O-GlcNAcylation of myosin phosphatase targeting subunit 1 (MYPT1) dictates timely disjunction of centrosomes

The role of O-linked N-acetylglucosamine (O-GlcNAc) modification in the cell cycle has been enigmatic. Previously, both O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) disruptions have been shown to derail the mitotic centrosome numbers, suggesting that mitotic O-GlcNAc oscillation needs to be in concert with mitotic progression to account for centrosome integrity. Here, using both chemical approaches and biological assays with HeLa cells, we attempted to address the underlying molecular mechanism and observed that incubation of the cells with the OGA inhibitor Thiamet-G strikingly elevates centrosomal distances, suggestive of premature centrosome disjunction. These aberrations could be overcome by inhibiting Polo-like kinase 1 (PLK1), a mitotic master kinase. PLK1 inactivation is modulated by the myosin phosphatase targeting subunit 1 (MYPT1)–protein phosphatase 1cβ (PP1cβ) complex. Interestingly, MYPT1 has been shown to be abundantly O-GlcNAcylated, and the modified residues have been detected in a recent O-GlcNAc–profiling screen utilizing chemoenzymatic labeling and bioorthogonal conjugation. We demonstrate here that MYPT1 is O-GlcNAcylated at Thr-577, Ser-585, Ser-589, and Ser-601, which antagonizes CDK1-dependent phosphorylation at Ser-473 and attenuates the association between MYPT1 and PLK1, thereby promoting PLK1 activity. We conclude that under high O-GlcNAc levels, PLK1 is untimely activated, conducive to inopportune centrosome separation and disruption of the cell cycle. We propose that too much O-GlcNAc is equally deleterious as too little O-GlcNAc, and a fine balance between the OGT/OGA duo is indispensable for successful mitotic divisions.


The role of O-linked N-acetylglucosamine (O-GlcNAc) modification in the cell cycle has been enigmatic. Previously, both O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA)
disruptions have been shown to derail the mitotic centrosome numbers, suggesting that mitotic O-GlcNAc oscillation needs to be in concert with mitotic progression to account for centrosome integrity. Here, using both chemical approaches and biological assays with HeLa cells, we attempted to address the underlying molecular mechanism and observed that incubation of the cells with the OGA inhibitor Thiamet-G strikingly elevates centrosomal distances, suggestive of premature centrosome disjunction. These aberrations could be overcome by inhibiting Polo-like kinase 1 (PLK1), a mitotic master kinase. PLK1 inactivation is modulated by the myosin phosphatase targeting subunit 1 (MYPT1)-protein phosphatase 1c␤ (PP1c␤) complex. Interestingly, MYPT1 has been shown to be abundantly O-GlcNAcylated, and the modified residues have been detected in a recent O-GlcNAcprofiling screen utilizing chemoenzymatic labeling and bioorthogonal conjugation. We demonstrate here that MYPT1 is O-GlcNAcylated at Thr-577, Ser-585, Ser-589, and Ser-601, which antagonizes CDK1-dependent phosphorylation at Ser-473 and attenuates the association between MYPT1 and PLK1, thereby promoting PLK1 activity. We conclude that under high O-GlcNAc levels, PLK1 is untimely activated, conducive to inopportune centrosome separation and disruption of the cell cycle. We propose that too much O-GlcNAc is equally deleterious as too little O-GlcNAc, and a fine balance between the OGT/OGA duo is indispensable for successful mitotic divisions.
The centrosomes are the primary microtubule-organizing centers that nucleate the mitotic spindle apparatus to ensure subsequent faithful sister chromatid segregation during mitosis. The centrosome cycle is tightly coordinated with other cell cycle events (1), and its aberrancy could culminate in chromosome segregation defects and aneuploidy (2). The entire centrosome cycle encompasses centrosome duplication during S phase, disjunction in late G 2 phase, and further separation during prophase or prometaphase, and eventual segregation into the two daughter cells.
Previous investigations have identified that MYPT1 is also subject to O-linked N-acetylglucosamine (O-GlcNAc) modifications (26). O-GlcNAcylation is an emerging post-translational modification (PTM) that integrates the metabolic signals with transcription, nutrient sensing, stress responses, and cell cycle events (27,28). It is catalyzed by the sole transferase O-GlcNAc transferase (OGT) and reversed by the only O-GlcNAcase (OGA) (27). Chemical inhibitors of OGT (acetyl-5S-GlcNAc (5S-G)) and OGA (Thiamet-G (TMG)) have been developed to interrogate various biological processes (29). During the cell cycle, O-GlcNAcylation levels fluctuate as the cells proceed through different stages (30). In particular, overproduction of both OGT and OGA results in multipolar spindles (31). However, myriad targets of O-GlcNAc and its quintessential functions remain largely unexplored. Here, we identify the O-GlcNAc-modified residues of MYPT1. We show that O-GlcNAcylation of MYPT1 antagonizes pSer-473 and results in its dissociation from PLK1. Elevated O-GlcNAc levels thus fuel PLK1 activity towards centrosomes and render ill-timed centrosome separation, disrupting the mitotic cell cycle.

O-GlcNAc promotes aberrant centrosome separation via PLK1
Previously, overproduction of both OGT and OGA has been linked with the multipolar spindle (31). We sought to identify whether O-GlcNAc could also be linked with centrosome dynamics. Strikingly, when HeLa cells were treated with TMG (OGAi), the inter-centrosomal distance was significantly augmented 4-fold (Fig. 1A), reminiscent of the phenotype of Nek2A overexpression or overactivation (4,32). As the centrosome cycle is tightly governed by PLK1, we attempted to inhibit PLK1. When BI2536 (PLK1i) was adopted in conjuncture with TMG, the centrosomal distances shortened considerably ( Fig.  1, A-C). When BI2536 was utilized alone (Fig. 1A), the cells reduced centrosomal distances as reported previously (4). These cytological studies suggest that high O-GlcNAc culminates in premature centrosomal separation, probably via PLK1.
We validated the interaction between MYPT1 and OGT through biochemical assays. As shown in Fig. 2A, GST-OGT pulled down HA-MYPT1 from cell extracts. Then both OGT and MYPT1 proteins were purified from Escherichia coli. Upon incubation, His-OGT pulled down GST-MYPT1 (Fig. 2B), suggesting that the interaction is direct.

O-GlcNAc of MYPT1 regulates centrosome dynamics
utilized in pulldown experiments, and the FL, F2, and F3 MYPT1 pulled-down Myc-OGT (Fig. 2D), suggesting that the potential modification sites could be residing in F2 and F3.
A recent quantitative proteomic analysis of protein O-GlcNAc sites using an isotope-tagged cleavable linker (isoTCL) strategy identified the potential O-GlcNAc sites of MYPT1 to be Thr-577, Ser-585, Ser-589, and Ser-601 (Fig. 3, A-D) (33), all of which locate on F2 and F3. We constructed the T577A/ S585A/S589A/S601A (4A) mutant accordingly and assessed its effect. When HA-MYPT1-WT and 4A plasmids were transfected into cells, the 4A mutant significantly abrogated O-GlcNAcylation (Fig. 4A), suggesting that these four amino acids are major O-GlcNAc sites. Considering that MYPT1 is abundantly O-GlcNAcylated, and other proteomic screens have also identified extra glycosylation sites (34), our results do not exclude the possibility that there could be more O-GlcNAcylated residues on MYPT1.

O-GlcNAcylation of MYPT1 antagonizes CDK1-dependent phosphorylation at Ser-473
Because CDK1 phosphorylates MYPT1 at Ser-473 during mitosis and creates a binding motif between MYPT1 and the PBD of PLK1 (18), we surmised that O-GlcNAc of MYPT1 might interplay with pSer-473. To address this possibility, we used a phospho-specific antibody targeting pSer-473 that has been previously described and utilized (25).

O-GlcNAcylation of MYPT1 enhances PLK1 activity
As the MYPT1 associates PLK1 to target PP1c␤ to dephosphorylate and deactivate PLK1 (18), stronger affinity could signify less activity. We took advantage of the IP-phosphatase assay to examine PLK1 activity (22,24). Cells were transfected with Flag-MYPT1 and treated with Noc. Cells were also supplemented with TMG ϩ Glu to enrich for O-GlcNAc or were not treated. When the anti-FLAG immunoprecipitates were incubated with recombinant PLK1, the relatively-low O-GlcNAc group efficiently dephosphorylated PLK1, as detected by IB with PLK1-pThr-210 antibodies, but not the high O-GlcNAc group (Fig. 6A).
MYPT1-4A mutants were then directly exploited in the IP-phosphate assay. In the absence of Noc, MYPT1-WT decreased PLK1 pThr-210 levels, and the MYPT1-4A completely abolished PLK1-pThr-210 levels (Fig. 6B). This is consistent with our results in Fig. 5, C and D, that MYPT1-4A partners with PLK1 in the absence of Noc treatment.
Then to directly measure PLK1 kinase activity, we utilized IPkinase assay (Fig. 6C). Cells were transfected with Flag-PLK1 and then treated with TMG plus glucose or left untreated. Then Flag-Plk1 was IPed from cellular extracts and incubated with GSTmethylenetetrahydrofolate reductase (MTHFR), as our lab has demonstrated that MTHFR is a PLK1 substrate (37). Using the phospho-specific pThr-549 antibody, we observed

O-GlcNAc of MYPT1 regulates centrosome dynamics
an increase of kinase activity after TMG plus glucose treatment. Collectively, our biochemical assays suggest that O-GlcNAcylated MYPT1 disjoins PLK1 and promotes its kinase activity.

MYPT1-4A suppresses the TMG-induced centrosome disjunction defects
Because the aforementioned results suggest that MYPT1 O-GlcNAcylation is a pivotal regulator in centrosome separation, we undertook shMYPT1 to knock down endogenous MYPT1 (Fig. 7A), so that the effects of MYPT1-4A could be directly measured and observed after TMG incubation. As shown in Fig. 7B, the premature centrosome separation phenotype is discernable in the shMYPT1 cells that bear MYPT1-WT plasmids. But in the cells transfected with MYPT1-4A plasmids, the aberrancy is suppressed (Fig. 7C), in line with previous reports that PLK1 sequestration culminates in duplicated but unseparated centrosomes (38,39). Taken together, the 4A mutant fails to show the untimely centrosome separation phenotype, probably due to PLK1 suppression.
MYPT1 is one of the most abundant O-GlcNAcylated proteins, and its modification sites have been unveiled time and again in distinct proteomic studies (34,40,41). Perhaps O-GlcNAc sites might not be conserved between humans and mice. MYPT1 is found to be O-GlcNAcylated at Ser-564, Ser-566, Thr-570, and Ser-578 by MS in the mouse brain (41), but our data show in HeLa cells that O-GlcNAc occurs in Thr-577, Ser-585, Ser-589, and Ser-601. Previously, p53 is identified to be O-GlcNAcylated at Ser-149 (42), which is not conserved in mice either. The same also holds true for phosphorylation. For instance, ataxia telangiectasia-mutated (ATM), a vital sensor protein for DNA damage signaling, is phosphorylated at Ser-  O-GlcNAc of MYPT1 regulates centrosome dynamics (the mouse equivalent) does not hinder ATM function in mice (43,44). Therefore, extrapolating data across species needs extra caution, as the function and sites of PTMs could be context-dependent.
Along the same vein, there could be more O-GlcNAc sites on MYPT1, as the proteomic studies were carried out under disparate circumstances and using different click chemistry methodologies (33,34,40,41). As the O-GlcNAc modification is highly dynamic, distinct sites could be modified in response to environmental cues.
O-GlcNAc could have multifaceted effects on the centrosome. Both OGT and OGA overexpressions result in multipolar spindles, which could be repressed by TMG treatment (31). Nuclear mitotic apparatus protein (NuMA) is indispensable for spindle pole formation and regulates spindle pole cohesion (45,46). NuMA is O-GlcNAcylated, and its localization was led astray by OGT overexpression (47). Further investigations reveal that O-GlcNAcylated NuMA interacts with Galectin-3, which is a prerequisite for mitotic spindle cohesion and proper NuMA localization (48). Here, we reveal that centrosome dynamics is also governed by O-GlcNAcylation levels. As the centrosome is pivotal for the mitotic process, O-GlcNAc is bound to modulate other aspects of centrosome function.
The interaction between OGT and PLK1 has been examined before (47): cytologically, PLK1 partially colocalizes with OGT during mitosis; OGT overproduction will decrease PLKl protein levels, without affecting its localization patterns; and OGT overproduction does not alter PLK1 pThr-210 levels or pThr-210 localization.
Our results add a layer of regulation between OGT and PLK1 and indicate that O-GlcNAcylated MYPT1 attenuates interaction with PLK1 and thus promotes PLK1 activity. It is intriguing that overall PLK1 pThr-210 levels remain unaltered in OGT or OGA overproduction cells (31,47). We did not detect a discernable difference either, in cells supplemented with TMG plus glucose (data not shown). This may seem paradoxical at first, but considering the versatile roles of PLK1 during mitosis (4,(14)(15)(16), we could entertain the possibility that only a small pool of PLK1 is regulated by MYPT1. First, although the overall activity of PLK1 is up-regulated during mitosis, CDK1 actually dampens PLK1 activity via MYPT1 in a mitosis-specific fashion (18). Second, the pool of PLK1 responsible for kinetochoremicrotubule attachment actually contains low PLK1 kinase activity during metaphase so that microtubules could be dynamic (49). Therefore, irrespective of the overall elevation of mitotic activity, the mitotic master kinase PLK1 is perhaps indeed fine-tuned in space and time. And, O-GlcNAc could be the sweet icing on the cake.
Peroxidase-conjugated secondary antibodies were from Jackson ImmunoResearch. Blotted proteins were visualized using the ECL detection system (Amersham Biosciences). Sig-

O-GlcNAc of MYPT1 regulates centrosome dynamics
nals were detected by a LAS-4000 and quantitatively analyzed by densitometry using the MultiGauge software (FUJIFILM). All Western blottings were repeated at least three times.

Cell culture treatment
For chemical utilization, Noc at 100 ng/ml was used for 16 h; Ro 3306 (CDK1 inhibitor) was used at 2 M for the time indicated; BI2536 (PLK1 inhibitor) was used at 100 nM for 2 h; TMG (OGA inhibitor) was used at 5 M for 24 h; 5S-G (OGT inhibitor) was used at 100 M (prepared at 50 mM in DMSO) for 24 h. TMG ϩ Glu treatment was utilized as before (35,36). Specifically, cells were incubated with TMG (5 M) for 24 h and glucose (30 mM) for 3 h.

Indirect immunofluorescence
Indirect immunofluorescence staining was performed as described before (51). Dilutions of primary antibodies were 1:1000 for mouse anti-␥-tubulin. Cell nuclei were stained with DAPI. Quantitation was performed with the software ImageJ.
Author contributions-C. L., Y. S., Jie Li, X. L., and Jing Li data curation; C. L., Y. S., Jie Li, X. L., and Z. X. investigation; Y. S., Jie Li, X. C., and Jing Li formal analysis; Z. X. and X. C. resources; Z. T., X. C., and Jing Li conceptualization; Z. T. and Jing Li funding acquisition; Z. T. and Jing Li methodology; X. C. and Jing Li supervision; Jing Li writing-original draft; Jing Li project administration.