Re-examination of siRNA specificity questions role of PICH and Tao1 in the spindle checkpoint and identifies Mad2 as a sensitive target for small RNAs
The DNA-dependent adenosine triphosphatase (ATPase) Plk1-interacting checkpoint helicase (PICH) has recently been implicated in spindle checkpoint (SAC) signaling (Baumann et al., Cell 128(1):101–114, 2007). Depletion of PICH by siRNA abolished the SAC and resulted in an apparently selective loss of Mad2 from kinetochores, suggesting a role for PICH in the regulation of the Mad1–Mad2 interaction. An apparent rescue of SAC functionality by overexpression of PICH in PICH-depleted cells initially seemed to confirm a role for PICH in the SAC. However, we have subsequently discovered that all PICH-directed siRNA oligonucleotides that abolish the SAC also reduce Mad2 mRNA and protein expression. This reduction is functionally significant, as PICH siRNA does not abolish SAC activity in a cell line that harbors a bacterial artificial chromosome driving the expression of murine Mad2. Moreover, we identified several siRNA duplexes that effectively deplete PICH but do not significantly affect SAC functionality or Mad2 abundance or localization. Finally, we discovered that the ability of overexpressed PICH to restore SAC activity in PICH-depleted cells depends on sequestration of the mitotic kinase Plk1 rather than ATPase activity of PICH, pointing to an underlying mechanism of “bypass suppression.” In support of this view, depletion or inhibition of Plk1 also rescued SAC activity in cells harboring low levels of Mad2. This observation suggests that a reduction of Plk1 activity partially compensates for reduced Mad2 levels and argues that Plk1 normally reduces the strength of SAC signaling. Collectively, our results question the role of PICH in the SAC and instead identify Mad2 as a sensitive off target for small RNA duplexes. In support of the latter conclusion, our evidence suggests that an off-target effect on Mad2 may also contribute to explain the apparent role of the Tao1 kinase in SAC signaling (Draviam et al., Nat Cell Biol 9(5):556–564, 2007).
The DNA-dependent adenosine triphosphatase (ATPase) Plk1-interacting checkpoint helicase (PICH) was discovered as a binding partner and substrate of the mitotic kinase Plk1 (Baumann et al. 2007). During early mitosis, PICH is concentrated at the centromere/kinetochore (KT) region of mitotic chromosomes. In response to inhibition or depletion of Plk1, PICH spreads over chromosome arms, indicating that its localization is controlled by Plk1 activity (Baumann et al. 2007). Conversely, PICH apparently contributes to the recruitment of Plk1 to chromosome arms (Santamaria et al. 2007; Leng et al. 2008). Most interestingly, PICH associates with ultrafine DNA bridges (UFBs) that often connect the KTs of separating sister chromatids (Baumann et al. 2007; Wang et al. 2008). After inactivation of the SAC and cleavage of centromere-associated cohesin complexes by separase (Hauf et al. 2001; Uhlmann et al. 1999), PICH-positive UFBs elongate concomitant with sister chromatid separation and often reach lengths of several microns before they are finally resolved, presumably through the action of topoisomerase II (Baumann et al. 2007; Wang et al. 2008). While PICH presently constitutes the most universal marker for UFBs, it is intriguing that Bloom syndrome RecQ helicase as well as its complex partners RMI1 and topoisomerase III also associate with some PICH-positive structures (Chan et al. 2007; Bachrati and Hickson 2008). Moreover, a subset of noncentromeric UFBs was recently found to connect fragile site loci associated with Fanconi anemia proteins FANCD2 and FANCI, and the available evidence suggests that these latter UFBs reflect abnormal intertwined DNA structures induced by replicative stress (Chan et al. 2009; Naim and Rosselli 2009).
The conspicuous localization of PICH to the centromere/KT region originally suggested a role for this DNA-dependent ATPase in the SAC. In support of this idea, siRNA-mediated depletion of PICH by two different siRNA oligonucleotides (PICH-1 and PICH-2) abolished the checkpoint (Baumann et al. 2007), and results consistent with SAC failure were subsequently observed after depletion of PICH by a third siRNA oligonucleotide (hereafter referred to as PICH-CC; Leng et al. 2008). Furthermore, siRNA-mediated PICH depletion caused the apparently selective loss of the checkpoint protein Mad2 from KTs, while the localization of the Mad2-binding partner Mad1 was unaffected (Baumann et al. 2007). Independently, an apparently selective loss of Mad2 from KTs was reported in response to siRNA-mediated depletion of the protein kinase Tao1, whose apparent role in the SAC was discovered in a functional genomic screen (Draviam et al. 2007). At face value, these observations suggested that both PICH and Tao1 are required for SAC activity and that these two proteins might cooperate in regulating the Mad1–Mad2 interaction at KTs.
The present study was undertaken with the aim of elucidating the molecular mechanism underlying the purported checkpoint function of PICH. Although this function appeared to be supported by concordant results obtained with three different siRNA oligonucleotides, the studies described here lead us to conclude that the checkpoint failure formerly attributed to the depletion of PICH most likely reflects an off-target effect that causes the lowering of Mad2 transcript and protein. Our data further suggest that an off-target effect may similarly explain the purported checkpoint function of the Tao1 kinase. The direct (off) target of the various siRNA oligonucleotides examined here is unlikely to be the Mad2 transcript itself; instead, our results suggest that the regulatory network controlling Mad2 expression represents a particularly sensitive target for small RNAs. Interestingly, the off-target effect of a PICH-directed siRNA oligonucleotide could be rescued by overexpression of PICH, which a priori would seem to demonstrate specificity. However, we discovered that this apparent rescue required the ability of the overexpressed protein to sequester the mitotic kinase Plk1. A similar, albeit partial “rescue” could be accomplished by depletion or inhibition of Plk1, thus providing an example for the apparent “rescue” of a siRNA phenotype via bypass suppression.
Results and discussion
Re-examination of the proposed SAC function of PICH
If the previously observed loss of Mad2 protein from KTs in PICH-depleted cells reflected a depletion of Mad2 protein rather than a selective disruption of the Mad1–Mad2 interaction at KTs, we reasoned that it might be possible to restore Mad2 localization to KTs by overexpression of exogenous Mad2 protein in PICH siRNA-treated cells. This was indeed the case, as a Mad2–GFP fusion protein readily localized to KTs in a PICH-1 siRNA background (Supplementary Fig. 2a). In contrast, overexpressed Mad2–GFP did not localize to KTs in cells depleted of Aurora B (Supplementary Fig. 2b), consistent with previous reports (Ditchfield et al. 2003; Vigneron et al. 2004). These observations thus support the conclusion that the PICH siRNA phenotype reported previously might reflect a change in Mad2 abundance rather than localization.
To investigate the mechanism underlying the reduction in Mad2 protein, we next asked whether this reduction was also reflected at the mRNA level. To this end, PICH, Mad1, and Mad2 mRNAs were measured by quantitative real-time polymerase chain reaction (qRT-PCR) in HeLa cells depleted of either PICH or Mad1 for control. As summarized in Fig. 4c, all three PICH siRNAs displayed almost identical knockdown efficiencies with regard to PICH mRNA. Remarkably, all three PICH siRNAs also caused a significant decrease of Mad2 mRNA when compared to the siGL2 control, whereas Mad1 mRNA was not detectably affected (Fig. 4c). As expected, Mad1 siRNA depleted Mad1 mRNA but did not reduce either PICH or Mad2 mRNA (Fig. 4c). Taken together, these results suggest that treatment of HeLa cells with any one of the three previously published PICH siRNA oligonucleotides reduces not only PICH but also Mad2 mRNA levels, suggesting that the previously observed loss of checkpoint functionality might actually reflect a reduction, albeit partial, of cellular Mad2 protein. This in turn implies that either PICH protein somehow regulates Mad2 expression or, alternatively, that the three PICH-directed siRNA oligonucleotides all display off-target effects that cause a lowering of Mad2 expression.
Re-examination of the role of Tao1 kinase in the spindle checkpoint
Recently, a genome-wide siRNA screen identified a requirement for the Tao1 kinase (also known as microtubule affinity-regulation kinase kinase (MARKK); Johne et al. 2008; Timm et al. 2003) in the spindle checkpoint (Draviam et al. 2007). Moreover, Tao1 depletion was reported to cause a selective loss of Mad2 but not Mad1 from KTs (Draviam et al. 2007), highly reminiscent of the early data obtained after PICH siRNA (Baumann et al. 2007). Intrigued by this similarity, we originally suspected that PICH and Tao1 cooperate in a regulatory step controlling Mad2 localization. However, having obtained evidence that the apparent requirement for PICH in Mad2 localization is likely to reflect an off-target effect, we wondered whether a similar explanation might pertain to Tao1. To explore this possibility, we compared different Tao1-directed siRNA oligonucleotides for their efficiency to deplete Tao1, their effects on Mad2 localization, protein, and transcript levels, as well as their ability to abolish SAC activity. This survey included the most effective previously published Tao1 siRNA oligonucleotide (Tao1-NCB3) as well four new oligonucleotides (Tao1-2 to Tao1-5) and a smart pool comprising Tao1-2 to Tao1-5 (Tao1-SP).
Rescue of PICH and Tao1 siRNA phenotypes by Mad2 expression from a bacterial artificial chromosome
Uncovering of a regulatory influence of Plk1 on Mad2 function
We also analyzed the effect of Plk1 depletion or inhibition on cells that were treated with Tao1-NCB3 siRNA (Fig. 10b). Whereas control cells exhibited the expected mitotic timing, Tao1-NCB3 siRNA-treated cells separated their chromosomes in less than 20 min without forming a metaphase plate. The simultaneous depletion of Plk1 by siRNA or its inhibition by TAL increased the timing in Tao1-NCB3 siRNA-treated cells to about 35 min, similar to the duration seen in control cells but far below the duration of the mitotic arrest seen after depletion of Plk1 alone (Fig. 10b). This suggests that Plk1 depletion or inhibition only marginally restored SAC activity in Tao1-NCB3 siRNA-treated cells, presumably because residual Mad2 levels were too low (see below).
To demonstrate that residual Mad2 was critical for the restoration of SAC activity by Plk1 depletion/inhibition in PICH-1 or Tao1-NCB3 siRNA-treated cells, we also examined the consequences of combining Plk1 inhibition with direct siRNA-mediated depletion of Mad2. As shown in Fig. 10c, TAL addition could not restore normal mitotic timing to cells that were extensively depleted of Mad2, arguing that the absence of Plk1 activity cannot rescue SAC functionality in cells from which the bulk of Mad2 has been depleted. These results support the view that residual Mad2, as it persists after PICH-1 or Tao1-NCB3 siRNA treatment, is critical for the restoration of some SAC activity upon depletion or inhibition of Plk1. As a final test of this conclusion, different amounts of Mad2 siRNA oligonucleotides were used to reduce cellular Mad2 to levels within the range of those seen as a consequence of PICH-1 or Tao1-NCB3 siRNA treatment (Fig. 10d). Then, live-cell imaging was used to analyze mitotic timing in TAL-treated cells as a function of Mad2 levels. As shown in Fig. 10e, the mitotic timing (NEBD to anaphase onset) observed in response to Plk1 inhibition showed a strong correlation with Mad2 levels. Furthermore, Tao1-NCB3 siRNA resulted in a more severe depletion of Mad2 than PICH-1 siRNA and (Fig. 10d), concomitantly, a more pronounced advancement of mitotic timing (Fig. 10e), in excellent agreement with the data shown above (Fig. 10a, b). On the basis of these results, we propose that restoration of SAC activity by depletion or inhibition of Plk1 depends on residual levels of Mad2, implying that depletion or inhibition of Plk1 enhances the functionality of Mad2. In turn, these observations point to the conclusion that—under physiological circumstances—Plk1 antagonizes that SAC signaling function of Mad2.
Several conclusions emerge from this study. First and foremost, our results question the previously reported requirements for PICH (Baumann et al. 2007) and Tao1 (Draviam et al. 2007) for SAC activity. In the case of PICH, we find it difficult to escape the conclusion that the previous siRNA-based implication of this protein in the SAC reflects an off-target effect on Mad2. At present, the precise physiological function of PICH awaits further study, but considering its conspicuous localization to centromeres, KTs, and UFBs, a role in chromosome structure and segregation appears likely. This view is supported by the physical interaction between PICH and Plk1 and the influence the two proteins exert on each other’s localization (Baumann et al. 2007; Santamaria et al. 2007; Leng et al. 2008). Our study on Tao1 is less extensive, and some of our results differ from those reported in the original study, notably with regard to the effect of siRNA on Mad2 and the advancement of mitotic timing (Draviam et al. 2007). Also, we emphasize that Draviam and coworkers reported an effective rescue of Tao1 depletion by active but not inactive kinase (Draviam et al. 2007). Thus, further study will be required to definitively validate or exclude a role for Tao1/MARKK in the SAC. In any case, considering the role of this kinase in the regulation of microtubules (Johne et al. 2008; Timm et al. 2003), an important contribution to mitotic events is to be expected.
Finally, our study illustrates that even rescue experiments, the accepted “gold standard” for assessing siRNA specificity, must be interpreted with caution, unless rescuing proteins are re-expressed at physiological levels. In this study, we have in fact discovered that the apparent “rescue” of a siRNA phenotype depended on the ability of the “rescuing” PICH protein to sequester the mitotic kinase Plk1. This intriguing case of bypass suppression suggests that Plk1 normally reduces the strength of the Mad2-dependent inhibitory SAC signal. In future studies, it will clearly be interesting to explore the molecular mechanism(s) by which Plk1 affects SAC functionality.
Materials and methods
HeLa, U2OS, and HeLa cells stably expressing H2B–GFP were grown under standard conditions in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Cell cycle synchronizations were performed by a single thymidine (2 mM) block 10–12 h posttransfection with siRNA duplexes. For microscopic analysis of mitotic cell cycle stages, cells were released from thymidine for 12 h before fixation. Mitotic cells were isolated by mitotic shake off. The stable 293T TREX-shPICH cell lines were prepared under selective conditions using blasticidin (5 µg/ml) and puromycin (1 µg/ml) and grown under standard conditions. Stable clones were picked and analyzed by Western blotting and immunofluorescence microscopy before and after tetracycline induction.
HeLa cells (pools) containing bacterial artificial chromosomes (BACs) expressing C-terminally LAP-tagged mouse PICH (Ercc6L, BAC-ID RP24-326A8) or mouse Mad2L1 (BAC-ID RP23-84G11) were produced as described previously (Poser et al. 2008). In brief, BACs were obtained from the BACPAC Resources Center (http://bacpac.chori.org), the LAP (EGFP-IRES-Neo)-cassette was amplified by PCR using primers that carry homology to the C terminus of Ercc6L and Mad2L1, respectively, prior to transfection of the modified BAC.
Transfection of plasmids, siRNA, and plasmid constructions
Plasmid and siRNA transfections were performed using TransIT-LT1 transfection reagent (Mirus) and Oligofectamine reagent (Invitrogen), respectively, according to the manufacturer’s instructions. The siRNA sequences of GL2 (Elbashir et al. 2001), Mad2, PICH-1, PICH-2 (Baumann et al. 2007), PICH-CC (Leng et al. 2008), Tao1-NCB3 (Draviam et al. 2007), Aurora B (Klein et al. 2006), Mps1, and Mad1 (Stucke et al. 2002) have been published previously. The following siRNA target sequences were purchased from Dharmacon (ON-TARGETplus) unless otherwise stated: PICH-3: 3′-AGUAGGUGGUGUCGGUUUA; PICH-4: 3′-GGAUAGAGUUUACC GAAUU; PICH-5: 3′-CCAGAAACCUCAAUCGGAU; PICH-6: 3′-ACUUUAAGACAUUGCGAAU; the PICH oligo SMARTpool contained a mixture of PICH-3, PICH-4, PICH-5, and PICH-6 siRNAs; Tao1-2: 3′-CCAAGUAUCUCGUCACAAA; Tao1-3: 3′-GUAAUAUGGUCCUUU CUAA; Tao1-4: 3′-CUAAAGUGAUGUCCAAUGA; Tao1-5: 3′-GCUGUGAGUUGAUCAGAUU; the Tao1 oligo SMART pool contained a mixture of Tao1-2, Tao1-3, Tao1-4, and Tao1-5 siRNAs; Bub1: 3′-CCAGTGAGTTCCTATCCAATT (Qiagen).
To clone PICH constructs for expression in HeLa cells, PICH cDNA was amplified by PCR using pEGFP-C1-PICH (WT and T1063A; Baumann et al. 2007) as a template and the following primers: M3863-TCTCCCCGGGATGGAGGCATCCCGAAGGTTTC and M3864-ATAAGAATGCGGCCGCTCAATTGTTATTAAGTTGC. The Walker A mutant (K128A) was generated by site-directed mutagenesis using the following two oligonucleotides: M3821-GATGATATGGGATTAGGGGCGACTGTTCAAATCATTGCT and M3822-AGCAATGATTTGAACAGTCAGTCGCCCCTAATCCCATATCATC. For rescue experiments, mCherry-tagged PICH constructs were used. The cDNA for mCherry was PCR-amplified from pRSET-B-mCherry (a kind gift from Roger Y Tsien), using the following primers: M3914-CGGGGTACCGCCACCATGGTGAGCAAGGGCGAGGAGGAT and M3915-GCGATATCCTTGTACAGCTCGTCCATGCCG. The mCherry construct was then subcloned into KpnI and EcoRV sites of the pcDNA4/TO vector. Both WT and mutant PICH cDNAs were subcloned into the pcDNA4/TO-mCherry vector. To make constructs resistant against PICH-1 siRNA (Baumann et al. 2007), the above constructs were then subjected to site-directed mutagenesis using the following primers: M3344-GAGGGTGAGAAACAAGACTTATCCAGTATAAAGGTG and M3345-CACCTTTATACTGGATAAGTCTTGTTTCTCACCCTC. For construction of the shPICH plasmid, the following oligonucleotides were annealed in annealing buffer (200 mM Tris pH 7.5, 100 mM MgCl2 and 500 mM NaCl) for 1 h at 37°C and cloned into pTER +vector using BglII and HindIII restriction sites: M3666-GATCCCCAAGATCTCTCCAGTATA TTCAAGAGATTATACTGGAGAGATCTTGTTTTTGGAAA and M3667-AGCTTTTCCAAAAACAAGATCTCTCCAGTATAATCTCTTGAATTATACTGGAGAGATCTTGGG. (Italic letters indicate the siRNA sequence published for siPICH-1 (Baumann et al. 2007)). Full-length Mad2 was amplified by PCR and subcloned into pEGFP-C2 through BamH1 and Sal1 restriction sites. All constructs were verified by sequencing.
For imaging experiments, HeLa cells stably expressing histone H2B–GFP were used (Sillje et al. 2006). They were cultured in eight-well chamber slides (Ibidi) with 300 μl medium per well and transfected with siRNA oligonucleotides and/or plasmid constructs, as appropriate. Ten hours posttransfection, they were synchronized using 2 mM thymidine for 24 h and then released into fresh medium for 8 h prior to imaging. Live-cell imaging was performed using a Zeiss Axiovert 2 microscope equipped with a Plan Neofluar 20x1.6 optivar/NA = 0.75 objective, an environmental chamber, and a CoolSNAP-ES2 digital camera system. Movies were acquired over at least 12 h, with pictures taken at time intervals of 3 min. Images were captured with 2.5–5 ms exposure time for histone H2B–GFP, 30 ms for mCherry and 50 ms for DIC. Metaview software (Visitron Systems) was used for data collection and analysis.
Whole cell extracts were prepared using lysis buffer (20 mM Tris/HCl pH 7.4, 150 mM NaCl, 40 mM β-glycerophosphate, 10 mM NaF, 0.5% IGEPAL, 2 mM Pefabloc (Roth), 0.3 mM NaVO3, 100 µM ATP, 100 µM MgCl2, 100 nM okadaic acid, one protease inhibitor tablet per 10 ml lysis buffer (Roche)). Proteins were resolved by sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (Millipore), and these were probed with primary antibodies overnight at 4°C. Membranes were washed with 0.5% Tween/phosphate-buffered saline (PBS) before incubation with secondary antibodies. Antibodies used were rabbit anti-Mad2 (Bethyl), rabbit anti-Tao1 (Bethyl), rabbit anti-PICH (Baumann et al. 2007), monoclonal anti-α-Tubulin (DM1A, Sigma), mouse anti-Mad1 (Sigma), mouse anti-Aurora B (AIM1; BD Transduction), monoclonal anti-Mps1 (Stucke et al. 2002). Monoclonal anti-Bub1 and monoclonal anti-GFP antibodies were kindly provided by Andreas Uldschmidt.
Cells were grown on coverslips, fixed with PTEMF for 10 min at room temperature (Stucke et al. 2002), and incubated for 20 min at room temperature in 3% bovine serum albumin/PBS. Antibody incubations were carried out for 1 h at room temperature, followed by three washes in 0.5% TritonX-100/PBS. DNA was stained with 1 µg/ml 4,6-diamidino-2-phenylindole (DAPI). For image acquisition, a Deltavision microscope (Applied Precision) on a Nikon Eclipse TE200 base (Applied Precision) equipped with an Apo 60/1.4 oil immersion objective and a CoolSnap HQ camera (Photometrics) was used. Samples were examined with optical sections acquired 0.4 µm apart in the z-axis, deconvolved for each focal plane and projected into a single-plane image using the Softworx software (Applied Precision). For immunostaining, rabbit anti-Mad2 (Bethyl), rat anti-PICH (Baumann et al. 2007), and mouse anti-Aurora B (AIM1; BD Transduction) were used. Primary antibodies were detected by Cy3-conjugated donkey antibodies (Dianova) and Alexa-Fluor-647-conjugated goat antibodies (Invitrogen).
Quantitative real-time PCR
To monitor mRNA levels, HeLa cells were depleted of various proteins by siRNA and synchronized by a single thymidine block. To analyze the expression levels of Mad1, Mad2, PICH, and Tao1 transcripts, total RNA was extracted using an RNeasy Mini kit according to manufacturer’s protocol (Qiagen) and quality-controlled for integrity by capillary electrophoresis on Agilent 2100 Bioanalyzer. Transcript levels were determined by qRT-PCR. First-strand cDNA synthesis was carried out from 1 µg of total RNA by using SuperScript II Reverse Transcriptase and random primers following the manufacturer’s instructions (Invitrogen Life Technologies). The amplicons were designed with the program Primer Express 2.0 (Applied Biosystems, Foster City, CA, USA) by using default parameters such that they spanned exon boundaries. Primer sequences are available upon request. Amplicon sequences where checked by BLAST against the human genome to ensure that they were specific for the gene being assayed. The specificity of each primer pair as well as the efficiency of the amplification step was tested by assaying serial dilutions of cDNA. PCR reactions were carried out in triplicate by using an SDS 7900 HT instrument (Applied Biosystems) and Power SYBR Green PCR master mix kits (Applied Biosystems). Normalization genes were selected using the geNorm script as published (Vandesompele et al. 2002). Raw Ct values obtained with SDS 2.0 (Applied Biosystems) were imported into Excel (Microsoft, Redmond, WA, USA) to calculate the normalization factor and the fold changes with the geNorm (Vandesompele et al. 2002).
We thank Roger Tsien and Andreas Uldschmid for reagents, Christelle Barraclough for help with qRT-PCR, and Rainer Malik for help with bioinformatics. We also thank Christoph Baumann, Viji Draviam, Steve Elledge, and all members of the Nigg lab for helpful discussions. This work was supported by the Max-Planck-Society. L.H. W. acknowledges a fellowship from the Taiwan Merit Scholarships program (NSC-095-SAF-I-564-627-TMS). NCH gratefully acknowledges the Ph.D. fellowship from the Boehringer Ingelheim Fonds.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
- Poser I, Sarov M, Hutchins JR, Heriche JK, Toyoda Y, Pozniakovsky A, Weigl D, Nitzsche A, Hegemann B, Bird AW, Pelletier L, Kittler R, Hua S, Naumann R, Augsburg M, Sykora MM, Hofemeister H, Zhang Y, Nasmyth K, White KP, Dietzel S, Mechtler K, Durbin R, Stewart AF, Peters JM, Buchholz F, Hyman AA (2008) BAC TransgeneOmics: a high-throughput method for exploration of protein function in mammals. Nat Methods 5(5):409–415CrossRefPubMedGoogle Scholar
- Santamaria A, Neef R, Eberspacher U, Eis K, Husemann M, Mumberg D, Prechtl S, Schulze V, Siemeister G, Wortmann L, Barr FA, Nigg EA (2007) Use of the novel Plk1 inhibitor ZK-thiazolidinone to elucidate functions of Plk1 in early and late stages of mitosis. Mol Biol Cell 18(10):4024–4036CrossRefPubMedGoogle Scholar