Research ResourceMitosis

Mitotic phosphotyrosine network analysis reveals that tyrosine phosphorylation regulates Polo-like kinase 1 (PLK1)

See allHide authors and affiliations

Sci. Signal.  13 Dec 2016:
Vol. 9, Issue 458, pp. rs14
DOI: 10.1126/scisignal.aah3525

Tyrosine phosphorylation controls mitosis after all

The cell cycle is a carefully controlled process in which serine/threonine kinases play a large role. Abnormal progression or attenuation of cell cycling is implicated in the pathogenesis of various diseases, such as cancer, myocardial infarction, stroke, atherosclerosis, infection, inflammation, and neurodegenerative disorders. Caron et al. analyzed public databases for information about protein localization and tyrosine phosphorylation status in mitotic cells and devised a mitosis-associated tyrosine phosphorylation network. The extent of this network predicted that tyrosine-targeted phosphorylation plays a larger role in mitosis than previously appreciated. For example, in their network generated from data mining and in cultured cells, tyrosine phosphorylation decreased activation of Polo-like kinase 1 (PLK1), a serine/threonine kinase that promotes chromosome separation during anaphase and is often excessively abundant in cancers. The network provides a wealth of targets for exploration into cell cycle control in physiology and disease.

Abstract

Tyrosine phosphorylation is closely associated with cell proliferation. During the cell cycle, serine and threonine phosphorylation plays the leading role, and such phosphorylation events are most dynamic during the mitotic phase of the cell cycle. However, mitotic phosphotyrosine is not well characterized. Although a few functionally-relevant mitotic phosphotyrosine sites have been characterized, evidence suggests that this modification may be more prevalent than previously appreciated. Here, we examined tyrosine phosphorylation in mitotic human cells including those on spindle-associated proteins.? Database mining confirmed ~2000 mitotic phosphotyrosine sites, and network analysis revealed a number of subnetworks that were enriched in tyrosine-phosphorylated proteins, including components of the kinetochore or spindle and SRC family kinases. We identified Polo-like kinase 1 (PLK1), a major signaling hub in the spindle subnetwork, as phosphorylated at the conserved Tyr217 in the kinase domain. Substitution of Tyr217 with a phosphomimetic residue eliminated PLK1 activity in vitro and in cells. Further analysis showed that Tyr217 phosphorylation reduced the phosphorylation of Thr210 in the activation loop, a phosphorylation event necessary for PLK1 activity. Our data indicate that mitotic tyrosine phosphorylation regulated a key serine/threonine kinase hub in mitotic cells and suggested that spatially separating tyrosine phosphorylation events can reveal previously unrecognized regulatory events and complexes associated with specific structures of the cell cycle.

INTRODUCTION

Throughout the cell cycle, phosphorylation events are thought to be most dynamic during mitosis (1), and seminal mitotic phosphoproteomic studies have provided a wealth of data on serine (Ser) and threonine (Thr) phosphorylation (112). However, phosphotyrosine (pTyr) is comparatively rare and is underrepresented and poorly annotated, even in data sets from large-scale mitosis-specific phosphoproteomic studies. Under basal conditions, and in the absence of specific signaling events, Tyr phosphorylation is maintained at substoichiometric levels (2), presumably owing to the high activity of endogenous protein Tyr phosphatases (13, 14). Consistently, phosphoamino acid and global phosphoproteomic analyses concur that protein phosphorylation occurs primarily at Ser and Thr residues (83% and 15% of sites, respectively), whereas Tyr phosphorylation represents only 0.5 to 2% of the total phosphoproteome (15). More recently, quantitative, high-resolution, and phosphoproteome-wide studies have identified a number of Tyr-phosphorylated peptides during mitosis, although they have received little attention in subsequent functional analyses (13, 512). Despite this, evidence that mitotic Tyr phosphorylation may be more prevalent and physiologically relevant than was previously appreciated is accumulating.

The best understood pTyr site regulated during mitosis is arguably the inhibitory phosphorylation of cyclin-dependent kinase 1 (CDK1) at Tyr15, which is catalyzed by the kinases WEE1 and MYT1, and whose dephosphorylation serves as the basis for the switch-like activity of CDK1–cyclin B complex at mitotic entry [reviewed in (16)]. By exploiting the monoclonal 4G10 pTyr antibody, Tyr phosphorylation was originally detected at kinetochores and centrosomes in the mammalian PtK1 cell line more than 20 years ago (17). More recent evidence in both meiotic and mitotic cells demonstrates that SRC family intracellular kinases (SFKs) are activated in mitosis, likely in a CDK1-dependent manner, and are then inactivated upon mitotic exit [reviewed in (18)]. SFKs are associated with, and activated at, the spindle and cortex during meiosis in mice, Xenopus, and zebrafish (1921), and inhibition of SFKs blocks meiotic progression (2224), prevents the formation of microtubules, and reduces the amount of Tyr-phosphorylated proteins in γ-tubulin complexes alongside disrupted spindle architecture. In mitosis, the SFK FYN colocalizes with the spindle and centrosomes in immune cells (25) and coimmunoprecipitates with γ-tubulin (26). More recently, inhibition of either SRC or the ABL Tyr kinase was reported to induce spindle orientation and tilting defects in early prometaphase (27, 28), and FYN is also implicated in promoting spindle assembly and mitotic progression (29). Finally, activation of epidermal growth factor receptor (EGFR) at the cell surface has been shown to determine the timing of centrosome separation (30) and can selectively activate downstream signaling components, including SRC, during mitosis (31). Thus, although pTyr signaling appears relevant to spindle formation and function, the mitotic substrates and signaling pathways regulated by Tyr phosphorylation in human cells are not yet defined. We postulate that pTyr signaling in mitosis is more prevalent than previously appreciated, prompting us to identify and explore Tyr phosphorylation in the human mitotic phosphosignaling space.

RESULTS

pTyr signals during mitosis

To determine the relative extent of pTyr modification during the cell cycle, we ascertained global pTyr profiles by immunoblotting HeLa S3 cell lysates synchronously progressing through the cell cycle with a pTyr-specific antibody (4G10). Whereas basal pTyr signals did not change substantially between cells in G1/S and those in G2, a clear shift in the pTyr pattern was observed in mitotic cells (at the 10-hour time point; Fig. 1A), similar to that in cells that were artificially arrested in mitosis with the microtubule poisons nocodazole and taxol. This pattern was closely mimicked by two other pTyr antibodies, PY20 and pY100 (fig. S1A), verifying these findings.

Fig. 1 Spindle-specific pTyr signals during mitosis.

(A) Immunoblot detection of pTyr using 4G10 antibody (top blot) on HeLa S3 lysates arrested at different stages of the cell cycle. Extracts were prepared from asynchronous cells (Asynch), serum-starved cells (presumably in G0), thymidine-arrested cells (presumably in G1/S; 0 hours), cells released from thymidine for the indicated time until G2/mitosis (10 hours), and cells presumably arrested in mitosis with nocodazole (N) or taxol (T). Membranes were reprobed for cyclin B (middle blot) and α-tubulin (bottom blot). Blot is representative of three independent experiments. Bottom: Intensity profile plot for the Asynch and 10-hour lanes. a.u., arbitrary units. (B) HeLa S3 cells were permeabilized and incubated with λ-phosphatase (λ-PPase) or control buffer (CNTL), fixed, and stained with pTyr antibody 4G10 (red), γ-tubulin antibody (green), and α-tubulin antibody (blue). Bottom: Interphase cells from the same experiments. Scale bar, 5 μm. (C) Cells were treated with 20 μM bpV(phen) for 10 min before fixation and stained with 4G10 (red), α-tubulin antibody (green), and DAPI (4′,6-diamidino-2-phenylindole; blue). (Right) Quantitation of the spindle-specific 4G10 staining. Scale bar, 5 μm. *P = 0.026, Mann-Whitney test. (D) U20S cells were treated with DMSO, nocodazole (10 ng/ml), or taxol (20 nM) before being fixed and stained with anti-pTyr (4G10; red), anti–γ-tubulin (green), and anti–α-tubulin (blue). The final panel shows an enlargement of the volumetric rendering of the mitotic spindle (blue), the spindle poles (green), and pTyr (red). Quantitation of the spindle-specific 4G10 staining is shown for the different conditions. Data are means ± SE of 10 to 15 cells from a representative experiment of three independent experiments. **P = 0.0011, one-way analysis of variance (ANOVA) with Dunnett’s multiple comparison test.

We next evaluated the subcellular localization of pTyr signals during mitosis. Confocal imaging revealed robust pTyr staining along the cell perimeter and at pole-proximal regions of the spindle (Fig. 1B, top). Identical staining patterns were observed with two independent pTyr antibodies (pY100 and 27B10; fig. S1, B and C). To examine the specificity of the 4G10 antibody for immunofluorescence, we permeabilized the cells and then treated them with λ-phosphatase to induce dephosphorylation before fixation and staining, which resulted in an almost complete loss of pTyr antibody signals (Fig. 1B, bottom). As an additional control, the converse experiment was performed. To stimulate Tyr phosphorylation, we inhibited the activity of endogenous protein Tyr phosphatases using the potent pervanadate derivative bpV(phen). As expected, an overall increase of pTyr signal was observed in cells cultured with bpV(phen), including a significant increase of the pTyr signals that were directly associated with spindles (Fig. 1C).

To further investigate pTyr at the spindle and associated structures, we treated U2OS (Fig. 1D) and HeLa S3 (fig. S1D) cells with dimethyl sulfoxide (DMSO; vehicle) or synchronized in mitosis with low doses of nocodazole and taxol before fixation and staining with pTyr antibodies. We found that a substantial amount of pTyr signal remained associated with the diminished spindle upon treatment with taxol and nocodazole, as confirmed by quantitation of spindle-associated pTyr signals (Fig. 1D, middle and bottom). Moreover, enhanced stabilization of the spindle with taxol appeared to significantly increase the associated pTyr signal (Fig. 1D, right), but no significant change was observed in nocodazole-treated cells. A three-dimensional reconstruction of pTyr staining of an intact spindle confirmed these results (movie S1). Collectively, these observations support and extend previous observations (17), and demonstrate that a substantial amount of Tyr phosphorylation in mitotic cells is concentrated at the cell cortex, around the centrosomes, and on the mitotic spindle.

Collation of mitotic and spindle-specific pTyr sites from phosphoproteomic studies

The systematic identification and cataloging of mitotic phosphorylation events from large-scale proteomic studies have yielded remarkable insight into mitotic progression (112). However, these studies largely excluded pTyr sites from in-depth analysis. To explore these sites in a systematic manner, we used the PhosphoSitePlus (PSP) database (32) established by Cell Signaling Technology (CST) to generate a data set of Tyr-phosphorylated proteins (and annotated sites of phosphorylation) that were originally identified in disparate mitosis-specific, shotgun mass spectrometry studies. The data curated by PSP use a common standard of analysis for all sites, whether from publications or from CST experimental data sets, which permitted internal normalization of independent data sets. Only sites with localization scores of P ≤ 0.05 or A-score > 13 qualify for inclusion in PSP.

The workflow for identifying mitotic pTyr sites (Fig. 2) involved initial curation of information from published mitotic studies (table S1) from HeLa cell extracts, including those from HeLa S3 cells. We focused on HeLa cells because they are the most commonly used human model for studies of mitosis, especially phosphoproteomic data set generation. Data from a total of 104,921 published phosphosites at the time of collation include 1507 nonredundant pTyr peptides from 1117 nonredundant proteins (table S2), representing 2% of phosphorylation sites. CST has independently identified pTyr sites from mitotically enriched HeLa cells in a number of studies that have also been deposited in PSP (table S3). Using the same criteria as above, we probed the CST curation data set in parallel and identified an additional 596 nonredundant pTyr peptides derived from specific pTyr enrichment approaches, which resulted in an 81% pTyr peptide content in these experiments (Fig. 2A and table S4). A small number of pTyr sites (33) were identified from mitotic HeLa cells that were specifically enriched for pSer/Thr (tables S5 and S6). Finally, the two data sets were merged to create a list of 1344 nonredundant proteins and a corresponding list of 1950 unique Tyr-phosphorylated peptides from mitotic cells (table S7). The curated and CST-generated data sets overlapped by 154 proteins, which represents about 8 and 25% of each set, respectively (Fig. 2B). Surprisingly, about 4% of total phosphosites identified in mitotic studies correspond to pTyr, which is somewhat larger than the predicted 0.5 to 2% (Fig. 2C) (34). We suggest that this skew is likely introduced by pTyr enrichment approaches used in a number of the studies. Collectively, however, our analysis suggests that pTyr signaling may be much more prevalent during mitosis than previously thought.

Fig. 2 Workflow of the mitotic Tyr phosphorylation data set construction.

(A) Mitotic data set information corresponding to pTyr identifications related to HeLa cells downloaded from PSP. S, Ser; T, Thr; Y, Tyr; pY, phosphorylated Tyr. (B) Venn diagram of mitotic pTyr site overlap between published and CST-curated phosphorylation according to PSP. (C) Phosphorylation site distribution from mitotic HeLa studies in PSP. (D) Cross-reference between nonredundant mitotic pTyr protein identification found in PSP and the spindle protein identification responding to spindle-associated definition according to PSP. For (B) to (D), the total number of sites per category is indicated in brackets. Data discussed in the text are highlighted in gray.

Our immunofluorescence observations demonstrate marked Tyr phosphorylation at the spindle, in agreement with previous studies (17, 29). By interrogating our final definitive data set, we next searched for and annotated Tyr-phosphorylated proteins found at the spindle and associated structures. We first generated a list of spindle proteins from PSP associated with the search terms kinetochore, centrosome, spindle, centriole, centromere, and pericentriolar material, thereafter referred to as “spindle proteins.” This resulted in a total of 605 protein identifications, which were cross-referenced with our pTyr data set to confirm mitotic Tyr phosphorylation of 116 spindle proteins (8.6%) corresponding to 203 pTyr peptides (10.4%) (tables S8 and S9 and Fig. 2D). This suggests that a substantial proportion of pTyr signaling occurs on proteins found at the mitotic spindle and associated structures.

Functional annotation of the mitotic pTyr network

To elucidate the signaling networks and biological processes that are regulated by pTyr during mitosis, we performed network analysis of 1344 mitotic Tyr-phosphorylated proteins. We first gathered protein-protein interaction data from the STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) database (35) and then visualized them in Cytoscape (36). The resulting collated network is unexpectedly dense and highly interconnected with 805 nodes and 4384 edges, although a number of outstanding nodes are also apparent (fig. S2A). Gene Ontology (GO) analysis of the entire network revealed enrichment of proteins involved in processes such as phosphosignaling, cytoskeletal organization, and chromosome organization, as expected (fig. S2, B and C). We then analyzed densely connected clusters within the network of protein interactions using the Molecular Complex Detection (MCODE) algorithm (33) to highlight nodes predicted to be involved in specific cellular processes. This approach yielded four prominent clusters, termed clusters 1 to 4 (Fig. 3). Mapping of the “spindle” network (yellow filled circles) onto the larger mitotic network (blue filled circles) confirmed the presence of a major spindle-associated subnetwork (cluster 2) and also revealed the integration of known spindle proteins into other pTyr-containing clusters. As further described below, GO analysis of each of the four major clusters (table S10) indicated their enrichment in biological processes and cellular components (37, 38).

Fig. 3 Protein-protein interaction subnetworks.

Subnetworks depicting the connectivity of highly connected clusters in the protein-protein interaction network from pTyr sites characterized in this study. The pTyr mitotic protein interaction network was determined in STRING, and the highest ranking clusters, enriched in (A) mRNA processing, splicing and translation proteins, (B) spindle organization proteins, (C) Tyr kinases, and (D) glucose metabolism-associated proteins, were identified using MCODE plug-in in Cytoscape. Proteins within the spindle network are marked in yellow. Other key proteins are marked according to the legend. Node size represents the number of pTyr sites identified on the protein.

Cluster 1 (Fig. 3A) included pTyr proteins implicated in the spliceosome, mRNA processing and splicing, and components of protein translation regulation. With the exception of a few cases (39), basic nucleolar functions, such as splicing, are thought to be shut down during mitosis, although this does not exclude a mitosis-specific regulatory role for some splicing factors, as has been recently proposed (39). However, Tyr phosphorylation of spliceosome and mRNA processing components has, to the best of our knowledge, not been previously recognized. This modification may be involved in inactivating the processing machinery during mitosis or in regulating protein-protein interactions during splicing or spliceosome formation, as described for Ser/Thr phosphorylation (40).

Cluster 2 (Fig. 3B) was highly enriched for pTyr-containing proteins involved in mitotic spindle organization. It includes proteins involved in spindle assembly (such as centrosome-associated proteins, kinesin family member 2B, and tubulin subunits) and proteins involved in centromere integrity and kinetochore assembly and function [such as inner centromere protein (INCENP), Shugoshin 1 (SGO1), wings apart-like (WAPL), cytoplasmic linker-associated protein 2 (CLASP2), centromere protein F (CENP-F), and kinetochore null-1 (KNL1)]. Moreover, a number of mitotic Ser/Thr kinases with well-established roles in spindle assembly and regulation were enriched in this cluster. These included CDK1, Aurora A and Aurora B, and Polo-like kinase 1 (PLK1). With the exception of the inhibitory CDK1-pTyr15, very little is known regarding the potential role of Tyr phosphorylation for any of these mitotic kinases.

Cluster 3 (Fig. 3C) was enriched in Tyr kinases, including several members of the EphA receptor family and, rather conspicuously, nonreceptor Tyr kinases such as SFK members implicated in mitotic spindle dynamics (see Discussion). This cluster also included the Tyr phosphatase SHP-2 (SRC homology region 2 domain–containing phosphatase-2, also known as PTPN11), which is known to be Tyr-phosphorylated and whose deficiency reportedly causes errors in kinetochore-microtubule attachments, chromosomal congression defects, and missegregation, likely through inappropriate activation of PLK1 and Aurora B [reviewed in (41)].

Finally, cluster 4 (Fig. 3D) was enriched in proteins involved in glucose metabolism. These include the tumor-specific pyruvate kinase M2 (PKM2), which is instrumental in both aerobic glycolysis and gene transcription and is also required for proper mitotic progression. PKM2 specifically binds to the spindle assembly checkpoint protein budding uninhibited by benzimidazoles 3 (BUB3) and phosphorylates it at Tyr207 during mitosis (42). This phosphorylation is required for BUB3-BUB1 complex interaction with KNL1 and thus promotes kinetochore loading of the checkpoint machinery, chromosome congression, and the spindle assembly checkpoint. Strikingly, all three pTyr sites identified in PKM2 are in the protein kinase domain, suggesting that they may be directly involved in modulating PKM2 catalytic activity (43).

pTyr substrate motif and domain analysis

To determine whether specific pTyr motifs are overrepresented in the mitotic pTyr data set, and in an attempt to identify any conserved sequence motifs known to be recognized by protein-Tyr kinases or protein-protein interaction domains, we performed motif analysis (44) on the 1950 mitotic and 203 spindle-associated pTyr sites. We exploited both the human International Protein Index (IPI) as a reference data set to determine enrichment of the pTyr motifs in our data set and the entire pTyr cohort in PSP to specifically determine motif enrichment over and above global pTyr signaling; both approaches gave similar results (Fig. 4A). Among the mitotic set, the most enriched motifs were pYXXSP and SPXpY motifs. The observed enrichment of (Ser/Thr) proline-directed phosphorylation motifs among our pTyr data set is important because a C-terminal Pro residue represents the minimal motif for phosphorylation of Ser/Thr by mitogen-activated protein kinases and CDKs, including the master regulator of mitosis CDK1. However, the observation that Tyr phosphorylation may occur in the vicinity of potential CDK sites has not previously been reported. We found that, in 46 of the 47 pYXXSP motifs and in 34 of the 39 SPXpY motifs, the Ser residue was also reported to be phosphorylated in the PSP database. Moreover, in 11 of 46 pYXXSP motifs and in 6 of 34 sites where the phosphorylation was found on both Tyr and Ser residues, a doubly phosphorylated peptide was identified in the mitotic phosphoproteomic studies (table S11), suggesting a possible cross-talk between mitotic pSer-Pro and neighboring pTyr residues in the same polypeptide. In addition, we found that both mitotic and (in particular) spindle-specific pTyr sets were highly enriched for pYXXP. The SH2 domains of a number of proteins, including those of intracellular kinases such as SFKs and ABL, all share a general preference for the pYXXP motif (45). A number of basophilic motifs also emerged from our analysis. In particular, robust enrichment of a Lys at the +5 or +6 position (pYX4K and pYX5K) was observed in the spindle data set. Notably, a recent report suggested that modified Lys residues tend to occur more frequently in the vicinity of phosphorylated residues and that this tendency was particularly strong for pTyr, suggesting that cross-talk between modified Lys and pTyr residues may be occurring in mitosis (2). Consistently, in the spindle data set, for 11 of the 26 pYX4K motifs, Lys modification was also reported in PSP, whereas this number dropped to 2 of 20 for the pYX5K motif. Finally, the spindle set revealed >2.5-fold enrichment of phosphosites with a Gly at the −2 position (GXpY). This generic motif is thought to be phosphorylated by a number of distinct Tyr kinases, including the insulin receptor, insulin-like growth factor 1 receptor, and SRC (46, 47). Collectively, our analysis identified novel pTyr motifs and a combinatorial pattern of phosphorylation in the mitotic phosphospace, suggesting that tyrosyl site modification may be guided by other mitotic inputs, such as phosphorylation of SP motifs and/or covalent Lys modifications.

Fig. 4 Motif and NetworKIN analysis of mitotic pTyr sites.

(A) Enriched phosphorylation site motifs of phosphopeptides identified in both mitotic and spindle networks using the entire human proteome as the reference data set (left) and using the entire pTyr cohort reported in PSP as the reference data set (right). (B) NetworKIN identification of kinase-substrate relationships for the mitotic networks (P = 3.8 × 10−20, χ2 test). (C) As in (B) but for spindle networks. For the most common kinase identifications, the observed and expected hits are indicated (P = 8.4 ×10−6, χ2 test). (D) HeLa cells were treated with DMSO or 20 μM PP2 for 30 min before fixation and staining with pTyr antibody (4G10, red), anti–γ-tubulin antibody (green), and anti–α-tubulin antibody (blue). Scale bar, 5 μm. Volumetric quantitation of pTyr signal intensity from 12 to 15 cells directly associated with the spindle is indicated on the graph. Data are means ± SE of 9 to 13 cells from one representative experiment (n = 3 independent experiments). P < 0.026, unpaired t test. (E) Percentage of pTyr sites observed and expected within individual domains. P < 0.0001, χ2 test.

We next exploited NetworKIN (48) to evaluate potential kinase-substrate relationships in both the mitotic and the spindle data sets. For 954 phosphosites, we identified a potential kinase. In the mitotic data set, we observed some enrichment of sites potentially phosphorylated by Abl kinase, Eph receptor, and EGFR (albeit with low counts), whereas enrichment of SFK sites was significantly higher (Fig. 4B). A similar pattern was observed in the spindle-associated data set (Fig. 4C). These results support and extend previous observations suggesting mitotic activity of intracellular kinases and a number of receptor Tyr kinases. The enrichment of potential SFK and ABL sites above the background set is in line with previous observations, implicating these kinases in mitotic regulation, in particular at the spindle (49).

Considering that the target sites for the SFKs were considerably enriched in both the mitotic and spindle data set, we next sought to determine whether pTyr signals associated with the spindle were sensitive to chemical SFK inhibition. We treated cells with the relatively specific SFK inhibitor PP2 before fixing and staining with anti-pTyr antibodies. Confocal imaging and volumetric quantitation of pTyr directly associated with the spindle revealed a significant reduction in pTyr signals (Fig. 4D), suggesting that SFKs, or other PP2-sensitive Tyr kinases, phosphorylate one or more spindle-associated proteins.

Phosphorylation within protein domains is known to regulate domain function, the best example being reversible phosphorylation within kinase domains [see, for example, (50)]. We thus determined whether the mitotic/spindle pTyr sites collated in our study resided within or outside domains in their respective proteins. We established that, in the mitotic data set, 47% of pTyr sites were within known domains, similar to the case when all pTyr sites in PSP were taken into consideration, where 50% were found to be in domains. We found that protein kinase domains were considerably more Tyr-phosphorylated when compared to other domains annotated in our data set; 59 of 829 were found in protein kinase domains in the mitotic set, and 434 of 16,858 in the nonmitotic set (Fig. 4E and table S12). In summary, our observations suggest that multiple Tyr kinases, and in particular SFKs, are functionally active during mitosis and may contribute to accurate and timely progression, likely through phosphorylation of many of the proteins collated in this study. One such Tyr-phosphorylated target is the key mitotic master Ser/Thr kinase, PLK1.

Functional validation of PLK1 Tyr217 phosphorylation during mitosis

Somewhat surprisingly, our data analysis revealed Tyr phosphorylation within the catalytic domain as a common regulatory modification in a number of mitotic Ser/Thr kinases. We were particularly intrigued by PLK1 phosphorylation at Tyr217 (table S8) (5). This position is highly conserved among Ser/Thr kinases, and most other kinases have either a Phe or a Trp at this position, suggesting that a bulky, aromatic residue at this position is likely essential for kinase activity (table S13). To address this in the case of PLK1, we initially measured total Tyr phosphorylation of PLK1 Y217F and a potential phosphomimetic mutant Y217E, and compared them with PLK1 wild-type and a catalytically inactive D194A mutant [hereafter referred to as kinase-deficient (KD)]. Western blotting confirmed that pTyr PLK1 abundance was further enhanced by treatment with bpV(phen) and that basal pTyr abundance was constant between wild-type, KD, Y217F, and Y217E PLK1, suggesting either that Tyr217 is one of several mitotic pTyr sites on PLK1 or, alternatively, that Tyr217 phosphorylation occurs on a small subpopulation of PLK1 in the cell (fig. S3A). The P+1 loop is a critical point of interaction between the substrate and the kinase (51), and phosphorylation in the P+1 loop of several mitotic Ser/Thr kinases, including MPS1 (52) and BUB1 (53), has been shown to be important for modulation of their activity. We therefore measured the catalytic activity of Y217F and Y217E PLK1 mutants using 32P-based kinase assays. Whereas Y217F retained significant kinase activity when compared side-by-side with PLK1, we could not detect activity in exogenous full-length PLK1 Y217E, suggesting that, whereas a Phe at this position of the P+1 loop can be tolerated, the negative charge introduced by the Glu substitution completely abrogates catalytic activity (Fig. 5, A and B). Recent work shows that partial PLK1 activity is required to support proper spindle formation and mitotic progression through phosphorylation of a number of specific substrates at centrosomes and kinetochores (5457). To confirm the effect of PLK1-Tyr217 substitutions on mitotic progression, we generated inducible isogenic stable cell lines expressing small interfering RNA (siRNA)–resistant wild-type, KD, Y217F, or Y217E PLK1 constructs (fig. S3, B to D). Induction of exogenous PLK1 was coupled to siRNA depletion of endogenous PLK1 (Fig. 5C). We next measured two different outputs of PLK1 function: mitotic progression and phosphorylation of a known PLK1 substrate. First, we determined mitotic arrest and progression, which reflects a requirement for PLK1 in proper spindle assembly and functioning. As expected, PLK1 depletion in the parental cell line resulted in a significant increase in the percentage of cells arrested in mitosis from about 5% in control cells up to 49% after depletion, and this arrest was relieved by the reexpression of wild-type and Y217F PLK1 but not KD or Y217E mutants (Fig. 5D). In agreement with this finding, quantification of mitotic stages demonstrated a significant number of cells arrested at prometaphase in PLK1-depleted cells and also in cells rescued with KD or Y217E PLK1 (Fig. 5E). We next assayed the phosphorylation of the endogenous canonical PLK1 kinetochore substrate BUBR1 on Ser676 (58). As expected, BUBR1-Ser676 phosphorylation was abolished upon PLK1 depletion but was significantly restored by reexpression of wild-type and Y217F PLK1 but not KD or Y217E PLK1 (Fig. 5F). In agreement with these findings, another report demonstrated that retinal pigment epithelium cells expressing PLK1 Y217F exhibited subtle mitotic defects that did not affect long-term cell viability, in contrast to those expressing KD PLK1 (57). Together, these observations suggest that introduction of a negative charge on Tyr217 through PLK1 phosphorylation might inactivate the kinase during mitosis.

Fig. 5 Characterization of the effect of PLK1 Tyr217 phosphorylation.

(A) In vitro kinase assay of full-length PLK1 wild type (WT), KD, Y217F, and Y217E expressed in 293T cells. (B) Quantification of the PLK1 and casein phosphorylation from (A), normalized to WT. Data are means ± SE from n = 3 independent experiments. *P = 0.0005 for PLK1 autophosphorylation, randomized block ANOVA. (C) Experimental timeline for the synchronization of the PLK1 stable cell lines. DsiRNA, dicer-substrate siRNA. (D) Mitotic index of parental HeLa T-REx cells transfected with control (GL2) or PLK1 siRNA or the stable PLK1 lines transfected with PLK1 siRNA. Data are means ± SE from n = 3 independent experiments, with ≥100 cells counted per experiment. *P = 0.001, randomized block ANOVA. (E) Percentage of cells in different stages of the mitotic cycle, treated as in (D). Data are means ± SE of ≥100 cells counted per experiment from n = 3 independent experiments. P = 0.0001, randomized block ANOVA. (F) Immunofluorescence of cells fixed after treatment, as in (C). Fixed cells were stained with anti–BUBR1-pSer676 antibody (red), anti-BUBR1 antibody (green), and anti–CENP-C antibody (blue). Scale bar, 2 μm. Quantitation of the normalized BUBR1-pSer676 signal is shown. Data are means ± SE of normalized signal from 15 cells with >30 kinetochores per cell, representative of three independent experiments. *P = 0.0001, heterogeneous one-way ANOVA.

We next sought to understand how Tyr217 phosphorylation might inhibit PLK1 activity mechanistically. First, we generated recombinant PLK1 kinase domain variants (Y217F, Y217E, and KD) in a T210D background, which mimics the activating phosphorylation generated by the Aurora kinases, locking PLK1 in an active conformation (fig. S4A). Measurement of thermal stability of wild-type and mutant PLK1 proteins using differential scanning fluorimetry indicated that they were folded and exhibited differential thermal stability, with Y217E and Y217F markedly less stable than the T210D mutant in the absence of ligands (fig. S4B). In the presence of ligand, we found that PLK1 T210D+Y217E and PLK1 T210D+Y217F, but not KD PLK1 T210D+D194A (hereafter T210D+KD) mutant, bound to Mg/adenosine 5′-triphosphate (ATP) complexes, as indicated by an increase in their thermal stability (Fig. 6A). Furthermore, all four proteins interacted strongly with the PLK1 inhibitor BI2536 and the broad-spectrum kinase inhibitor staurosporine, but not the Aurora kinase inhibitor VX 680 (Fig. 6A). These data collectively confirm that the ATP binding site is intact and accessible in these proteins and that the loss of catalytic activity associated with Y217E is not likely due to an inability to bind ATP.

Fig. 6 Phosphotransferase activity of PLK1 Tyr217 mutants.

(A) ΔTm values for PLK1 proteins calculated upon ATP or kinase inhibitor incubations. Mean ΔTm values ± SD (n = 2 independent experiments) were calculated by subtracting the control Tm value (no ligand) from the measured Tm value. (B) PLK1 autophosphorylation and α-casein phosphorylation were visualized after autoradiography using [γ-32P]ATP as cofactor (top). Equal loading of assayed protein was confirmed by Naphthalene blue staining of the nitrocellulose membrane (bottom two panels) from n = 2 independent experiments. (C) Activity of T210D (red), T210D+Y217F (blue), and T210D+Y217E (black) mutant PLK1 proteins measured in the presence of increasing concentrations of ATP using a direct fluorescent peptide–based kinase assay. Data are mean Km [ATP] values ± SD (n = 2 experiments). (D) PLK1 in vitro kinase assays as in (B) but in the T210A PLK1 background (n = 2 independent experiments). (E) PLK1 phosphotransfer activity toward an optimized PLK1 substrate peptide. Data are means ± SD from n = 2 independent experiments. (F) WT and mutant MYC-PLK1 were expressed in 293T cells, immunoprecipitated (IP), resolved by SDS–polyacrylamide gel electrophoresis (PAGE), and blotted for PLK1-pThr210. Blots were reprobed for MYC (n = 3 independent experiments).

We next sought to determine whether any relationship might exist between multisite phosphorylation at Thr210 and Tyr217. In agreement with kinase assays using full-length PLK1 isolated from human embryonic kidney (HEK) 293 cells, the recombinant PLK1 kinase domain was active if either Tyr or Phe was present at position 217, whereas Y217E PLK1 could not autophosphorylate or phosphorylate α-casein under identical conditions (Fig. 6B). In addition, T210D or T210D+Y217F PLK1 phosphorylated an optimized peptide substrate with similar affinity, inferred from Km [ATP] values, suggesting functional equivalence, whereas PLK1 T210D+Y217E failed to detectably phosphorylate this peptide (Fig. 6C). This finding reveals that mimicking phosphorylation of Tyr217 with Glu inhibits kinase activity even in the presence of a hyperactivating, stoichiometric T210D mutation. Next, we tested the potential for cross-talk between Thr210 and Tyr217. To accomplish this, we generated PLK1 T210A, which abolished the constitutive negative charge stationed at this position in PLK1 T210D. Consistently, although PLK1 T210A was less active than PLK1 T210D, the activity of T210A+Y217F PLK1 was further reduced when compared to T210D+Y217F PLK1, whereas Y217E PLK1 remained inactive regardless of the Thr210 status (Fig. 6, D and E; compare also Fig. 6, B and D). Thus, a negative charge at Thr210 upon phosphorylation might be particularly important for catalytic activity in the context of PLK1 Y217F. Finally, to determine whether Tyr217 phosphorylation affects Thr210 phosphorylation in cells, we measured the relative phosphorylation of Thr210, confirming that whereas a small fraction of active wild-type and Y217F PLK1 was phosphorylated on Thr210, a greater proportion of KD and Y217E PLK1 was phosphorylated on Thr210 (Fig. 6E), suggesting that Tyr217 phosphorylation may influence activation segment phosphorylation, which culminates in Thr210 hyperphosphorylation as a response to Tyr-mediated blockade of substrate binding.

DISCUSSION

Tyr phosphorylation during mitosis was established more than 30 years ago. SFKs, and more recently a number of other Tyr kinases such as ABL, EGFR, and the metabolic kinase PKM2, have all been reported to be active during mitosis (31, 43, 49). Nevertheless, only a handful of mitotic Tyr phosphorylation sites have been studied to date, and the field is lacking a concerted data set for analysis of mitotic pTyr dynamics. This is partly due to the low stoichiometry (about 10%) in the absence of direct stimulation and the localized and temporal nature of the relatively small but highly selective pool of Tyr phosphorylated targets (2). Here, we have taken an integrated approach that includes bioinformatics, cell biology, and biochemistry to explore the depth and role of pTyr-dependent signaling during mitosis. Our analysis reveals enrichment of pTyr-specific signals at the plasma membrane and, most importantly, around the mitotic spindle and centrosomes in HeLa cells, in agreement with previous reports (17, 29). By exploiting the PSP database, we compiled a resource of 1950 unique pTyr sites in 1344 proteins identified in mitotic cells through large-scale phosphoproteomic studies. The specific and general dynamics of Tyr phosphorylation between interphase and mitosis currently remains unclear but represents an important future goal for quantitative phosphoproteomic approaches. Nevertheless, to the best of our knowledge, this work represents the first systematic compilation and analysis of pTyr signaling during mitosis and therefore presents a valuable resource and rationale for further exploration of mitotic pTyr signaling.

Bioinformatics analysis of mitotic pTyr sites

Network analysis of mitotic proteins containing Tyr phosphorylation sites revealed new and known biological processes and pathways that are likely to be regulated by Tyr phosphorylation. For example, our analysis identified multiple Tyr phosphorylation substrates in a subnetwork that contains many of the known components of the mitotic spindle apparatus. In addition, we verified a subnetwork enriched for SFKs, which agrees with our pTyr motif analysis, demonstrating strong enrichment of SH2 domain binding motifs and SFK target sites. The high number and relative proportion of sites predicted to be targeted by SFKs is not unexpected because these kinases are functionally pleiotropic and involved in fundamental processes including mitosis and cytokinesis (59). In agreement with our bioinformatics analysis and our observations of spindle-specific pTyr signals that are lost upon chemical SFK inhibition, both SRC and FYN kinases are directly implicated in spindle function. In meiotic cells, activated SFKs are localized to the spindle microtubules (19, 21), whereas in somatic cells, FYN is reported to localize to these mitotic structures (25). SRC promotes proper spindle orientation in early prometaphase through centrosome-mediated aster formation (27), whereas FYN is necessary for spindle formation and kinetochore-fiber stability in a manner dependent on kinase activity (29). Very recently, and in agreement with our own observations of PP2-sensitive pTyr signals at the spindle, PP2 treatment was shown to decrease FYN-induced pTyr signals around the centrosome (29). SFKs may also regulate cytokinesis through modulation of the cleavage furrow and abscission. The hyperactive, oncogenic form of SRC induces delocalization of MKLP2, the chromosomal passenger complex, and subsequently MKLP1 from the spindle midzone, leading to cytokinesis failure (60, 61). Moreover, a recent computational analysis of experimentally derived phosphosites revealed that the pYXXP motif (which is highly enriched in our data set) and its variations were found to be one of the most common pTyr motifs in the phosphospace (62) and are known to be targeted by ABL (47) and a number of SFKs. Notwithstanding these studies, very little is known about SFK substrates in mitosis, and proteins within our network are attractive candidate substrates for these Tyr kinases during mitosis, which will be explored in the future.

An interesting observation from our current analysis is the presence of pTyr sites alongside other potential sites of Ser/Thr or Lys covalent modification. Cross-talk between different types of posttranslational modifications (PTMs) is an emerging theme in cell signaling and may well be a common concerted mechanism of protein regulation (63). In addition to the new pTyr motifs identified in the mitotic phosphospace, our data suggest the possibility of cross-talk between Tyr phosphorylation and pSP (CDK) motifs as well as Lys modifications during mitosis. Previous work has demonstrated that pTyr is preferentially found close to modifiable Lys residues, and that the frequency of comodification is inversely proportional to the distance between these PTMs (2). Consistently, our data reveal enrichment of Lys-containing motifs at sites between 5 and 6 amino acids C-terminal to pTyr (Fig. 4). Whether Lys modifications can coexist on the same proteoforms, representing a new form of regulatory switch, remains to be established. Although the biological significance of potential PTM cross-talk remains unclear in the context of pTyr, one attractive hypothesis is that it may be used to assign functional relevance to distinct modifications, potentially implicating co-occurrence of phosphorylated (in the case of Ser) and acetylated/ubiquitylated residues (in the case of Lys) for the multisite regulation of protein function (6466). Along the same lines, it is attractive to speculate that the presence of pSP motifs in the vicinity of pTyr may guide modification of the latter residue in mitosis. We observed that the vast majority of Ser residues within pYXXSP and SPXpY motifs are known to be phosphorylated in PSP, and in a number of cases, mass spectrometric evidence for the doubly phosphorylated peptide does exist in mitotic samples.

Inhibitory phosphorylation of PLK1

Phosphorylation in the activation segment of protein kinases is a common mechanism of kinase regulation. However, activation loop phosphorylation of many kinases generally induces activating structural changes by repositioning key structural elements that permit substrate and cofactor binding and efficient catalysis (51). Although no common mechanism has been proposed for negative regulation of protein-Ser/Thr kinases, phosphorylation of several of the CDKs within the subdomain I GXGXXG motif at the Thr14 and Tyr15 (human CDK1 numbering) are known to be inhibitory (6769), and acetylation of the ATP coordinating Lys has been shown to reduce the kinase activity of CDK9 (70).

Here, we establish for the first time that mimicking phosphorylation of PLK1 on Tyr217 in the P+1 loop completely inhibits detectable kinase activity, likely through inhibition of substrate binding, although we cannot formally rule out the possibility that the effect is due to the Glu substitution rather than a phosphomimicking effect per se (51). Nevertheless, and in agreement with this assertion, neither Mg/ATP nor kinase inhibitor binding was affected in PLK1 Y217F and Y217E, whereas PLK1 autophosphorylation and in-trans α-casein protein phosphorylation were selectively abolished in the Y217E mutant. Further evidence from the literature demonstrates the relevance of modifications at this position. A recent report demonstrated very high conservation of PLK1 Tyr217 homologous residues among human Ser/Thr kinases (71). Tyr is found at the position equivalent to PLK1 Tyr217 in 272 of the 393 “typical” human protein-Ser/Thr kinases, and phosphorylation at this residue has been reported for a number of these (71). For example, phosphorylation of the DNA damage response kinase CHK2 at the homologous Tyr390 site is markedly decreased after exposure to ionizing radiation, suggesting that phosphorylation at this site is likely to be inhibitory to enzyme activity (72). Moreover, phosphorylation of Tyr210 in the P+1 loop of extracellular signal–regulated kinase 1 (ERK1) has recently been proposed to be important for proper conformation of the active site, and neither ERK1 Y210F nor Y210E mutants were phosphorylated at Thr202 and Tyr204 by mitogen-activated protein kinase kinase 1 (MEK1) in vitro (71). Of particular interest, PLK4 has also been reported to be phosphorylated on Tyr177, the equivalent P+1 loop residue (73, 74). Finally, it is noteworthy that a Tyr residue at the homologous position is not conserved in any of the human protein-Tyr kinases examined, perhaps to prevent nonspecific inhibitory Tyr autophosphorylation in the activation segment (71). Collectively, our data suggest that P+1 loop Tyr phosphorylation (or in other motifs essential for substrate phosphorylation) may serve as a selective mechanism for inhibition of protein-Ser/Thr kinase activity and may be more common than previously appreciated. We highlight phosphorylation of Tyr148 of Aurora A in our compiled data set, situated just after the Gly-rich loop. This modification has very recently been shown to be catalyzed by Golgi-associated SRC and appears to be required for targeting of Aurora A to the centrosome (75).

We were initially surprised to find that mimicking PLK1 Tyr217 phosphorylation was inhibitory, considering that this site was originally identified in phosphoproteomic studies of mitosis and considering the prominent role of PLK1 in mitosis. We believe that this site is unlikely to be an example of “accidental” or “nonfunctional” phosphorylation, as has been noted with some pSer/Thr sites (76). As noted above, pTyr sites are stringently regulated, and multiple identifications of this site have been reported in proteomic databases, suggesting that it is likely to be physiologically relevant. Moreover, Tyr residues at this position are phosphorylated (and relevant to catalysis) in other kinases. Last, inhibitory phosphorylation of the master mitotic regulator CDK1 at Tyr15 is also commonly observed in mitotic studies (6769), suggesting that pools of inhibited mitotic kinases must exist.

Instead, we propose that phosphorylation at this site may occur at very low stoichiometry to control selected populations of PLK1 or may be a rapid mechanism to maintain balanced PLK1 activity during the stages of cell division. Unfortunately, we could not generate phosphospecific antibodies to Tyr217, in line with a very low and localized stoichiometry, and consistent with complex multisite phosphorylation in the PLK1 activation segment (55). Glutamate mimicking of pTyr at this site did not support activity even in the background of the hyperactive PLK1 T210D, suggesting a dominant substrate blockade. This suggests complex structural and regulatory requirements for kinase activation/inactivation that extend beyond PLK1 T-loop Thr210 phosphorylation and may explain the remarkable specificity of PLK1 toward its substrates despite its highly pleiotropic localization. A more complete understanding of PLK1 inhibitory phosphorylation during mitosis will require detailed quantitative and spatiotemporal investigation of Tyr217 phosphorylation during interphase and mitosis, and identification of the kinase responsible for this modification. Initial experiments demonstrate that SRC is not likely to be a physiological PLK1 Tyr kinase because incubation of recombinant PLK1 with active SRC fails to induce marked phosphorylation or to directly modulate PLK1 activity in vitro (fig. S5).

In conclusion, our study represents a valuable new resource for researchers in the mitosis and Tyr kinase signaling fields. Our findings indicate that this modification may be much more important in mitosis than previously thought. Quantitative proteomics approaches are needed to establish how changes in pTyr that occur during mitosis (and in interphase) are specifically linked to the cell cycle. This will provide a more complete picture of the dynamic mitotic phosphoproteome and furnish researchers with substrates linked to cancer-associated Tyr kinases.

MATERIALS AND METHODS

Cell culture

All cells were maintained in Dulbecco’s modified Eagle’s medium with high glucose (HyClone) containing 10% bovine growth serum (BGS), 1% l-glutamine, and 1% penicillin/streptomycin and cultured at 37°C in a 5% CO2. For Western blotting and immunofluorescence, cells were synchronized in mitosis after release from a thymidine-mediated G1/S arrest. For 293T cells, transfection was performed with polyethylenimine (2 μg/ml) and plasmid DNA (300 ng/ml). Depletion of endogenous PLK1 was performed using jetPRIME (2 μl/ml; Polyplus) with 3.3 μM DsiRNA (IDT) targeting a previously validated site (5′-CGAGCUGCUUAAUGACGAGUUCUTT-3′) (4). DsiRNA-resistant PLK1 stable lines were generated by cotransfection of the pOG44 plasmid (1.3 μg/ml; Thermo Fisher Scientific) and pcDNA5/TO/FRT/PLK1 wild-type and mutants (Thermo Fisher Scientific) using TransIT LT-1 (3.5 μl/ml; Mirus Bio). Forty-eight hours after transfection, cells were trypsinized and cultivated in Dulbecco’s modified Eagle’s medium containing 10% BGS, 1% l-glutamine, 1% penicillin/streptomycin, and hygromycin B (400 μg/ml). Clones were isolated after 3 to 4 weeks of selection, and expression was verified by Western blotting after induction with tetracycline (50 ng/ml; Sigma-Aldrich) for 24 hours.

Drug treatments and antibodies

Drug concentrations (unless otherwise indicated) were 3.3 μM nocodazole (Sigma-Aldrich), 200 nM paclitaxel (taxol, Millipore), 2 mM thymidine (Thermo Fisher Scientific) (all three for 16 hours), 20 μM PP2 (Abcam), and 20 μM peroxovanadium derivative bpV(phen) (Calbiochem).

The following antibodies were used in this study: anti-pTyr clone 4G10 (Millipore), clone PY20 (Enzo Life Sciences), clone pTY100 (CST), and clone 27B10 (Cytoskeleton Inc.); anti–α-tubulin (11H10) Alexa Fluor 647 (CST); anti–γ-tubulin (Sigma-Aldrich); anti-MYC (9E10, Santa Cruz Biotechnology); anti–CENP-C (MBL International); anti–PLK1-pThr210 (Rockland); and anti–BUBR1-pSer676 (58). Anti–immunoglobulin G (IgG) rabbit/mouse/guinea pig secondary antibodies for immunofluorescence and IgG horseradish peroxidase–conjugated secondary antibodies were purchased from Jackson ImmunoResearch.

Immunoblotting

Mitotic cells were collected by shake-off, rinsed in 1× phosphate-buffered saline, and directly lysed in radioimmunoprecipitation assay buffer [10 mM tris (pH 7.5), 150 mM NaCl, 1% NP-40, 10 mM NaF, 0.1% sodium deoxycholate, 1 mM sodium orthovanadate, 20 mM β-glycerophosphate, 10 mM sodium pyrophosphate, leupeptin (10 μg/ml), aprotinin (10 μg/ml), and 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF)] under constant agitation for 30 min at 4°C before centrifugation (15 min, 20,000g) at 4°C. Protein concentrations of the clarified lysates were determined using the bicinchoninic acid assay (Pierce Thermo Fisher Scientific), and equalized lysates were then resolved by SDS-PAGE and subsequent immunoblotting.

Immunofluorescence and volumetric quantification of pTyr signals by Imaris

Cells grown on coverslips were gently washed with tris-buffered saline (TBS) containing 100 μM sodium orthovanadate and then fixed and permeabilized simultaneously using PTEMF [0.2% Triton X-100, 20 mM Pipes (pH 6.8), 1 mM MgCl2, 10 mM EGTA, and 4% formaldehyde] for 10 min. Cells were blocked (TBS: 0.02% Tween, 3% bovine serum albumin, and 100 μM Na3VO4) for 30 min before staining with primary and secondary antibodies and mounting on slides using Fluoromount-G mounting medium (Thermo Fisher Scientific). Cells were imaged by confocal microscopy on an inverted Olympus IX80 microscope equipped with a WaveFX-Borealin-SC Yokagawa spinning disc (Quorum Technologies) and an ORCA-Flash4.0 camera (Hamamatsu). Image acquisition was performed using MetaMorph software (Molecular Devices). Images were acquired by 0.5-μm z steps at a depth of 5 μm using a 100× oil objective. Voxel size was 0.0529 μm × 0.0529 μm × 0.5 μm. Nine to 12 cells were imaged per experimental conditions. Three-dimensional reconstruction was performed using Imaris software (version 7.6.1, Bitplane). Briefly, isosurfaces were generated in the surpass module to segment the α-tubulin structures, as published previously (77, 78). The sum of intensities of pTyr signals was quantified within the segmented α-tubulin volume. Results are expressed as ratios of pTyr intensity per unit of volume of α-tubulin (μm3).

λ-Phosphatase assay

Cells grown on coverslips were rinsed with TM buffer [50 mM tris (pH 7.5) and 5 mM MgSO4]. A 5-min extraction was realized using TM buffer with 1% CHAPS, 1 mM dithiothreitol (DTT), and protease inhibitor cocktail. Cells were rinsed with TM buffer. Dephosphorylation was allowed for 10 min using TM buffer containing the λ-phosphatase, 1 mM DTT, and the protease inhibitor cocktail. Cells were rinsed with TM buffer, incubated in TM buffer + 5 mM N-ethylmaleimide for 5 min, and rinsed again in TM buffer. Cells were then fixed using PTEMF and processed for immunofluorescence, as described earlier.

Kinase assays

Mitotic cells were prepared in NP-40 lysis buffer [75 mM Hepes (pH 7.5), 150 mM KCl, 1.5 mM EGTA, 10% glycerol, 0.075% NP-40, 1 mM sodium orthovanadate, 20 mM β-glycerophosphate, 10 mM sodium pyrophosphate, leupeptin (10 μg/ml), aprotinin (10 μg/ml), and 1 mM AEBSF] under constant agitation for 30 min at 4°C. Lysates were then centrifuged (15 min, 20,000g) at 4°C, and protein concentrations were determined using the Bradford assay (Bio-Rad). Anti-MYC immunoprecipitations were performed on 500 μg of protein with 10 μl of protein G–agarose (Thermo Fisher Scientific) overnight at 4°C. Beads were washed three times in lysis buffer and once in kinase reaction buffer (KRB) [50 mM tris-HCl (pH 7.5), 10 mM β-glycerophosphate, and 100 μM sodium orthovanadate]. Samples were resuspended in 25 μl of KRB complemented with 0.5 mM DTT, 10 mM MgCl2, 50 μM cold ATP, 5 μCi of [32P]ATP, and casein (2 μg per sample) (Sigma-Aldrich) before incubation for 30 min at 30°C with agitation. Reactions were stopped, and samples were resolved by SDS-PAGE. Gels were stained with Coomassie brilliant blue, destained, and dried. Signals were revealed on BioMax MR-1 Film (Thermo Fisher Scientific) and quantified in the linear range on ImageJ.

For protein substrate analysis with recombinant PLK1, 10 μg of α-casein was incubated with 2 μg (~60 pmol) of wild-type, Y217E, and Y217F T210D or T210A PLK1 in 50 mM tris (pH 7.4), 100 mM NaCl, 1 mM DTT, and 100 mM imidazole at 30°C in the presence of 200 μM ATP (2 μCi of [32P]ATP per assay) and 5 mM MgCl2. At the indicated time points, the reactions were terminated by denaturation after boiling in SDS sample buffer, before separation by SDS-PAGE and transfer to nitrocellulose membranes. 32P-phosphotransfer to PLK1 and casein was detected by autoradiography, and the total amount of protein present in each lane was evaluated by Naphthalene blue staining of the membrane. To compare the specific activity of T210A and T210D PLK1 proteins, we analyzed linear phosphate incorporation into PLK1 peptide using 1 μg of the appropriate protein in the presence of 25 mM Hepes (pH 7.4), 2 μM PLK1 substrate peptide, 0.5 mM ATP, 5 mM MgCl2, and 1 mM DTT at 30°C. After 25 min, the extent of peptide phosphorylation was calculated by integrating the area under the phosphosubstrate peak, calculating percentage conversion, and converting to picomoles of phosphate transferred per minute of the reaction. The data shown represent the means ± SD from two independent experiments carried out in duplicate.

PSP data set

Mitotic pTyr identifications from HeLa (and HeLa S3) cells were downloaded from PSP. First, all curated information from published mitotic studies were downloaded and manually verified with the corresponding publication to only conserve the mitotic identifications adhering to the following criteria: false discovery rate < 1%, A-score ≥ 13, and localization probability ≥ 0.96. This corresponded to 1507 nonredundant peptide related to 1117 nonredundant protein. Data sets from CST curation set corresponding only to mitotic extracts from HeLa cells (Tyr- or Ser/Thr-enriched and nocodazole-treated), resulting to 596 nonredundant peptides that corresponded to 408 nonredundant proteins, were also downloaded. These data sets were merged to create a master list of uniformly nonredundant mitotic pTyr sites and proteins. The mitotic peptide list corresponds to all nonredundant pTyr sites based on their unique site group ID (according to PSP), with the exception of some isoforms sharing the same group IDs that were then sorted on the basis of their unique peptide sequence. Next, all pTyr identifications from PSP associated with these GO terms kinetochore, centrosome, spindle, centriole, centromere, and pericentriolar material (including subcellular location, biological process, molecular function, and cellular component; total 605 protein identifications) were downloaded from PSP and cross-referenced with the mitotic list to confirm mitotic pTyr in 116 pTyr proteins corresponding to 203 pTyr peptides named spindle protein and peptide ID, respectively. Finally, all pTyr peptide and protein identification (corresponding to 36,176 peptides from 11,959 proteins) on HeLa cells from PSP were downloaded and used as background information throughout bioinformatics analyses, unless otherwise stated.

pTyr network analyses

Functional and physical protein-protein interaction network were generated based on the mitotic (1344 unique proteins) and the spindle (116 unique proteins) protein list. The UniProtKB accession numbers corresponding to our mitotic and spindle PSP protein list were searched against the STRING database version 10 for protein-protein interaction (experimental and databases only; high confidence score, 0.700) (35). The interactions were visualized on the Cytoscape platform (version 3.2.0) (36). Self-loops were removed, and only interactions between the proteins belonging to the data set were selected. The resulting interactome was analyzed for highly connected complexes with the theoretical clustering algorithm MCODE (33). The most highly ranked subnetworks were extracted for further analysis and rendering. GO term enrichment analyses were performed using the DAVID (Database for Annotation, Visualization and Integrated Discovery) Bioinformatics Resources version 6.7 (medium stringency) (79). Functional clustering was performed using the following database: GO FAT, KEGG pathway, SMART, InterPro, Biocarta, and UniProt sequence feature. The statistical significance of the GO terms associated with mitotic or spindle protein ID was estimated by comparing it to our background (all pY from PSP). Kinase predictions were performed with NetworKIN (48) version 3.0 (KinomeXplorer) using the high-throughput workflow option. A cutoff NetworKIN score of 1.5 was used. Phosphorylation motifs were extracted from prealigned Tyr-phosphorylated peptide using Motif-X version 1.2 (44). Analyses were performed with a significance cutoff of 0.001 and at least 20 occurrences of the motif in the data set. Both the IPI human and complete PSP pTyr data sets were used as background data sets with the built-in function, as indicated. Data on protein domains were obtained on 11 June 2015. The domains were parsed so that protein domains completely comprised within another one were eliminated, the longest being conserved. Each residue in the mitotic pTyr data set was then assigned to a protein domain, if applicable.

Recombinant protein expression and purification

Human PLK1 (1–369) T-loop mutants (T210D and T210A) were produced as N-terminal His6-tag fusion proteins in BL21 (DE3) pLysS Escherichia coli (Novagen) and purified by immobilized metal affinity chromatography and gel filtration chromatography, where appropriate (80). Y217E, Y217F, and D194A point mutations were introduced by polymerase chain reaction (PCR) site-directed mutagenesis as required in T210D and T210A backgrounds and purified as above. Proteins were ~80% pure before analysis, behaved as monomeric species after gel filtration, and were folded on the basis of their thermal denaturation profiles.

PLK1 turnover analyzed by microfluidic mobility shift assay using a fluorescent peptide substrate

To measure PLK1 peptide substrate phosphorylation, we developed a direct microfluidic phosphorylation-based assay, which permitted determination of a Km value for ATP. Purified recombinant wild-type, Y217F, or Y217E T210D PLK1 (1 μg) was assayed individually in the presence of the novel PLK1 fluorescent peptide substrate (5-FAM-AEEISDELMEFSLKDQEA-CONH2; 5 μM) in the presence of the indicated concentrations of ATP and 5 mM MgCl2. Peptide phosphorylation was calculated directly by measuring the conversion of the ratio of the peptide to the phosphopeptide after 1 hour of assay at room temperature. Phosphate incorporation into the peptide was linear at this time point, and incorporation was limited to ~30% to prevent depletion of ATP at low concentrations. Km [ATP] values were determined by nonlinear regression analysis using GraphPad Prism software.

Differential scanning fluorimetry

Thermal shift assays were performed with a StepOnePlus RT-PCR machine (Life Technologies) using a thermal ramp between 25° and 95°C (0.3°C per data point). Thermal denaturation of purified PLK1 proteins (5 μM) was detected by the fluorescent emission of SYPRO Orange dye (1:4000 final dilution). Assays included 1 mM ATP ± 10 mM MgCl2 or the indicated concentrations of known PLK1 inhibitors (BI2536 and staurosporine) and a negative control compound (VX-680). For all assays, the final concentration of DMSO was 4%. Mean ΔTm values ± SD from duplicate experiments were calculated by subtracting the control Tm value (calculated using the Boltzmann equation) from the Tm value measure in the presence of ligand, as previously described (81). Kinase inhibitor binding was evaluated as previously described (82).

Statistical analysis

All statistical analysis was performed in GraphPad Prism (version 6) or custom script written in R. Statistical differences between two groups were assessed with Student’s t test after verification of normal distribution by the Shapiro-Wilk test or the Mann-Whitney test for nonparametric data. For three or more groups, one-way (with Dunnett’s multiple comparison test) or randomized block (with Tukey-Kramer correction for multiple comparisons) ANOVA was used. To determine the significance of differences between observed and expected incidences of domain or phosphorylation motifs in our data set, we used the χ2 test. For all test, the significance threshold was set to P < 0.05.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/9/458/rs14/DC1

Fig. S1. pTyr patterns are reproducible with multiple pTyr antibodies.

Fig. S2. Network analysis of mitotic pTyr sites.

Fig. S3. MYC-PLK1 expression and stable cell line generation.

Fig. S4. Characterization of T210D PLK1 (1–369) proteins (wild type, D194A, Y217F, and Y217E).

Fig. S5. Analysis of PLK1 phosphorylation by SRC in vitro.

Table S1. Phosphoproteomic reference publications for collation of mitotic pTyr sites.

Table S2. List of nonredundant pTyr sites identified from the publications in table S1.

Table S3. CST mitotic data sets using pY enrichment approaches used for collation of mitotic pTyr sites.

Table S4. List of nonredundant pTyr sites identified from the data sets in table S3.

Table S5. CST mitotic data sets using general phosphoenrichment approach used for collation of mitotic pTyr sites.

Table S6. List of nonredundant pTyr sites identified from the data sets in table S5.

Table S7. Complete list of nonredundant mitotic pTyr sites collated in this study.

Table S8. List of Tyr-phosphorylated spindle-associated proteins.

Table S9. List of nonredundant pTyr sites identified on spindle proteins in this study.

Table S10. Raw data from GO analysis of the individual clusters in Fig. 3.

Table S11. Incidence of comodification of pTyr and other PTMs in Fig. 4.

Table S12. List of Tyr-phosphorylated kinases identified among the mitotic and spindle data sets.

Table S13. Alignment of the activation and P+1 loops of the human kinase complement showing conservation of the PLK1 Tyr217 residue.

Movie S1. Three-dimensional reconstruction of pTyr signals at the mitotic spindle.

REFERENCES AND NOTES

Acknowledgments: We would like to thank R. Malik and members of the Elowe laboratory for helpful discussions. We also thank G. Daigle for guidance with statistical analyses. Funding: Research in the Elowe laboratory was funded by grants for the Canadian Institutes for Health Research (CIHR; #244442), the National Science and Engineering Research Council (05841), and the Canadian Cancer Society (19140). S.E. holds a CIHR New Investigator award. P.A.E. acknowledges funding from the North West Cancer Development Fund and project grant CR1088. C.R.L. was funded by grants 299432 and 324265 from CIHR and holds the Canada Research Chair in Cell and Systems Biology. Author contributions: D.C., C.R.L., and S.E. performed the data mining and bioinformatics analysis. D.C. and P.T. performed all immunofluorescence and Western blotting experiments. D.S. quantified the Imaris data. D.P.B. performed in vitro PLK1 assays. S.E. wrote the manuscript with help from D.C. and P.A.E. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article