A Functional Analysis of MELK in Cell Division Reveals a Transition in the Mode of Cytokinesis During Xenopus Development

958 Research Article A functional analysis of MELK in cell division reveals a transition in the mode of cytokinesis during Xenopus development

Yann Le Page*,‡, Isabelle Chartrain*, Caroline Badouel§ and Jean-Pierre Tassan¶ UMR 6061 CNRS Université de Rennes 1, IFR140 GFAS, Equipe Développement et Polarité Cellulaires, 2 avenue du Professeur Léon Bernard, CS 34317, 35043 Rennes CEDEX, France *These authors contributed equally to this work ‡Present address: CNRS UMR 6026, 35042 Rennes CEDEX, France §Present address: Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, ON M5G1XT, Canada ¶Author for correspondence ([email protected])

Summary MELK is a serine/threonine kinase involved in several cell processes, including the cell cycle, proliferation, apoptosis and mRNA processing. However, its function remains elusive. Here, we explored its role in the Xenopus early embryo and show by knockdown that xMELK (Xenopus MELK) is necessary for completion of cell division. Consistent with a role in cell division, endogenous xMELK accumulates at the equatorial cortex of anaphase blastomeres. Its relocalization is highly dynamic and correlates with a conformational rearrangement in xMELK. Overexpression of xMELK leads to failure of cytokinesis and impairs accumulation at the division furrow of activated RhoA – a pivotal regulator of cytokinesis. Furthermore, endogenous xMELK associates and colocalizes with the cytokinesis organizer anillin. Unexpectedly, our study reveals a transition in the mode of cytokinesis correlated to cell size and that implicates xMELK. Collectively, our findings disclose the importance of xMELK in cytokinesis during early development and show that the mechanism of cytokinesis changes during Xenopus early development.

Key words: Cytokinesis, Early embryos, Mitosis, Xenopus, MELK

Introduction complex. The subcellular localization of MELK is also cell cycle The maternal embryonic leucine zipper kinase [MELK (Heyer et dependent as a fraction of MELK protein is redistributed at the cell

Journal of Cell Science al., 1997)], also known as pEg3 in Xenopus, is a cell-cycle- cortex specifically during anaphase and telophase (Chartrain et al., dependent protein kinase of the KIN1/PAR-1/MARK family (Blot 2006). Together, these results indicate that mitosis is crucial in et al., 2002; Davezac et al., 2002; Tassan and Le Goff, 2004). MELK regulation; however, the function of Xenopus MELK MELK is involved in diverse functions, including the cell cycle (xMELK) remains elusive. (Davezac et al., 2002), cell proliferation (Nakano et al., 2005), In this study, we asked how MELK is involved in mitosis. We apoptosis (Jung et al., 2008; Lin et al., 2007) and RNA processing investigated the role of xMELK in early embryos in which very (Vulsteke et al., 2004). Its expression is specific to cells engaged rapid cell cycles comprising only S- and M-phases occur in the in the cell cycle and becomes undetectable upon cell differentiation absence of gene transcription until the mid-blastula transition (Badouel et al., 2010). Recently, MELK has emerged as an (MBT) (Newport and Kirschner, 1982a; Newport and Kirschner, important gene in the field of cancer research. Indeed, several 1982b). xMELK is encoded by a maternal mRNA, and its studies have shown that MELK expression is dramatically increased expression, phosphorylation and catalytic activity are tightly in a large spectrum of cancers (Gray et al., 2005; Marie et al., controlled during oocyte maturation and early embryonic cleavage 2008; Nakano et al., 2008). Moreover, a high level of MELK (Badouel et al., 2006; Blot et al., 2002; Paris and Philippe, 1990). mRNA is correlated with malignancy grade in brain tumors (Marie These fine controls suggested that xMELK might have an important et al., 2008; Nakano et al., 2008) and is associated with poor role in early embryos. prognosis in breast cancer (Pickard et al., 2009). Our results reported here show that perturbing xMELK In human cells and Xenopus embryos, MELK catalytic activity expression by either knockdown or overexpression leads to abortive is correlated with its phosphorylation and is maximal during mitosis cell division in embryos. Shortly before cells start cytokinesis, (Blot et al., 2002; Davezac et al., 2002). We have shown that, in endogenous xMELK as well as other essential cytokinetic proteins M phase, MELK is phosphorylated at multiple sites (Badouel et accumulate in a narrow band at the equatorial cortex. A FRET- al., 2006) and that the two major mitotic kinases, MPF (cyclin-B– based probe shows that a conformational change correlating with CDK1 complex) and the mitogen-activated protein kinase (MAPK) xMELK activation accompanies relocalization of the kinase at the ERK2, participate in these phosphorylations and enhance MELK cell cortex. Overexpression of active xMELK, in contrast to a activity in vitro. MELK is potentially involved in cell cycle kinase-dead mutant, leads to defective constriction of the progression through association with CDC25 (Davezac et al., 2002; cytokinetic ring. xMELK overexpression induces a marked decrease Mirey et al., 2005). This protein phosphatase controls entry into in accumulation of activated Rho GTPase at the cleavage furrow, mitosis by dephosphorylating and activating the CDK1–cyclin-B thus providing a direct explanation for the xMELK-induced failure xMELK in cytokinesis of Xenopus embryos 959

of cytokinesis. Finally, we show that xMELK is associated with fertilization, xMELK is rapidly dephosphorylated (Blot et al., anillin, a crucial cytokinetic protein, and that the two proteins 2002), and, during the first cell cycle, its levels are transiently colocalize at the equatorial cortex. Interestingly, our study also decreased by approximately 50%. In cleaving embryos, the levels reveals that a developmentally regulated transition involving of xMELK are similar to those present in eggs and remain xMELK occurs in the mode of cytokinesis during early unchanged during early development. Such sophisticated expression embryogenesis. controls suggested that xMELK could have a function during the cleavage period that follows fertilization. Therefore, to determine Results this function, we altered xMELK expression in early development. xMELK knockdown before fertilization leads to cell As Xenopus eggs contain a supply of maternal xMELK protein division defect in embryos accumulated during oocyte maturation (Fig. 1A), we performed During Xenopus oogenesis and early embryogenesis, xMELK xMELK knockdown before fertilization. This relies on the expression, phosphorylation and activity are tightly controlled inhibition of xMELK neosynthesis during oocyte maturation by (Badouel et al., 2006; Blot et al., 2002; Paris and Philippe, 1990). inducing specific mRNA degradation with two antisense Indeed, during oocyte maturation, xMELK protein levels increase phosphorotioate-modified deoxyoligonucleotides (AS9 and AS11 by approximately 2.5 fold, and xMELK is phosphorylated, inducing xMELK). The host transfer method (Heasman et al., 1991) was a decrease in its electrophoretic mobility (Fig. 1A). After used to obtain embryos from in-vitro-cultured full-grown oocytes injected with AS deoxyoligonucleotides. These interfered neither with the oocyte maturation (Table 1 and supplementary material Fig. S1) nor with the extrusion of the first (data not shown) and the second polar body after release of meiosis II by parthenogenic activation (supplementary material Fig. S1). When compared with uninjected oocytes and oocytes treated with a control AS (‘AS Co’), only AS deoxyoligonucleotides against xMELK induced xMELK mRNA degradation in oocytes (Fig. 1B), leading to a marked decrease of xMELK protein levels in embryos (Fig. 1C). Time-lapse video-recording showed that the large majority of embryos injected with AS Co cleaved as expected (Fig. 1D and Table 1). By contrast, AS9- and AS11-xMELK-treated embryos either did not cleave or, for the majority, presented abortive cleavages (Fig. 1D and Table 1). Abortive cleavage furrows of AS- xMELK-injected embryos appeared at the same time as in controls. Moreover, surface contraction waves, a marker of cell-cycle progression (Hara et al., 1980), occurred with the same frequency as controls, indicating that the cell cycle progression is not impaired

Journal of Cell Science in AS-xMELK embryos. Altogether, these results demonstrate that xMELK is a maternal-effect gene required for early embryonic cleavages and that xMELK knockdown impairs the progression, but not the initiation, of the cleavage furrow.

Cytokinesis progression is inhibited by xMELK overexpression To gain insight into the xMELK function, we also overexpressed xMELK and examined the effects of increased levels in early embryos. In-vitro-transcribed mRNAs coding xMELK were microinjected into a single blastomere of two-cell-stage embryos. xMELK overexpression efficiently induced inhibition of blastomere division, leading to the appearance of abnormally large cells in all Fig. 1. Abortive cell divisions induced by xMELK knockdown. (A)xMELK embryos (Fig. 2A). The expressivity of the xMELK-induced expression was followed by western blot with anti-xMELK antibodies during phenotype was clearly dose dependent. Accordingly, accumulation in vitro oocyte maturation released by progesterone [Pg, prophase arrested (PI) of xMELK protein increased with the amount of injected xMELK and metaphase arrested (MII) oocytes] and at different times post-fertilization mRNA (Fig. 2A). Inhibition of cell division was not observed (Pf.). xMELK to -tubulin (used as a loading control) ratios are indicated. In when mRNAs encoding enhanced green fluorescent protein (EGFP; MII oocytes, xMELK is phosphorylated and, consequently, its electrophoretic Fig. 2A) or the two Xenopus xMELK-related protein kinases xPAR- mobility is decreased compared with that of PI and embryos (Blot et al., 2002; 1A and xPAR-1BX were used (Ossipova et al., 2005) (data not Badouel et al., 2006). (B)Northern blot analysis with an xMELK probe of shown). Moreover, expression of an xMELK mutant (xMELK oocytes treated with 2 or 4 ng of AS Co, AS9, AS11 oligos or untreated (–). K/R), the activity of which is largely reduced compared with that Eg7 was used as a loading control. (C)Western blot analysis with anti-xMELK antibodies of individual AS Co (a and b), AS11 (c, d and e) and AS9 (f and g) of the wild-type xMELK (Blot et al., 2002) did not affect cell MELK-treated embryos. xMELK to actin (used as a loading control) ratios are division (Fig. 2A). This indicates that, upon xMELK indicated. (D)Time-lapse images of AS Co, AS9 and AS 11 embryos. overexpression, the kinase activity is necessary for inhibition of Arrowheads point to the furrows of the first and second abortive divisions. cell division. In numerous instances, it has been reported that Time is in minutes. expression of a dead kinase has a dominant-negative effect. This 960 Journal of Cell Science 124 (6)

Table 1. Knockdown of xMELK does not affect oocyte maturation but inhibits embryo cleavage

Percentage of Number of video-recorded embryos Percentage oocyte maturation No furrow Abortive cleavage Fully cleaved Total fully cleaved AS11 xMELK 88 (n445) 8 25 5 38 13 AS9 xMELK 81 (n176) 4 6 1 11 9 AS Co control 91 (n425) 1 2 11 14 78

Percentages of oocyte maturation are given in the left column; the numbers of oocytes studied are indicated in brackets. Video-recorded embryos were classified with respect to their phenotype. The percentages of fully cleaved embryos relative to total embryos are indicated for the two xMELK antisense constructs AS9 and AS11 and the control AS Co.

is generally due to the titration by the mutant of an activating Analyses by confocal microscopy revealed that large undivided partner, leading to inactivation of the endogenous kinase, or to the cells contain multiple nuclei and supernumerary centrosomes, sequestration of its substrate. Therefore, the absence of a dominant- indicating that these cells present a cytokinesis defect, whereas the negative effect of xMELK K/R suggests that xMELK does not cell cycle is not impaired (Fig. 2B). To define more precisely the stably associate with an activator or a substrate to perform its role induced defect, embryos overexpressing xMELK were followed in cell division or that the point mutation (K/R) directly or indirectly by time-lapse confocal microscopy. To visualize the periphery of destabilizes a putative interaction. the cells, GFP–ABD, a fluorescent protein having affinity for actin (Lenart et al., 2005), was coexpressed with xMELK. In these embryos, the cleavage furrows started to ingress but failed to complete cell division and finally regressed (Fig. 2C). This shows that xMELK overexpression, as well as xMELK knockdown, inhibits the progression, rather than the initiation, of cytokinesis. Collectively, the results of knockdown and overexpression experiments demonstrate that xMELK is important for the progression of cytokinesis and that its levels must be finely tuned for successful cytokinesis.

Endogenous xMELK is localized at the cell cortex and at the division furrow As alterations in the levels of xMELK induce cytokinesis failure, we examined endogenous xMELK localization during cell division in Xenopus laevis embryos with a previously characterized affinity- purified antibody against xMELK (Blot et al., 2002; Chartrain et

Journal of Cell Science al., 2006). During the first and the second embryonic divisions, xMELK was detected at the cell cortex and at the division site (Fig. 3A). As expected, the small GTPase RhoA was also detected at the cleavage furrow. Interestingly, xMELK colocalizes with RhoA (Fig. 3A, merge panels), indicating that its accumulation at the division site is an early event during cytokinesis. A second antibody against xMELK gave the same result (data not shown). To confirm that the signal detected with antibodies is specific to xMELK, we performed immunofluorescence on embryos treated with either AS Co or AS11 xMELK and fixed when the first division furrow started to ingress. In AS11 xMELK treated embryos, the signal detected with the antibody to xMELK at the cell cortex and the division furrow was largely reduced compared with that of AS Co embryos. In these conditions, even if the furrow was not formed perfectly in AS11 embryos, RhoA was still localized Fig. 2. Overexpression of xMELK induces abortive cell divisions. at the division site (Fig. 3B). This demonstrates that the signal (A)Embryos microinjected with 5 ng or 1 ng of xMELK, 1 ng of xMELK K/R detected with the antibody to xMELK is specific and that its or 5 ng of EGFP mRNAs. Blastulas are shown on the left and gastrulas on the accumulation at the furrow is not necessary for RhoA accumulation. right (only the gastrula stage is shown for xMELK K/R). Dotted lines indicate Later in development, at blastula stage 7 (Nieuwkoop and Faber, large undivided cells. xMELK and xMELK K/R constructs are indicated: N, 1956), xMELK was localized at the lateral cell cortex independently catalytic; M, median; and C, C-terminal domains. Western blots show the of the cell cycle phase (Fig. 3C, c2). xMELK was also concentrated levels of expressed proteins. (B) Uninjected (Co) and xMELK-overexpressing  in a narrow band at the surface of blastomeres during anaphase (xMELK) embryos were fixed and stained for DNA (blue), F-actin (red) and (Fig. 3C, c1). The xMELK band was already detectable at the lamin or -tubulin (green), as indicated. Scale bars: 10m. (C)Still frames of time-lapse confocal microscopy showing embryos coinjected with xMELK equatorial cortex before ingression of the furrow started (Fig. 3D, and the GFP–ABD fluorescent probe mRNAs. Projections of six confocal 1- left), demonstrating that xMELK is an early marker of the cleavage m sections are shown. Arrowheads indicate division furrows that ultimately site. As the furrow ingresses, xMELK stays associated with the regress (asterisks). Time is indicated in minutes. Scale bar: 50m. furrow and moves inwards (Fig. 3D, middle and right). In post- xMELK in cytokinesis of Xenopus embryos 961

Fig. 3. Endogenous xMELK is localised at the cell cortex and division furrow. (A)Albino embryos were fixed when the first (top raw) and second (bottom raw) division furrows started to ingress. Indirect immunofluorescence was performed with antibodies against xMELK (green) and RhoA (red). Projections of 22 confocal 0.5-m sections are shown. Pictures were merged to visualize colocalization of xMELK with RhoA at the division furrow (merge). The dotted line indicates the first division. Scale bar: 100m. (B)AS-Co- and AS11-treated embryos were fixed when the first division furrow started to ingress. Indirect immunofluorescence was processed as in (A). For each condition, projections of 26 confocal 0.5-m sections are shown. Scale bar: 100m. (C)Blastula stage 7 (c1 and c2), blastula stage 9 (c3 and c4) and gastrula stage 11 (c5 and c6) albino embryos were fixed, processed for indirect immunofluorescence with antibody against xMELK (green) and stained for DNA (blue). Surface (Z0, c1, c3 and c5) and deeper (distance inm relative to Z0; c2, c4 and c6) single optical sections are shown. Asterisks indicate dividing cells in stage-9 and -11 embryos. At the right is shown an orthogonal projection of a dividing cell (c5–c6). The plan of orthogonal projection is symbolized by white lines on the two confocal planes shown in panel c5 and c6, and reversely white dotted lines on the orthogonal projection indicate the position of the two confocal planes shown in panels c5 and c6. (D)Single optical sections of the surface of three distinct blastomeres (blastula stage 7) at several stages of cytokinesis fixed as in C. White lines represent the plan of orthogonal projections shown under pictures. Black and open arrowheads indicate, respectively, cell division sites and lateral cortex. Scale bar: 50m. Journal of Cell Science MBT embryos at the blastula stage 9 and the gastrula stage 11, of these proteins in embryos at stages 7, 9 and 11. The results show xMELK was still localized at the cell cortex, but the xMELK that the xMELK level in gastrula embryos still represents 51% of equatorial band was no longer detectable (Fig. 3C, c5). Instead, a the level in blastula (supplementary material Fig. S3A). By contrast, slightly more intense xMELK signal was sometimes observed at the level of F-actin did not change, and the level of MHC even the site of cell constriction in the late blastula (stage 9, Fig. 3C, slightly increased. Accordingly, in living gastrula embryos, EGFP– arrow in c4). This is not specific only to xMELK as F-actin and xMELK was not concentrated at the equatorial cortex (see above the myosin heavy chain (MHC) have an identical behaviour and supplementary material Fig. S3B). This demonstrates that, in (supplementary material Fig. S2A and S2B), demonstrating that a gastrula embryos, the absence of F-actin, MHC and xMELK change in the mode of cytokinesis progression occurs during early concentration at the division site is not directly correlated to their development concomitantly with the MBT. We also observed that expression levels and indicates that a currently unknown an inversion in the ingression direction of furrows occurs between mechanism(s) controls their localization during development. stage 9 and stage 11. Indeed, at blastula stage 9, whereas the apex Altogether, these results show that xMELK localization at the of blastomeres has already ingressed, more deeply in cells, the equatorial cortex is precisely controlled during development and basolateral membrane only shows the beginning of a constriction that the equatorial band is specific to large cells of pre-MBT (Fig. 3C, compare c3 and c4). Conversely, at the gastrula stage 11, embryos. The precise localization of xMELK at the division furrow, the basolateral membrane has already ingressed, whereas their in addition to the division defect induced by xMELK knockdown surface is not yet cleaved (Fig. 3C, compare c5 with c6 and see the and overexpression, strongly suggested a role for xMELK in corresponding orthogonal view and supplementary material Movie cytokinesis in cleaving embryos. 1). This was observed for all proteins analyzed (supplementary material Fig. S2B), demonstrating that this is not an artefact due to xMELK localization at the cell cortex and at the division xMELK antibodies. At the MBT, many maternal proteins are furrow is highly dynamic replaced by zygotically expressed ones, and this is frequently To determine the dynamics of xMELK localization, EGFP–xMELK accompanied by a change in their expression levels. Therefore, a fusion proteins were expressed in embryos and followed by time- decrease in F-actin, MHC and xMELK levels in stage 11 embryos lapse confocal microscopy. To examine xMELK dynamics under could explain why these proteins are no more concentrated at the conditions where cell division proceeds normally, we first used the equatorial cortex. So, we analysed by western blotting the levels EGFP–XMELK K/R mutant that does not perturb cytokinesis. 962 Journal of Cell Science 124 (6)

of cytokinesis were observed (Fig. 4B and supplementary material Movie 3). The fast relocalization of EGFP–XMELK at the cell cortex was also observed in gastrula embryos. In agreement with results obtained by immunofluorescence on fixed embryos (Fig. 3C), the basolateral membrane was already divided, whereas the surface was not cleaved and EGFP–XMELK was not accumulated at the cell division site (supplementary material Fig. S3C). Together, these results further demonstrate that, in early embryos, xMELK marks the division furrow position before ingression and that its relocalization at the cortex and at the division site is a highly dynamic event.

xMELK localization correlates with its conformational rearrangement It was previously shown that MELK is auto-inhibited by its C- terminal domain and that deletion of this domain results in an increased catalytic activity in vitro (Beullens et al., 2005) (data not shown). The widespread mechanism of kinase auto-inhibition by intramolecular folding was proposed to apply for MELK (Beullens et al., 2005) and Kin2, a MELK-related kinase in yeast (Elbert et al., 2005). Therefore, we asked whether conformational changes in xMELK occur during cell division in Xenopus embryos and we explored the possibility of a correlation between xMELK relocalization at the cleavage furrow and a conformational rearrangement by using intramolecular fluorescence resonance energy transfer (FRET). A FRET-based probe corresponding to xMELK K/R sandwiched between the cyan fluorescent protein (CFP) donor and the yellow fluorescent protein (YFP) acceptor was constructed (Fig. 5A). The FRET probe was expressed in embryos and followed by time-lapse confocal microscopy. The highest YFP:CFP emission ratio, reflecting a closed configuration, was observed when the protein was localized in the cytoplasm and it did not change across the cell cycle (Fig. 5B). A smaller YFP:CFP ratio was observed for the protein localized at the cell cortex, Fig. 4. Dynamics of xMELK localization. (A,B)Still frames of time-lapse Journal of Cell Science confocal microscopy showing, respectively, blastomeres expressing EGFP– either at the cell surface outside of the division site or located more XMELK K/R (A) and EGFP–XMELK (B). Projections of 10 confocal 2-m deeply in cells. Again, the ratio stayed stable during cell cycle sections are shown. Arrowheads point to the xMELK band that coincides with progression. By contrast, a marked decrease of the YFP:CFP ratio, the cleavage furrow. Arrows point to unfocused xMELK. Asterisks indicate reflecting an open configuration, occurred when the protein was cells that failed to divide. Time is indicated in minutes. Scale bar: 50m. localized at the cleavage furrow. These differences were not (C)Blastomeres expressing EGFP alone. Arrowheads point to cell division observed with CFP–MELK K/R alone or with a mix of CFP– sites. Time is in minutes. Scale bar: 50m. MELK K/R and YFP–MELK K/R, in spite of their accumulation at the cleavage furrow (Fig. 5C), thus excluding the occurrence of FRET due to auto-fluorescence or dimerization. These results Before cells started to divide, fluorescence emitted by the EGFP– demonstrate that xMELK conformational rearrangements occur XMELK K/R was diffuse, resembling that of EGFP-expressing during cell division and are correlated with its subcellular embryos (compare Fig. 4A with Fig. 4C). Just before the cells localization. started dividing, a thin fluorescent line corresponding to the cell cortex appeared in addition to a faint equatorial band (arrowheads Overexpression of xMELK impairs accumulation of in Fig. 4A and supplementary material Movie 2). While cells activated Rho GTPase at the division furrow progress through mitosis, the fluorescence intensity of this EGFP– Indirect immunofluorescence and live imaging demonstrated that XMELK K/R band increased. The EGFP–XMELK K/R band xMELK marks the division site shortly before furrow ingression. corresponds exactly to the division site and moves inward with This regulation of xMELK localization was reminiscent to that ingressing furrows. These results are in agreement with endogenous described for the activated small GTPase Rho (Bement et al., xMELK immunolocalization in fixed embryos. When cells 2005; Miller and Bement, 2009). Indeed, using a fluorescent probe completed division, the fluorescence came back to the initial state allowing specific detection of activated Rho [GFP–rGBD (Benink (Fig. 4A). Thus, the dynamic study in whole embryos shows that and Bement, 2005)], it has been shown, in echinoderms and xMELK is already detectable at the presumptive division site Xenopus embryos, that active Rho GTPase accumulates as a narrow approximately two minutes before membrane ingression. Wild- band at the equatorial cell cortex presaging the division furrow type EGFP–XMELK presented an identical dynamic localization before it ingresses (Bement et al., 2005). The similarity between at the cleavage furrow and at the cell cortex; however, the EGFP– the spatio-temporal localization of active Rho and xMELK XMELK band became unfocused and, correlating with this, failures prompted us to examine the effect of elevating xMELK levels on xMELK in cytokinesis of Xenopus embryos 963

Fig. 5. Spatio-temporal xMELK conformational changes. (A)Intramolecular FRET for YFP–XMELK K/R-CFP. Fluorescent proteins are fused to the same xMELK K/R molecule allowing FRET from the donor CFP to the acceptor YFP. FRET depends on distance and orientation and thus indicates conformational changes. CFP is excited by 433 nm

Journal of Cell Science light, and then, after energy transfer, light emitted at 527 nm by YFP is detected. (B,C)Embryos were injected with YFP– XMELK K/R-CFP (B), xMELK K/R-CFP or a mix of xMELK K/R-CFP plus YFP–XMELK K/R (C). Surface views for CFP and colour-coded images of the YFP:CFP emission ratio (‘YFP/CFP’) are shown. White arrowheads indicate cleavage furrows. A confocal section at a distance of 18m relative to the blastomere surface is shown for time 18 (t18). Mean values of YFP:CFP emission ratios measured for three dividing cells at several subcellular locations are plotted versus time; error bars represent the s.d. The longitudinal brackets indicate the period of cytokinesis. Time is indicated in minutes. Scale bars: 50m.

activated Rho localization. As described previously, the GFP– provide a direct explanation for the cytokinesis defect induced by rGBD fluorescent probe localized at the presumptive cleavage site overexpression of xMELK. before furrow ingression (Fig. 6). When coexpressed with xMELK K/R, which does not induce defective cell division, active Rho xMELK copurifies with anillin, an essential cytokinesis accumulated at the division site (Fig. 6). By contrast, when component coexpressed with active xMELK, the GFP–rGBD band was not xMELK and active Rho accumulate in a band that marks the detected at the furrow (Fig. 6) and blastomeres still initiated furrow presumptive division site shortly before furrow ingression. To analyse ingression that ultimately regressed. The fact that cells still present further this discrete cellular localization, we examined the localization signs of furrow ingression suggests that either a low amount of of anillin, a key protein involved in cytokinesis, which has been cortically localized Rho is sufficient to allow the initiation of shown to be localized at the equatorial cortex during anaphase in membrane ingression or that an alternative pathway, independent Caenorhabditis elegans (Maddox et al., 2005), Drosophila (Field of Rho, allows initiation of cytokinesis. This issue is discussed and Alberts, 1995), Xenopus cultured cells (Straight et al., 2005) and further below. These data show that overexpression of xMELK embryos (Miller and Bement, 2009) as well as human cultured cells perturbs accumulation of activated Rho at the cleavage furrow and (Piekny and Glotzer, 2008). First, we tested by immunoprecipitation 964 Journal of Cell Science 124 (6)

Fig. 6. Accumulation of activated Rho at the division furrow is impaired by overexpression of xMELK. Embryos microinjected with mRNA encoding GFP–rGBD (top row), a fluorescent probe sensing activated Rho, or a mix of GFP–rGBD mRNA with xMELK K/R (middle row) or wild-type xMELK (bottom row) mRNAs. Projections of five confocal 2-m sections are shown. Arrowheads indicate the cell division site. Time is indicated in minutes. Scale bars: 50m.

whether the two endogenous proteins could be copurified. As expected, endogenous xMELK and anillin were immunoprecipitated, respectively, by affinity-purified antibodies against MELK and against Xenopus anillin and were not detectable when pre-immune Fig. 7. Copurification of xMELK with anillin. (A)Endogenous xMELK and serum was used (Fig. 7A). We found that endogenous xMELK was anillin co-immunoprecipitate. Protein extracts were prepared from cytokinetic specifically co-immunoprecipitated with anillin and, reciprocally, one-cell embryos (input). Proteins were immunoprecipitated (IP) with anillin was co-immunoprecipitated with xMELK. Anillin and antibodies against xMELK and anillin or pre-immune immunoglobulins (‘Pi’) and blotted with antibodies against anillin, xMELK or -tubulin.

Journal of Cell Science xMELK were not detected in immunoprecipitations made with pre- (B)Endogenous anillin is localized at the equatorial and lateral cell cortex in immune serum, and, in addition, -tubulin was absent from all  Xenopus embryos. Blastula stage-7 albino embryos were fixed and processed immunoprecipitates, demonstrating that xMELK and anillin co- for immunofluorescence with antibody to anillin. Single optical sections are immunoprecipitations are specific. Next, we examined the shown from the cell surface (Z0, left) towards deeper planes (the distance distribution of endogenous anillin in Xenopus early embryos with an relative to Z0 is indicated inm). The cell surrounded by a dotted line is affinity-purified antibody against Xenopus anillin (Straight et al., further magnified and shown in the row below. Scale bars: 100m. 2005). In the early blastula, anillin presented a distribution very (C)Colocalization of xMELK K/R-CFP and EGFP–anillin at the division similar to that observed for xMELK, active Rho, F-actin and MHC. furrow. Pictures were merged to visualize colocalization of xMELK K/R-CFP Indeed, anillin was concentrated in a narrow band at the blastomere and EGFP–anillin at the division furrow (merge). Scale bar: 100m. surface in addition to cortex (Fig. 7B). The band was already (D)Schematic representation of the transition in the mode of cytokinesis detectable at the equatorial cortex before furrow ingression started, during early development. Cubes represent cytokinetic embryonic cells. In large cells of blastula embryos, xMELK, F-actin, active Rho (‘act. Rho’), demonstrating that anillin accumulates early at the cleavage site in MHC (all in red) and anillin (green) are localised at the cell cortex and at the Xenopus embryos. EGFP–anillin expressed in embryos confirmed cleavage furrow. The furrow is an arc that expands in an apical-to-basal immunofluorescence observations and showed that accumulation at direction (arrows). In contrast, in smaller cells of gastrula, only anillin forms a the equatorial cortex is highly dynamic and precedes furrow ring at the equatorial cortex corresponding to the furrow which ingresses in a ingression (data not shown). As the antibodies against both xMELK basal to apical direction. At this developmental stage, xMELK is still localized and anillin were raised in rabbits, the two endogenous proteins could at the cortex. not be examined simultaneously in the same cell. But coexpression of EGFP–anillin and xMELK K/R–CFP shows that the two proteins are colocalized at the equatorial cortex (Fig. 7C). Interestingly, in whereas the cell surface had only started to contract (supplementary gastrula embryos, endogenous anillin (supplementary material Fig. material Fig. S4). Altogether, these results show that xMELK S2B) and EGFP–anillin (supplementary material Fig. S4) were still colocalizes and interacts with anillin, a crucial cytokinesis regulator, detected as an equatorial band at the apical surface of dividing and reinforce the link between xMELK and cytokinesis. epithelial cells. This contrasted with the clear absence of xMELK, F-actin and MHC which, in the gastrula, were detected only at the Discussion cell cortex. Following EGFP–anillin localization in living embryos In this study, we investigated the function of the xMELK protein further showed that the basolateral membrane is already divided, kinase in cleaving Xenopus embryos. We show that xMELK marks xMELK in cytokinesis of Xenopus embryos 965

the division site, copurifies with the cytokinesis factor anillin and function is compromised (Kanada et al., 2005). We hypothesize demonstrate that alterations in its levels of expression affect the that, in adherent cells of the Xenopus embryo, a similar mechanism progression of cytokinesis through mislocalization of RhoA. could operate to allow the formation of an abortive furrow when By knockdown of xMELK before fertilization, we show that accumulation of active Rho at the division site is compromised by xMELK is the product of a maternal-effect gene necessary for overexpression of xMELK. completion of cytokinesis. Surprisingly, overexpression of xMELK In cleaving embryos, xMELK shares remarkable properties with also results in abortive cytokinesis. Numerous studies have active RhoA, actin, myosin heavy chain and anillin. One of the documented the unstable properties of cytokinesis in animal cells. most noteworthy resemblances between these proteins is their The reversibility of cell division highlights the requirement to accumulation at the presumptive division site before the furrow maintain constriction until the daughter cells are topologically starts to ingress. Indeed, like the discrete RhoA activity zone that distinct. The actomyosin-based contractile ring produces the force prefigures the division site (Bement et al., 2005), we have shown necessary for ingression of the plasma membrane. It is established that xMELK and anillin dynamically accumulate in a narrow band that cytokinesis requires polymerization of actin filaments (F-actin) at the equatorial cell cortex shortly before furrow ingression. Later as well as depolymerization. This requirement for F-actin dynamics in development, in gastrula embryos, cells still accumulate anillin is illustrated by the deleterious effects of actin-depolymerizing at the equatorial cortex in contrast to xMELK, actin and MHC, drugs (Merriam et al., 1983) or inactivation of cofilin, an F-actin- which are not concentrated at this site, although they accumulate depolymerising and -severing factor (Abe et al., 1996). Accordingly, at the cell cortex. Thus, the equatorial band is not specific for deregulation of signalling pathways that operate on the F-actin active RhoA and appears as a specialized structure that concentrates polymerization–depolymerization equilibrium eventually affects several cytokinesis molecules in rapidly dividing large cells. We cytokinesis. As demonstrated in Xenopus embryos, both increasing have precisely determined that xMELK is no more concentrated at and decreasing the activity of RhoA, leading, respectively, to the equatorial cortex between stage 7 and stage 9. Therefore, the increasing and decreasing the amount of F-actin, inhibit cytokinesis decrease in xMELK concentration at the equatorial cortex correlates (Drechsel et al., 1997). The fact that xMELK knockdown impairs with the MBT, during which important modifications occur, such furrow ingression indicates that xMELK is important for cytokinesis as cell cycle remodelling and the commencement of zygotic gene after its initiation. The overexpression of xMELK is also deleterious expression. Interestingly, bundles of microtubules that form a for cell division. This suggests that a misregulated phosphorylation structure, termed the furrow microtubule array [FMA (Danilchik et of a putative xMELK substrate involved in cytokinesis could block al., 1998)], at the base of the cleavage furrow that are probably cytokinesis progression and that xMELK must be inactivated for involved in furrow progression were observed until stage 7. cytokinesis to proceed. The finding that perturbing the levels of Together, these data indicate that profound modifications in the xMELK leads to a default in furrow ingression rather than a failure mode of cytokinesis occur at the MBT. The change in xMELK of cytokinesis initiation suggests that xMELK participates in the accumulation at the equatorial cortex might depend on a factor maintenance of the cytokinetic ring constriction. Moreover, the and/or a posttranslational modification regulated at the MBT. The fact that overexpression of xMELK leads to a default in high xMELK concentration at the equatorial cortex appears to be accumulation at the division site of active RhoA, which is crucial specific to large cells. Indeed, during the cleavage period, the size

Journal of Cell Science for the function of the acto-myosin ring, could explain the of embryonic cells decreases by about two orders of magnitude, detrimental effect of overexpression of xMELK on the constriction ranging from more than a millimetre in one-cell embryos to ~20 m of the cytokinetic ring. in gastrula cells. The preferential accumulation of xMELK, F- Interestingly, we noticed that, in spite of a large decrease in the actin, MHC, anillin and active RhoA at the equatorial cortex in accumulation of active Rho at the division site induced by blastomeres might reflect the higher forces necessary for membrane overexpression of wild-type xMELK, blastomeres still initiated ingression in large compared with small embryonic cells, as cytokinesis. We cannot rule out the possibility that a low level of proposed by Wang (Wang, 2005). Interestingly, in C. elegans and active Rho localized at the equatorial cortex is sufficient to initiate sea urchin embryos, it was shown that the duration of cytokinesis cell division; however, this would be insufficient to maintain is independent of the original cell size, with large cells dividing in constriction. By contrast, this observation suggests that, in Xenopus the same amount of time as exhibited by small cells (Carvalho et embryos, an alternative Rho-independent pathway might contribute al., 2009; Mabuchi, 1994). Thus, the mechanism of cytokinesis to early steps of cytokinesis. In LLC-Pk1 tissue-culture cells that scales with cell size, leading to a model of structural memory of adhere tightly to the substratum, it was observed that the the contractile ring (Carvalho et al., 2009). Nevertheless, the Clostridium botulinum C3 transferase, which ADP-ribosylates and molecular base of the structural memory is unknown. However, it thereby inhibits Rho, does not prevent cytokinesis in the majority was proposed that the cytokinesis ring comprises contractile units. of treated cells (Murthy and Wadsworth, 2008). Although The number of these units, the base of which would be F-actin, surprising, this result was in agreement with a similar observation would be proportional to the original cell size and they would reported previously (O’Connell et al., 1999). Moreover, Yoshizaki shorten during constriction of the ring. Contractile units could and colleagues (Yoshizaki et al., 2004) have shown by using explain the structural memory and the proportionality between the mammalian cells (HeLa and Rat1A) that activation of RhoA is constriction rate and the initial cell size observed during cleavage tissue specific, leading to differential effects of C3 transferase of embryonic cells (Carvalho et al., 2009). In large cells, some treatment on cytokinesis. In Dictyostelium discoideum myosin-II- molecules might be specifically recruited at the cell division furrow, null cells, cytokinesis can still proceed by furrow formation, which which could contribute to potentiate the furrow constriction rate. relies on traction forces when cells are anchored on the substratum As xMELK is accumulated preferentially at the equatorial cortex (Neujahr et al., 1997; Zang et al., 1997). It was also reported that specifically during cleavage and as modification of its levels alters certain adherent mammalian cells can divide by a contractile-ring- the progression of cytokinesis, it might be specifically involved in independent, adhesion-dependent process when the contractile ring sustaining cytokinesis in large embryonic cells. 966 Journal of Cell Science 124 (6)

We also established in this current study that an inversion in Materials and Methods furrow ingression occurs between late-blastula and gastrula stages Preparation of Xenopus embryos (Fig. 7D). In gastrula cells, the division furrow progresses from the Xenopus laevis albino and wild-type adults were obtained from the Biological Resources Centre (CRB, Rennes, France). Embryos were prepared as described basolateral towards the apical membrane, whereas, in the blastula, previously (Blot et al., 2002). it is the opposite. This in vivo observation is in agreement with the asymmetric ingression already described in mammalian epithelial xMELK knockdown Knockdown before fertilization was performed as described by Heasman and cells cultured in vitro (Reinsch and Karsenti, 1994). In C. elegans, colleagues (Heasman et al., 1991). Briefly, full-grown oocytes of pigmented frogs it was shown that asymmetric ingression of the division furrow were manually dissected in oocytes culture medium [OCM; Liebovitz L-15 medium during the first embryonic division depends on several factors, (Sigma) diluted 1:2 with sterile deionized water, 0.32 g/l bovine serum albumin including anillin, septin and the RNA-binding protein CAR-1 (Sigma), 5% penicillin–streptomycin (Sigma)]. Oocytes were microinjected with 2 ng/oocyte (volume of injections was 13.8 nl) of HPLC-purified antisense (Audhya et al., 2005; Maddox et al., 2007). Our study shows that oligos (Eurogentec) AS9: GTCTCTGCTCTACAAAGAG, AS11: CGCTCTT - asymmetric furrow ingression also occurs in a vertebrate organism TCTCGATCCGGAC and AS Co: CAAGTTATAGTATTCTGATT (underlined are and therefore suggests that it is potentially important. The reason phosphorothioate-modified bases) and cultured at 16°C. The day after injection, for the inversion in ingression polarity during development is oocytes were placed at 21°C and maturation was induced by treatment with 1 M progesterone (Sigma) overnight. Matured oocytes were stained with vital dyes to unknown, and it will be of great interest to test whether this feature distinguish the different set of oocytes and transferred into the same host laying was conserved through evolution. female. Finally, laid eggs were in vitro fertilized and processed for subsequent Anillin is a key protein involved in cytokinesis that stabilizes analysis.

the division furrow (Hickson and O’Farrell, 2008a). Our finding Indirect immunofluorescence that, in Xenopus embryos, anillin accumulates at the equatorial Embryos obtained by in vitro fertilization of eggs laid by albino or wild-type females cortex before the furrow starts to ingress is in agreement with a (knockdown) with sperm of wild-type males were grown to the appropriate stages previous report showing that anillin plays a role early in and fixed in PBS containing 2% TCA (for xMELK, anillin, myosin heavy chain and RhoA) or 3.7% formaldehyde (for tubulin and F-actin) for 2 hours at room cytokinesis (Hickson and O’Farrell, 2008b). Interestingly, temperature. Embryos were devitellinated, permeabilized in PBS plus 1% Triton X- similarly to active RhoA and xMELK, accumulation of EGFP– 100 for 20 minutes at room temperature, incubated in PBS plus 0.1% Triton X-100 anillin at the equatorial cortex is very dynamic, appearing (PBST) for 20 minutes and then for 1 hour in PBST with 5% foetal bovine serum immediately as a band of defined width and propagating along (FBS). Embryos were incubated with primary antibodies overnight at 10°C. The following antibodies were used: affinity-purified anti-xMELK (Blot et al., 2002), the equatorial cortex. This contrasts with EGFP–anillin in final concentration 1 ng/l; anti-RhoA (26C4, Santa-Cruz, 1:20); affinity-purified Drosophila S2 cells, where two steps in its cortical distribution anti-anillin (Straight et al., 2005), final concentration 2 ng/l; affinity-purified anti- have been identified: first, anillin localizes to the cell cortex at myosin heavy chain (Straight et al., 2005), final concentration 2.7 ng/l and anti- metaphase and is focused to the equator; and, second, anillin tubulin (TUB2.1, Sigma, 1:200). Rhodamine-conjugated-phalloidin (Molecular probes, 1:100) was used to stain F-actin. For lamin staining, wild-type embryos were disappears from the poles during anaphase. This raised the fixed in Dent’s fixative [20% dimethyl sulphoxide (DMSO): 80% methanol (Dent et possibility that the mechanism leading to anillin concentration at al., 1989)] for 7 hours at room temperature. Indirect immunofluorescence was the equatorial cortex in insect cells is not conserved in Xenopus performed as described above except that 1% DMSO was added to PBST. Embryos or, more likely, that the equatorial band of Xenopus early embryos were incubated with anti-lamin (L6 5D5, a kind gift from Reimer Stick, 1:20) and anti-actin (AC-40, Sigma, 1:200) in PBST plus 5% FBS. Secondary antibodies were is not strictly the equivalent of the equatorial band reported in anti-rabbit-alexa-488 or anti-mouse-alexa-555 (Molecular Probes, 1:100). DNA was

Journal of Cell Science Drosophila. Contrary to F-actin, MHC and xMELK, anillin is stained with DAPI (0.5 g/ml). Embryos were mounted in Vectashield (Vector) for still localized at the equatorial cortex in post-cleavage embryos, observations.

suggesting a sustained requirement for a high concentration of Construction of plasmids anillin at the division site for cytokinesis, whereas other pT7T-xMELK(Eg3) was constructed previously (Blot et al., 2002). pT7T-xMELK cytokinesis molecules are no more preferentially accumulated at K/R was obtained by PCR from the pET21a-Eg3 K/R plasmid (Blot et al., 2002) this site. Studies utilizing diverse organisms have reported that with oligos xMELK-20 and xMELK-21 (all primers are listed in supplementary material Table S1). PCR product was cut with BglII and XbaI restriction enzymes anillin interacts with several molecules involved in cytokinesis, and ligated into pT7T cut with BglII and SpeI. pT7T-NEGFP-xMELK allowing including actin (Field and Alberts, 1995), non-muscular myosin expression of EGFP–XMELK was constructed by PCR using the pEGFPN1-XlEg3 II MHC (Straight et al., 2005), RhoA (Piekny and Glotzer, 2008) (Chartrain et al., 2006) plasmid with the primers xMELK-71 and pEGFP-C1-3Ј. The and MgcRacGAP (D’avino et al., 2008; Gregory et al., 2008). PCR product was digested with NruI and XbaI and ligated into pT7T vector (Krieg and Melton, 1984) at EcoRV and SpeI sites. To construct the pT7T-NEGFP-xMELK Interestingly, xMELK and anillin co-immunoprecipitate, K/R plasmid, a DNA fragment containing the K/R mutation was excised from the demonstrating that the two proteins interact either directly or pT7T-xMELK K/R plasmid and was exchanged in the pT7T-NEGFP-xMELK indirectly. Altogether, our results indicate that, during cleavage plasmid. To construct the pT7T-NEGFP-Xlanillin, the Xenopus anillin open reading of the large cells of early embryos, anillin probably contributes frame was PCR amplified from pAFS217 (Straight et al., 2005) using anillin-1 and anillin-2 primers. The PCR product was cut with SpeI and ligated in pT7T digested to the concentration and/or stabilization of cytokinesis factors, with EcoRV and SpeI to generate pT7T-Xlanillin. EGFP ORF amplified from including xMELK, at the equatorial cortex. pEGFPN1 using GFP-5 and GFP-7 primers and cut with BglII and EcoRV and was In conclusion, our study demonstrates that important ligated into pT7T-Xlanillin cut with BglII and EcoRV. To obtain the plasmid allowing modifications in the cytokinesis mechanism occur during the early expression of the FRET-based probe, pT7T-YFP-xMELK K/R-CFP, YFP was PCR- amplified from pEYFP vector (Clontech) with EYFP-1 and EYFP-2 primers, xMELK development of a vertebrate organism (Fig. 7D). The specialization K/R ORF was PCR-amplified from pET21a-XlEg3 K/R plasmid with xMELK-72 of the cytokinetic process has been reported in distinct differentiated and xMELK-75 primers and CFP was PCR-amplified from pECFP vector (Clontech) cell types of post-cleaving Xenopus embryos (Kierserman et al., with ECFP-3 and ECFP-2 primers. Note that primers EYFP-2 and ECFP-3 each 2008). Here, we show that important modifications occur in introduce five consecutive glycine residues, allowing free orientation of fluorescent proteins. The PCR products were cut, respectively, with BglII, BglII and XhoI and early-cleaving embryos. Localization at the equatorial cortex, XhoI and SpeI. Fragments were ligated into pT7T open with BglII and SpeI restriction copurification with the cytokinesis factor anillin and the deleterious enzymes. The pT7T-xMELK K/R-CFP was constructed following the same strategy effect on the progression of cytokinesis of altered xMELK levels except that the YFP fragment was omitted in the ligation. The pT7T-YFP-xMELK all indicate that xMELK participates in cytokinesis in large K/R plasmid was obtained following the same strategy with xMELK K/R ORF amplified with xMELK-72 and xMELK-21 primers. All constructs were subsequently embryonic cells. verified by sequencing. xMELK in cytokinesis of Xenopus embryos 967

Microinjection of in vitro transcribed mRNAs Bement, W. M., Benink, H. A. and von Dassow, G. (2005). A microtubule-dependent In vitro transcriptions were performed with mMessage mMachine transcription kits zone of active RhoA during cleavage plane specification. J. Cell Biol. 170, 91-101. following the manufacturer’s instructions (Ambion). mRNAs were microinjected in Benink, H. A. and Bement, W. M. (2005). Concentric zones of active RhoA and Cdc42 one blastomere of two-cell stage embryos (volume of injections was 13.8 nl) and around single cell wounds. J. Cell Biol. 168, 429-439. placed at 16°C until observation. Beullens, M., Vancauwenbergh, S., Morrice, N., Derua, R., Ceulemans, H., Waelkens, E. and Bollen, M. (2005). Substrate specificity and activity regulation of protein kinase Imaging MELK. J. Biol. Chem. 280, 40003-40011. Only embryos obtained from albino females were used for live imaging. Imaging Blot, J., Chartrain, I., Roghi, C., Philippe, M. and Tassan, J.-P. (2002). Cell cycle was performed on a Leica SP2 confocal microscope using a ϫ20 HC PL APO-ON regulation of pEg3, a new Xenopus protein kinase of the KIN1/PAR-1/MARK family. ϫ ϫ Dev. Biol. 241, 327-338. 0.7, 40 HC Plan-APO-ON 1.25 and a 63 HCX Plan-APO-ON 1.4 oil-immersion Carvalho, A., Desai, A. and Oegema, K. (2009). Structural memory in the contractile objective lens (Microscopy platform, IFR140). Images and movies were obtained ring makes the duration of cytokinesis independent of cell size. Cell 137, 926-937. by projections of optical sections using the ImageJ software (Rasband, W. S., Chartrain, I., Couturier, A. and Tassan, J. P. (2006). Cell cycle dependent cortical http://rsb.info.nih.gov/ij/). Figures were assembled in Adobe Photoshop and Adobe localization of pEg3 protein kinase in Xenopus and human cells. Biol. Cell 98, 253-263. Illustrator (Adobe Systems). D’Avino, P. P., Takeda, T., Capalbo, L., Zhang, W., Lilley, K. S., Laue, E. D. and Glover, D. M. (2008). Interaction between Anillin and RacGAP50C connects the Extraction of proteins actomyosin contractile ring with spindle microtubules at the cell division site. J. Cell Oocytes or embryos were homogenized in EB buffer (10 mM Hepes, pH 7.7, 100 Sci. 121, 1151-1158. mM KCl, 2 mM MgCl2, 5 mM, EGTA, 5 mM DTT, 1% IGEPAL CA-630, 5% Danilchik, M. V., Funk, W. C., Brown, E. E. and Larkin, K. (1998). Requirement for glycerol) supplemented with pepstatin, leupeptin, chymostatin and PMSF at 10 M microtubules in new membrane formation during cytokinesis of Xenopus embryos. each, 40 mM NaF, 40 mM b-glycerophosphate, and 2.5 M okadaic acid. Extracts Dev. Biol. 194, 47-60. were then centrifuged at 14,000 g for 15 minutes at 4°C for further analysis. Davezac, N., Baldin, V., Blot, J., Ducommun, B. and Tassan, J. P. (2002). Human pEg3 kinase associates with and phosphorylates CDC25B phosphatase: a potential role for Western blot analysis and immunoprecipitations pEg3 in cell cycle regulation. Oncogene 21, 7630-7641. Proteins separated by SDS–PAGE were electrotransferred to polyvinylidene difluoride Dent, J. A., Polson, A. G. and Klymkowsky, M. W. (1989). A whole-mount membrane (Immobilon-P, Millipore). Membrane blocking and antibody incubations immunocytochemical analysis of the expression of the intermediate filament protein vimentin in Xenopus. Development 105, 61-74. were performed in TBST (25 mM Tris–HCl, pH 8.0, 150 mM, NaCl, 0.05% Tween Drechsel, D. N., Hyman, A. A., Hall, A. and Glotzer, M. (1997). A requirement for Rho 20) containing 5% skimmed milk. Anti-xMELK [L2 (Blot et al., 2002)], anti-anillin, and Cdc42 during cytokinesis in Xenopus embryos. Curr. Biol. 7, 12-23. anti-GFP (Roche), anti-actin (Sigma) and anti-tubulin (TUB 2.1, Sigma) antibodies Elbert, M., Rossi, G. and Brennwald, P. (2005). The yeast par-1 homologs kin1 and kin2 were diluted 1:1000. Secondary anti-rabbit and anti-mouse alkaline phosphatase- show genetic and physical interactions with components of the exocytic machinery. coupled or peroxidase-coupled immunoglobulins were from Jackson. Mol. Biol. Cell 16, 532-549. For immunoprecipitations, 2 g of either affinity-purified anti-xMELK [3G (Blot Field, C. M. and Alberts, B. M. (1995). Anillin, a contractile ring protein that cycles from et al., 2002)], affinity-purified anti-anillin antibodies (Straight et al., 2005) or protein- the nucleus to the cell cortex. J. Cell Biol. 131, 165-178. A-purified immunoglobulins of pre-immune serum were incubated at 4°C for 16 Gray, D., Jubb, A. M., Hogue, D., Dowd, P., Kljavin, N., Yi, S., Bai, W., Frantz, G., hours with 10 l of protein-A–sepharose beads (Pharmacia LKB Biotechnology) in Zhang, Z., Koeppen, H. et al. (2005). Maternal embryonic leucine zipper kinase/murine PBS (140 mM Na2HPO4, 1.8 mM, KH2PO4, 138 mM NaCl, 2.7 mM KCl, pH 7.5). protein serine-threonine kinase 38 is a promising therapeutic target for multiple cancers. After incubation at 4°C for 2 hours with protein extract, beads were washed Cancer Res. 65, 9751-9761. extensively with EB buffer and proteins eluted in sample buffer. Gregory, S. L., Ebrahimi, S., Milverton, J., Jones, W. M., Bejsovec, A. and Saint, R. (2008). Cell division requires a direct link between microtubule-bound RacGAP and Second polar body extrusion Anillin in the contractile ring. Curr. Biol. 18, 25-29. To prepare full-grown oocytes (stage VI), fragments of Xenopus ovaries were Hara, K., Tydeman, P. and Kirschner, M. (1980). A cytoplasmic clock with the same period as the division cycle in Xenopus eggs. Proc. Natl. Acad. Sci. USA 77, 462-466. incubated in OR2 medium (10 mM Hepes, pH 7.6, 82.5 mM NaCl, 2.5 mM KCl, 1 Heasman, J., Holwill, S. and Wylie, C. C. (1991). Fertilization of cultured Xenopus mM CaCl2, 1 mM MgCl2) containing 0.4 mg/ml of Dispase II (Boehringer) for 3 oocytes and use in studies of maternally inherited molecules. Methods Cell Biol. 36, hours at 21°C. Next, ovary fragments were incubated in OR2 containing 100 units/ml 2+ 213-230. of collagenase (Sigma) in the absence of Ca for 1 hour. Recovered oocytes were Heyer, B. S., Warsowe, J., Solter, D., Knowles, B. B. and Ackerman, S. L. (1997). New

Journal of Cell Science then washed extensively in OR2. Full-grown oocytes (stage VI) were microinjected, member of the Snf1/AMPK kinase family, Melk, is expressed in the mouse egg and as described above. In vitro maturation was triggered by 15 µM of progesterone preimplantation embryo. Mol. Reprod. Dev. 47, 148-156. (Sigma). Matured oocytes were then prick activated and fixed in methanol 30 Hickson, G. R. and O’Farrell, P. H. (2008a). Anillin: a pivotal organizer of the cytokinetic minutes later. To stain DNA, DAPI was added at a final concentration of 10 µg/ml machinery. Biochem. Soc. Trans. 36, 439-441. for 1 hour. After washing in PBS, oocytes were mounted in Vectashield (Vector) for Hickson, G. R. and O’Farrell, P. H. (2008b). Rho-dependent control of anillin behavior observations. Imaging was performed on a DMRXA Leica microscope coupled to a during cytokinesis. J. Cell Biol. 180, 285-294. Q550 CW image-analysis system (Leica) at the Microscopy platform, IFR140. Jung, H., Seong, H. A. and Ha, H. (2008). Murine protein serine/threonine kinase 38 activates apoptosis signal-regulating kinase 1 via Thr 838 phosphorylation. J. Biol. Chem. 283, 34541-34553. We thank A. F. Straight for anillin cDNA, purified anillin, anti- Kanada, M., Nagasaki, A. and Uyeda, T. Q. (2005). Adhesion-dependent and contractile anillin and anti-myosin heavy chain antibodies; R. Stick for anti-lamin ring-independent equatorial furrowing during cytokinesis in mammalian cells. Mol. antibody; J. Ellenberg for GFP–ABD and W. M. Bement for GFP– Biol. Cell 16, 3865-3872. rGBD. We are grateful to M. Tramier for helpful discussions on FRET, Kieserman, E. K., Glotzer, M. and Wallingford, J. B. (2008). Developmental regulation of central spindle assembly and cytokinesis during vertebrate embryogenesis. Curr. to Y. Audic and C. Wylie for their help with the transfer method and Biol. 18, 116-123. to J. Kubiak, H. McNeill and Ankush Garg for critical reading of the Krieg, P. A. and Melton, D. A. (1984). Functional messenger RNAs are produced by SP6 manuscript. This work was supported by the Ligue Départementale in vitro transcription of cloned cDNAs. Nucleic Acids Res. 12, 7057-7070. contre le Cancer (22 et 35), the ARC and INCa. Lenart, P., Bacher, C. P., Daigle, N., Hand, A. R., Eils, R., Terasaki, M. and Ellenberg, J. (2005). A contractile nuclear actin network drives chromosome congression in Supplementary material available online at oocytes. Nature 436, 812-818. http://jcs.biologists.org/cgi/content/full/124/6/958/DC1 Lin, M. L., Park, J. H., Nishidate, T., Nakamura, Y. and Katagiri, T. (2007). Involvement of maternal embryonic leucine zipper kinase (MELK) in mammary carcinogenesis through interaction with Bcl-G, a pro-apoptotic member of the Bcl-2 References family. Breast Cancer Res. 9, R17. Abe, H., Obinata, T., Minamide, L. S. and Bamburg, J. R. (1996). Xenopus laevis Mabuchi, I. (1994). Cleavage furrow: timing of emergence of contractile ring actin actin-depolymerizing factor/cofilin: a phosphorylation-regulated protein essential for filaments and establishment of the contractile ring by filament bundling in sea urchin development. J. Cell Biol. 132, 871-885. eggs. J. Cell Sci. 107, 1853-1862. Audhya, A., Hyndman, F., McLeod, I. X., Maddox, A. S., Yates, J. R., 3rd, Desai, A. Maddox, A. S., Habermann, B., Desai, A. and Oegema, K. (2005). Distinct roles for and Oegema, K. (2005). A complex containing the Sm protein CAR-1 and the RNA two C. elegans anillins in the gonad and early embryo. Development 132, 2837-2848. helicase CGH-1 is required for embryonic cytokinesis in Caenorhabditis elegans. J. Cell Maddox, A. S., Lewellyn, L., Desai, A. and Oegema, K. (2007). Anillin and the septins Biol. 171, 267-279. promote asymmetric ingression of the cytokinetic furrow. Dev. Cell 12, 827-835. Badouel, C., Korner, R., Frank-Vaillant, M., Couturier, A., Nigg, E. A. and Tassan, J. Marie, S. K., Okamoto, O. K., Uno, M., Hasegawa, A. P., Oba-Shinjo, S. M., Cohen, P. (2006). M-phase MELK Activity is Regulated by MPF and MAPK. Cell Cycle 5, T., Camargo, A. A., Kosoy, A., Carlotti, C. G., Jr, Toledo, S. et al. (2008). Maternal 883-889. embryonic leucine zipper kinase transcript abundance correlates with malignancy grade Badouel, C., Chartrain, I., Blot, J. and Tassan, J. P. (2010). Maternal embryonic leucine in human astrocytomas. Int. J. Cancer 122, 807-815. zipper kinase is stabilized in mitosis by phosphorylation and is partially degraded upon Merriam, R. W., Sauterer, R. A. and Christensen, K. (1983). A subcortical, pigment- mitotic exit. Exp. Cell Res. 316, 2166-2173. containing structure in Xenopus eggs with contractile properties. Dev. Biol. 95, 439-446. 968 Journal of Cell Science 124 (6)

Miller, A. L. and Bement, W. M. (2009). Regulation of cytokinesis by Rho GTPase flux. Ossipova, O., Dhawan, S., Sokol, S. and Green, J. B. (2005). Distinct PAR-1 proteins Nat. Cell Biol. 11, 71-77. function in different branches of Wnt signaling during vertebrate development. Dev. Mirey, G., Chartrain, I., Froment, C., Quaranta, M., Bouche, J. P., Monsarrat, B., Cell 8, 829-841. Tassan, J. P. and Ducommun, B. (2005). CDC25B phosphorylated by pEg3 localizes Paris, J. and Philippe, M. (1990). Poly(A) metabolism and polysomal recruitment of to the centrosome and the spindle poles at mitosis. Cell Cycle 4, 806-811. maternal mRNAs during early Xenopus development. Dev. Biol. 140, 221-224. Murthy, K. and Wadsworth, P. (2008). Dual role for microtubules in regulating cortical Pickard, M. R., Green, A. R., Ellis, I. O., Caldas, C., Hedge, V. L., Mourtada- contractility during cytokinesis. J. Cell Sci. 121, 2350-2359. Maarabouni, M. and Williams, G. T. (2009). Dysregulated expression of Fau and Nakano, I., Paucar, A. A., Bajpai, R., Dougherty, J. D., Zewail, A., Kelly, T. K., Kim, MELK is associated with poor prognosis in breast cancer. Breast Cancer Res. 11, R60. K. J., Ou, J., Groszer, M., Imura, T. et al. (2005). Maternal embryonic leucine zipper Piekny, A. J. and Glotzer, M. (2008). Anillin is a scaffold protein that links RhoA, actin, kinase (MELK) regulates multipotent neural progenitor proliferation. J. Cell Biol. 170, and myosin during cytokinesis. Curr. Biol. 18, 30-36. 413-427. Reinsch, S. and Karsenti, E. (1994). Orientation of spindle axis and distribution of Nakano, I., Masterman-Smith, M., Saigusa, K., Paucar, A. A., Horvath, S., Shoemaker, plasma membrane proteins during cell division in polarized MDCKII cells. J. Cell Biol. L., Watanabe, M., Negro, A., Bajpai, R., Howes, A. et al. (2008). Maternal embryonic 126, 1509-1526. leucine zipper kinase is a key regulator of the proliferation of malignant brain tumors, Straight, A. F., Field, C. M. and Mitchison, T. J. (2005). Anillin binds nonmuscle including brain tumor stem cells. J. Neurosci. Res. 86, 48-60. myosin II and regulates the contractile ring. Mol. Biol. Cell 16, 193-201. Neujahr, R., Heizer, C. and Gerisch, G. (1997). Myosin II-independent processes Tassan, J. P. and Le Goff, X. (2004). An overview of the KIN1/PAR-1/MARK kinase in mitotic cells of Dictyostelium discoideum: redistribution of the nuclei, re- family. Biol. Cell 96, 193-199. arrangement of the actin system and formation of the cleavage furrow. J. Cell Sci. Vulsteke, V., Beullens, M., Boudrez, A., Keppens, S., Van Eynde, A., Rider, M. H., 110, 123-137. Stalmans, W. and Bollen, M. (2004). Inhibition of spliceosome assembly by the cell Newport, J. and Kirschner, M. (1982a). A major developmental transition in early cycle-regulated protein kinase MELK and involvement of splicing factor NIPP1. J. Xenopus embryos: I. characterization and timing of cellular changes at the midblastula Biol. Chem. 279, 8642-8647. stage. Cell 30, 675-686. Wang, Y. L. (2005). The mechanism of cortical ingression during early cytokinesis: Newport, J. and Kirschner, M. (1982b). A major developmental transition in early thinking beyond the contractile ring hypothesis. Trends Cell Biol. 15, 581-588. Xenopus embryos: II. Control of the onset of transcription. Cell 30, 687-696. Yoshizaki, H., Ohba, Y., Parrini, M. C., Dulyaninova, N. G., Bresnick, A. R., Nieuwkoop, P. D. and Faber, J. (1956). Normal Table of Xenopus laevis (Daudin). Mochizuki, N. and Matsuda, M. (2004). Cell type-specific regulation of RhoA activity Amsterdam: North-Holland Publishing Company. during cytokinesis. J. Biol. Chem. 279, 44756-44762. O’Connell, C. B., Wheatley, S. P., Ahmed, S. and Wang, Y. L. (1999). The small GTP- Zang, J. H., Cavet, G., Sabry, J. H., Wagner, P., Moores, S. L. and Spudich, J. A. binding protein rho regulates cortical activities in cultured cells during division. J. Cell (1997). On the role of myosin-II in cytokinesis: division of Dictyostelium cells under Biol. 144, 305-313. adhesive and nonadhesive conditions. Mol. Biol. Cell 8, 2617-2629. Journal of Cell Science