a better understanding of the molecular details of the function of the WD-containing ␥TuRC subunit we turned to the Xenopus member of this family. Xenopus egg extracts, pioneered by Lohka and Masui (Lohka and Masui, 1983), have been used extensively to study spindle assembly and ␥TuRC recruitment. These extracts have the advantage that they can be experimentally manipulated such that centrosome-dependent and centrosomeindependent processes can be studied independently. Moreover, the extracts can be made to support interphase or mitotic processes. Thus, the contributions of a given protein of interest to interphase or mitotic processes (most notably, centrosome-and microtubuleassembly) can be examined in the absence of other cell cycle stages. In a process that mimics fertilization, addition of The centrosome serves as the major microtubule-nucleating and -organizing center in animal cells. It is composed of hundreds of proteins. The molecular details of how centrosomal proteins contribute to centrotome function are only beginning to emerge. Members of the neuron-precursor-cell-expressed developmentally downregulated protein 1 (NEDD1) family of conserved proteins have recently been implicated in recruiting ␥-tubulin and its associated proteins, which together make up the ␥-tubulin ring complex (␥TuRC), to the centrosome. Human NEDD1 and its Drosophila ortholog Dgp71WD are WD-repeat proteins that interact with the ␥TuRC. Experimental knockdown of human NEDD1 was recently shown to result in loss of ␥-tubulin from the centrosome. By contrast, however, Dgp71WD knockdown has no effect on targeting the ␥TuRC to NEDD1 organizes microtubules in Xenopus demembranated sperm heads to the egg extracts initiates centrosome assembly from sperm-derived centrioles and eggderived pericentriolar material (including ␥TuRC) (Felix et al., 1994). The centrosomes, once competent to nucleate microtubules, initially assemble microtubule asters; subsequent structural rearrangements of the asters result in 'half-spindles', which then fuse to form bipolar spindles (Lohka and Masui, 1983;Murray, 1989;Sawin and Mitchison, 1991). These processes can be used to functionally characterize proteins for their contribution to ␥TuRC recruitment, centrosome assembly, microtubule stability or spindle assembly. More recently, Xenopus egg extracts have also been used to assemble asters and spindle-like structures in the absence of centrosomes by the addition of RanGTP or RanGTP mimicking mutant proteins (reviewed in Kahana and Cleveland, 1999;Dasso, 2002). Moreover, protein requirements for microtubule organization in the absence of centrosomes can also be studied in microtubule asters induced by the addition of the microtubule stabilizing reagent, taxol (e.g. Verde et al., 1991).
In this study, we used Xenopus egg extracts to investigate the functions of Xenopus NEDD1 in ␥TuRC recruitment to centrosomes and in centrosome-independent processes. Our results suggest that NEDD1 plays a minor (and possibly indirect) role in recruiting ␥TuRC during sperm-centriole-templated centrosome assembly. However, our data also strongly suggest that, in Xenopus, NEDD1 has functions in microtubule organization that are independent of its role in recruiting the ␥TuRC to the centrosome.

Results
An alignment of the NEDD1 primary sequence with its human and Drosophila (Dgp71WD) orthologs (supplementary material Fig. S1) shows that Xenopus NEDD1 shares sequence similarity with human NEDD1 and Dgp71WD over large portions of the N-terminal WD-40 domain. Except for a ~100 amino acid stretch of unknown function near the C-terminus that is shared by Xenopus NEDD1 and human NEDD1 there is only limited similarity among human NEDD1, Xenopus NEDD1 and Dgp71WD outside of the WD domain.
To learn more about the function of Xenopus NEDD1, we began our analysis by raising polyclonal antibodies in rabbits against the C-terminal half of NEDD1 (amino acids 301-655) and using these antibodies to characterize NEDD1 by western blotting and immunofluorescence microscopy (Fig. 1). The affinity-purified antibodies recognized a band of ~75 kDa in interphase Xenopus egg extracts and a tight doublet migrating at ~85 kDa in mitotic extracts (Fig. 1A). The slower migration of NEDD1 from mitotic extracts is, at least in part, due to phosphorylation, because treatment with phosphatase resulted in faster migration of NEDD1 (Fig. 1Ab). This is consistent with a previous report that human NEDD1 is phosphorylated during mitosis (Lüders et al., 2006). We speculate that the doublet in mitotic extracts probably represents two NEDD1 gene products that are less-well resolved in interphase extracts, because Xenopus laevis is pseudo-tetraploid. Consistent with a role for NEDD1 in centrosome function, our antibodies stained centrosomes in Xenopus egg extracts (Fig. 1B,C) and in Xenopus tissue culture cells during all stages of the cell cycle (Fig. 1D,E). The NEDD1 antibodies also stained spindle microtubules in mitotic extracts (Fig. 1B) and cells (Fig. 1D,E). Similar centrosome and microtubule staining was seen in tissue culture cells transiently expressing GFP-tagged NEDD1 (supplementary material Fig. S2).
Human NEDD1 has been proposed to function in centriole assembly, because disruption of NEDD1 in human cells in culture leads to centriole defects (Haren et al., 2006). Consistent with the idea that NEDD1 family members might function at the centriole, antibodies against Xenopus NEDD1 stained two distinct spots on sperm heads that presumably correspond to sperm-associated centrioles (supplementary material Fig. S3).
NEDD1 disruption causes defects of aster-and spindleassembly in Xenopus egg extracts To test whether NEDD1 is required for the assembly of microtubule structures, we immunodepleted NEDD1 from mitotic Xenopus egg extracts and examined the depleted extracts (supplemented with a small amount of rhodamine-labeled tubulin to allow visualization of microtubules formed in the extract; see Materials and Methods) for their ability to assemble microtubule asters around exogenously added sperm heads within 15 minutes of incubation. Two rounds of immunodepletion reduced the levels of NEDD1 to ~10-20% ( Fig. 2A). This reduction in NEDD1 levels had profound effects on aster assembly: the number of sperm heads able to assemble asters ('firing' sperm heads) was reduced to 34% of control levels. In addition, the microtubule density per aster (aster size; measured by the amount of rhodamine fluorescence associated with each aster) was reduced by ~80% compared with controls ( Fig. 2C,D). Both types of defect (number of sperm heads that fired and aster size) could be rescued by the addition of bacterially expressed recombinant Xenopus NEDD1 (Fig. 2C-E), suggesting that the defects were caused by the absence of NEDD1 rather than by the absence of co-depleted proteins. Indeed, the defects could be rescued by XNEDD1-C, a truncation mutant that lack the N-terminal WD-40 repeats (lacking amino acid residues 1-300), but not by XNEDD1-N, which comprises the WD-40 domain (lacking amino acid residues 301-655) ( To determine whether NEDD1 was also required for spindle assembly, we examined the microtubule structures formed in egg extracts at later time points. Spindles assembled within 90 minutes in NEDD1-depleted extracts showed one of two phenotypes: they either had seemingly normal numbers of disorganized microtubules and misaligned chromosomes, or they had substantially fewer microtubules that were organized into bipolar spindles but whose chromosomes were misaligned (Fig. 2F,G). This suggests that NEDD1 function is required for spindle assembly.
Addition of 300 nM full-length NEDD1 to egg extractsequivalent to a sixfold excess of NEDD1 compared with ~50 nM of endogenous NEDD1 levels -resulted in spindle disruption but had little effect on asters. Consistent with the inability of XNEDD1-N to rescue the NEDD1 depletion, addition of NEDD1-N to egg extracts had no detectable effect on microtubules. By contrast, overexpression of NEDD1-C showed a dominant-negative phenotype: both microtubule asters (Fig. 3A,C) and spindles ( Fig.  3B) were severely disrupted. Interestingly, the microtubules appeared highly bundled, suggesting a role for NEDD1 in microtubule crosslinking. This is further discussed below. Overexpression of XNEDD1-C also partially disrupted the centrosomes assembled around sperm centrioles, because the amount of ␥-tubulin recruited to sperm heads was reduced by 50% (Fig. 3D,E). Together, these observations suggest that NEDD1 is required for the correct organization of microtubules assembled in egg extracts.
NEDD1 is not strictly required for ␥TuRC recruitment to centrosomes Smaller and fewer sperm asters in NEDD1-depleted extracts could result from defects in either microtubule nucleation or stability, or both. A decrease in the amount of ␥-tubulin at the centrosome in the presence of exogenous XNEDD1-C suggested that NEDD1 might be involved in recruiting ␥-tubulin to the centrosome. To test this more directly, we first examined whether NEDD1 depletion Journal of Cell Science 121 (5) affected ␥-tubulin recruitment to sperm centrosomes. Reactions were supplemented with the microtubule-destabilizing drug nocodazole to eliminate ␥-tubulin staining along the microtubules to facilitate visualization of centrosome-associated ␥-tubulin. Sperm heads were incubated with mock-depleted or NEDD1depleted extracts, spun onto coverslips and then processed for ␥tubulin immunofluorescence to measure the amount of ␥-tubulin present at the reconstituted sperm centrosome. We found that the Western blots of mitotic or interphase extracts treated with (+) alkaline phosphatase (AP) or control buffer (-). (B) Immunofluorescence micrographs of 'cycled' microtubule asters (top panels) or spindles (bottom panels) induced by addition of sperm chromatin to egg extracts. Overlays show XNEDD1 in green and ␣tubulin in red. Scale bar, 10 m. (C) Immunofluorescence micrographs of sperm tips (top panels) incubated in egg extract in the presence of nocodazole to prevent microtubule assembly, or sperm-induced asters (bottom panels) triple-labeled with antibodies against XNEDD1 or ␥-tubulin, or stained for DNA, as indicated above the panels. Scale bars, 1 m (top panels) and 10 m (bottom panels). Overlays show XNEDD1 in green, ␥-tubulin in red and DNA in blue. (D,E) Xenopus tissue culture cells at various cell cycle stages (indicated on the left) triple labeled for XNEDD1, ␣-tubulin or ␥-tubulin and DNA, as indicated above the panels. Overlays in D show XNEDD1 in green, ␣-tubulin in red, DNA in blue. Overlays in E show ␥-tubulin in green, XNEDD1 in red, DNA in blue. Scale bars, 10 m. NEDD1 organizes microtubules in Xenopus total amount of ␥-tubulin at sperm centrosomes in NEDD1depleted extracts varied significantly between experiments and ranged between 54% in one experiment and 96% in another, with an average of 78±6.5% (±s.e.m.; n=6 independent experiments, with 50 sperm tips measured per sample per experiment) of the amount of ␥-tubulin at centrosomes assembled in mock-depleted extracts (Fig. 4).
Because the presence of ␥-tubulin on sperm centrioles prior to the experiment (supplementary material Fig. S3) potentially introduces 'background noise' to the experiment, we sought an Examples of sperm-induced asters assembled in mock-depleted (top panels) or XNEDD1-depleted extracts (all other panels). XNEDD1-depleted extracts were supplemented with buffer or full-length XNEDD1, XNEDD1-N or XNEDD1-C, as indicated. All extracts were supplemented with small amounts of rhodamine-tubulin to visualize microtubules (red in overlays). Signals in the rhodamine channel (left panels) or UV to visualize DNA (middle panels) were acquired separately and digitally overlaid. Scale bar, 10 m. (D,E) Quantification of the (D) fluorescence intensity and (E) asters assembled around sperm centrioles for the experiments shown in C, expressed as percent of mock-depleted control. Error bars give the standard error (± s.e.). (F) Examples of spindles assembled in mock-depleted (top two rows of panels) or XNEDD1-depleted extracts (all other panels). Scale bar, 10 m. (G) Quantification of the relative abundance of bipolar spindles, spindles with at least one disorganized pole and spindles with greatly reduced microtubule numbers in mock-depleted (light bars) and XNEDD1-depleted (dark bars) egg extracts. Error bars give the standard error (± s.e.). independent way to measure the amount of ␥-tubulin recruited to centrosomes in NEDD1-depleted extracts. We therefore tested NEDD1-depleted extracts for their ability to complement saltextracted centrosomes. In this centrosome-complementation assay, centrosomes are rendered unable to nucleate microtubules in vitro by exposing them to a chaotropic agent such as 1 M potassium iodide (KI) (Moritz et al., 1998). The microtubule-nucleating capacity of the centrosome can then be restored by incubating the salt-extracted centrosomes in Xenopus egg extract (Popov et al., 2002). We found that NEDD1-depleted extracts were able to restore full activity to salt-stripped centrosomes (Fig. 4C). Consistent with this observation, the amount of ␥-tubulin recruited to saltstripped centrosomes in NEDD1-depleted extracts was indistinguishable from the amount of ␥-tubulin recruited to centrosomes complemented with control mock-depleted extracts (Fig. 4D). Interestingly, the most noticeable effect of NEDD1 depletion was on the amount of ␥-tubulin associated with microtubules, which was reduced to ~30% in NEDD1-depleted extracts ( Fig. 4F,G).
Although we could not rule out that the small amount of NEDD1 that remains in the extract after immunodepletion is sufficient to recruit ␥-tubulin to centrosomes, or that NEDD1/Dgp71WD was Journal of Cell Science 121 (5) not efficiently removed from the centrosomes by salt extraction, our results suggest that ␥-tubulin can be recruited to the centrosome in the absence of most of the endogenous NEDD1. Consistent with this idea, depletion of NEDD1 from the egg extracts had no effect on the levels of ␥-tubulin that remained in the extracts (Fig. 4E).
Recombinant NEDD1 can recruit the ␥TuRC from extracts Next, we wanted to further investigate a potential interaction between NEDD1 and the ␥TuRC. We began by asking whether NEDD1 can recruit the ␥TuRC. For this, we reasoned that recombinant NEDD1 that had been immobilized on beads should gain the ability to nucleate microtubules when incubated in untreated egg extracts, but should be unable to do so when incubated in ␥TuRC-depleted extracts. Furthermore, ␥TuRC subunits should be able to associate with immobilized NEDD1 incubated in egg extract. Using the NEDD1-coated beads as bait, we found that beads coated with full-length NEDD1 or XNEDD1-C gained the ability to nucleate microtubules in vitro from purified bovine brain tubulin when incubated in egg extract and isolated again (Fig. 5A), but not if the extracts had been depleted of ␥TuRC. Western blot analysis showed that full-length NEDD1 and XNEDD1-C were able to recruit the ␥TuRC subunits ␥-tubulin, Xgrip210 and Xgrip109 from with buffer (top panels) or 300 nM XNEDD1-C (bottom panels) and labeled with antibodies against ␥-tubulin (green in the overlays) or acetylated tubulin (to mark the position of the sperm centrioles; red in the overlays), and stained to visualize the DNA (blue in the overlays). Scale bar, 1 m. (E) Quantification of the amount of ␥-tubulin immunofluorescence associated with sperm heads for the experiment shown in D, expressed as percent of mock-depleted control. Error bars give the standard error (± s.e.). NEDD1 organizes microtubules in Xenopus the extracts (Fig. 5B). The interaction between NEDD1 and the ␥TuRC was further confirmed by reciprocal immunoprecipitation experiments, which showed that ␥-tubulin immunoprecipitates contained NEDD1, and NEDD1 immunoprecipitates contained ␥tubulin and Xgrip109 (Fig. 5C). These observations suggested that NEDD1 recruited the ␥TuRC to nucleate microtubules in vitro. Fig. 4. Xenopus NEDD1 (XNEDD1) depletion reduces the amount of ␥-tubulin that associates with the centrosome but does not abolish ␥-tubulin recruitment. (A,B) Micrographs of ␥-tubulin staining at sperm heads incubated in buffer (top panels), mock-depleted extract (middle panels), or XNEDD1-depleted extract (bottom panels). Signals from the individual fluorescein (␥-tubulin, left column; green in overlay), rhodamine (␣-tubulin, center column; red in overlay), or UV (DNA stain; blue in overlay) are shown as overlays in the right column. Scale bar, 1 m. (B) Quantification of the amount of ␥-tubulin associated with sperm centrioles for the experiment shown in A, normalized against the ␥-tubulin immunofluorescence for sperm tips incubated in mock-depleted extract. Error bars, SE. (C) Drosophila centrosomes were rendered inactive by treatment with KI, and were then incubated with mock-depleted (top row) or XNEDD1-depleted (bottom row) extract (see text for detail). The extract was then washed off and reconstituted centrosomes were incubated with a solution of pure tubulin (containing a small amount of rhodamine-tubulin) to assay their ability to nucleate microtubules. Microtubules were fixed, and visualized under the microscope. Centrosomes reconstituted with mitotic (left panels) or interphase (right panels) extract are shown. Scale bar, 5 m. (D) Quantification of ␥-tubulin recruited to reconstituted centrosomes treated as described in C, except that they were fixed and stained for ␥-tubulin immunofluorescence instead of being tested in the microtubule nucleation assay. ␥-tubulin immunofluorescence intensity is reported as percent of mock-depleted control. Error bars give the standard error (± s.

NEDD1 exists in a complex that is distinct from the ␥TuRC
Our results thus far suggested that NEDD1 can indeed interact with the ␥TuRC, but our experiments could not determine whether this interaction was direct. Our finding that ␥-tubulin recruitment to sperm centrosomes was only partially impaired in NEDD1depleted extracts suggested that the interaction between NEDD1 and ␥-tubulin might not be important. Furthermore, we noticed that, although NEDD1 and ␥-tubulin colocalized to centrosomes they generally did not colocalize very well within the centrosome ( Fig. 1C; for more examples, see supplementary material Fig. S3). Taken together with the observation that depletion of NEDD1 had little effect on the levels of ␥-tubulin remaining in the egg extract (Fig. 4E), we wondered whether NEDD1 is truly a subunit of the ␥TuRC, as has been suggested for its homologs in flies and humans (Gunawardane et al., 2003;Vérollet et al., 2006;Lüders et al., 2006;Haren et al., 2006). To examine this question further, we fractionated untreated or XNEDD-depleted Xenopus egg extracts on 5%-to 40%-sucrose gradients and probed the fractions for NEDD1 and ␥-tubulin (Fig. 6B,D). This analysis revealed that depletion of NEDD1 had no effect on the migration of ␥-tubulin on the gradient, confirming earlier reports that the NEDD1 subunit is not an integral part of the ␥TuRC (Vérollet et al., 2006;Lüders et al., 2006;Haren et al., 2006). We were surprised to find that the distributions of NEDD1 and ␥-tubulin were largely nonoverlapping: NEDD1 peaked in fractions 11-13, whereas ␥tubulin peaked in fractions 14-16 (Fig. 6C). This suggested that NEDD1 exist in a complex that is distinct from the ␥TuRC. A comparison of NEDD1 and ␥-tubulin immunoprecipitates by Coomassie staining supported this idea (Fig. 6A). Although NEDD1 antibodies co-immunoprecipitated faint bands that presumably corresponded to ␥TuRC subunits, these antibodies also co-precipitated proteins that are not part of the ␥TuRC (marked by asterisks in Fig. 6A). The molecular identity of these NEDD1interacting proteins is not yet known.
Journal of Cell Science 121 (5) NEDD1 is required for centrosome-independent microtubule organization Our finding that the majority of NEDD1 exists outside of the ␥TuRC led us to examine whether NEDD1 was required for microtubuledependent processes that do not involve recruiting the ␥TuRC to the centrosome. For this, we tested the effects of NEDD1 depletion or overexpression on microtubule assembly induced by the addition of the RanGTP mimetic RanL43E, which does not require centrosomes, to Xenopus egg extracts. Microtubule aster formation without centrosomes can also be induced by addition of the microtubule stabilizing drug taxol to egg extracts. Surprisingly, we found that NEDD1 depletion had no effect on microtubule formation in the egg extract, but completely abolished the ability of the microtubules to organize into focused arrays ( Fig. 7B; see also supplementary material Fig. S5B). Addition of recombinant XNEDD1-C (but not full-length NEDD1 or XNEDD1-N) to RanGTP-treated extracts similarly disrupted microtubule organization and led to formation of highly bundled microtubule arrays (Fig. 7C). This indicates that NEDD1 is dispensable for microtubule nucleation, but is required for microtubule organization in the egg extracts. Consistent with a role in focusing microtubule arrays, NEDD1 was concentrated at the centers of RanGTP or taxolinduced asters, as well as along astral microtubules (Fig. 7A, supplementary material Fig. S5A). Based on these observations, we conclude that NEDD1 is required for microtubule organization independent of its activity in recruiting ␥-tubulin to the centrosome.

XNEDD1-C bundles microtubules
To test whether NEDD1 interacted directly with microtubules, we examined the effects of adding bacterially expressed NEDD1 to in vitro microtubule-assembly reactions. Microtubules assembled in the presence of 300 nM XNEDD1-C but not full-length NEDD1 or XNEDD1-N (both at 3 M), and appeared highly bundled when examined under the light microscope (Fig. 8A). XNEDD1-C, or ␥-tubulin antibodies were incubated with pure tubulin to assay their ability to nucleate microtubules. Beads were either used directly (left panels; not exposed to extract) or were first incubated in extract (middle panels, exposed to extract) or in ␥-tubulin-depleted extract (right panels, exposed to extract but ␥TuRC-⌬). Representative pictures are shown for each condition. Beads were detected by autofluorescence; microtubules appear as spiky halos surrounding the beads. Scale bar, 10 m. (B) Western blot of proteins that associate with the beads coated as in A and incubated in extract. The blot was probed with antibodies directed against the ␥TuRC subunits Xgrip210, Xgrip109 or ␥-tubulin, as indicated on the right. Protein 'baits' are indicated above the lanes. Extr, extract control; GST, GST-coated beads; FL, full-length GST-XNEDD1; N, GST-XNEDD1-N; C, GST-XEDD1-C; IP, beads coated with antibodies against ␥-tubulin. (C) Western blot of imunoprecipitations from egg extracts using antibodies directed against ␥-tubulin or XNEDD1, as indicated above the blot. Input represents 1.5 l of extract; immunoprecipitations were from 100 l of extract. similarly bundled preformed taxol-stabilized microtubules (Fig. 8C), and we determined that it bound to microtubules with an apparent K d of ~1.43 M. Consistent with a role in crosslinking microtubules, XNEDD1-C localized along the length of bundled microtubules and pelleted with tubulin in pelleting assays (Fig. 8B). These results suggest that XNEDD1-C can bundle both polymerizing and already polymerized microtubules. We propose that NEDD1 affects microtubule organization through its bundling activity.

NEDD1 localizes to kinetochore fibers
In addition to localizing to the centrosome, NEDD1 also localized to spindle microtubules during mitosis (Fig. 1). The microtubulebundling activity of NEDD1 prompted us to test whether NEDD1 is involved in organizing kinetochore fibers, a highly bundled subset of spindle microtubules that are resistant to cold treatment (reviewed by Compton, 2000). To visualize kinetochore fibers, we placed Xenopus tissue culture cells on ice to depolymerize non-kinetochore microtubules. The cells were then fixed and processed for immunfluorescence with NEDD1-specific antibodies. This showed that NEDD1 localized to fibers that are stable in the cold ('coldstable fibers'; Fig. 8D,E). To confirm that these cold-stable fibers are indeed kinetochore fibers, we double-stained cold-treated cells with NEDD1 antibodies and antibodies against the kinetochore protein Bub1 (Boyarchuk et al., 2007). We found that many NEDD1-positive fibers ended at the kinetochores (Fig. 8E). Based on these observations, it is very intriguing to hypothesize that NEDD1 functions as a microtubule bundling protein at the kinetochore fibers.

Discussion
Centrosomal targeting of the ␥TuRC presumably requires at least two components: a ␥TuRC 'receptor' on the centrosome, and a ␥TuRC subunit (or set of subunits) that act as ligand(s) for this receptor. To identify the molecular mechanisms that bring the  ␥TuRC to the centrosome, we thought it reasonable to first identify the centrosome-targeting subunit of the ␥TuRC and to then use this subunit as bait to identify potential ␥TuRC receptor(s). Five of the six known ␥TuRC subunits are members of a conserved protein family that share one or two regions of sequence similarity known as 'grip motifs' (reviewed in Wiese and Zheng, 2006). The sixth ␥TuRC subunit, NEDD1 (initially identified in flies as Dgp71WD), lacks both grip motifs but instead possesses a WD domain. Being unrelated to the other ␥TuRC subunits, it was tempting to speculate that the WD-domain protein may be the centrosome-targeting subunit of the ␥TuRC. The work presented here suggests that the Xenopus ortholog of Dgp71WD, NEDD1, is a centrosomal protein that contributes to microtubule organization. We were surprised to find, however, that the WD-domain protein may not be a bona fide subunit of the ␥TuRC in frogs. This hypothesis is supported by several observations: for one, depletion of NEDD1 from egg extracts results in co-depletion of only small amounts of ␥-tubulin, and vice versa. Second, NEDD1 exists in a complex that does not contain ␥-tubulin. And, third, NEDD1 is not required for the assembly or the maintenance of the ␥TuRC.

Journal of Cell Science 121 (5)
The molecular mechanisms that regulate how ␥-tubulin is targeted to various sites within the vertebrate cell remain poorly understood. The ␥TuRC resides mainly at the centrosome during both interphase and mitosis, but centrosomal ␥TuRC dynamically exchanges with a much larger cytoplasmic pool (Khodjakov and Rieder, 1999;Wiese and Zheng, 2006). ␥-Tubulin has been reported to associate with both the centrioles and the pericentriolar material (Fuller et al., 1995;Moritz et al., 1995), which together make up the centrosome. During mitosis, ␥-tubulin also localizes within the spindle (Lajoie-Mazenc et al., 1994;Stearns and Kirschner, 1994), presumably either by lateral interactions with centrosomal microtubules or by minus-end capping of non-centrosomal microtubules, or both. Moreover, ␥-tubulin is found at noncentrosomal microtubule anchoring sites in polarized cells (Buendia et al., 1990;Rizzolo and Joshi, 1993;Meads and Schroer, 1995;Salas, 1999;Oriolo et al., 2007). How recruitment and anchoring of ␥-tubulin complexes differs under these different conditions remains a mystery.
Depletion of 80-90% of the endogenous NEDD1 from egg extracts resulted at most in a 50% reduction of centrosomal ␥- Recombinant XNEDD1-C binds to microtubules along their length. (a) GST-XNEDD1-C was allowed to bind to microtubules, and the samples were processed for immunofluorescence with antibodies against GST (XNEDD1-C; green in overlay) or ␣-tubulin (red in overlay). Scale bar, 5 m. (b) Microtubule-pelleting assay using a constant amount of recombinant XNEDD-C and increasing amounts of taxol-stabilized microtubules. Supernatant (S) and pellet (P) of each sample were loaded on a gel and stained with Coomassie Blue. Only the part of the gel corresponding to XNEDD1 (upper panels) or tubulin (lower panels) are shown. (C) Recombinant XNEDD1-C bundles preformed microtubules. Recombinant XNEDD1-C (without the GST used to purify the protein) was incubated with taxolstabilized microtubules. An aliquot of the reaction mixture was spotted onto a microscope slide and viewed under the microscope. Microtubules appear highly bundled in the presence of 300 nM XNEDD1-C. Scale bar, 10 m. (D) XNEDD1 binds to kinetochore fibers.
(Upper panels) Immunofluorescence shows that XNEDD1-positive fibers end at kinetochores stained by Bub1. XNEDD1, green; Bub1, red; DNA, blue. Scale bar, 5 m. (Lower panels) Immuno-localization of XNEDD1 at kinetochore fibers. XNEDD1, green; kinetochore fibers, red; DNA, blue. Scale bar, 5 m. (E) Higher magnification image of the boxed area shown in D. NEDD1 organizes microtubules in Xenopus tubulin -often substantially less. One explanation for this reduction is the possibility that ␥-tubulin is depleted together with NEDD1. However, NEDD1 depletion had little effect on the overall levels of ␥-tubulin that remained in the extracts, making co-depletion a very unlikely possibility. It is possible that the small amount (10-20%) of endogenous NEDD1 that remains in the depleted extract can efficiently recruit ␥-tubulin to the centrosome. However, it is worth considering that only a small fraction of the total NEDD1 appears to be associated with ␥-tubulin in the egg extract, and the amount of NEDD1 available for ␥-tubulin recruitment is likely to be very small. Thus, we interpret our results to suggest that NEDD1 depletion results in only minor defects in ␥TuRC recruitment. Consistent with this, RNAi of Dgp71WD in flies does not inhibit ␥-tubulin recruitment to the centrosome (Vérollet et al., 2006).
Why does NEDD1 depletion affect sperm-centrosome assembly more strongly than centrosome reconstitution after salt extraction in our in vitro assay? Salt extraction of centrosomes leaves an insoluble 'centromatrix' onto which centrosomal proteins can assemble to reform a functional centrosome. Sperm centrioles, however, recruit virtually all of the pericentriolar-material proteins from the egg extract and thus, presumably, start from an earlier step in the assembly process. These observations raise the interesting possibility that NEDD1 may be involved in the early stages of centrosome assembly. Once past this point in the assembly process, NEDD1 is no longer required to recruit the ␥TuRC. Its depletion, however, fails to provide the scaffold onto which other centrosomal components can be attached. Two observations support this idea: (1) human NEDD1 is required for centriole assembly in human cells (Haren et al., 2006) and, (2) NEDD1 is associated with the sperm centrioles (this study). However, a potential role for NEDD1 in centriole assembly needs yet to be explored.
␥-Tubulin localizes to spindle microtubules in animal cells (Lajoie-Mazenc et al., 1994) and along microtubules in Xenopus egg extracts (Stearns and Kirschner, 1994). Although NEDD1 depletion had only minor effects on ␥-tubulin recruitment to centrosomes, it strongly affected overall microtubule organization. The most noticeable effects of NEDD1 disruption were a 70% reduction in the amount of ␥-tubulin associated with astral microtubules and, in the case of NEDD1 overexpression, highly bundled microtubules. Interestingly, knockdown of Dgp71WD or human NEDD1 in RNAi experiments was also reported to abolish the association of ␥-tubulin with spindle microtubules (Haren et al., 2006;Lüders et al., 2006), suggesting that the NEDD1 family of proteins might mediate an interaction between the ␥TuRC and microtubules. A role for NEDD1 in microtubule organization is further supported by our observations that microtubules assembled by addition of RanGTP or taxol to egg extracts are unable to organize into asters but instead form disorganized heaps of microtubules. Finally, our observation that NEDD1 binds to kinetochore fibers in mitotic cells suggests a mechanism by which the microtubulebundling activity of NEDD1 is involved in organizing kinetochore fibers. Alternatively, NEDD1 might recruit a subset of ␥-tubulin to kinetochore fibers to nucleate and organize kinetochore microtubules. In support of a role for NEDD1 in kinetochoremicrotubule organization, chromosomes were often misaligned on spindles assembled in NEDD1-depleted extracts.
Consistent with a direct interaction between NEDD1 and microtubules, we found that XNEDD1-C binds to taxol-stabilized microtubules with an apparent K d of 1.43 M. Association of ␥tubulin with astral and/or spindle microtubules as a mechanism to nucleate new microtubules has been described in fission yeast and plants. For example, the S. pombe protein mto2p recruits ␥-TuCs to the side of existing microtubules to nucleate new microtubules (Janson et al., 2005), and ␥-tubulin has been proposed to localize to existing microtubules to nucleate new ones in plant cells (Murata et al., 2005). However, it is also possible that ␥-tubulin localization to astral microtubules is an indirect effect of the microtubule bundling activity of NEDD1. In this scenario, ␥-tubulin localizes to the side of one microtubule because it is associated with the minus end of a second microtubule that is held close to the first one by the bundling activity of NEDD1. Although ␥TuRC-independent roles for Dgp71WD or human NEDD1 have not been described, our findings are consistent with the findings that Dgp71WD is required for ␥TuRC recruitment to spindle microtubules but not to centrosomes. This hypothesis is further supported by the observations by Lüders et al. that phosphorylation of human NEDD1 is required for ␥TuRC recruitment to spindle microtubules but not for centrosome recruitment (Lüders et al., 2006).
In summary, we draw four main conclusions from this work in Xenopus: (1) NEDD1 exists in a complex that is distinct from the ␥TuRC and includes only a small, if any, amount of ␥-tubulin.
(2) NEDD1 depletion has only minor effects on recruitment of ␥-tubulin to centrosomes. (3) NEDD1 binds directly to microtubules. (4) NEDD1 is required for the organization of microtubules. However, several major questions regarding NEDD1 remain. For example, it is not yet clear whether the centrosome-independent functions of NEDD1 involve ␥-tubulin, or who the binding partners of NEDD1 are. The interaction between NEDD1 and the ␥TuRC is entirely unexplored and it might be indirect. It will be interesting to see whether members of the NEDD1 family of proteins in different species serve conserved or distinct functions, and what role the WD repeats play in NEDD1 function.

Recombinant proteins
Xenopus NEDD1 was identified in database searches as the ortholog of Dgp71WD (Gunawardane et al., 2003) and human NEDD1 (Lüders et al., 2006;Haren et al., 2006). Full-length Xenopus NEDD1 cDNA was generated by PCR amplification of an expressed sequence tag (EST; IMAGE clone 6633797; purchased from Open Biosystems, Huntsville, AL). To generate a fusion protein with glutathione-Stransferase (GST), full-length Xenopus NEDD1 was subcloned directionally into pGEX-6P2 (Amersham Biosciences, Piscataway, NJ) using BamH1 and XhoI as cloning sites. The following primers (restriction sites are underlined) were used: 5Ј: ATGGATCCCAGGATAACATCAG; 3Ј: ATCTCGAGCTAAAAATTGGCCCG. XNEDD1-C, a fusion protein of GST and the C-terminus of Xenopus NEDD1 (amino acid residues 301-655) was generated using primers 5Ј: ATGGATCCCACAGCAACATTTAC; 3Ј: ATCTCGAGCTAAAAATTGGCCCG that introduced 5Ј BamHI and 3Ј XhoI sites to amplify the desired region of Xenopus NEDD1 by PCR and subcloned into the BamHI-XhoI sites of the pGEX-6P2 vector. Fusion protein XNEDD1-N, in which the GST was fused to the N-terminus of Xenopus NEDD1 (amino acid residues 1-300) was cloned using primers 5Ј: ATGGATCCCAGGATAACATCAG; 3Ј: ATCTCGAGCTATTAAACTGTATAC that introduced BamH1 and XhoI sites, respectively.
To generate a fusion protein with green-fluorescent-protein (GFP), full-length Xenopus NEDD1 was subcloned into the XhoI-BamHI-site of the pEGFPC3 vector (Clontech). A fusion protein in which GFP is fused to the C terminus of Xenopus NEDD1 (amino acid residues 301-655) was generated using primers (5Ј: ATCTCGAGCACAGC AACATTTAC; 3Ј: ATGGATCCCTAAAAATTGGCCCG) that introduced 5Ј XhoI and 3Ј BamHI restriction sites (underlined) to amplify the desired region of Xenopus NEDD1 by PCR and subcloned into the XhoI-BamHI sites of the pEGFPC3 vector. The GPF fusion protein with the N terminus of Xenopus NEDD1 (amino acid residues 1-300) was generated using a similar approach: using primers (5Ј: ATCTCGAGCAGGATAACATCAG; 3Ј: ATGGATCCTCCCTAT -TGAAACTGTATAC) that introduced 5Ј XhoI and 3Ј BamHI sites.
The following fusion proteins were expressed in E. coli strain C41(DE3) (Miroux and Walker, 1996) and purified by affinity chromatography using standard protocols for purification of GST-fusion proteins: GST-RanL43E (see Wilde and Zheng, 1999), GST-XNEDD1-FL, GST-XNEDD1-N and GST-XNEDD1-C (diagrammed in Fig.  2B). All purified recombinant proteins were concentrated to >10 mg/ml, dialyzed against XB (10 mM K-HEPES, 100 mM KCl, 1 mM MgCl 2 , 0.1 mM CaCl 2 , 50 mM sucrose, pH 7.6), flash-frozen in liquid nitrogen as small aliquots, and stored at -80°C for later use. Full-length XNEDD (aa 1-655), XNEDD-N (aa 1-300) or XNEDD-C (aa 301-655) were expressed as glutathione S-transferase (GST) fusion proteins and purified using GST agarose. For the experiments shown in Figs 2, 3, 7, 8A and 8C, the GST portion of the fusion protein was proteolytically removed prior to the addition of the purified protein to the reaction. The uncleaved protein was used for the experiments shown in Figs 5 and 8B.

Xenopus egg extract preparation
CSF-arrested Xenopus egg extracts were prepared as described (Murray, 1991) and were supplemented with ⌬90-cyclinB to arrest them in mitosis .

Cell culture
XLKWG cells (Martin et al., 1998) were cultured in R1640 medium (GIBCO) supplemented with 10% FBS and 2% penicillin and streptomycin at 32°C in a 5% CO 2 atmosphere. For immunofluorescence, cells were fixed in 4% formaldehyde (in PBS) for 20 minutes at room temperature and permeabilized in TBS with 0.1% Triton X-100 for 5 minutes.

Microtubule polymerization assays
Microtubule polymerization assays were performed as previously described (Oegema et al., 1999) with the following modifications: microtubule-assembly reactions (5 l) containing various concentrations of purified XNEDD1-C (0 mg/ml, 0.1 mg/ml, 0.2 mg/ml and 0.8 mg/ml) and 32.5 M bovine brain tubulin (supplemented with a small amount of rhodamine-labeled tubulin) were incubated at 30°C for 10 minutes, fixed with ten volumes of 1% glutaraldehyde in BRB80 at 23°C for 3 minutes, and diluted with 250 l of 70% glycerol in BRB80. A 3-l aliquot of the reaction was then mounted on a microscope slide and viewed in the microscope.

Covalent attachment of GST fusion proteins to activated Dynabeads
GST-fusion proteins or GST alone were covalently linked to Dynabeads (Dynal Biotech, Oslo, Norway) according to the manufacturer's instructions. Coupled beads (10 l) were then incubated with 100 l Xenopus egg extracts for 1 hour at 4°C under rotation. Beads were collected on a magnet and were washed three times with H100 (50 mM HEPES, 1 mM EGTA, 1 mM MgCl 2 , 100 mM NaCl, pH 7.6). Washed beads were resuspended in 100 l of XB buffer; 1 l of resuspended beads was then tested for microtubule-nucleation activity with 5 l of microtubule assembly reaction (see above). Alternatively, the washed beads were boiled in SDS-sample buffer and the eluate was analyzed by western blot on 10% SDS-PAGE gels.

Aster-and spindle-assembly assays
Xenopus egg extracts were supplemented with energy mix (7.5 mM phosphocreatine, 1 mM ATP, 100 M EGTA, 1 mM MgCl 2 ) and ⌬90-cyclinB, and incubated at room temperature for 30 minutes. To induce aster assembly, either RanL43E (1 mg/ml) or sperm chromatin (150 sperm per l) were added to extracts and incubated at room temperature for 15 minutes. To induce taxol asters, taxol was added to the extract to a final concentration of 0.1 M and the taxol-treated extract was incubated at room temperature for 30 minutes. To induce cycled asters and spindles, fresh mitotic extract was driven into interphase by addition of sperm chromatin and 0.3 mM Ca 2+ , and were then cycled back into mitosis by addition of an equal volume of mitotic extract. Cycled extracts were incubated at room temperature for 15 minutes to assemble asters, or 60 minutes for spindles. In all cases, extracts were supplemented with rhodaminelabeled tubulin (final concentration 0.2 mg/ml) to allow visualization of microtubule structures. To test the effects of recombinant proteins on aster assembly, the proteins were added to the reaction to a final concentration of 300 nM unless otherwise noted. In all cases, the reactions were diluted with 30% glycerol in BRB80 at the end of the incubation time. The reactions were then spun onto coverslips and the microtubules were fixed in -20°C methanol as described (Evans et al., 1985). For quantification of aster numbers, the number of sperm with asters per total number of sperm were counted in 50 random fields. For quantification of aster intensity, asters were photographed at the same exposure, an area was drawn around the aster and fluorescence intensity was measured using MetaMorph software. The average intensity of a background area measured near the aster was subtracted from the 'raw' aster intensity.
Journal of Cell Science 121 (5) Immunofluorescence and microscopy Asters or spindles induced in Xenopus egg extracts were spun onto coverslips, fixed with -20°C MeOH for 5 minutes, and processed for immunofluorescence as described (O'Brien et al., 2005). To visualize the kinetochore fiber staining of Xenopus NEDD1, XLKWG cells were incubated on ice for 10 minutes to depolymerize non-kinetochore microtubules. Cells were then fixed with 4% paraformaldehyde/0.2% Triton X-100 in TBS (150 mM NaCl, 20 mM Tris-HCl, pH 7.4) for 20 minutes at room temperature. Samples were first blocked in Abdil (TBS, 0.1% Triton X-100, 2% BSA, 0.1% sodium azide) followed by primary and secondary antibodies diluted in Abdil in a humidified chamber at room temperature as follows: rabbit anti-NEDD1 of Xenopus, 1:10,000; mouse anti-␣-tubulin (DM1␣), 1:1000, XenC rabbit anti-␥-tubulin, 1:1000; mouse anti-acetylated tubulin, 1:1000; rabbit anti-GST 1:500; chicken anti-Bub1 1:100; all secondary antibodies, 1:1000. DNA was visualized by incubating the coverslips in a solution of 200 g/ml bisbenzimide H33342 trihydrochloride fluorophore ('Hoechst dye'; Calbiochem) in water for 10 seconds, rinsing them in water, and mounting them in mounting medium (Sigma). Spindles and asters were photographed with a Photometrics CoolSnap HQ cooled CCD camera (Roper Scientific, Inc.) through a ϫ60/1.4 NA plan apo objective mounted on a Nikon Eclipse E800 fluorescence microscope equipped with MetaMorph software. Images were obtained using MetaMorph and processed using Adobe Photoshop software. Pictures of kinetochore fiber staining were deconvolved using the deconvolution function of MetaMorph.

Assays for ␥-tubulin recruitment
To study ␥-tubulin recruitment, sperm chromatin was incubated with Xenopus egg extract at room temperature for 5 minutes in the presence of nocodazole to prevent microtubule assembly, which allows visualization of ␥-tubulin recruitment to sperm centrioles. Samples were spun onto coverslips, fixed in -20°C MeOH, and doublelabeled with anti-acetylated tubulin and XenC to locate centrosome and ␥-tubulin. The amount of ␥-tubulin recruited to the centrosome was quantified by measuring the ␥-tubulin immunofluorescence intensity.

Centrosome complementation assay
The centrosome complementation assay was done essentially as described by Moritz et al. (Moritz et al., 1998). To quantify ␥-tubulin recruitment, immunofluorescence was carried out for acetylated-tubulin and ␥-tubulin. A circle slightly bigger than the centrosome was drawn around each centrosome, and the integrated intensity of ␥tubulin was measured. A circle of the same size was then drawn at the background to measure the background intensity. The intensity of ␥-tubulin was subtracted by the background intensity.

Sucrose-gradient sedimentation
Sucrose gradients (5-40%) were prepared as described in O'Brien et al. (O'Brien et al., 2005). Briefly, clarified egg extracts (33 l) were fractionated on parallel 2-ml-sucrose gradients in a 4 hour spin at 55,000 rpm in a Beckman TLS 55 rotor. Sixteen 125-l fractions were collected and 25 l of each fraction were loaded on a 10% SDS-PAGE gel. Proteins were visualized by western blotting.