Mouse Emi2 as a distinctive regulatory hub in second meiotic metaphase

The meiotic cell cycle of fertilizable oocytes in vertebrates is typically restrained at the second metaphase (mII) by a cytostatic factor (CSF) to prevent development without a paternal genome (parthenogenesis). The underlying mechanisms of mII arrest and exit are best understood in Xenopus and relatively poorly in mammals. In both, mII arrest correlates with the kinase activity of maturation promoting factor (MPF), a heterodimer of Cyclin B (CycB) and the cyclin-dependent kinase Cdc2 (Masui and Markert, 1971; Gautier et al., 1989; Gautier et al., 1990; Perry and Verlhac, 2008). MPF is active in both mitotic and meiotic cell cycles in vertebrates, but its prolonged stabilization by CSF is unique to mII and results in mII arrest.

Exit from mII occurs when CycB undergoes destruction box-(D-box-) dependent ubiquitylation by the anaphase-promoting complex, APC, an E3 ubiquitin ligase; this targets CycB for 26S proteasomal hydrolysis and eliminates MPF, thereby inducing metaphase exit (Glotzer et al., 1991; Peters, 2006). Arrest at mII is achieved by suspending APC activity, which is the function of CSF. One CSF responsible for this inhibition is the endogenous meiotic inhibitor 2, Emi2, the activity of which is essential for mII arrest as independently revealed in Xenopus (Schmidt et al., 2005) and the mouse (Shoji et al., 2006). Depletion of Emi2 from intact mouse oocytes causes mII release in a manner that requires the APC activator, Cdc20; one explanation of this is that Emi2 prevents Cdc20 from activating the APC (Shoji et al., 2006; Amanai et al., 2006).

Xenopus Emi2 (xEmi2) is stabilized during mII by phosphorylation from xMos to xMek to xMAPK to xRsk to xEmi2 (Sagata et al., 1989; Bhatt and Ferrell, 1999; Gross et al., 2000; Inoue et al., 2007; Nishiyama et al., 2007a) (Fig. 1). xRsk phosphorylates xEmi2 at S335, T336, S342 and S344. Phosphorylation at S335 and T336 facilitates the binding of protein phosphatase 2A (xPP2A), which in turn dephosphorylates phospho-residues at T545 and T551, and S213, T239, T252 and T267 (Wu et al., 2007b). Dephosphorylation of T545/T551 enhances binding of the xEmi2 C-terminal domain to the APC core component, xCdc27 (xAPC3) to inhibit the APC (Wu et al., 2007b) whereas dephosphorylation of the S213-T267 cluster stabilizes xEmi2 (Wu et al., 2007a). In Xenopus, xPP2A activity towards xEmi2 is thus stimulated by xMos via xRsk to promote mII arrest (Fig. 1).

In the mouse, Mos null oocytes fail to activate the MAPK pathway but nevertheless often arrest or pause at mII with MPF activity initially unaffected, or progress through mII and then ‘collapse’ back to mIII (Verlhac et al., 1996; Choi et al., 1996). Oocytes from Mos-null mice contain anomalously long, interphase-like microtubules during mI to mII and mII to mIII transitions (Verlhac et al., 1996). Emi2-depleted oocytes undergo aberrant cytokinesis (Shoji et al., 2006), but the relationship between Mos and Emi2 in the mammalian meiotic cell cycle remains unknown and meiotic coordination with spindle dynamics poorly understood.

Fertilization triggers an increase in the oocyte concentration of intracellular ‘free’ calcium, [Ca²⁺]_i (reviewed by Runft et al., 2002). In Xenopus oocyte extracts, this activates the Ca²⁺-dependent enzymes calmodulin kinase II (CaMKII) and calcineurin (CaN) (Fig. 1). It is unclear whether xCaN regulates the APC directly through xEmi2, with support both for (Nishiyama et al., 2007b) and against (Mochida and Hunt, 2007). Activated xCaMKII phosphorylates xEmi2 at threonine 195 (T195) of its canonical motif, RXST (Rauh et al., 2005). xEmi2 phosphorylated at T195 is a favoured substrate for polo-like kinase, Plx1 (the counterpart of mammalian Plk1), which then phosphorylates xEmi2 at S33/S38 in the phosphodegron motif DSGX₃S, targeting xEmi2 for xβTrcp- (Trcpb-) dependent proteasomal destruction (Schmidt et al., 2005; Rauh et al., 2005).

These details await analysis in mammalian Emi2 but it already seems clear that mouse and Xenopus (x)Emi2 differ. The N-terminal Plx1 phosphodegron does not have an N-terminal mouse Emi2 counterpart (Rauh et al., 2005; Perry and Verlhac, 2008). Moreover, Xenopus xRsk links the Mos-MAPK cascade to xEmi2 but mouse oocytes lacking Rsk (those of Rsk1, Rsk2, Rsk3 triple null mice) present a stable mII arrest (Dumont et al., 2005). Although it has recently been shown (Chang et al., 2009; Backs et al., 2010) that the γ isoform of CaMKII (CaMKIIγ) transmits the Ca²⁺ signal during mouse fertilization, the nature of this relay to Emi2 – whether, for example, it is direct – is unknown and mammalian Emi2 lacks the canonical RXX^T/_S CaMKII phosphorylation target used in xEmi2 (Rauh et al., 2005; Perry and Verlhac, 2008).

Here, we report a detailed molecular dissection of mammalian Emi2 and mII arrest. Xenopus and mouse Emi2s are functionally non-equivalent. Both N- and C-terminal Emi2 domains possess metaphase modulating activity, the latter through a potential TSS phosphorylation, D-box, Zn²⁺-binding motif and RL domain. Oocytes contain equimolar isoforms of CaMKIIγ, although neither directly phosphorylates Emi2 in vitro.

Eight- to 12-week-old B6D2F₁ females (SLC, Shizuoka-ken, Japan) were superovulated by standard serial intraperitoneal injection of pregnant mare serum gonadotropin (PMSG) followed 48 hours later by human chorionic gonadotropin (hCG). Oviductal mII oocytes were collected typically 12 to 14 hours after hCG injection and cumulus cells removed following hyaluronidase treatment as previously described (Yoshida et al., 2007; Suzuki et al., 2010). Fully grown germinal vesicle stage (GV) oocyte-cumulus cell complexes were collected from 8- to 10-week-old B6D2F₁ females 44 to 48 hours after PMSG injection. After 1 hour of culture in Waymouth medium (Waymouth, 1959) supplemented with 10% (v/v) foetal calf serum (FCS), cumulus cells were displaced from associated GV oocytes by repeated pipetting in TaM medium (Miki et al., 2006) supplemented with 10% FCS and 150 μM isobutylmethylxanthine (IBMX). GV oocytes were held in TaM medium containing 150 μM IBMX until micromanipulation and IBMX washed out or incubation in IBMX-containing medium continued, depending on the experiment (see Fig. S1 in the supplementary material). Culture of mII oocytes was typically in kalium simplex optimized medium (KSOM) (Erbach et al., 1994). All oocyte and embryo culture was in humidified CO₂ [5% (v/v) in air] at 37°C. Parthenogenetic activation was by incubating mII oocytes in Ca²⁺-free CZB (Chatot et al., 1989) supplemented with 10 mM SrCl₂ in humidified CO₂ [5% (v/v) in air] at 37°C for 1 hour. Oocytes were then washed in KSOM and incubation continued at 37°C.

Sperm were demembranated as described previously, with minor modifications (Yoshida et al., 2007). Briefly, cauda epididymidal spermatozoa from 12- to 30-week-old male B6D2F₁ mice were triturated in nuclear isolation medium [NIM: 125 mM KCl, 2.6 mM NaCl, 7.8 mM Na₂HPO₄, 1.4 mM KH₂PO₄, 3.0 mM EDTA (pH 7.45)] containing 1.0% (w/v) 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) at room temperature (25°C) for 1 minute and washed twice in ambient temperature NIM to give control demembranated heads – cHds. Where appropriate, sperm suspensions were incubated at 48°C for 30 minutes, with trituration after 15 minutes, to generate ‘inactivated’, iHd preparations. Sperm were mixed with 1 to 2 volumes of 15% (w/v) PVP₃₆₀ (average M_r=360,000; Kanto Chemical, Tokyo, Japan) and microinjected as described (Yoshida and Perry, 2007), typically within 60 minutes of PVP mixing.

Standard immunoblotting (IB) was with rabbit polyclonal anti-CycB1 (Santa Cruz Biotechnology, CA, USA), anti-Emi2 (Shoji et al., 2006), anti-Tubγ (γ-tubulin; Abcam, MA, USA), rabbit monoclonal anti-phosphoErk1/2 (Cell Signaling Technology, MA, USA), rabbit polyclonal anti-MAPK1/2 (Cell Signaling), rabbit polyclonal anti-Mos (Santa Cruz), rabbit polyclonal anti-IP₃R (Calbiochem, USA), rabbit polyclonal anti-CaM (Zymed, CA, USA), rabbit polyclonal anti-Plk1 (Santa Cruz), goat polyclonal anti-CaMKIIγ (calmodulin kinase II γ; Santa Cruz), rabbit polyclonal anti-Cdc2 (Delta Biolabs, CA, USA), rabbit polyclonal anti-phosphoY15Cdc2 (R&D Systems, MN, USA) or rabbit polyclonal anti-Tuba (alpha-tubulin; Abcam) primary antibodies, and anti-rabbit or -goat IgG (Invitrogen) secondary antibodies.

Immune complex detection, IB of oocytes and early embryos, and immunoprecipitation (IP) from NIH 3T3 cell extracts, were performed essentially as described previously (Itoh et al., 1998; Shoji et al., 2006). NIH 3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM, GIBCO) containing 10% (v/v) fetal bovine serum (FBS, Gemini-Bioscience) and penicillin/streptomycin (Sigma). Lipofectamine-mediated transfection (Invitrogen) of NIH 3T3 cells used 0.4 μg of each construct in pCI-Neo plasmid DNA (Shoji et al., 2006) per six-well plate. Constructs encoded bacterial alkaline phosphatase (BAP) or Emi2 (or deletions thereof) as C-terminal fusions to (FLAG)₃ for Fig. 4B, or mCherry (control) or Emi2-mCherry for Fig. 4F. Cells were collected 48 hours after transfection, and ~5×10⁶ cells lysed in 300 μl incubation buffer [150 mM NaCl, 2 mM Na₂EDTA, 100 mM PMSF, 5 mM DTT, 50 mM Tris (pH 7.5)] supplemented with 0.05% (w/v) NP-40, on ice for 10 minutes. Where appropriate, TPEN was added to cells at a final concentration of 100 μM 1 hour prior to harvesting. Lysates were cleared by centrifugation at 20,000 g at 4°C for 30 minutes. Cleared lysates (~5%) were stored for analysis by SDS-PAGE, and the remainder incubated with 20 μl of anti-Cdc27 (clone AF3.1, Sigma) or anti-DsRed (Clontech Laboratories, CA, USA) antibody overnight at 4°C and immunoprecipitates (IPs) incubated with protein A-agarose beads [12.5 μl of 50% (v/v) suspension] for 1 hour at 4°C, washed three times with ice-cold incubation buffer, and resuspended in a sample buffer. To generate FLAG IPs, lysates were applied to a mouse anti-FLAG M2 affinity matrix (Sigma) overnight with agitation at 4°C, unbound material removed, the matrix washed five times in 150 mM NaCl, 0.01% (v/v) NP-40, 10% (v/v) glycerol, and bound protein eluted at 4°C for 2 hours in a solution containing 5 mg/ml 3× FLAG peptide (Sigma) in 1 M NaCl, 0.5 M Tris (pH 7.5). Immunoblotting was with mouse anti-FLAG [1:1000 (v/v)] overnight and anti-mouse IgG secondary antibody [1:100,000 (v/v)] for 1 hour. Approximately 50% of each eluate was analyzed by SDS-PAGE. Labelled protein detection was performed as described previously (Shoji et al., 2006).

The different injection protocols adopted for in this work are illustrated in Fig. S1 in the supplementary material. Constructs containing PCR-generated Emi2 or Mos cDNA NheI-XhoI fragments cloned into pCI-Neo-mCherry (Shoji et al., 2006) were used to produce cRNA-encoded mCherry fusions at the C terminus of Emi2. Venus constructs were similarly constructed using a pCI-Neo-Venus scaffold, either as NheI-XhoI (β-Catenin) or NheI-XbaI (Emi2, Bora) fragments (see Fig. S11A in the supplementary material). For the transfections of Fig. 4B, (FLAG)₃-BAP or -Emi2 cDNA (i.e. encoding FLAG fused at the N terminus) were cloned into pNEBR-X1-Hygro (New England Biolabs, MA, USA) as XhoI-NotI fragments. For transfections in Fig. 4F, mCherry or Emi2-mCherry cDNA were cloned as XhoI-NotI fragments into pNEBR-X1-Hygro and the AAT preceding the start codon in Emi2-mCherry altered to ACC by site-directed mutagenesis. Site-directed mutagenesis was performed in the Quick-Change system (Stratagene, CA, USA). cRNAs were synthesized in vitro from linear plasmid DNA template and 5′-capped and polyadenylated in the same reaction using an mSCRIPT mRNA Production System (Epicentre Biotechnologies, WI, USA) according to the instructions of the manufacturer. cRNAs were dissolved in nuclease-free distilled water, quantified and stocked in aliquots at −80°C. Double-stranded siRNAs (iGENE Therapeutics, Tsukuba, Japan) were designed as described previously (Amanai et al., 2006) and stored in aliquots at −80°C.

RNA stock solutions were diluted with sterile DEPC-treated water to the desired concentration and 5~10 pl injected at default concentrations of 0.5 to 1.0 mg/ml (‘Lo’) or occasionally at higher concentrations (2.0 to 2.5 mg/ml; ‘Hi’) in Fig. 4G and Fig. S4 in the supplementary material, for cRNA and 25 μM for siRNA within 1 hour of thawing, either through a fine needle (tip inner diameter ≤1 μm) into GV oocytes in IBMX-containing M2 medium, or via a piezo-actuated micropipette (tip inner diameter 6~7 μm) into mII oocytes in M2 medium. Where appropriate, post-injection levels of cRNA were confirmed by semi-quantitative PCR (qPCR; not shown).

Ratiometric quantification of mRNAs (qPCR) was essentially as described previously, with cDNA prepared by Superscript (Invitrogen, CA, USA) from RNA from sperm or GV or mII oocytes (Amanai et al., 2006; Shoji et al., 2006) using the following primer pairs (5′ to 3′): H2afz, GCGTATCACCCCTCGTCACTTG and TCTTCTGTTGTCCTTTCTTCCCG; H3f3a, CCATGCCAAACGTGTAACAA and TACCTTTGACCCCATGGAAA; mCherry, TGAAGGTGACCAAGGGTGGC and AAGTAGTCGGGGATGTCGGC; CaMKIIa, AGCACCAACACCACCATTGAGG and GGTCGCACATCTTCGTGTAGGACTC; CaMKIIb, TCTTCCGACAGCACCAACACAAC and GGGTCACAGATTTTCGCATAGGC; CaMKIIg, TGAGCCAGAAGCCCTTGGTAAC and GTTTAGGATGGTGGTGTGGATAGG; CaMKIId, TGGAAGGGATGGACTTTCACAG and GAGCCGAATGTATGCGATGC; CaM, TCAGAACCCAACAGAAGCCG and TCCCATCCTTGTC AAACACTCG. Data were normalized with respect to H3f3a or H2afz.

To evaluate the role of CaN in mouse mII exit, oocytes were pre-incubated and injected in medium containing 100 μM FK506 (Sigma) and/or 2 μM cyclosporin A (CsA; Sigma) and incubation continued in inhibitor-containing medium. The CaN inhibitor peptide (iPep) ITSFEEAKGLDRINERMPPRRDAMP (Sigma) was co-injected with sperm at a pipette concentration of 500 μM. For sperm injection in the presence of inhibitors, sperm were pre-incubated in FK506 and/or CsA and/or iPep and injected so that the inhibitor concentrations were, respectively, 100, 200 or 500 μM within the pipette. Oocytes were scored for activation 5.25±0.25 hours after treatment in Fig. S8D. For the inhibition of calmodulin (CaM) and protein kinase C (PKC), oocytes were respectively incubated in media containing 100 μM CGS9343B (Sigma) or chelerythrine (Sigma) at the concentrations shown, each from 10 mM stocks in DMSO stored at −20°C. Where appropriate, preincubation was for 1 hour prior to SrCl₂ treatment and continued until reading.

Immunocytochemistry, differential interference contrast microscopy (DIC) and epifluorescence imaging were essentially as described previously (Yoshida et al., 2007). Images of live oocytes following cRNA injection were captured via a BioZero-8000 microscope/detector (Keyence, Osaka, Japan) and analyzed with BZ-Analyzer software (Keyence). Excitation at 540/25 nm was used with a TRITC (red) filter system for mCherry fluorescence detection and at 480/30 nm with a GFP (green) filter system to detect Venus epifluorescence.

Spindle behaviour during oocyte aging was visualized by time-lapse microscopy of transgenic oocytes containing a Venus-tubulin-alpha (Tuba) fusion protein whose expression was driven by the ZP3 promoter on a C57BL/6 × C3H background (subsequently back-crossed to C57BL/6). The 4.5 kb BciVI-MluI pZP3→Venus-Tuba transgene fragment was generated by inserting a 2019 bp pZP3-containing XhoI-KpnI genomic DNA fragment upstream of a 710 bp BamHI-BsrGI Venus fragment linked to a 1633 bp BsrGI-MluI fragment from pEGFP-Tuba, which encodes human tubulin-α (Clontech Laboratories, CA, USA). Oocytes were placed in a KSOM droplet under mineral oil on a glass-bottomed dish on the stage of a TE2000 inverted microscope (Nikon, Japan) equipped with a CSU10 confocal scanning unit (Yokogawa, Japan) and a humidified chamber [5% (v/v) CO₂ in air] at 37°C. DIC images and fluorescent (488 nm) images (typically 13 focal planes, step size=2 μm) were captured at 5-minute intervals by a C9100-13 ImagEM EM-CCD camera (Hamamatsu Photonics, Shizuoka, Japan) driven by MetaMorph (Molecular Devices, CA, USA) image analysis software.

H1 protein kinase assays of MPF activity for Fig. S2C in the supplementary material were as described previously (Shoji et al., 2006). For CaMKII phosphorylation assays, CAMKIIγ3, CaMKIIγJ, mouse Emi2 and xEmi2 were translated as Myc-His fusions in vitro using a TNT rapid coupled transcription/translation (IVTT) system (Promega) charged with 2 μg template DNA, and purified using the MagZ Protein Purification System (Promega, Cat. No. V8830). As a positive control, we employed preparation of bovine brain CaMKII containing all four isoforms (Upstate Biotechnology, NY, USA). CaMKII kinase assays were performed using the CaMKII Assay Kit (Upstate) as instructed by the manufacturers. Where appropriate, 9 μl of IVTT-purified mouse Emi2 or xEmi2 plus 1 μl of 400 ng/μl CaM replaced 10 μl CaMKII substrate cocktail (500 μM autocamtide peptide plus 40 ng/μl CaM). Where both autocamtide and Emi2 substrates were present in the same reaction, 10 μl of IVTT-purified (x)Emi2 and 10 μl of CaMKII substrate cocktail (500 μM autocamtide plus 40 ng/μl CaM) were added to give a final volume of 60 μl. As a negative control, ADBI buffer was included instead of CaMKII and/or Emi2. To visualize phosphorylation, 14 (out of 50) μl of each reaction was subjected to SDS-PAGE, fixed in a mixture of 30% (v/v) methanol plus 10% (v/v) acetic acid for 10 minutes, incubated in crack-proof solution for 10 minutes and dried prior to autoradiography using a BAS-2500 image detection system.

Student's t-tests were applied to comparative unpaired analyses. Data for each experiment were collected on at least 2 days. Emi2 protein sequences for mouse and species orthologues were obtained from Ensemble (http://www.ensemble.org/index/html) and the multiple alignments shown in Fig. 3A and Fig. 4C optimized via the CLUSTALW algorithm (http://align.genome.jp).

In Xenopus, the classical cystostatic factor, xMos, stabilizes xEmi2 via xMAPK (Inoue et al., 2007), but roles for Mos in establishing and maintaining mII arrest in mammals have not been clearly delineated. To address this, we injected mouse germinal vesicle stage (GV) oocytes with complementary RNA (cRNA) encoding Mos-mCherry and cultured them for 24 hours in the phosphodiesterase inhibitor, IBMX, followed by in vitro maturation (IVM) for 16 hours (the IB24 protocol; see Fig. S1A in the supplementary material). Injected oocytes underwent 97.6±2.4% mI arrest, whereas the corresponding value dropped to 38.5±3.6% for the dominant-negative, Mos^S3A/S105A-mCherry and 35.7±14.3% for mCherry alone (Fig. 2A; see Fig. S2A in the supplementary material). Mos overexpression in IB24 oocytes induced the appearance of active, phospho-MAPK, but this was markedly reduced by the MAPKK (Mek) inhibitor U0126 (Fig. 2B), which also reduced the establishment of mI arrest by Mos (Fig. 2A). In corroborating earlier work relating Mos-MAPK to mouse cytostatic arrest (Verlhac et al., 1996), these findings confirm the fidelity of the IVM system.

The relationship between Mos and Emi2 in mammals was next investigated by co-injecting GV oocytes with siEmi2 siRNA and Mos-mCherry cRNA in the IB24 system. Only 5.9±3.9% (n=52) of oocytes arrested at mI (Fig. 2A; see Fig. S2A in the supplementary material), suggesting that metaphase establishment by Mos requires Emi2. The Xenopus xMos-xMAPK pathway enhances xEmi2 stability and activity by phosphorylating S335 and T336 (Inoue et al., 2007). To test whether T336 of xEmi2 corresponds to T327 in the mouse, we injected Emi2^T327A-mCherry cRNA in the si16 system (see Fig. S1C in the supplementary material). High concentrations of Emi2^T327A-mCherry-encoding cRNA induced mI arrest (Fig. 2C), but only weakly compared with the same amount of cRNA encoding full-length Emi2-mCherry (compare Fig. 3D with ‘T327A Lo’ in Fig. 2C). This mirrors the result of a similar functional assay using xEmi2^T336A in Xenopus extracts (Inoue et al., 2007), consistent with an analogous role for T327 in mouse Emi2 to T336 in xEmi2.

Few data thus far distinguish between the role of Mos in the establishment versus the maintenance of mII arrest. We investigated this in aging oocytes, which from 18 to >36 hours post-hCG retained the morphological features of mII, although metaphase plate cortical protuberances eventually decreased in size and number (Fig. 2E). The spindles of Venus-tuba transgene-expressing mII oocytes elongated and occasionally became decoupled from the cortex during aging (see Fig. S2B in the supplementary material) (Webb et al., 1986). Immunoblotting showed that levels of meiotic exit signalling proteins changed little; phospho-MAPK exhibited a pronounced decline after only 32 hours, the inactive Cdc2 modification Y15-phospho-Cdc2 was undetectable, and H1 kinase activity, a marker of MPF (i.e. mII) persisted (see Fig. S2C in the supplementary material). By contrast, Mos levels declined markedly and were undetectable by ~28 hours post-hCG (Fig. 2D; see Fig. S2C in the supplementary material). Exposure to the parthenogenetic agent, SrCl₂ activated >95% of oocytes up to 32 hours post-hCG, and ~50% thereafter. The proportion undergoing aberrant cytokinesis following SrCl₂ treatment exhibited a pronounced increase after 32 hours (Fig. 2E).

These findings suggest that Mos is not sufficient in itself to establish mouse meiotic metaphase but requires Emi2, to which it signals via MAPK. The disappearance of Mos during mII oocyte aging correlates with aberrant spindle behaviour in oocytes that are competent to exit mII, indicating that Mos is not required for mII maintenance.

These results raise the possibility that mammalian and Xenopus Emi2s are also functionally distinct. Species alignment of predicted Emi2 sequences conserves multiple regulatory elements of xEmi2 (Fig. 3A) but it is not known whether they function outside Xenopus. To address this, we compared xEmi2 and mouse Emi2 behaviour in mouse oocytes.

Mouse GV oocytes were injected with cRNA encoding mCherry and subjected to IVM for 16 hours (referred to as IVM16; see Fig. S1B in the supplementary material). Of these mCherry-injected controls, 90±1.9% (n=61) underwent normal meiotic progression and arrested at mII, whereas those injected with Emi2-mCherry cRNA precociously arrested at mI (see Fig. S3A in the supplementary material). Emi2 localized to mI spindles (Fig. 3B); no first polar body (Pb₁) extrusion was observed by videomicroscopy (not shown). Segments of Emi2 (residues 1-250 and 251-C-terminus) localized nebulously to spindles (see Fig. S3B in the supplementary material). When GV (i.e. nuclear) breakdown was suspended by IBMX, Emi2-mCherry was efficiently targeted to the GV, localizing to a nuclear domain outside the nucleolus-like body (Bouniol-Baly et al., 1999; Fig. 3C; see Fig. S3C in the supplementary material) and to nuclei in NIH 3T3 cells (not shown), even though Emi2 lacks a consensus nuclear localization motif (Lee et al., 2006). Similar GV localization was exhibited by Emi2_251-C-ter – but not by the N-terminal Emi2 domain of Emi2_1-250-mCherry (see Fig. S3C in the supplementary material).

These experiments did not distinguish between the metaphase-imposing activities of Emi2-mCherry and native Emi2. To address this, we employed the si16 system of RNAi (see Fig. S1C in the supplementary material) to deplete native Emi2, by injecting siEmi2-5′UTR#1 (Fig. 3D; see Fig. S3D in the supplementary material) or #2 (not shown), which are specific to native Emi2 mRNA (Shoji et al., 2006; Amanai et al., 2006). As expected (Shoji et al., 2006), almost none (1.9±1.9%) of the oocytes injected with mCherry cRNA in the si16 system established mI arrest, whereas mouse Emi2-mCherry cRNA caused 95.6±2.6% mI arrest (Fig. 3D), showing that the Emi2 knock-down phenotype had been rescued.

If xEmi2 is mechanistically equivalent to mouse Emi2, it should have similar properties. However, in contrast to Emi2, xEmi2-mCherry was excluded from the GV of IBMX-treated oocytes, suggesting that it lacked a nuclear targeting signal (Fig. 3C). When native Emi2 was depleted in the si16 system, xEmi2-mCherry elicited metaphase (mI and mII) at a lower efficiency than that achieved by its mouse counterpart, Emi2-mCherry (Fig. 3D; see Fig. S3D in the supplementary material). These findings indicate that mouse and Xenopus Emi2s are not functionally interchangeable and that they participate in distinct regulatory mechanisms. We therefore sought a mechanistic dissection of mouse Emi2 in vivo.

Injecting GV oocytes with cRNA encoding either Emi2_1-250 or Emi2_251-Cter as mCherry fusions induced ~100% mI arrest following IVM16 in a manner analogous to full-length Emi2-mCherry (Fig. 4A; see Fig. S4 in the supplementary material). Analogous to xEmi2, functional MAPK-dependent phosphorylation, D-box and ZBR motifs all reside in Emi2_251-Cter (Fig. 3A), so Emi2_1-250 was not expected to possess cytostatic activity. In mitotic cells, full-length Emi2 and Emi2_251-Cter co-complexed with APC subunits Cdc20 and Cdc27, as judged by co-immunoprecipitation (Fig. 4B). We did not observe interactions between Emi2_1-250 and Cdc27, but occasionally detected an interaction between Emi2_1-250 and Cdc20 (Fig. 4B). Emi2_251-Cter retained the ability to induce mI arrest in si16 oocytes but Emi2_1-250 did not (Fig. 4A; see Fig. S4 in the supplementary material), suggesting mechanisms that are, respectively, independent of and dependent on native Emi2.

We wished to define more specifically the Emi2 N-terminal cytostatic function. Residues 1-100 induced mI arrest in 42.3±3.6% (n=26) of IVM16 oocytes (Fig. 4A), suggesting that residue 100 lies within the cytostatic N-terminal region. Consistent with this, Emi2_80-115 induced 100% IVM16 mI arrest, but did not in si16 oocytes (Fig. 4A; see Fig. S4 in the supplementary material). Residues 80-115 of ectopically expressed Emi2 are therefore not inherently sufficient for metaphase establishment but enhance native Emi2-dependent mI arrest; the region is conserved among mammals (Fig. 4C).

The Emi2 C terminus contains a putative ZBR, but although a single ZBR mutant of Xenopus xEmi2 lacked cytostatic activity in cell-free extracts, its stability was not reported (Schmidt et al., 2005). To address the relationship between Emi2 and Zn²⁺ in mII maintenance, we injected mII oocytes with cRNA and exposed them ~4 hours later to the highly specific Zn²⁺ chelator, N,N,N′,N′-tetrakis-(2-pyridylmethyl)-ethylenediamine (TPEN), which has been shown to induce meiotic exit (Suzuki et al., 2010). Injection with cRNA encoding Emi2-mCherry, but not mCherry or Emi2^C573A-mCherry (a mutant within its putative ZBR; see below), reduced the efficiency of TPEN-induced mII exit (Pb₂ extrusion; Fig. 4D). Injection of cRNA encoding ZBR mutants in the si16 system generally produced mCherry fluorescence intensities lower than Emi2-mCherry controls (see Fig. S5 in the supplementary material), even though qPCR confirmed higher mutant [cRNA] in oocytes (average, 1.23±0.38; control [Emi2-mCherry]=1.0±0.0). To compensate, more cRNA was injected and the resultant mCherry epifluorescence level related to its corresponding ability to impose metaphase arrest (Fig. 4E; see Fig. S6 in the supplementary material). These experiments indicate a pronounced contribution from ZBR residues C573, C591, C601 to Emi2 cytostatic activity in vivo.

Pre-treatment with TPEN had no discernable effect on Emi2 complex formation with the APC core subunit, Cdc27 in NIH 3T3 cells (Fig. 4F), consistent with separable APC- and Zn²⁺-binding domains (Miller et al., 2006; Tang et al., 2010). We also found that whereas Emi2_551-Cter was unable to impose metaphase arrest in si16 oocytes, Emi2_491-Cter did so efficiently; the results were mirrored in the IVM16 system (Fig. 4G; see Fig. S7A in the supplementary material). This lead us to investigate the candidate D-box RXXL, which lies between 491 and 551. The double mutant Emi2^R511A,L514A-mCherry induced mI arrest only weakly in si16 oocytes (22.5±2.5%, n=29), even when expressed at high levels (Fig. 4G; see Fig. S7A in the supplementary material), indicating that R511 and/or L514 are required by Emi2-mCherry for full cytostatic activity.

These studies predict autonomous meiotic regulatory activity in the Emi2 C terminus. Indeed, when cRNA encoding Emi2_551-Cter was injected into mII oocytes, ~100% underwent meiotic exit (Fig. 4H; see Fig. S7B in the supplementary material). This was probably not due to Zn²⁺ sequestration by the ZBR, as Emi2_551-Cter^C573A also efficiently induced meiotic exit (Fig. 4H). Recently, an RL domain at the C terminus has been shown in Xenopus to mediate APC binding (Ohe et al., 2010). Accordingly, cRNA encoding Emi2_551-Cter^ΔRL, in which the RL domain had been deleted, failed to elicit mII exit (Fig. 4H). However, Emi2_491-Cter, which contains the RLAL D-box in addition to the ZBR and RL domain (Fig. 3A), only modestly (13.3±13.3%, n=30) induced mII exit (Fig. 4H), suggesting that it contained information not present in Emi2_551-Cter that tended to impose mII.

Collectively, these data support the conclusion that residues 491 to the C terminus represent the (near) minimal region of Emi2 required for it to establish and maintain metaphase arrest. Meiotic stabilization by Emi2 involves its D-box (residues 511-514), ZBR (573~613) and RL domain (640-641) and possibly an indirect mechanism involving a novel N-terminal domain (80-115).

We next investigated signalling processes implicated in mII exit to test how they converged on Emi2. The calmodulin (CaM) inhibitor, CGS9343B inhibited Pb₂ emission and ectopic Emi2 degradation following parthenogenetic activation (see Fig. S8A,B in the supplementary material). CaM-dependent enzymes calcineurin (CaN, also called PP2B) and calmodulin kinase II (CaMKII) mediate Xenopus mII exit and it has been suggested that they achieve this by regulating xEmi2 degradation (Rauh et al., 2005; Nishiyama et al., 2007b). However, mouse oocytes lacked the CaN catalytic A subunit above a detection limit of ~700 fg/oocyte (43 nM; see Fig. S8C in the supplementary material), 10- to 100-fold lower than CaN concentrations in other tissues (Sharma et al., 1979; Stewart et al., 1983). ICSI-induced mII exit was slowed but not prevented by a combination of the CaN inhibitors CsA, FK506 or CaN-inhibitory peptide (see Fig. S8D in the supplementary material). Thus, in contrast to Xenopus, we found no evidence that CaN is essential for mouse mII exit, suggesting that the principal effect of CGS9343B was indirectly to reduce CaMKII activity.

In keeping with recent reports (Chang et al., 2009; Backs et al., 2010), GV and mII oocytes contained mRNA for CaMKIIγ (CaMKIIγ) but not the three other isoforms; we also found CaMKIIγ transcripts in spermatozoa (Fig. 5A). Mature mII oocytes contained two CaMKIIγ mRNA variants, encoding the previously reported 67 kDa γ3 (γ3) isoform (Chang et al., 2009) and the J isoform (γJ), not previously described in the mouse, containing a predicted 21 amino acid variable domain insertion (AAL69956.1; Fig. 5A) (Gangopadhyay et al., 2003). CaMKIIγ3 and γJ isoforms were approximately equimolar in mII oocytes (Fig. 5A).

Injecting a mixture of active bovine brain CaMKII isoforms into mII oocytes did not induce mII exit (not shown), leading us to challenge Xenopus and mouse (x)Emi2 proteins with active bovine brain CaMKIIγ or mouse oocyte CaMKIIγ3 and γJ isoforms in a phosphorylation assay in vitro. All CaMKII preparations tested phosphorylated a control substrate and exhibited autophosphorylation, indicative of holoenzyme activity (Fig. 5B). However, there was little or no phosphorylation of either mouse or Xenopus (x)Emi2 (Fig. 5B). The absence of detectable Emi2 phosphorylation in vitro is unlikely to be because CaMKII functions as a non-CaMKIIγ hetero-multimer, as neither was phosphorylated by the cocktail of neuronal CaMKII isoforms (Fig. 5B). We therefore investigated whether this lack of phosphorylation in vitro might reflect a predisposing requirement for protein kinase C (PKC), which has also been implicated in oocyte activation (Halet, 2004). Combining CGS9343B with the PKC inhibitor chelerythrine markedly inhibited mII exit (see Fig. S8E in the supplementary material), but chelerythrine alone only modestly inhibited Emi2-mCherry degradation in mII oocytes (see Fig. S8F in the supplementary material) suggesting that PKC does not potentiate CaMKIIγ signalling that might lead to Emi2 degradation.

To address whether CaMKII transduces the Ca²⁺ signal at fertilization to degrade mouse Emi2, mature mII oocytes were injected with cRNA encoding truncation mutant Emi2 mCherry fusions, followed by Sr²⁺ exposure [mII(Sr)] or ICSI [mII(ICSI); see Fig. S1D,E in the supplementary material]. Taking fluorescence decrease as a meter of degradation, Sr²⁺ induced loss of Emi2_1-300- and Emi2_50-300-, but not Emi2_1-250- or Emi2_251-Cter-mCherry (Fig. 5C; see Fig. S9, S10 in the supplementary material). Thus, residues 50-300 are apparently required for Ca²⁺-dependent degradation of mouse Emi2, but not the region corresponding to the Plx N-terminal phosphodegron in xEmi2 (Fig. 3A). Analogous to Xenopus, residues 50-300 contain a candidate, non-canonical CaMKII phosphorylation target, T176 (Fig. 3A); full-length xEmi2 was efficiently degraded in response to Sr²⁺ in mouse oocytes (Fig. 5D). Mouse Emi2^T176A-mCherry did not undergo Sr²⁺-induced degradation in mII oocytes (Fig. 5C). A similar phenotype was exhibited by D274A and S275N mutants, which lay within a consensus phosphodegron, DSGX_nS (Fig. 5C); the corresponding phosphodegron is located differently in xEmi2, and is essential for destruction downstream of Ca²⁺-dependent xCaMKII phosphorylation (Rauh et al., 2005).

Finally, we investigated the activity responsible for Emi2 removal in response to fertilization-induced Ca²⁺ mobilization. The degradation of Emi2 is highly selective in that Ca²⁺ mobilization does not result in the targeting of Ca²⁺/Plk1-regulated substrates, Emi1, β-Catenin or Bora for destruction (see Fig. S10, Fig. S11A in the supplementary material). The pathway responsible did not require the APC, as depletion of its activator, Cdc20, by RNAi (Amanai et al., 2006) had little, if any, effect on Emi2 degradation (see Fig. S11B in the supplementary material). Moreover, degradation of Emi2 occurred efficiently in the oocytes of gene-targeted mice lacking the E3 ubiquitin ligase SCF component Trcpb1 (Fig. 5E), implying that Trcpb1 is not essential.

This work reveals in intact mouse oocytes, essential domains and residues, protein interactions and degradative mechanisms required by Emi2 to establish, maintain and exit from meiotic metaphase (Fig. 6). We also show that the establishment of metaphase by Mos requires Emi2 and suggest that they are linked by MAPK, as in Xenopus (Inoue et al., 2007). This begs the question of precisely how, as Xenopus MAPK signals to xEmi2 via xRsk, but Rsk is dispensable in the mouse (Dumont et al., 2005). It was therefore important to establish whether mouse Emi2 possessed a functional site corresponding to the Xenopus xRsk phosphorylation target. It does: T327 (which corresponds to the xRsk target, T336 in Xenopus) is essential for mouse Emi2 cytostatic activity. This suggests that Mos-MAPK signals to Emi2 either directly or via a similar pathway to that in Xenopus, omitting Rsk and passing through an unknown intermediate.

Our work implies that, once established, Mos is not required to maintain mammalian mII in that it autonomously disappears during mouse mII oocyte aging. Rather, the disappearance of Mos occurs during an increase in spindle abnormalities (Webb et al., 1986; Verlhac et al., 1996) (Fig. 2). Loss of active MAPK also correlates with spindle lengthening (see Fig. S2B,C in the supplementary material), implicating the mouse oocyte spindle regulators MISS and Doc1R, both of which are MAPK substrates that do not mediate mII arrest (Lefebvre et al., 2002; Terret et al., 2003). It is therefore possible that, after mII arrest is established, the principal role of Mos is to maintain spindle integrity, possibly in conjunction with spindle regulators such as MISS and Doc1R.

Mouse Emi2 and Xenopus xEmi2 possess distinguishable activities in mouse oocytes; xEmi2 did not fully compensate for mouse Emi2 in meiotic regulation and localization patterns of xEmi2 were different from those of mouse Emi2. Differences may (in part) reflect the temperatures at which mouse and Xenopus fertilization occurs: 36.9°C and ~23°C, respectively. Species alignment of Emi2 sequences reveals both conservation and variation of domains and amino acid residues shown to be important for xEmi2 function (Fig. 3A), suggesting that non-conserved regions have evolved in different species to perform these functions.

It has already been shown that mouse Emi2 has the ability to arrest meiotic oocytes at metaphase (Shoji et al., 2006). When expressed in GV oocytes, this cytostatic activity was found principally to reside in a large C-terminal segment (Emi2_251-Cter), which contains domains that are important for Emi2 regulation by analogy to Xenopus, including PP2A-interacting domains, a D-box, ZBR and RL domain (Fig. 3A) (Schmidt et al., 2005; Wu et al., 2007a; Nishiyama et al., 2007a; Ohe et al., 2010). We provide evidence here and in complementary work (Suzuki et al., 2010), that Zn²⁺ is required for mII arrest, working through Emi2; we identify residues that contribute to the putative Zn²⁺-coordinating lattice needed for Emi2 cytostatic activity. Taken together, these results suggest that a Zn²⁺-containing functional Emi2 ZBR is essential for it to inhibit the APC in vivo.

A fragment of Emi2 from 551-Cter elicited mII exit with or without the ZBR mutation, C573A (Fig. 4H; see Fig. S7B in the supplementary material). This fragment contains a C-terminal RL motif shown in Xenopus to mediate Emi2-APC binding, the overexpression of which causes native Emi2 to dissociate from the APC, causing meiotic progression (Ohe et al., 2010). A similar mechanism is likely in the mouse, as Emi2_551-Cter^ΔRL failed to induce mII exit (Fig. 4H). In contrast to 551-Cter, Emi2 residues 491-Cter, which also contain the D-box and RLAL, only weakly induced mII exit (Fig. 4H). The low activity relative to 551-Cter is probably not because the D-box turned the fragment into an APC substrate; expression levels were comparably high (see Fig. S7B in the supplementary material). Rather, we favour a model in which the D-box, when combined with ZBR and RL domains, effectively inhibits the APC, consistent with the behaviour of xEmi2 (Tang et al., 2010). We tested this in the IVM16 and si16 systems and found that, unlike Emi2_551-Cter, which possessed only limited cytostatic activity, Emi2_491-Cter was sufficient for potent metaphase establishment and maintenance (Fig. 4G; see Fig. S7A in the supplementary material). This is in keeping with the situation in Xenopus (Wu et al., 2007b; Tang et al., 2010) and indicates that in mammals a region of Emi2 from a position between residues 491 and 551 to the C terminus is sufficient to establish and maintain mII arrest.

Unexpectedly, an N-terminal fragment, Emi2_1-250, also exhibited metaphase-inducing activity, as evidenced by its ability to impose mI arrest in GV oocytes (Fig. 4A; see Fig. S4 in the supplementary material). This activity required native Emi2, implying an indirect mechanism. Deletion mutagenesis mapped the activity to a region (residues 80-115) that lacks consensus phosphorylation or other regulatory sites (Marchler-Bauer et al., 2009) but is conserved among several species, including mammals (Fig. 4C). It has been suggested (Wu and Kornbluth, 2008) that smooth mI-to-mII transition relies on Emi2 degradation in mI, and it is possible that Emi2_80-115 is required for mI inactivation and/or degradation of Emi2, such that ectopically expressed 80-115 interferes with this negative regulation.

Separable modules of mouse Emi2 orchestrate mII metaphase establishment and exit. In Xenopus, it has been shown that S33/38 is essential for xEmi2 degradation during mII exit, but although there are signs that this degron (DSGX_nS) is conserved in lizards and the platypus, there is no conserved motif in the N-terminal region of Emi2 from most other species, including eutherian mammals (Fig. 3A). Xenopus possesses the more C-terminal sequence, DSAFHS, that is involved in degradation independently of mII exit (Nishiyama et al., 2007a) and this position corresponds to ²⁷⁴DSGFCS²⁷⁹ in the mouse; we find that D274 and S275 are essential for Ca²⁺-dependent mouse Emi2 degradation (Fig. 5C; see Fig. S10B in the supplementary material). Consistent with this, the ²⁷⁴DSGFCS²⁷⁹ motif is located among residues 51-300, which approximately define the minimal region of Emi2 sufficient for Ca²⁺-dependent degradation (Fig. 5C). This suggests that the motif has become a Ca²⁺-dependent phosphodegron in the mouse and other species in which it is conserved (Fig. 3A) and that the position of this Emi2 phosphodegron has shifted during evolution. The reasons for this transposition will remain unknown until the function of the N-terminal domain of Emi2 becomes clear. The evolutionary retention of an N-terminal extension in mammals, even though it is no longer required for mII exit, implies that it plays one or more additional roles, such as that played by residues 80-115 or even roles that do not directly modulate the cell cycle.

We found no evidence that CaN is essential for mII exit in the mouse: a cocktail of inhibitors did not prevent meiotic exit and the CaN catalytic subunit was undetectable in mII oocytes. This implies that in mammals the Ca²⁺ signal is primarily transduced by CaMKIIγ to downregulate Emi2 activity (Chang et al., 2009; Backs et al., 2010). We identified two isoforms of oocyte CaMKIIγ, present in equimolar amounts, raising the question of whether CaMKIIγ mediates dual regulation of Emi2 or only one of the isoforms is essential. In other cell types, CaMKII splice variants exhibit distinctive protein interactions that influence CaMKII activity directly and/or by influencing subcellular targeting (Bayer et al., 1998; Hudmon and Schulman, 2002; Gangopadhyay et al., 2008). CaMKIIγ splice variants that affect protein binding include G-2, which associates specifically with the PP2C-type phosphatase SCP3; the G-2 isoform shares a variant region with the γJ isoform present in oocytes (Gangopadhyay et al., 2003; Gangopadhyay et al., 2008). Interactions have been shown to modulate CaMKII activity through allosteric mechanisms that promote substrate phosphorylation in vivo but only poorly in vitro (e.g. Lu et al., 2003). This may explain why CaMKIIγ did not phosphorylate Emi2 in vitro, with the clear implication that one or more accessory proteins are required. One goal of future work will be to validate this and identify the adaptor molecule(s) responsible.

The machinery responsible for Ca²⁺-induced Emi2 degradation in mouse oocytes is highly specialized. Neither Emi1, β-Catenin nor Bora – which undergo Plk- and Ca²⁺-dependent proteasomal degradation in response to cell cycle progression – were degraded in oocytes treated with SrCl₂ (see Fig. S11A in the supplementary material). Emi2 degradation was not sensitive to the removal of Cdc20; Cdc20 is an APC activator required for mII exit (see Fig. S11B in the supplementary material) (Shoji et al., 2006). This is consistent with the situation in Xenopus, in which xEmi2 degradation requires βTrcp, a component of the E3 ubiquitin ligase, SCF (Tung et al., 2005). Using mII oocytes from gene-targeted females that lack one of the two mouse βTrcp isoforms, Trcpb1, we provide strong evidence that Trcpb1 is dispensable for Emi2 degradation (Fig. 5E). As Trcpb1^−/− null female mice exhibit normal fertility (Nakayama et al., 2003), it is possible either that Trcpb1 plays no physiological role in Emi2 removal during fertilization, or that Trcpb2 and/or additional E3 ubiquitin ligases can efficiently compensate when it is absent.

We are grateful to Professor Keiko Nakayama for the provision of homozygous Trcpb1 knockout mice, to Heide Oller and to members of LARGE for their help looking after mice, and to a Wellcome Trust Value In People Award. This work was funded by RIKEN. Deposited in PMC for release after 6 months.