Control of the pericentrosomal H₂O₂ level by peroxiredoxin I is critical for mitotic progression

Activation of the Cdk1–cyclin B complex occurs first at the centrosome during prophase, and its amplification through multiple feedback loops involving cyclin B, Cdc25B, Cdc25C, Plk, and Aurora A also occurs at this organelle (Jackman et al., 2003; Bonnet et al., 2008; Lindqvist et al., 2009). Successful cell cycle progression requires that many cell cycle regulators—including cyclins A and B, Plk1, and Aurora A—be degraded in a timely manner. Degradation of these regulators by the 26S proteasome results from their ubiquitination by the multisubunit ubiquitin ligase anaphase-promoting complex/cyclosome (APC/C). Activation of APC/C occurs at the centrosome and requires Cdc20 or Cdh1 as an activator protein (Peters, 2006; Pesin and Orr-Weaver, 2008; van Leuken et al., 2008; Wurzenberger and Gerlich, 2011). Cdh1 is prevented from interaction with APC/C when Cdh1 is phosphorylated by Cdks. APC/C–Cdh1 activity thus depends on both Cdks as well as Cdk-opposing phosphatases. The dual-specificity protein tyrosine phosphatase (PTP), Cdc14B, and the Ser/Thr phosphatases, PP1 and PP2A, have been proposed to function as Cdk1-opposing enzymes in mammalian cells (Bassermann et al., 2008; Mochida et al., 2009; Wu et al., 2009; Mocciaro and Schiebel, 2010; Schmitz et al., 2010; Domingo-Sananes et al., 2011).

A fraction of each of APC/C, Cdc20, Cdh1, and Cdk1-opposing phosphatases (Cdc14B, PP1, and PP2A) is present at the centrosome (Leach et al., 2003; Cho et al., 2005; Peters, 2006; Wu et al., 2008; Schmitz et al., 2010), as are Cdk1, Cdc25, cyclin B, Plk1, and Aurora A. At the onset of mitosis, Cdk1–cyclin B activity begins to increase as a result of positive feedback loops including cyclin B, Cdc25, Plk1, and Aurora A. The low level of incipient Cdk1 activity is likely insufficient to allow the accumulation of phosphorylated Cdh1 at the centrosome in the absence of concurrent suppression of the activity of Cdk1-opposing phosphatases, which, together with Cdh1, are enriched at this organelle. In the absence of such suppression of centrosomal phosphatase activity, further activation of Cdk1 would not be expected to occur because of the premature degradation of cyclin B, Plk1, and Aurora A.

Here, we find that the centrosomal levels of cyclin B, Plk1, and Aurora A as well as mitotic entry are likely regulated by the local concentration of H₂O₂ around the centrosome. We were led into this study by our previous observation that PrxI is inactivated when phosphorylated on Thr⁹⁰ by purified Cdk1–cyclin B (Chang et al., 2002). Peroxiredoxins (Prxs) are a major class of H₂O₂-eliminating enzymes (Rhee et al., 2012). Mammalian cells express six Prx isoforms (PrxI to PrxVI), which are implicated in a variety of cellular processes.

Whereas high H₂O₂ levels induce cell cycle arrest, low H₂O₂ levels are required for G₁–S and G₂–M phase transitions (Havens et al., 2006; Yamaura et al., 2009). The molecular mechanisms by which H₂O₂ modulates cell cycle progression have remained unclear, however. To examine the possible link between the role of H₂O₂ in cell cycle regulation and PrxI phosphorylation on Thr⁹⁰, we monitored this latter event during the cell cycle in HeLa cells that had been synchronized at the G₁–S border (0 h) with a double thymidine block and then released for various times. Phosphorylated PrxI (pPrxI) appeared slightly earlier than did the mitotic marker phosphorylated histone H3 (pHH3), and it disappeared in parallel with pHH3 (Fig. 1 A). When HeLa or U2OS cells arrested in prometaphase with nocodazole were released from the arrest, pPrxI disappeared rapidly, with the rate of its loss being slightly greater than that for cyclin B1 or pHH3 (Fig. S1 A).

The amount of pPrxI in prometaphase HeLa cells was estimated to be ∼0.4% of total PrxI (Fig. S1 B). We therefore reasoned that PrxI phosphorylation is likely a localized event, and we searched for its location in asynchronously growing HeLa cells using confocal microscopy. pPrxI was found to colocalize with the centrosome marker γ-tubulin at early stages of mitosis (prometaphase and metaphase) but not during interphase or late mitotic stages (anaphase, telophase, and cytokinesis; Fig. 1 B). The centrosome is composed of a pair of centrioles surrounded by pericentriolar material. Structured illumination microscopy (SIM; Fig. 1 C) revealed that although pPrxI is colocalized with Cep192, a pericentriolar material marker of mitotic cells (Kim et al., 2013), there are also regions of Cep192 that do not exhibit pPrxI. This may reflect unphosphorylated PrxI, although this remains to be verified. Release of HeLa cells from G₁–S arrest revealed that pPrxI was present at early and late prophase, prometaphase, and metaphase, but not during G₂, anaphase, telophase, or cytokinesis (Fig. S1 C). The appearance of pPrxI specifically at mitotic centrosomes was also detected in A431 and MCF7 cells (Fig. S1 D). The specificity of the antibodies to pPrxI was demonstrated using PrxI-deficient cells (Fig. S1, E and F). The centrosome-specific localization of pPrxI was also shown using biochemically isolated centrosomes (Fig. 1 D). Among the four cytosolic Prx enzymes (PrxI, II, V, and VI), only PrxI was detected at the centrosome (Fig. 1 D and Fig. S1 G).

When mouse embryonic fibroblasts (MEFs) were partially synchronized at mitosis and then released, the number of mitotic cells, estimated based on pHH3, was increased in PrxI-deficient MEFs compared with that in wild-type MEFs, and the levels of cyclin B1, Plk1, and Aurora A were higher in PrxI-deficient MEFs than in wild-type MEFs (Fig. S1 H).

To examine the effect of exposure of the centrosome to cytoplasmic H₂O₂, which would be expected to occur as a result of PrxI phosphorylation (inactivation), we infected HeLa cells with a retroviral vector encoding a modified form of catalase either with a centrosome-targeting pericentrin-AKAP450 centrosomal targeting (PACT) sequence (Cat-PACT) or with the corresponding sequence lacking the centrosome-targeting core (Cat-delPACT; Fig. 2 A; Gillingham and Munro, 2000). The presence of Cat-PACT and the absence of Cat-delPACT at the centrosome were verified by confocal immunofluorescence microscopy (Fig. 2 B). HeLa cells that had been infected with the empty, Cat-PACT, or Cat-delPACT vectors were synchronized at G₁–S by thymidine treatment and released in the presence of nocodazole for 10 h (Fig. 2 C). Early mitotic cells were scored on the basis of cell rounding (Fig. 2 D) and chromosome condensation (Fig. 2 E). The number of mitotic cells (mean of values from Fig. 2, D and E) for cells expressing Cat-PACT was reduced by ∼45% and ∼35% compared with that for cells infected with the empty vector or the vector encoding Cat-delPACT, respectively. Immunoblot analysis of the infected HeLa cells showed that the level of inactive (Tyr¹⁵ phosphorylated) Cdk1 in Cat-PACT cells was about twice that in Cat-delPACT cells and that the level of pHH3 in Cat-PACT cells was ∼50% of that in Cat-delPACT cells (Fig. 2, F and G). Such analysis also revealed that Cat-PACT and Cat-delPACT were expressed stably at low levels relative to endogenous peroxisomal catalase, suggesting that any effect of Cat-PACT present in the cytosol instead of at the centrosome as a result of excessive expression was minimal (Fig. 2 F). We also monitored mitotic entry by fluorescence imaging of live cells after the release of Cat-PACT and Cat-delPACT cells from G₁–S arrest. Mitotic entry of Cat-PACT cells was delayed by a mean of ∼2–3 h compared with that of Cat-delPACT cells (Fig. 2 H; and Videos 1 and 2). These results thus supported the notion that exposure of the centrosome to H₂O₂ is required for normal mitotic entry.

We also found that the centrosomal levels of cyclin B1, Plk1, and Aurora A in Cat-PACT cells were reduced by 55, 30, and 40%, respectively, compared with those in Cat-delPACT cells, suggesting that exposure of the centrosome to H₂O₂ is likely required to protect them from degradation (Fig. 2, I–K).

Reactive oxygen species including H₂O₂ increase gradually during cell cycle progression. There are multiple potential sources of H₂O₂ during mitosis, which include NADPH oxidase 4 (Yamaura et al., 2009) and arachidonic acid metabolism coupled to cPLA₂ (cytosolic phospholipase A₂; van Rossum et al., 2001; Cho et al., 2011). Indeed, we found that diphenyleneiodonium (DPI; NADPH oxidase inhibitor), AACOCF3 (cPLA₂ inhibitor) and methyl arachidonyl fluorophosphonate (MAFP; cPLA₂ inhibitor), and nordihydroguaiaretic acid (NDGA; lipoxygenase inhibitor) each attenuated mitotic entry in HeLa cells (Fig. 3, A–D). NADPH oxidase 4 and cPLA₂ are localized to the membranes of ER, which accumulates around the centrosome before nuclear breakdown (Whitaker, 2006; Lassègue et al., 2012). It is therefore possible that sources of H₂O₂ are arranged specifically to channel H₂O₂ to the centrosome in mitotic cells.

The role of H₂O₂ as a signaling molecule that regulates various biological processes has been well established. The best characterized H₂O₂ effectors are members of the PTP family such as PTP1B and PTEN (phosphatase and tensin homolog), which are reversibly inactivated by H₂O₂ produced in various cells stimulated with growth factors (Kwon et al., 2004; Tonks, 2005; Rhee, 2006). The redox sensitivity of these enzymes is attributable to a low-pK_a cysteine residue located within the PTP signature motif (HCX₅R). Although PTP1B, PTEN, and Cdc14 act on distinct substrates, they are structurally closely related, whereas Cdc25 phosphatases have a different structure, which is similar to that of rhodanese (Mustelin, 2007).

Oxidation of PTEN renders its structure more compact as a result of the formation of an intramolecular disulfide bond, with the oxidized protein migrating faster than the reduced form during nonreducing SDS-PAGE (Kwon et al., 2004). Like PTEN, Myc-tagged Cdc14B showed a reversible H₂O₂-induced increase in electrophoretic mobility under nonreducing conditions when tested using HeLa cells expressing Myc-tagged Cdc14B (Fig. S2, A and B), whereas no such mobility shift was observed with Myc-tagged forms of Cdc25B or Cdc25C (Fig. S2 B). Cdc14B appeared to be more sensitive to H₂O₂ than was PTEN (Fig. S2 B). Human Cdc14B contains nine Cys residues in addition to the catalytic Cys³¹⁴. Analysis of various Cys mutants of Cdc14B showed that only Cys²²⁸ and Cys³¹⁴ among the 10 Cys residues are required for the H₂O₂-dependent mobility shift (Fig. S2 C), suggesting that oxidized Cys³¹⁴ forms a disulfide bond with Cys²²⁸.

Although oxidation of Cys residues does not necessarily result in a gel mobility shift, oxidized Cys can be reliably detected by alkylation of reduced Cys residues followed by reduction of the oxidized cysteines and subsequent biotinylation of the newly formed thiols. We applied the biotinylation method to the cells that had been exposed to exogenous H₂O₂ maintained at low micromolar levels by incubation of cells with glucose and glucose oxidase. Such analysis also revealed that Cdc14B and PTEN, but not Cdc25B and Cdc25C, are sensitive to oxidation by H₂O₂ (Fig. S2 D). Collectively, these results suggested that centrosomal Cdc14B might be oxidatively inactivated during early mitosis. Direct evaluation of the oxidation status of centrosomal Cdc14B at different stages of the cell cycle was not possible because biochemical isolation of the centrosome requires treatment with 2-mercaptoethanol (Wu et al., 2008), which would reduce the disulfide bond of oxidized Cdc14B. The redox sensitivity of PP1 and PP2, which contain a redox-sensitive bimetallic center, has also long been described (Rusnak and Reiter, 2000; Pieri et al., 2003; Wright et al., 2009). Nevertheless, their redox regulation in cells has yet to be demonstrated.

The activation of APC/C–Cdh1 has been proposed to occur initially at the centrosome (Raff et al., 2002). Our results suggested that PrxI phosphorylation at the onset of mitosis might give rise to a series of events including an increase in the pericentrosomal concentration of H₂O₂, inactivation of Cdk1-opposing phosphatases and accumulation of phosphorylated Cdh1. To test this hypothesis, we produced rabbit antibodies to the Ser⁴⁰-phosphorylated form of human Cdh1 (pCdh1). Phosphorylation of Cdh1 is known to occur at as many as 8–11 sites in yeast, Xenopus laevis, and mammalian cells (Jaspersen et al., 1999; Kramer et al., 2000). All of these potential Cdk phosphorylation sites were identified indirectly with the use of site-directed mutants. Among them, the amino acid sequence surrounding Ser⁴⁰ in human Cdh1 is best conserved (Fig. S3 A). The specificity of the antibodies was demonstrated using Cdh1 phosphorylated by recombinant Cdk1–cyclin B (Fig. S3 B) and Cdh1-depleted HeLa cells (Fig. S3 C). We next examined whether Ser⁴⁰ phosphorylation is important for the binding of Cdh1 to APC/C. HA-tagged forms of wild-type Cdh1 or mutants thereof in which Ser⁴⁰ is replaced with alanine or aspartic acid (S40A or S40D, respectively) were expressed in HeLa cells, and the amount of HA-Cdh1 that coimmunoprecipitated with the APC/C complex was estimated in nocodazole-arrested cells. The amount of coprecipitated S40D (phosphomimetic mutant) was only ∼30% and ∼20% of that of the wild-type or S40A (nonphosphorylatable mutant) proteins, respectively (Fig. S3, D and E), suggesting that Ser⁴⁰ phosphorylation inhibits Cdh1 binding to APC/C.

Using the antibodies to pCdh1, we investigated the effect of Cat-PACT expression on Ser⁴⁰ phosphorylation in nocodazole-arrested cells. The amount of pCdh1 in HeLa cells expressing Cat-PACT was reduced by ∼30% compared with that in cells infected with the empty vector or expressing Cat-delPACT (Fig. 4, A and B), consistent with the notion that Cdh1 phosphorylation is highly influenced by pericentrosomal H₂O₂ level and that pCdh1 undergoes turnover at the centrosome (Raff et al., 2002).

The role of PP1 and PP2A in determining the phosphorylation state of PrxI and Cdh1 was tested in HeLa cells treated with okadaic acid (OA), a potent inhibitor of these phosphatases or silenced for both PP1 and PP2A. The abundance of both pPrxI and pCdh1 was increased by OA treatment (Fig. 4 C) or by silencing PP1 and PP2A (Fig. 4 D), indicating that phosphorylation of PrxI and Cdh1 by Cdk1 is reversed by PP1 and PP2A. Given that Cdc14B is highly concentrated at the centrosome, it is also possible that the OA effect on Cdh1 phosphorylation was mediated indirectly through the inactivation of Cdc14B by H₂O₂ that accumulates around the centrosome as a result of PrxI phosphorylation. As observed before (Picard et al., 1989), OA also promoted mitotic entry as indicated by an increase in the abundance of pHH3 (Fig. 4 C).

A series of events that are proposed to occur at the centrosome in connection to H₂O₂ is shown schematically in Fig. 5, beginning with PrxI phosphorylation. Whereas the intracellular concentration of H₂O₂ increases as the cell cycle progresses to G₂ phase, the centrosome is shielded from the high tide of H₂O₂ by the peroxidase activity of PrxI associated with the organelle. At the onset of mitosis, Cdk1–cyclin B phosphorylates PrxI, which results in exposure of the centrosome to H₂O₂ and consequent inactivation of Cdk1-opposing phosphatases represented by Cdc14B. Among the many Cdk1-phosphorylated proteins in mammalian cells, at least pCdh1 appears to be a substrate of Cdc14B in vitro and in vivo (Bassermann et al., 2008; Schindler and Schultz, 2009; Mocciaro and Schiebel, 2010). In late mitosis, PrxI is dephosphorylated, and the centrosome is consequently again shielded from H₂O₂, allowing sequential reactivation of phosphatases, dephosphorylation of Cdh1, and activation of APC/C-Cdh1. This newly discovered circuit provides a means to integrate input from diverse H₂O₂-generating cellular processes. The presence of PrxI at the centrosome might also serve as a safeguard to prevent unscheduled mitotic entry as a result of an accidental increase in H₂O₂ abundance.

Nocodazole, N-ethylmaleimide (NEM), and TCA were obtained from Sigma-Aldrich; DPI and OA were purchased from EMD Millipore; AACOCF3 was obtained from Santa Cruz Biotechnology, Inc.; and MAFP and NDGA were obtained from Cayman Chemical. Rabbit polyclonal antibodies specific for human PrxI, PrxII, PrxV, or PrxVI were described previously (Kang et al., 1998a,b; Seo et al., 2000), as were those specific for PrxI phosphorylated on Thr⁹⁰ (rabbit IgG; epitope, phosphopeptide CHLAWVNpTPKKQG, which corresponds to residues 83–95 of human Prx I; Chang et al., 2002). Rabbit polyclonal antibodies to Cdc14B, mouse monoclonal antibodies to Plk1, and rabbit antibodies to GFP were obtained from Invitrogen. Antibodies to cyclin A (rabbit IgG; H-432), to cyclin B1 (rabbit IgG; H-20), to Plk1 (mouse IgG; F-8, for immunofluorescence analysis), to HA (mouse IgG; F-7), to Cdk1 (mouse IgG; sc-54), to PP1 (mouse IgG; E-9), and to c-Myc (mouse IgG; 9E10) were obtained from Santa Cruz Biotechnology, Inc. Rabbit monoclonal antibodies to Cdc25C (5H9) and to Aurora A (1G4), rabbit polyclonal antibodies to Tyr¹⁵-phosphorylated Cdk1, rabbit monoclonal antibodies to PP2A C subunit (52F8), and mouse antibodies to cyclin B1 (for immunofluorescence analysis) were purchased from Cell Signaling Technology. Antibodies to β-actin (mouse IgG; AC-74) and to γ-tubulin (mouse IgG; GTU-88) were obtained from Sigma-Aldrich, mouse monoclonal antibodies to Cdh1 (Ab-2) were obtained from EMD Millipore, rabbit polyclonal antibodies to Ser¹⁰-pHH3 were obtained from EMD Millipore, rabbit monoclonal antibodies to PTEN (C-terminal) were from Epitomics, rabbit antibodies to catalase were from Young In Frontier, rabbit antibodies to pericentrin were obtained from Abcam, and mouse antibodies to HSP90 were obtained from BD. HeLa cells stably expressing histone H2B–GFP (a human histone H2B gene subcloned into pEGFPN1 vector; the expression was driven by the EF-1α promotor, and the vector contained the blasticidin-resistant gene) were provided by J.H. Lee (Ajou University, Suwon, South Korea). The pMIN retrovirus plasmid (a derivative of MPIN plasmid, the entire residual pol coding sequence was deleted; Yu et al., 2000) was obtained from Viromed, the pHyPer-Cyto plasmid (catalog no. FP941, cytomegalovirus promoter) was from Evrogen, and pHyPer-C199S (H₂O₂-sentive Cys¹⁹⁹ was replaced with serine) was described previously (Poburko et al., 2011). MEFs derived from PrxI knockout or catalase knockout mice were prepared at embryonic day 13.5 from embryos obtained by mating PrxI^+/− mice (the sequence from exon 1 to exon 6 of a Prx I gene was replaced with the neomycin gene; provided by D.Y. Yu, Korea Research Institute of Biology and Biotechnology, Daejon, South Korea; Han et al., 2012) or Cat^+/− mice (the BamHI genomic DNA fragment of the mouse catalase gene, containing parts of intron 4 and exon 5, was replaced by a neomycin resistance cassette in the gene targeting vector; provided by Y.S. Ho, Wayne State University, Detroit, MI; Ho et al., 2004).

HeLa, U2OS, A431, and MCF7 cells as well as MEFs were cultured in DMEM supplemented with 10% FBS (Gibco) and penicillin-streptomycin (Hyclone). Cell lines were subjected to transient transfection with expression plasmids with the use of the Effectene (QIAGEN) or FuGENE 6 (Roche) reagents, and synthetic siRNAs were introduced into cells with the use of the Oligofectamine reagent (Invitrogen). MEFs were transfected by the Neon transfection method (Invitrogen). Synchronized cells were prepared and released as described previously (Fang et al., 1998; Whitfield et al., 2002), with minor modifications. In brief, for synchronization at the G₁–S border, HeLa cells were grown for 18 h in complete medium containing 2 mM thymidine (Sigma-Aldrich), washed with PBS, incubated for 8 h in fresh medium without thymidine, and cultured again for 18 h in medium containing 2 mM thymidine. The cells were then transferred to fresh medium, and samples were harvested every 1–3 h. For arrest in prometaphase, HeLa cells were treated with 2 mM thymidine for 18 h, released into fresh medium for 3 h, and treated with 100 ng/ml nocodazole (Sigma-Aldrich) for 10 h. The cells were then collected every 0.5–2 h after their transfer to fresh medium. MEFs were synchronized as described previously (García-Higuera et al., 2008), with some modifications. In brief, cells were arrested in G₀ by serum deprivation (0.5% FBS) for 72 h, released in medium containing 10% serum, and in the presence of 0.2 µg/ml aphidicolin (Sigma-Aldrich) for 24 h for G₁–S synchronization, and then treated with 500 ng/ml nocodazole and 5 µM paclitaxel (Sigma-Aldrich) for 12 h for mitotic synchronization.

A full-length Cdc14B cDNA was amplified by PCR from a human testis cDNA library (Takara Bio Inc.). The sequences of PCR primers for the first and second reactions were 5′-CCCCCTGACGGGCCGC-3′ (first, forward), 5′-TTCCTTCAGCTCTGGTCACA-3′ (first, reverse), 5′-AATTGGCCGGCCCATGAAGCGGAAAAGCGAGC-3′ (second, forward), and 5′-ATATGGCGCGCCTTAACGCAAGACTCTTTTAGTC-3′ (second, reverse). The amplified coding region was then cloned into the Fse1 and Asc1 sites of the pCS2-Myc vector (cytomegalovirus promoter, 6 Myc epitope; provided by D. Turner, University of Michigan, Ann Arbor, MI) for expression of a Myc epitope–tagged protein. Point mutations of Cdc14B were generated with the use of a QuikChange Site-Directed Mutagenesis kit (Agilent Technologies) with pCS2-Myc-Cdc14B as the template. The plasmids pCS2-HA-Cdh1(S40A) and pCS2-HA-Cdh1(S40D) were constructed with the same kit and with pCS2-HA-Cdh1 as the template. Mutations were verified by DNA sequencing. Double-stranded siRNA oligonucleotides for PrxI and Cdh1 were synthesized by GE Healthcare and were previously validated (Lee et al., 2011; Song et al., 2011); the target sequences are 5′-AAACTCAACTGCCAAGTGATT-3′ and 5′-CGCTTGTCTGAGGATTACG-3′ for PrxI and 5′-GGAACACGCTGACAGGACA-3′ (siCdh1-1) and 5′-GAAGAAGGGTCTGTTCACG-3′ (siCdh1-2) for Cdh1. A control siRNA for GFP (5′-GTTCAGCGTGTCCGGCGAGTT-3′) was obtained from Samchully Pharma. Pan-PP1 siRNA (for the α, β, and γ isoforms; sc-43545) were obtained from Santa Cruz Biotechnology, Inc. Double-stranded siRNA oligonucleotides for PP2A-Cα and PP2A-Cβ were synthesized from GE Healthcare. SMARTpool siRNA target sequences consist of four siRNA targeting multiple sites on PP2A-Cα and PP2A-Cβ. The siRNA sequences for PP2A-Cα are 5′-CCGGAAUGUAGUAACGAUU-3′, 5′-ACAUUAACACCUCGUGAAU-3′, 5′-UCAUGGAACUUGACGAUAC-3′, and 5′-CAGGUAGAGCUUAAACUAA-3′. The siRNA sequences for PP2A-Cβ are 5′-CACGAAAGCCGACAAAUUA-3′, 5′-UUUAGUAGAUGGACAGAUA-3′, 5′-CCAGAACGCAUUACAAUAU-3′, and 5′-GAACCAGGCUGCUAUCAUG-3′. Human Cdc25B cDNA (GenBank accession no. BC051711) was obtained from GE Healthcare and cloned into pCS2-Myc, and human His₆-Cdc25C cDNA (plasmid 10964) was obtained from T. Finkel (National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD) through the Addgene repository.

Asynchronously growing HeLa cells were fixed with methanol and immunostained with antibodies to pPrxI or with antibodies to Alexa Fluor 647–conjugated Cep 192 C (Kim et al., 2013). SIM image was obtained using a microscope (ELYRA PS.1; Carl Zeiss) with a 63×/1.4 NA objective and an electron-multiplying charge-coupled device (1,024 × 1,024 pixels, 8 × 8–µm pixel size, 65% quantum efficiency; iXon 885; Andor Technology), for a maximum field of view of 80 × 80 µm of the sample. To get line intensity histograms and a 3D-rendering image, acquired images of mitotic HeLa cells were analyzed with Zen software (Carl Zeiss) and NIS elements AR 3.0 software (Nikon).

For generation of the pCat-PACT plasmid, the PACT cDNA was amplified from dsRed-PACT (provided by D.S. Hwang, Seoul National University, Seoul, South Korea) and inserted into a plasmid for human catalase (provided by S.W. Kang, Ewha Woman’s University, Seoul, South Korea) from which the DNA sequence encoding the C-terminal KANL residues had been removed. For generation of pCat-delPACT, a region of the PACT sequence (K594–W675) containing two motifs essential for centrosome targeting was removed from pCat-PACT.

Rabbits were injected with a keyhole limpet hemocyanin–conjugated phosphopeptide (SPVSpSPSKHG) corresponding to amino acid residues 36–45 of human Cdh1. Antibodies that recognize the nonphosphorylated peptide were removed from serum with the use of an affinity resin conjugated with the nonphosphorylated peptide. Specific antibodies were then further purified with the use of a resin conjugated with the phosphopeptide.

Reduced and oxidized forms of Cdc14B were identified on the basis of a mobility shift on nonreducing SDS-PAGE as described previously (Kwon et al., 2004), with minor modifications. After treatment with H₂O₂, HeLa cells (10⁶ cells) were scraped into 1 ml PBS and mixed with 0.2 ml of ice-cold 50% TCA. The mixture was centrifuged at 2,000 g for 5 min at 4°C, and the resulting pellet was washed with ice-cold acetone and then solubilized in 0.2 ml of 100 mM Tris-HCl, pH 6.8, containing 2% SDS and 40 mM NEM. Solubilized proteins (20 µg) were subjected to SDS-PAGE under nonreducing conditions and then transferred to a nitrocellulose membrane for immunoblot analysis with antibodies to Cdc14B or to the Myc epitope tag.

Recombinant human PrxI was expressed in and purified from bacteria as previously described (Seo et al., 2000). Complementary DNA corresponding to PrxI protein was expressed in Escherichia coli and PrxI was purified by ammonium sulfate fractionation (40–60% saturation) and sequential chromatography on DEAE-Sephacel ion exchange and HPLC TSK heparin-5PW columns. The protein (10 µg) was phosphorylated by incubation for 1 h at 37°C with 8 µg Cdk1–cyclin B (New England Biolabs, Inc.) in a final volume of 50 µl of a solution containing 50 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 1 mM EGTA, 1 mM DTT, and 200 µM ATP. One-half of the product was used to estimate the proportion of Thr⁹⁰-phosphorylated PrxI by 2D-PAGE analysis, and the other half was used as a standard for the quantification of pPrxI in lysates of mitotic HeLa cells.

For detection of Cdc14B oxidation in Fig. 5 D, an enhanced method based on Cys biotinylation was performed as previously described (Kwon et al., 2004), with some modifications. In brief, cells (10⁶ per 100-mm dish) exposed to glucose oxidase in high-glucose medium were washed with ice-cold PBS, scraped into 1 ml of the same solution, transferred to a microfuge tube, isolated by centrifugation at 8,000 g for 1 min at 4°C, and rapidly frozen in liquid nitrogen. The frozen cells were incubated for 1 h at room temperature with 1 ml of oxygen-free extraction buffer (50 mM sodium phosphate, pH 7.0, 1 mM EDTA, 10 mM NEM, 10 mM iodoacetic acid, 1% Triton X-100, 5 mM NaF, 50 µg/ml leupeptin, 50 µg/ml aprotinin, and 1 mM 4-(2-aminoethyl)benzene-sulfonyl fluoride) in an anaerobic chamber. The samples were then centrifuged, SDS was added to the harvested supernatants to a final concentration of 1%, and the mixtures were incubated for 2 h at room temperature in the dark. 500-µg portions of the denatured proteins were precipitated by the addition of TCA to a final concentration of 10% and incubation for 1 h at room temperature. The precipitates were washed twice with acetone that had been cooled on dry ice and were then reduced by incubation for 30 min at 50°C in 0.1 ml of oxygen-free reducing buffer (50 mM Hepes-NaOH, pH 7.7, 1 mM EDTA, 2% SDS, and 4 mM DTT) in an anaerobic chamber. The reduced proteins containing free sulfhydryl groups were biotinylated by incubation for 30 min at 50°C with 0.9 ml of a solution containing 50 mM sodium phosphate, pH 7.0, 1 mM EDTA, and 1 mM biotin that had been conjugated to polyethylene oxide–maleimide (Thermo Fisher Scientific). The reaction was stopped by the addition of DTT to a final concentration of 1 mM, and proteins were precipitated by incubation with TCA at a final concentration of 10% for 1 h. The precipitates were washed with dry ice–chilled acetone and then solubilized in 0.2 ml of a solution containing 50 mM Hepes-NaOH, pH 7.7, 1 mM EDTA, and 2% SDS, with ultrasonic treatment for 15 s. The samples were diluted with 0.2 ml of the same solution without SDS, and 40-µl portions of the resulting mixtures were saved for immunoblot analysis. The remaining 360-µl portion of each mixture was diluted further with the same solution without SDS until the final concentration of SDS was 0.5%. Biotinylated proteins were then precipitated by incubation for 1 h at room temperature with 3 µl of packed UltraLink Immobilized NeutrAvidin (Thermo Fisher Scientific). The beads were washed five times with a solution containing 20 mM Hepes-NaOH, pH 7.7, 200 mM NaCl, 1 mM EDTA, and 0.5% SDS, and the biotinylated proteins were released from the beads by boiling in SDS-PAGE sample buffer and then subjected to immunoblot analysis with antibodies to Cdc14B. Biotinylated PTEN was detected on the stripped blots with antibodies to PTEN.

Subcellular fractionation was performed as described previously (Bornens et al., 1987), with minor modifications. Prometaphase-arrested HeLa cells (10⁷) obtained by treatment with thymidine and nocodazole were lysed in 5 ml of a hypotonic solution (1 mM Hepes-NaOH, pH 7.2, 0.5% NP-40, 0.5 mM MgCl₂, 0.1% 2-mercaptoethanol, proteinase inhibitor cocktail, 50 mM NaF, and 1 mM sodium orthovanadate). The cell lysate was centrifuged at 2,500 g for 10 min at 4°C to remove nuclei and debris, and the resulting supernatant was collected and filtered through a 50-µm nylon mesh for analysis as a total cell lysate without nuclei. Hepes and DNase were added to a portion of the latter preparation to final concentrations of 10 mM and 2 U/ml, respectively, and the mixture was incubated for 30 min on ice before layering on top of 0.5 ml of a solution containing 60% (wt/wt) sucrose, 10 mM Pipes-NaOH, pH 7.2, 0.1% Triton X-100, and 0.1% 2-mercaptoethanol in a 5-ml centrifuge tube. The gradient was centrifuged at 10,000 g for 30 min at 4°C, after which the bottom portion (1.5 ml) containing centrosomes was purified further by layering on top of a discontinuous sucrose gradient consisting of 0.5 ml of 70%, 0.3 ml of 50%, and 0.3 ml of 40% sucrose in the same solution followed by centrifugation at 120,000 g for 1 h at 4°C. 0.2-ml fractions were collected from the bottom of the tube, diluted with 1 ml of 10 mM Pipes-NaOH, pH 7.2, and centrifuged at 15,000 g for 10 min at 4°C. The resulting pellets were subjected to immunoblot analysis with antibodies to γ-tubulin and to HSP90. The two fractions containing the largest amounts of γ-tubulin were pooled and used as the centrosomal fraction. The final four fractions containing the largest amounts of HSP90 were pooled and used as the cytosolic fraction.

For analysis of cell cycle stage, cells (5 × 10⁵/ml) were washed twice with ice-cold PBS, fixed overnight at 4°C in 70% ethanol, and stained with 1 ml of a solution containing 50 µg/ml RNase and 50 µg/ml propidium iodide before flow cytometry with a FACSCalibur instrument (BD).

Cells were cultured in 12-well dishes containing coverslips (diameter, 12 mm) coated with poly-l-lysine for both live-cell imaging and immunofluorescence staining. For the latter, the cells were fixed for 10 min on ice in 100% methanol or 4% formaldehyde in PBS, exposed for 30 min at room temperature to PBS containing 5% horse serum (Gibco-BRL) and 0.1% Triton X-100, and then incubated for 30 min at room temperature with primary antibodies to phosphorylated PrxI, to cyclin B1, to Plk1, or to Aurora A in the same solution. After three 5-min washes with PBS, the cells were incubated for 30 min at room temperature with Alexa Fluor 488–conjugated goat secondary antibodies (Invitrogen) at a 1:1,000 dilution in PBS containing 5% horse serum and 0.1% Triton X-100. For coimmunostaining of a centrosome marker, the cells were again washed three times with PBS, exposed for 30 min to PBS containing 5% horse serum and 0.1% Triton X-100, and incubated for 30 min with primary antibodies to γ-tubulin or to pericentrin in the same solution before detection of immune complexes with Alexa Fluor 546–conjugated goat secondary antibodies (Invitrogen) at a 1:500 dilution. The cells were also stained with 0.2 µg/ml DAPI to detect DNA. Confocal images were acquired using a 60× Plan Apochromat VC objective, NA 1.40, by illuminating with a 488-nm multi-Ar laser (for excitation of Alexa Fluor 488 fluorochrome) or with a 561-nm diode-pumped solid-state laser (for excitation of Alexa Fluor 546 fluorochrome) with a microscope (A1R; Nikon) equipped with galvano detector, and images were processed with NIS elements AR 3.0 software. For live-cell imaging, cells were maintained in an incubation chamber (Chamlide TC; Live Cell Instrument) at 37°C under an atmosphere of 5% CO₂ in air.

HeLa cells (3 × 10⁶) transiently expressing HA-tagged wild-type or mutant (S40D or S40A) forms of human Cdh1 were synchronized at prometaphase by treatment with thymidine and nocodazole. The cells were then lysed by ultrasonic treatment in 1 ml of a solution containing 50 mM Tris-HCl, pH 7.7, 150 mM NaCl, 0.5% NP-40, 1 mM DTT, a proteinase inhibitor cocktail, 50 mM NaF, and 0.5 µM OA, the lysates were centrifuged at 15,000 g for 30 min at 4°C, and the resulting supernatants (800 µl) were incubated for 4 h at 4°C with 10 µl of protein A–conjugated beads that had been coated with 10 µg of antibodies to Cdc27 (H-300; Santa Cruz Biotechnology, Inc.). The beads were then washed five times with the cell lysis buffer containing 500 mM NaCl and twice with the original lysis buffer, after which bound proteins were eluted with SDS sample buffer and subjected to immunoblot analysis.

Quantitative data are presented as means ± SD (unless indicated otherwise) and were analyzed with Student’s t test. A P < 0.05 was considered statistically significant.

Fig. S1 demonstrates phosphorylation of PrxI on Thr⁹⁰ at the centrosome of mitotic cells. Fig. S2 shows the sensitivity of PTPs to H₂O₂. Fig. S3 shows identification of a potential Cdk phosphorylation site in Cdh1. Videos 1 and 2 show live-cell imaging of HeLa cells expressing histone H2B-GFP and Cat-delPACT or histone H2B–GFP and Cat-PACT, respectively. Additional data are available in the JCB DataViewer at http://dx.doi.org/10.1083/jcb.201412068.dv.

We thank Jung-Eun Park at the National Cancer Institute (Bethesda, MD) for providing all the technical assistance for SIM imaging.

This study was supported by grants from the Korean Science and Engineering Foundation (National Honor Scientist program grant 2006-05106 and Bio R&D program grant M10642040001-07N4204-00110 to S.G. Rhee) and the National Research Foundation of Korea (Basic Science Research program grants 2011-0010514 and 2012R1A1A2002268 to D. Kang).

The authors declare no competing financial interests.

Author contributions: D. Kang and S.G. Rhee conceived the study and supervised experiments. J.M. Lim, H.A. Woo, and D. Kang performed experiments and analyzed data. K.S. Lee provided critical reagents. D. Kang and S.G. Rhee wrote the paper. All authors discussed and commented on the manuscript.