Quantitative Analysis of the Human Spindle Phosphoproteome at

Aug 19, 2009 - formation, mitotic progression, and cytokinesis. In the recent past, mass spectrometry has been applied successfully to identify spindl...
0 downloads 0 Views 4MB Size
Quantitative Analysis of the Human Spindle Phosphoproteome at Distinct Mitotic Stages Rainer Malik,†,# Rene´ Lenobel,†,§,# Anna Santamaria,† Albert Ries,† Erich A. Nigg,†,‡ and Roman Ko ¨ rner*,† Max Planck Institute of Biochemistry, Department of Cell Biology, Am Klopferspitz 18, D-82152 Martinsried, Germany, and Biozentrum, Universita¨t Basel, Klingelbergstrasse 50/70, CH-4056 Basel, Switzerland Received April 27, 2009

During mitosis, phosphorylation of spindle associated proteins is a key regulatory mechanism for spindle formation, mitotic progression, and cytokinesis. In the recent past, mass spectrometry has been applied successfully to identify spindle proteomes and phosphoproteomes, but did not address their dynamics. Here, we present a quantitative comparison of spindle phosphoproteomes prepared from different mitotic stages. In total, we report the identification and SILAC based relative quantitation of 1940 unique phosphorylation sites and find that late mitosis (anaphase, telophase) is correlated with a drastic alteration in protein phosphorylation. Further statistical cluster analyses demonstrate a strong dependency of phosphorylation dynamics on kinase consensus patterns, thus, linking subgroups of identified phosphorylation sites to known key mitotic kinases. Surprisingly, we observed that during late mitosis strong dephosphorylation occurred on a significantly larger fraction of phospho-threonine than phospho-serine residues, suggesting a substrate preference of phosphatases for phospho-threonine at this stage. Taken together, our results constitute a large quantitative data resource of phosphorylation abundances at distinct mitotic stages and they provide insight into the systems properties of phosphorylation dynamics during mitosis. Keywords: mitosis • spindle • phosphorylation • kinase consensus • signaling

Introduction Error-free segregation of sister chromatids into the two daughter cells is essential to ensure genomic stability. Physically, this process is carried out by the mitotic spindle, a highly dynamic microtubule-based structure. After entry into mitosis, the major microtubule organizing centers in animal cells, the centrosomes, start to increase microtubule nucleation and move to opposite poles of the cell. Throughout prometaphase, microtubules emanating from centrosomes are captured by kinetochores, protein complexes assembled on centromeric chromosomal DNA. This eventually leads to the alignment of all chromosomes in a metaphase plate. Since proper bipolar attachment of chromosomes to spindle microtubules is essential for the correct segregation of chromosomes, this critical step is monitored by a signaling pathway, known as the spindle assembly checkpoint (SAC).1 This checkpoint is silenced only after all chromosomes have attached to the spindle in a bioriented fashion, resulting in the synchronous segregation of the sister chromatids during anaphase. Simultaneously, a * Corresponding author. Tel., +49 89 8578 3105; fax, +49 89 8578 3102; e-mail, [email protected]. † Max Planck Institute of Biochemistry. # These authors contributed equally to this work. § Present address: Palacky University & Institute of Experimental Botany, Laboratory of Growth Regulators, Slechtitelu 11, Olomouc 783 71, Czech Republic. ‡ Universita¨t Basel. 10.1021/pr9003773 CCC: $40.75

 2009 American Chemical Society

so-called central spindle is formed between the separating chromatids and the formation of a contractile ring initiates cytokinesis. Finally, in telophase, the chromosomes decondense and reassemble into nuclei, whereas remnants of the central spindle form the midbody, marking the site of abscission.2 Mitotic spindle dynamics are induced by numerous proteins, of which most are associated with kinetochores, centrosomes, or the microtubule-based core structure. In turn, the functions of many spindle proteins are regulated by protein phosphorylation at specific mitotic stages,3-6 underscoring the notion that phosphorylation plays a pivotal role in the control of mitotic progression. Prominent among the kinases that are crucial for spindle formation and functioning are Cyclin dependent kinase 1 (Cdk1), Polo-like kinase 1 (Plk1) and the Aurora kinases A and B.6 In addition to direct functional regulation of substrates, protein phosphorylation is also important for temporally targeting proteins to specific spindle structures. Plk1, for example, contains a C-terminally located polo-box domain (PBD), which acts as a phosphopeptide specific binding domain to target Plk1 to different locations in a temporally regulated manner.7 Thus, through the interconnection of regulatory kinases in time and space, highly complex phosphorylation networks are formed. Although our understanding of microtubule dynamics and spindle formation has significantly deepened over the past decade, the complexity and the dynamic behavior of the mitotic spindle continue to hamper the investigation of the structural Journal of Proteome Research 2009, 8, 4553–4563 4553 Published on Web 08/19/2009

research articles properties of this system. Mass spectrometry (MS) in combination with phosphopeptide enrichment methods8-13 has proven to be an excellent technique for the mapping of phosphoproteomes of cultured cells14-18 and purified subcellular structures,19-22 including the mitotic spindle.23 Furthermore, the current developments in quantitative MS18,24-29 allow for relative quantitation of protein and phosphorylation site abundance from different samples, thus, facilitating the investigation of dynamic changes. Quantitative phosphoproteome studies on HeLa cells arrested in S-phase and mitosis (prometaphase) have recently been published for cell lysates28 and purified kinomes,30 demonstrating a strong upregulation of most phosphorylation stoichiometries at the onset of mitosis. Here, we report the first comparative phosphoproteome analysis at distinct mitotic stages. To enrich for proteins involved in the regulation of mitosis, we performed this study on spindles purified from cultured human cells. In total, 1940 phosphorylation sites were quantified at late prometaphase, metaphase, and late mitosis (anaphase, telophase), respectively. Clustering of phosphorylation sites according to their dynamics revealed enrichment of Cdk1, Plk1, and Aurora kinase consensus motifs in specific clusters, thus, demonstrating kinase specific phosphorylation dynamics. Furthermore, we present the first in vivo evidence for a strong difference of serine versus threonine phosphorylation dynamics. In summary, the presented data set constitutes a comprehensive analysis of the changes in phosphorylation at specific mitotic stages and this repository in turn provides the basis for a systems modeling of the temporal regulation of mitotic progression.

Experimental Section Materials. Unless otherwise stated, chemicals were purchased from Sigma or Fluka. Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC) Media. Dulbecco’s Modified Eagle Medium (DMEM) high glucose medium deficient in amino acids arginine and lysine was supplemented with 5% dialyzed fetal calf serum (FCS), penicillin-streptomycin (100 IU/mL and 100 µg/mL, respectively), and either unlabeled L-arginine · HCl and L-lysine · HCl (SILAC light), L-arginine-U-13C6HCl (Cambridge Isotope Laboratories) and L-lysine-2H4 · 2HCl (Cambridge Isotope Laboratories) (SILAC medium), or L-arginine-U-13C615 N4 · HCl (Cambridge Isotope Laboratories) and L-lysine-U-13C615 N2 · HCl (Cambridge Isotope Laboratories) (SILAC heavy) at concentrations of 42 µg/mL (arginine) and 72 µg/mL (lysine). Media were kept at 37 °C before use. Cell Culture and Synchronization. HeLa S3 cells were grown at 37 °C in a humidified incubator with a 5% CO2 atmosphere. Cells were adapted to the appropriate SILAC medium for at least 5 passages to achieve complete incorporation of the isotopically labeled amino acids. For large-scale mitotic spindle isolation, each population of HeLa S3 labeled cells was propagated to 5 culture triple-flasks, with a total surface of 500 cm2. Cells were first presynchronized with thymidine (5 mM) for 20 h, then washed twice with PBS, and released from the thymidine block into drug-free, prewarmed appropriate SILAC medium. Seven hours after release, noscapine was added at a concentration of 25 µM. This was sufficient to block the cells in prometaphase with fairly normal bipolar spindles and almost all chromosomes aligned. SILAC light-labeled cells were harvested by shake-off 9 h after the addition of noscapine and pelleted by centrifugation (0 min time point). The rest of the 4554

Journal of Proteome Research • Vol. 8, No. 10, 2009

Malik et al. cells were centrifuged, washed twice with PBS, and released into drug-free, prewarmed appropriate SILAC medium, for 15 min (SILAC medium) and 60 min (SILAC heavy). Cells were then harvested by mitotic shake-off and pelleted by centrifugation at the respective time points. Finally, the three SILAC labeled cell populations were pooled and lysed as described previously.31 Optimization of the 15 and 60 min time points was achieved in prior pilot experiments during which progression of cells through mitosis was followed by immunofluorescence microscopy, after staining of DNA with 4′,6-diamidino2-phenylindole (DAPI) and microtubules with R-tubulin antibody, respectively (Supplementary Figure 1). Spindle Isolation. Taxol-stabilized mitotic spindles (including kinetochores and centrosomes) were isolated essentially as described previously,31 except that noscapine instead of nocodazole was used for the mitotic arrest (see above), and that okadaic acid (100 nM) was added to all buffers to prevent dephosphorylation of proteins by PP2A- and PP1-type protein phosphatases. In short, taxol (5 µg/mL) was added to prelysis and lysis buffers to stabilize spindles. Cells were then treated with DNases to partially remove chromosomes and latrunculin B to depolymerize the actin cytoskeleton. Finally, intermediate filaments were depolymerized in low-ionic-strength buffer and the mitotic spindles collected by centrifugation. The purity of isolated spindles was evaluated by differential interference contrast (DIC) light microscopy. 1D-Gel Electrophoresis and In-Gel Tryptic Digestion of Proteins. Glycine hydrochloride (pH 2) was added to the purified mitotic spindles to a final concentration of 0.1 M. The acidified spindles were further solubilized in an ultrasonic water bath for 30 s and stirred in a thermomixer (Eppendorf) at 1200 rpm for 10 min. The insoluble pellet was collected by centrifugation at 10 000g at 4 °C for 5 min and the solubilization step repeated once, before the supernatants containing solubilized spindle proteins were combined. Next, the protein concentration of the isolated soluble mitotic spindles was determined by a Bradford assay with bovine serum albumin (BSA) as a standard. At this stage, the isolated spindle proteins were split into two fractions of about 400 µg each for a replicate analysis (see below). For 1D-gel electrophoresis, a soluble spindle fraction was mixed with dithiothreitol (DTT) and NuPAGE LDS buffer solution (Invitrogen) according to the supplier’s recommendations, and separated on a NuPAGE Bis-Tris gel (4-12%) from which the spacers between the lanes had been removed to achieve maximal loading capacity. The separated proteins were stained with 0.5% Coomassie Brilliant Blue in 50% methanol/10% acetic acid and destained in 50% methanol/10% acetic acid. The gel was cut into 23 slices and in-gel digested using trypsin (Promega, sequencing-grade), essentially as described32 Titansphere Enrichment of Phosphopeptides. Phosphopeptides were enriched using TiO2 (Titansphere, GL Sciences) columns as described previously9,33 with minor modifications. Approximately 3 µL of Titansphere (GL Sciences, Japan) suspension (100 mg/mL resuspended in 80% ACN in Milli-Q water/2% TFA) was placed on top of homemade C8 (Empore, 3M) STAGE Tips34 in 200 µL GELoader (Eppendorf) pipet-tips. The columns were washed twice with 40 µL of LA1 solution (80% acetonitrile (ACN)/0.2% TFA and 300 mg/mL of lactic acid as a modifier as described in ref 35). Dried samples were reconstituted in LA2 solution (80% ACN/2% TFA and 300 mg/ mL of lactic acid as a modifier), loaded onto TiO2 columns, and slowly passed through twice. The TiO2 columns were

research articles

Dynamics of the Mitotic Spindle Phosphoproteome washed with the following solutions: 40 µL of LA1 solution, 40 µL of 80% ACN/0.2% TFA and 40 µL of water. Flow-through fractions and all solutions from the washing steps were collected and analyzed separately for the determination of protein ratios. Phosphorylated peptides bound on the TiO2 were eluted slowly with 60 µL of freshly prepared 0.6% ammonium hydroxide and 30 µL of 80% ACN/0.2% TFA. Phosphopeptide enriched eluates were immediately acidified with 12 µL of 5% formic acid (FA) and dried in a SpeedVac concentrator. The phosphopeptide samples and the flow-through samples were desalted and purified on C18 STAGE columns34 and resuspended in 5 µL of 0.5% FA for online nanoLC-Orbitrap analysis. Mass Spectrometry Analysis. Desalted Titansphere-eluates (enriched for phosphopeptides) and flow-through fractions (nonphosphorylated peptides) were analyzed by online C18 reversed-phase nanoscale liquid chromatography tandem mass spectrometry on a NanoAcquity UPLC system (Waters) connected to an Orbitrap (Thermo Electron) equipped with a nanoelectrospray ion source (Proxeon). The mass spectrometer was operated in data-dependent mode for software controlled switching between MS survey and MS/MS fragmentation. Desalted samples were loaded by the NanoAcquity autosampler directly onto a 13-cm pulled fused-silica capillary packed with ReproSil-Pur C18-AQ (Dr. Maisch GmbH) 3-µm reversedphase material at a flow of 350 nL/min for 30 min. This fritless capillary column had an internal diameter of 75 µm and a tip opening of 8 µm (NewObjective) and was prepared as described by Ishihama et al.36 Peptides were then separated by a stepwise 90 min gradient of 0-100% between buffer A (2% ACN, 0.5% FA) and buffer B (80% ACN, 0.5% FA) at a flow rate of 200 nL/ min. Orbitrap full scan MS spectra (from m/z 350-1500) were acquired by the Orbitrap at a resolution of 60 000 and the five most intensive ions were fragmented in the linear ion trap using collision-induced dissociation. Fragmented target ions selected for MS/MS were dynamically excluded for 60s. The total cycle time was approximately 1.5 s. Other Orbitrap parameters were spray voltage, 2.3 kV; no sheath and auxiliary gas flow; ion transfer tube temperature, 170 °C; normalized collision energy using wide band activation mode and multistage activation mode, on; collision energy for MS2, 35%; MS2 selection threshold, 250 counts; activation q, 0.25; activation time, 30 ms. Mass Spectrometry Data Analysis. MaxQuant (version 1.0.12.5)37 was used for identification and quantitation of phosphopeptides. A peptide false discovery rate (FDR) specification of 0.01 was used, whereas the precursor mass tolerance was set to 7 ppm, and refined during MaxQuant processing as described.37 To account for differences in protein levels at the three time points, all determined phosphopeptide ratios were normalized by the measured ratios of the corresponding proteins. For the determination of protein levels, at least two unmodified peptides were required. Otherwise, MaxQuant default parameters were used. Data were searched against IPI_human (version 3.48) using MASCOT (version 2.204). Results were filtered using Mascot Score g12 and phosphorylation site localization probabilities g75% (see Results) as criteria. The final false-positive rate for the data set was 0.96%. The Orbitrap raw data files and the MS/MS peaklists associated with this manuscript may be downloaded from the ProteomeCommons Tranche server (https://proteomecommons.org/ tranche/) using the passphrase Maliketal and the access hash: “A5Ywtla3FxIaMcT+PiNoX7mhgNz9bAidxvJ0y81qTkuO++lsBft4mYdo3PsaQAPe8q55Puol37zrAjSNaOqKedqfeFoAAAAAAABKww))”.

Figure 1. Scheme of the experimental setup of the SILAC experiment. To visualize chromosomes and spindles at the respective time points, immunofluorescence microscopy pictures are shown. DNA is stained in blue and R-tubulin in green, scale bars represent 10 µm.

Results and Discussion Experimental Strategy. To quantify phosphorylation sites in mitotic spindle proteins, an analytical SILAC38 strategy was combined with a previously established spindle isolation protocol31,39 (Figure 1). HeLa S3 cells were grown in media containing “heavy” (H), “medium” (M) and “light” (L) lysine and arginine and were arrested in mitosis using the opium alkaloid noscapine. Noscapine is a microtubule-interacting agent40 that binds to tubulin and alters its conformation, thus, reducing the dynamics of microtubule turnover.40 As a consequence,aSACdependentarrestisinducedinlateprometaphase. The arrested cells generally show a bipolar spindle arrangement with most chromosomes aligned at the metaphase plate, but a few remaining near the spindle poles. To enrich for specific mitotic stages, cells were released from a noscapine block and collected at three different time points: 0 min (L), 15 min (M) and 60 min (H). These represent late prometaphase (Figure 1, Supplementary Figure 1), a stage close to metaphase (Figure 1, Supplementary Figure 1), and late mitotic stages (i.e., anaphase and telophase, Figure 1, Supplementary Figure 1), respectively. Proteins extracted from the pooled isolated spindles were separated by one-dimensional gel electrophoresis, gels were cut into 23 slices, proteins were in-gel digested with trypsin, and extracted peptides were enriched for phosphopeptides using TiO2 based affinity purification. The TiO2-elution fraction (phosphorylated peptides) and the TiO2-flowthrough fraction (unphosphorylated peptides) were both analyzed using a Journal of Proteome Research • Vol. 8, No. 10, 2009 4555

research articles nanoLC-ESI-Orbitrap. To address the reproducibility of phosphorylation site identification and quantitation, half of the purified spindle protein sample was separated on a distinct 1D gel and subsequently analyzed as a technical replicate (see below). Synchronization. Proteome analysis of cells synchronized at late mitosis has, to the best of our knowledge, not previously been described. To ensure specific enrichment at the selected mitotic stages, an efficient synchronization protocol had to be established. In contrast to the commonly used drug nocodazole,41 noscapine does not depolymerize microtubules, but rather interferes with microtubule turnover.40 Therefore, the spindle structures remain organized upon noscapine block and can directly be isolated for phosphoproteome analysis. HeLa S3 cells treated with noscapine arrest efficiently in late prometaphase with nearly normal bipolar spindles (Figure 1, 0 min time point). In an initial experiment designed to assess cell synchrony after noscapine release, about 75% of the cells were found to be in late prometaphase or metaphase when analyzed 15 min after noscapine release (Supplementary Figure 1). Therefore, this time point was chosen to capture phosphorylation states immediately after release from the noscapine block. Clear morphological changes characterized by separated sister chromatids and predominant anaphase (21%) and telophase (75%) spindles were observed at the 60 min time point (Supplementary Figure 1 and Figure 1). Thus, noscapine allows efficient HeLa S3 cell synchronization at late mitotic stages, enabling the comparative analysis of specific changes in protein phosphorylation during mitotic progression. Identified Phosphorylation Sites and Phosphoproteins. Initially, 3887 phosphorylation sites on 1056 different proteins were identified using MaxQuant37 in combination with Mascot42 (see Experimental Section). Since individual protein levels on the isolated spindles may change significantly between the studied mitotic stages, all measured phosphopeptide ratios had to be normalized by the determined protein levels. Therefore, phosphopeptides for which unmodified peptides of the same protein could not be detected were discarded from the list. Next, we applied a Mascot score cutoff (g12) and only considered phosphopeptides for which the localization of the phosphorylated residue within the identified peptide could be assigned with high confidence, as determined by MaxQuant according to the previously used “class A site” definition.18 Next, all multiply phosphorylated peptides were removed for further analysis, as the dynamics of individual phosphorylation sites in multiply phosphorylated peptides cannot be confidently deduced. Since the fraction of multiply phosphorylated peptides was low (236 multiply phosphorylated sites in the original list, corresponding to 6% of the phosphorylated peptides and 31 multiply phosphorylated peptides after quality filtering, corresponding to 1.6% of the phosphorylated peptides), removing these peptides did not influence the overall results. The final high-quality data set (Supplementary Table 1) thus contains 1940 phosphorylation sites on 580 proteins at a peptide false discovery rate (FDR) of 0.96% and a mean phosphopeptide mass accuracy of 0.34 ppm. In total, 1632 (84.12%) phospho-serines, 302 phospho-threonines (15.57%) and 6 phospho-tyrosines (0.31%) were identified. Of the 1940 phosphorylation sites, 946 were not yet annotated in Uniprot (version 14.7), even though this database covers other recent mitotic phosphoproteome studies.23,28,30 Reproducibility of the Experiment. To check for the reproducibility of sample processing and SILAC based quantitation 4556

Journal of Proteome Research • Vol. 8, No. 10, 2009

Malik et al.

Figure 2. Grouping of the detected phosphorylation sites according to consensus motifs of key mitotic kinases. Sequences matching more than one consensus motif were assigned to the “Overlapping motifs” category.

by mass spectrometry, the experiment was carried out twice as a technical replicate (see Experimental Section). Next, we analyzed whether the phosphorylation sites were reproducibly quantified in the two experiments. We plotted the log2-ratios of the 60 versus 0 min time points against each other (Supplementary Figure 2), since the strongest variation was observed between the most divergent time points. A correlation coefficient of 1 represents perfect linear correlation. With the use of linear regression, we obtain a correlation coefficient (r-value) of 0.91 (r2-value: 0.83) which indicates a good reproducibility and is in agreement with other recent phosphoproteome studies.28,43 We further estimated the quantitation accuracy based on a SILAC mixture of HeLa cell from samples arrested in mitosis and found that 94.1% of the 1612 detected phosphorylation sites varied by less than 50% which is in line with our significance cutoffs of 0.66 and 1.5. The reproducibility of the biological sample preparation was mainly addressed by optimizing the robustness of synchronization and spindle morphology based on immunofluorescence microscopy (Figure 1, Supplementary Figure 1). However, based on data from three biological replicates on similarly synchronized samples (kinases enriched from HeLa cells in prometaphase, metaphase and telophase) (Dulla, K., et al., in preparation), we can estimate the biological reproducibility of the mitotic synchronization. In this experiment, we found that, for metaphase/prometaphase ratios and telophase/metaphase ratios, less than 4% and 7.5% of the phosphorylation sites varied by more than 50%, respectively. Importantly, many of the sites with variations above 50% followed the same trend (up- or down-regulation) so reflecting mainly varying synchronization efficiencies. Distribution of Phosphorylation Sites for Different Kinase Consensus Motifs. Since phosphorylation site dynamics are likely to be linked to the activity dynamics of individual kinases, we grouped the identified phosphorylation sites corresponding to the published consensus motifs of the key mitotic kinases Cdk1, Plk1, and Aurora-A/B44 (used motifs can be seen in Supplementary Table 2). Although kinase consensus motifs cannot be defined with ultimate precision and consensus motifs of different kinases may overlap (as is the case, for instance, for the consensus motifs of Aurora-A and Aurora-B), it is reasonable to assume that the consensus groups are enriched for substrates of the corresponding kinases. The distribution of phosphorylation consensus groups in our data set is depicted in Figure 2. When phosphorylation sites matched multiple consensus patterns, they were assigned to a group labeled “overlapping motifs” to avoid ambiguous classifications. Of the three analyzed consensus patterns, Cdk1 motifs con-

Dynamics of the Mitotic Spindle Phosphoproteome

research articles

Figure 4. Cumulative number of phosphorylation sites within the indicated kinase consensus groups as functions of their log2 ratios at the 60 min versus 0 min time point.

Figure 3. Global phosphorylation dynamics at (A) the 15 min time point and (B) the 60 min time point in comparison to the 0 min time point. Regulation levels above 50% were used as cutoffs for up- and downregulation. Phosphorylation sites with lower regulation levels were grouped in the “Same” category.

stituted the largest fraction (39%), as expected. This reflects the outstanding importance of Cdk1 as the “master kinase” of mitosis2 and is in agreement with previous mitotic phosphoproteome studies23,28 Phosphorylation sites, which did not match the consensus sequences of the above-mentioned kinases (labeled as “other motifs” in Figure 2) were analyzed using the Motif-X software tool.45 This program extracts statistically significant sequence patterns from large-scale data sets. The highest “motif score” and enrichment (“fold increase”, against IPI_human) was found for the pS/pT-D-X-E pattern, which may reflect the activity of Casein Kinase II (CKII)44 (Supplementary Figure 3). Phosphorylation Dynamics. In line with other recent SILAC based proteome studies,46-48 we considered quantitative changes above 50% as significant for regulated phosphorylation events. This corresponds to ratios of >1.5 (3/2) and