During cell division, microtubules organize a bipolar spindle to drive accurate chromosome segregation to daughter cells. Microtubules are nucleated by the γ-TuRC, a γ-tubulin complex that acts as a template for microtubules with 13 protofilaments. Cells lacking γ-TuRC core components do nucleate microtubules; however, these polymers fail to form bipolar spindles. NEDD1 is a γ-TuRC-interacting protein whose depletion, although not affecting γ-TuRC stability, causes spindle defects similar to the inhibition of its core subunits, including γ-tubulin. Several residues of NEDD1 are phosphorylated in mitosis. However, previously identified phosphorylation sites only partially regulate NEDD1 function, as NEDD1 depletion has a much stronger phenotype than mutation of these residues. Using mass spectrometry, we have identified multiple novel phosphorylated sites in the serine (S)557–S574 region of NEDD1, close to its γ-tubulin-binding domain. Serine to alanine mutations in S565–S574 inhibit the binding of NEDD1 to γ-tubulin and perturb NEDD1 mitotic function, yielding microtubule organization defects equivalent to those observed in NEDD1-depleted cells. Interestingly, additional mutations in the S557–T560 region restore the capacity of NEDD1 to bind γ-tubulin and promote bipolar spindle assembly. All together, our data suggest that the NEDD1/γ-tubulin interaction is finely tuned by multiple phosphorylation events in the S557–S574 region and is critical for spindle assembly. We also found that CEP192, a centrosomal protein similarly required for spindle formation, associates with NEDD1 and modulates its mitotic phosphorylation. Thus CEP192 may regulate spindle assembly by modulating NEDD1 function.
The fidelity of chromosome segregation during cell division relies on the formation of a microtubule-based bipolar spindle (O'Connell and Khodjakov, 2007). Microtubules are polar filaments nucleated by the γ-TuRC, a ring-shaped complex formed by γ-tubulin associated with GCP-2, GCP-3, GCP-4, GCP-5 and GCP-6 (Kollman et al., 2011). NEDD1 is a γ-TuRC-interacting protein that binds directly to γ-tubulin (Manning et al., 2010). Although NEDD1 is not necessary for the stability of the γ-TuRC, its depletion causes spindle assembly defects that resemble the depletion of core γ-TuRC components, where microtubules fail to organize into bipolar spindles. Analyses of mitotic and meiotic cells confirm the critical role of NEDD1 in spindle formation in human, Xenopus and Drosophila (Haren et al., 2006; Liu and Wiese, 2008; Lüders et al., 2006; Ma et al., 2010; Vérollet et al., 2006). NEDD1 also associates with the Augmin complex to allow γ-tubulin recruitment to spindle microtubules, thereby promoting microtubule amplification (Goshima et al., 2008; Lawo et al., 2009). The NEDD1/Augmin interaction depends on NEDD1 phosphorylation on S411 (Johmura et al., 2011; Lüders et al., 2006; Uehara et al., 2009). Phosphorylation on T550 creates a priming site for PLK1, which further phosphorylates NEDD1 to regulate microtubule assembly in the vicinity of chromosomes (Zhang et al., 2009).
Similar to NEDD1 and γ-tubulin, the centrosome protein CEP192, plays a critical role in centrosome biogenesis and spindle assembly. Depletion of CEP192 causes defects in centriole duplication and pericentriolar material (PCM) recruitment, thereby interfering with spindle pole organization (Fig. 1A) (Gomez-Ferreria et al., 2007; Gomez-Ferreria and Sharp, 2008; Zhu et al., 2008). Additionally, microtubules nucleated near the chromosomes fail to reorganize into robust bipolar spindles. CEP192 interacts with Aurora kinase A and regulates its activation in Xenopus egg extracts (Joukov et al., 2010). However, the mechanisms underlying the role of CEP192 in spindle formation in human cells remain poorly understood.
Here we show that CEP192 interacts with NEDD1 and regulates its mitotic phosphorylation, which suggests that CEP192 may control spindle assembly by modulating NEDD1 activity. Phosphorylation site mapping of mitotic NEDD1 reveals that multiple residues in the serine (S)557–S574 region are phosphorylated in vivo. This phosphorylation regulates the NEDD1/γ-tubulin interaction and is required for bipolar spindle assembly.
Results and Discussion
CEP192 interacts with NEDD1 and modulates its mitotic phosphorylation
To unravel the molecular mechanisms underlying CEP192 function in spindle assembly, we performed affinity purification followed by mass spectrometry (AP-MS). We found that CEP192 interacts with NEDD1, which was of particular interest since depletion of either protein yields similar defects in mitotic microtubule organization (Fig. 1A; supplementary material Fig. S1) (Haren et al., 2006; Lüders et al., 2006). Further analyses showed that this interaction depends on the N-terminal region of CEP192 (Fig. 1B).
In mitosis, NEDD1 displays a complex electrophoretic pattern where slower migrating bands correspond to differentially phosphorylated forms (Fig. 1C, bands b and c) (Haren et al., 2009; Johmura et al., 2011). Interestingly, NEDD1 mitotic phosphorylation is altered in CEP192-depleted cells, as levels of the hyperphosphorylated band are consistently reduced (Fig. 1C, band c). These data suggest that CEP192 may regulate spindle assembly by modulating NEDD1 phosphorylation in mitosis.
Mapping of NEDD1 phosphorylation sites in mitosis
We then mapped NEDD1 phosphorylated residues in mitotic control and CEP192 depleted cells by mass spectrometry (MS; Fig. 1D, upper panel; see also supplementary material Fig. S2; Table S1). This analysis identified novel NEDD1 phosphorylation sites and corroborated some previously described ones (Fig. 1D, lower panel) (Johmura et al., 2011; Lüders et al., 2006; Santamaria et al., 2011; Zhang et al., 2009).
A quantitative analysis of the spectra was performed using the MaxQuant software, which monitors abundance of phosphopeptides by determining their ion intensity (supplementary material Fig. S3; Tables S2, S3). Our results suggest that double phosphorylation of the peptide 503–527 and phosphorylation at S468, S586, S460, S325 and S332 are modulated by CEP192 (peptides B–F in supplementary material Fig. S3). The abundance of the double phosphorylated peptide 555–570 (A) decreases to 15% after CEP192 depletion. However, the low absolute ion intensity of this peptide does not definitively support that this double phosphorylation is in fact CEP192 dependent.
NEDD1 phosphorylation regulates spindle assembly
To investigate the role of global NEDD1 phosphorylation in spindle assembly, we initially generated the mutant NEDD1-A23, where the majority of phosphorylation sites identified in this or previous studies were replaced with alanine (Ala; Fig. 1D, underlined residues). We established a rescue assay where cells are depleted of endogenous NEDD1 using RNA interference (RNAi), and then transfected with GFP alone or RNAi-resistant NEDD1, either wild-type or mutant, fused to GFP (Fig. 2A). In GFP transfected cells, 90% of spindles are disorganized with dispersed PCM or small, while most of the cells transfected with wild-type NEDD1 have bipolar spindles (Fig. 2B,C). Cells expressing a mutant where Ala substitutes S411 and T550 (A411-550) show fragmented centrosomes and, as previously reported, defects in microtubule nucleation within the spindle, resembling depletion of Augmin subunits (HAUS) (Lüders et al., 2006; Zhang et al., 2009). This ‘HAUS-like’ phenotype is caused by the inability of this mutant to associate with the Augmin complex (Lawo et al., 2009; Uehara et al., 2009). By contrast, cells expressing the NEDD1-A23 mutant have more drastic spindle defects where 70% are disorganized or abnormally small with fragmented centrosomes.
We next analyzed the role of these residues in regulating the well-established oligomerization properties of NEDD1 and its ability to bind γ-tubulin (Haren et al., 2006; Lüders et al., 2006; Manning et al., 2010). Immunoprecipitation assays show that both GFP–NEDD1 and A411-550 mutant interact with γ-tubulin and endogenous NEDD1; however, mutations in NEDD1-A23 mutant clearly affect these interactions (Fig. 2D). Therefore, the additional substitutions in NEDD1-A23 impair NEDD1 function and ability to bind γ-tubulin to a greater extent than mutation of the previously characterized sites S411 and T550 (Zhang et al., 2009).
Phosphorylation in the S557–S574 region regulates spindle assembly
We then determined the function of specific NEDD1 phosphorylation sites, being initially interested in those whose phosphorylation is reduced in CEP192-depleted cells (supplementary material Fig. S3, peptides B–F). Mutation of S460, S468, S325, S332, S516 or S586 to Ala has no significant effect on the ability of NEDD1 to rescue bipolar spindle assembly in cells depleted of endogenous NEDD1 (supplementary material Fig. S4, and data not shown). Similarly, mutation of S493 and S516 (peptides L and M), whose phosphorylation is higher after CEP192 depletion, to aspartic and glutamic acid respectively, is compatible with NEDD1 activity (data not shown). However, phosphorylation of these sites could control other aspects of NEDD1 function, which may therefore be modulated by CEP192. Additionally, CEP192 may regulate microtubule organization by modulating NEDD1 phosphorylation at yet unidentified residues.
Interestingly, our MS analysis shows that the region [amino acids (aa)] 555–579 proximal to the γ-tubulin-binding domain of NEDD1 (aa 599–660) is phosphorylated on multiple residues in vivo (Fig. 3A; supplementary material Fig. S2B, Fig. S3A, peptides G–H). While the tryptic peptide 571–579 includes a single phosphorylation site at S574, the peptide 555–570 includes six Ser/Thr residues. Several spectra identify single phosphorylations at residues S565, S566 and S568 and a lower number shows phosphorylation at S557, S558 and T560; indeed phosphorylation of S557 and S558 was previously reported (Santamaria et al., 2011). We also detect double phosphorylated forms of this peptide, although with a very low intensity (supplementary material Fig. S3A, peptide A). Sequence alignment of this region shows that these phosphoresidues are conserved in different species (Fig. 3A).
In order to study the functional relevance of this region we started by mutating stretches of phosphorylated residues located in close proximity, hypothesizing that in hyperphosphorylated proteins, like NEDD1, functional regulation may depend on multiple phosphorylations located in a particular region rather than on unique sites. Using the RNAi-rescue system described earlier, we observed that Ser to Ala mutation in the S565–S574 region abrogates NEDD1 function, as 80% of spindles are either small or disorganized [Fig. 3B,C, mutant A4 (A565-566-568-574)]. Single substitution to Ala of S565, S566, S568 or S574 does not significantly interfere with NEDD1 function (supplementary material Fig. S5A), while mutation of S565-S566-S568 (mutant A3.b) has only a minor effect (Fig. 3C). Ser to Glu mutations in the S565–S574 region are compatible with NEDD1 function in mitotic spindle assembly, further supporting that these residues can be phosphorylated in vivo [Fig. 3D, phosphomimetic mutant E4 (E565-566-568-574)]. Interestingly, although mutation to Ala of S557-S558-T560 in the wild-type protein does not affect NEDD1 activity [mutant A3.a (A557-558-560)], these mutations in the A4 (A565-566-568-574) mutant background [mutant A7 (A557-558-560-565-566-568-574)] restore NEDD1 function (Fig. 3A,C). Further mutational analysis of the region S557–T560 revealed that mutation to Ala of T560 individually or the combination of S557-S558 partially restores A4 (A565-566-568-574) function (supplementary material Fig. S6A, mutants A560+A4 and A557-558+A4). This suggests that phosphorylation in the S557–T560 region has a negative effect on NEDD1 function: in mutant A3.a (A557-558-560), phosphorylation of the S565–S574 region may promote NEDD1 activity, while in mutant A4 (A565-566-568-574), phosphorylation of the S557–T560 region may inhibit NEDD1 function. In wild-type NEDD1, phosphorylation at S565–S574 would counter the negative effect of phosphorylation at S557–T560.
Interestingly, the drastic spindle defects observed with the mutant A4 (A565-566-568-574) are comparable to those observed when S636-Y637, in the γ-tubulin-binding domain of NEDD1, are replaced with Ala (Fig. 3B,C, mutant A636-637). Mutation of these two residues is reported to inhibit NEDD1/γ-tubulin interaction (Manning et al., 2010). We notice that mutants A4 (A565-566-568-574) and A636-637 show a weak localization to the centrosome (Fig. 3B,E). Consistently, these mutants do not rescue the localization of γ-tubulin to centrosomes upon NEDD1 depletion (supplementary material Fig. S7).
Phosphorylation in the S557–S574 region regulates γ-tubulin binding
Analysis of the capacity of these mutants to bind γ-tubulin and endogenous NEDD1 show that Ser to Ala mutation of the S565–S574 region [mutant A4 (A565-566-568-574)] inhibits these interactions. Single mutation to Ala of S565, S566 or S574 does not have a significant effect on γ-tubulin binding, and only a partial defect is observed for S568 (supplementary material Fig. S5B,C). Mutation of S557-S558-T560 in the A4 (A565-566-568-574) mutant background [A7 (A557-558-560-565-566-568-574)] partially rescues NEDD1/γ-tubulin interaction, which could explain the ability of mutant A7 to assemble bipolar spindles (Fig. 4A, Fig. 3C). Consistently with its capacity to form bipolar spindles, the phosphomimetic mutant E4 (E565-566-568-574) interacts with γ-tubulin and endogenous NEDD1 (Fig. 3D, Fig. 4B). As previously described, mutant A636-637 still forms oligomers but does not interact with γ-tubulin (Fig. 4A) (Manning et al., 2010). Similarly, the mutant E636-637, which is unable to rescue the formation of bipolar spindles, does not interact with γ-tubulin (Fig. 3D, Fig. 4B).
Mutant A7 (A557-558-560-565-566-568-574), although unable to interact with endogenous NEDD1, forms bipolar spindles, suggesting that NEDD1 oligomerization is not required for spindle assembly (Fig. 4A, Fig. 3C). Nevertheless, oligomerization is predicted to be a common feature in more than 35% of the proteins in the cell to better support function (Ali and Imperiali, 2005). Therefore, it is possible that in a more physiological condition, NEDD1 oligomerization positively contributes to spindle assembly.
Mutation to Ala of S565-S566-S568 [A3.b (A565-566-568)], although shows a minor effect in the RNAi-resistant rescue assays, clearly affects γ-tubulin binding (Fig. 3C, Fig. 4A). This mutant still retains some capacity to bind γ-tubulin, as compared to mutant A4 (A565-566-568-574) and A636-637 (Fig. 4A; supplementary material Fig. S8), which do not rescue bipolar spindle assembly. Thus, it is possible that A3.b (A565-566-568) residual capacity to bind γ-tubulin is enough in the appropriate cellular context to form bipolar spindles. Additionally, it is conceivable that the ability of A3.b (A565-566-568) to rescue spindle assembly, in contrast to A4 (A565-566-568-574), is due to its partial capacity to bind endogenous NEDD1. In rescue assays, depletion of endogenous NEDD1 is unlikely to be complete. In this scenario, A3.b (A565-566-568), by interacting with the remnant endogenous NEDD1, could form chimeric oligomers wild-type/mutant NEDD1 that bind γ-tubulin and promote bipolar spindle assembly.
Taken together, these data suggest that the critical function of NEDD1 in spindle assembly depends on its ability to bind γ-tubulin and that this interaction is regulated by phosphorylation in multiple residues at S557–S574 region. The fact that mutations at S557–T560 and S565–S574 have different effects, suggests that NEDD1 activity may be finely tuned through multiple phosphorylation events in the S557–S574 region. Furthermore, a recent proteomic screen reports NEDD1 ubiquitylation at K570 (Kim et al., 2011), thus a complex interplay between ubiquitylation and phosphorylation in this region may regulate NEDD1/γ-tubulin interaction.
We propose a model where NEDD1 phosphorylation in the region S565–S574, close to the γ-tubulin-binding domain, exposes this domain for interaction with γ-tubulin. This binding may activate the γ-TuRC to nucleate microtubules capable of assembling a bipolar spindle (Fig. 4C). Structural analyses of the γ-TuRC suggest that this complex is assembled in an ‘off’ state that requires activation to efficiently nucleate microtubules (Guillet et al., 2011; Kollman et al., 2011). In vivo, microtubules can be generated in absence of NEDD1 and γ-tubulin (Gomez-Ferreria et al., 2007; Haren et al., 2006); however, why these microtubules are unable to build a bipolar spindle is unclear. It is possible that they are uncapped and therefore highly prone to depolymerization (Wiese and Zheng, 2000). Moreover, they may lack the 13-fold symmetry established when the γ-TuRC acts as a template for nucleation (Kollman et al., 2011). In this scenario, MAPs (microtubule associated proteins) and molecular motors (dynein and kinesins) could be unable to crosslink and slide these unstable and/or asymmetric polymers to organize a functional bipolar spindle (Walczak and Heald, 2008). By contrast, phosphorylation at S411 would only regulate the Augmin-dependent microtubule amplification pathway (Fig. 4C) (Goshima et al., 2008; Lawo et al., 2009; Uehara et al., 2009). Finally, we notice that this model could explain the role of NEDD1 in spindle assembly in systems where NEDD1 does not determine γ-tubulin recruitment to centrosomes, as in meiotic cells, Xenopus and Drosophila (Liu and Wiese, 2008; Ma et al., 2010; Vérollet et al., 2006). We cannot discard, however, that NEDD1 regulates spindle assembly by yet unidentified γ-tubulin independent pathways.
Given the phenotypic similarities between CEP192 depletion and Ser to Ala mutations in the S565–S574 region, we expected that phosphorylation at these sites depends on CEP192. However, our semi-quantitative phosphoproteomic data suggest that CEP192 has, at best, a modulating influence on the phosphorylation of this region (supplementary material Fig. S3, peptides A, G and H). CEP192 may still positively regulate the NEDD1/γ-tubulin interaction by favoring phosphorylation towards the S565–S574 instead of S557–T560 region. Therefore, more detailed phosphoproteomic studies and the use of phospho-specific antibodies will be required to test this hypothesis. Nevertheless, the physical interaction we describe between NEDD1 and CEP192, the similar phenotypes observed upon depletion of either protein, the change in the electrophoretic motility of NEDD1 upon CEP192 depletion and the different phosphorylation status of NEDD1 in absence of CEP192 together provide overlapping lines of evidence that suggest an interplay between CEP192 and NEDD1 phosphorylation during mitotic spindle assembly.
Although important questions regarding the mechanism underlying the role of NEDD1 and γ-tubulin in mitotic spindle assembly remain to be answered, we now show that the interaction between these two proteins regulates this process and can be spatially and temporally controlled by phosphorylation in the S557–S574 region. Additionally, our data suggest that CEP192 may exert its critical role in bipolar spindle assembly by regulating NEDD1 phosphorylation.
Materials and Methods
CEP192 cDNA [1941 amino acids (Gomez-Ferreria et al., 2007)] or the different deletion mutants were cloned into pcDNA3 fused with FLAG. NEDD1 cDNA MGC:26881 (Open Biosystems, BC027605) was cloned into pcDNA3 fused to GFP. The cDNA sequence encodes a 667-amino-acid protein which corresponds to the long isoform of NEDD1. The mutated phosphorylated residues are named according to their position in the short isoform of NEDD1 (660 aa).
EsiRNA (endoribonuclease-prepared siRNA) was generated as described previously (Kittler et al., 2005). The targeted regions are encompassed by the following primers: CEP192 (NM_032142): 5′-TTTTCAAGGGCTAGTATGTCTGA-3′, 5′-GGATGTTATTCTGGGGTTCCT-3′; NEDD1 3′UTR (NM_001135175): 5′-TAAAAATTGTACAGTATGTCATCTACCCAATAG-3′, 5′-CAAAAATTGTCTTATCAAATGTACAATAAATAATA-3′; non-targeting esiRNA (AY_015988, luciferase): 5′-TGGTTTGGTTGTTGATGGAA-3′, 5′-GTGCCTGGTGAAACTTGGTT-3′.
NEDD1 phosphorylation sites mapping
A clonal cell line expressing FLAG–NEDD1 was generated with the Flp-In T-REx system (Invitrogen) in U2OS cells following the protocol from the manufacturer (Malecki et al., 2006). Cells were synchronized in mitosis (supplementary material Fig. S2A). Lysis was performed using a buffer with 10 mM Tris pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 10% glycerol, 0.1% SDS, 0.5% deoxycholate, 10 mM NaF, 50 mM β-glycerophosphate, 5 nM okadaic acid, 5 nM calyculin A, 1 mM DTT and protease inhibitors. FLAG–NEDD1 was immunoprecipitated with anti-FLAG M2 magnetic beads (Sigma, M8823) and digested with trypsin. Phosphopeptide enrichment was performed using Ga (III) resin (SwellGel Gallium-Chelated Disc; Pierce, 89853). Peptides were loaded on Zorbax C18 (Agilent ZorbaxSB, 3.5 µm) and analyzed by LC-MS/MS using a ThermoFinnigan Orbitrap. Files were searched with Mascot v2.2 against RefSeqV42. For the quantitative analysis of the data, the spectra corresponding to three experiments for control and CEP192 esiRNA-treated cells were analyzed with MaxQuant software v188.8.131.52 (Cox and Mann, 2008). False discovery rate was set to 1% and a retention time alignment window of 5 min was utilized. MS/MS tolerance was set to 0.5 Da. Ion intensities for oxidated, deamidated and mis-cleaved peptides were summed. The values shown in supplementary material Fig. S3A correspond to the raw ion intensities arbitrarily divided by 106.
Microscopy and automatic quantification of signal intensities
Three-dimensional images were acquired on a DeltaVision Core System (Applied Precision) equipped with an IX71 microscope (Olympus), a CCD camera (CoolSNAP HQ2 1024×1024; Roper Scientific) and 60×/1.42 NA objective (Olympus). Z-stacks (0.4 µm apart) were collected, deconvolved using the SoftWorx v4.0 (Applied Precision) and shown as maximum intensity projections. Automated analysis of fluorescence intensities was performed on 12-bit TIFF images using Acapella v2.18 (Perkin Elmer). Cellular and centrosomal masks were detected in the CEP192 channel using an adaptive threshold to specifically outline cytosolic and centrosomal regions. Fluorescence intensity of GFP–NEDD1 and γ-tubulin was analyzed using the detected masks.
HeLa cells were transfected with 0.4 µg of esiRNA targeting NEDD1 3′UTR. 24 h later, 1 µg of plasmids expressing wild-type NEDD1 or the different mutants were transfected for 36 h. For immunofluorescence, cells were processed as previously described (Lawo et al., 2009). For the spindle phenotypes quantification we show the average of at least three experiments where we counted 50 cells per condition. Error bars refer to standard deviation (s.d.).
Immunoprecipitation and western blotting
HEK293 cells were transfected with FLAG- or GFP-fused proteins for 36 h. For immunoprecipitation we used antibodies against FLAG (Sigma, A2220), GFP (kindly provided by D. Drechsel) and NEDD1 (Abcam, ab57336). For blotting we used antibodies against FLAG (Sigma, F7425), NEDD1, γ-tubulin (Sigma, T6557), AURKA (Abcam, ab13824), CEP192 (Zhu et al., 2008) or α-tubulin (Sigma, T9026).
We thank C. Yeh, S. Lawo and J. Goncalves for critical reading of the manuscript, and C. Holley and A. Tagliaferro for esiRNA production.
This work was funded by the Canadian Cancer Society [grant number 019562 to L.P.]; Fundacion Caja Madrid [to M.G.]. L.P. holds a Canada Research Chair in Centrosome Biogenesis and Function.