|Home | About | Journals | Submit | Contact Us | Français|
Centrosomes organize the bipolar mitotic spindle, and centrosomal defects cause chromosome instability. Protein phosphorylation modulates centrosome function, and we provide a comprehensive map of phosphorylation on intact yeast centrosomes (18 proteins). Mass spectrometry was used to identify 297 phosphorylation sites on centrosomes from different cell cycle stages. We observed different modes of phosphoregulation via specific protein kinases, phosphorylation site clustering, and conserved phosphorylated residues. Mutating all eight cyclin-dependent kinase (Cdk)–directed sites within the core component, Spc42, resulted in lethality and reduced centrosomal assembly. Alternatively, mutation of one conserved Cdk site within γ-tubulin (Tub4-S360D) caused mitotic delay and aberrant anaphase spindle elongation. Our work establishes the extent and complexity of this prominent posttranslational modification in centrosome biology and provides specific examples of phosphorylation control in centrosome function.
Phosphorylation is a reversible posttransla-tional modification that regulates most cellular processes, including the duplication of centrosomes to form the mitotic spindle, which functions in chromosome segregation. Protein kinases, such as cyclin-dependent kinase Cdk1 (Cdc28), Mps1, and Polo kinase (Cdc5) (1, 2), phosphorylate the centrosome, known in yeast as the spindle pole body (SPB; Fig. 1A). The 18 centrosomal proteins (10 have human homologs; Figs. 1B and and2)2) can be organized into five functional subcomplexes (1): the γ-tubulin complex (Tub4, Spc98, and Spc97), which nucleates mi-crotubules; the central core (Nud1, Spc42, Spc29, and Cnm67), which form the organelle’s structural foundation and precursor; the linker proteins connecting the core and γ-tubulin complexes; the membrane anchors; and the half-bridge components, where assembly begins. Previous studies examined phosphorylation of these components individually or within whole cell preparations (database S1, column 3). In contrast, we performed a comprehensive analysis of phosphorylation on enriched, intact centrosomes.
Centrosomal complexes were isolated from yeast cells by using a modified affinity purification (3) (fig. S1A), and copurifying proteins were analyzed by solution digest and mass spectrometry (MS). Phosphopeptides were enriched with a metal affinity column, processed by liquid chromatography tandem mass spectrometry (LC MS/MS) on an LTQ-Orbitrap mass spectrometer (Thermo Fisher Scientific, San Jose, CA), and identified with SEQUEST and DTASelect2 programs (4). DeBunker (5) and Ascore (6) programs were used to further validate phosphopeptides and phosphorylation assignments, respectively (database S2). Peptides from all 18 proteins were identified, with extensive peptide coverage for most proteins (Fig. 2). The centrosomal preparations (fig. S1B) were highly phosphorylated (fig. S1C), as observed by MS analysis (www.yeastrc.org/pdr/pages/front.jsp; search by protein name). In total, 297 phosphorylation events were mapped on 17 of the 18 yeast centrosomal proteins, of which 227 have not been previously reported. Among these are 49 potential Cdk1 sites [S/T-P, serine or threonine followed by proline, 5 are confirmed as Cdk sites (7)] and 22 tyrosines (Fig. 2). Combining data from this study (297 sites) and previous studies (29 sites), a total of 326 phosphorylation sites are now identified on the yeast centrosome (database S1).
Because phosphorylation regulates cell cycle events (8), including centrosome duplication and mitotic spindle formation, we explored phospho-rylation profile differences between centrosomes in cell cycle–arrested cells versus those growing asynchronously. Cells were arrested at an early step of centrosome duplication in late G1 phase with α-factor treatment or in mitosis after centrosome duplication and separation by depletion of the anaphase-promoting complex (APC) activator Cdc20 (fig. S2A). We detected 54 sites that were phosphorylated only in G1, 110 sites that were phosphorylated only in mitosis, and 68 sites phos-phorylated in both phases (Fig. 3, fig. S2B, and database S1). The latter 68 sites are likely to be constitutively phosphorylated (because 61 were also found in asynchronous preparations). Of all the subcomplexes, the central core contained the largest number of sites (46% of the total, Fig. 2) and also the highest percentage of all shared sites (72%) (fig. S2C). The 29 mitotic sites on Nud1 may affect its subsequent role in recruitment of cell cycle regulatory proteins required for mitotic exit (9). In contrast to the central core, the γ-tubulin complex, linkers, and half-bridge have few constitutive sites and a large number of sites phos-phorylated in mitosis.
Analysis of phosphorylated residues that are likely within binding sites or targets of specific kinases showed distinct cell cycle patterns. For example, the majority of sites within Polo (Cdc5) binding motifs (fig. S3A) were observed in mitosis when its activity peaks (10). Cell cycle–specific phosphorylation was also observed in 21 of 22 tyrosine sites. In contrast, over half (27 of 49) of the potential Cdk consensus sites were phosphorylated throughout the cell cycle. The constitutive Cdk phosphorylation in Spc42 appeared to be essential, because mutating the Cdk motifs to nonphosphorylatable residues [8 sites out of 32 total phosphorylation sites in Spc42 (fig. S3B)] was lethal. The lethality may result, in part, from the decrease in Spc42 assembly into the centrosome (Fig. 4A). Furthermore, phosphoryl-ation of these Cdk sites is critical for overall Spc42 phosphorylation, because phosphate incorporation decreased in the Spc42-8A mutant by 93% compared with the wild type (WT) (Fig. 4B).
Twelve centrosomal proteins are known substrates of Cdk1 or Mps1 (Fig. 2 and fig. S4A). We performed kinase reactions in vitro with either Cdk1 or Mps1 on our centrosome preparations and identified potential centrosomal substrates by in-gel digestion and MS analysis (fig. S4B). We observed kinase specificity on centrosomal substrates by distinct phosphorylation banding patterns, confirmed several substrates, and identified possible new substrates (Spc72 and Cnm67) for Mps1 and Cdk1, respectively.
Clustering of phosphorylation sites is a rare event that creates a charged region, which can affect protein interactions and contribute to structural integrity (11). A study of yeast Cdk phos-phorylation showed that a cluster of sites, rather than individual residues, can be evolutionarily conserved (7). Clustering was prominent within our centrosomal phosphoproteome, with 174 of the 297 mapped sites clustered in seven proteins [fig. S5, ≥5 sites within 50 residues (4)]. Twenty-nine of the 49 Cdk consensus sites were included within these clustered regions. The importance of phosphorylation site clustering is exemplified by analysis of the N terminus of Spc110, which interacts with Spc97 to stabilize the γ-tubulin complex (12, 13). Mutating even 2 out of the 18 phosphorylation sites in this region (fig. S5) is lethal when combined with spc97 mutations (14).
Individual residues that are functionally and structurally important are also conserved through evolution (15, 16). We therefore examined fungal orthologs of centrosomal proteins to determine evolutionary constraint values [measured by positional conservation (17, 18)] for the 297 sites and also regional conservation throughout the proteins [fig. S6; see Fig. 5, A and B, for γ-tubulin (Tub4)]. This analysis identified 59 highly constrained sites in 13 proteins (>80% conserved) and 14 fully conserved residues in 8 proteins (fig. S7, A and B), of which three sites, Tub4-Y445, Spc29-T18, and Spc29-T240 (19), are essential for cen-trosome function (20–22). Fourteen sites from this study are conserved in human centrosomal proteins (fig. S7C).
The phosphorylated residue, S360 (19), within γ-tubulin (Tub4) (fig. S8A) is fully conserved in fungi (Fig. 5B and fig. S6) and in humans (fig. S8B). γ-Tubulin is part of an evolutionarily conserved complex (γ-tubulin small complex, γ-tuSC) that nucleates microtubules for chromosome segregation. Phosphorylation of γ-tubulin has been shown to promote centrosome duplication and microtubule assembly (20, 23). Tub4-S360 lies within a Cdk motif and is phosphorylated by Cdk1 in vitro (fig. S8, C and D). This site is located within a surface loop available for protein-protein interactions, as viewed in the γ-tubulin crystal structure (Fig. 5C, star). Furthermore, structural analysis using cryo-electron microscopy of the yeast γ-tubulin complex places this loop directly between Spc98 and Spc97 (13). Therefore, we mutated S360 to either a nonphosphoryl-atable alanine (A) or an aspartic/glutamic acid (D or E) to mimic constitutive phosphorylation. The tub4-S360A allele did not affect growth; however, tub4-S360D and tub4-S360E caused growth defects at 25°C and mitotic arrest resulting in cell death upon shift to a higher temperature (37°C) (fig. S9, A to C). Also, tub4-S360D was lethal in combination with mutations in SPC98 (spc98-2), by deletion of the spindle checkpoint gene MAD2 (mad2Δ) that allows for correction of mitotic defects, and by deletion of the EB1 homolog BIM1 (bim1Δ), which is involved in microtubule dynamics (fig. S9A).
These genetic interactions suggested that tub4-S360D cells had defects in mitotic spindle assembly, which we analyzed by immunofluo-rescence microscopy and live cell imaging. At 25°C the majority (66%) of spindles in tub4-S360D large-budded mitotic cells had not extended past metaphase length [~1.5 μm (24)], whereas WT cells (91%) had normal elongated anaphase spindles [6 to 10 μm (24)] (Fig. 5D, 25°C). This phenotype was exacerbated at 37°C, with 98% of tub4-S360D cells containing either adjacent or unresolvable spindle poles (54%) or metaphase-length spindles (44%) (Fig. 5D), compared with WT cells (89% normal anaphase spindles). Live cell analysis of microtubules in tub4-S360D cells grown at 25°C revealed that spindles persisted longer in the fast phase (25) of anaphase spindle elongation (Fig. 5E, green lines), resulting in a transition to the slow phase (25) with longer anaphase spindles [6.6 μm ± 1.5 (SEM)] than WT cells [4.1 μm ± 0.4 (SEM); P < 0.001] (Fig. 5E, dashed green lines). In addition, large spindle length fluctuations were observed in tub4-S360D cells before anaphase (Fig. 5E, black lines; quantified in fig. S10) and in the slow phase of anaphase (Fig. 5E, red lines). Thus, phosphorylation of a single Cdk site in γ-tubulin appears to contribute to proper dynamics of ana-phase spindle microtubules.
Phosphoregulation of the centrosome is likely to be conserved, not only with respect to the protein kinases but also through specific residues in the respective human homologs. Illustrated by our analysis of γ-tubulin and Spc42, the conserved residues and phosphorylation patterns in yeast will be useful tools for studying the human cen-trosome, a much larger (>100 proteins) and more complicated microtubule organizing center.
We thank C. Pearson, A. Stemm-Wolf, and T. Su for comments on the manuscript; T. Davis and the Winey lab for helpful discussions; E. O’Toole for the electron micrograph; E. Nazarova for assisting with spindle dynamics analysis; M. Riffle for Yeast Resource Center assistance; and D. D’amours for the Cdc28 purification protocol. This work was supported by NIH U54 RR022220 and R01 GM062427 (M.P.R.), NIH R01 HG003039 (A.S.), NIH P41 RR011823 ( J.R.Y. III, T.N. Davis, principal investigator), CIHR MOP-64404 ( J.V.), and NIH GM51312 (M.W.). J.M.K. was supported by NIH F32 GM086038, and E.P.H. was supported by NIH T32 GM008759. Mass spectrometry data are provided at the Yeast Resource Center (www.yeastrc.org/pdr/pages/front.jsp).
Materials and Methods