|Home | About | Journals | Submit | Contact Us | Français|
The establishment of trophectoderm (TE) manifests as the formation of epithelium, and is dependent on many structural and regulatory components that are commonly found and function in many epithelial tissues. However, the mechanism of TE formation is currently not well understood. Prickle1 (Pk1), a core component of the planar cell polarity (PCP) pathway, is essential for epiblast polarization before gastrulation, yet the roles of Pk family members in early mouse embryogenesis are obscure. Here we found that Pk2−/− embryos died at E3.0–3.5 without forming the blastocyst cavity and not maintained epithelial integrity of TE. These phenotypes were due to loss of the apical-basal (AB) polarity that underlies the asymmetric redistribution of microtubule networks and proper accumulation of AB polarity components on each membrane during compaction. In addition, we found GTP-bound active form of nuclear RhoA was decreased in Pk2−/− embryos during compaction. We further show that the first cell fate decision was disrupted in Pk2−/− embryos. Interestingly, Pk2 localized to the nucleus from the 2-cell to around the 16-cell stage despite its cytoplasmic function previously reported. Inhibiting farnesylation blocked Pk2’s nuclear localization and disrupted AB cell polarity, suggesting that Pk2 farnesylation is essential for its nuclear localization and function. The cell polarity phenotype was efficiently rescued by nuclear but not cytoplasmic Pk2, demonstrating the nuclear localization of Pk2 is critical for its function.
In preimplantation mouse development, the first cell lineages to be established are the trophectoderm (TE) and inner cell mass (ICM) (Marikawa and Alarcón, 2009; Rossant and Tam, 2009; Sasaki, 2010; Zernicka-Goetz, 2009). In mouse, these lineages begin to diverge at the 8-cell stage, when blastomeres polarize during compaction. In this process, blastomeres acquire apical-basal (AB) polarity, typified by the apical localization of microvilli and acquisition of cytoplasmic polarity, including the asymmetric distribution of E-cadherin and reorganization of the microtubule (MT) network (Fleming and Johnson, 1988; Jonson et al., 1986; Maro et al., 1990). Next, the blastocoel, a fluid-filled cavity, is formed in the central region of the embryonic cell mass. Blastocoel formation requires two major interrelated features of TE differentiation: intracellular junction biogenesis and a directed transport system, mediated by Na+/K+ ATPase (Eckert and Fleming, 2008). These findings suggest that the functional polarity of embryonic cells is essential for proper blastocyst cavity formation. The cell polarity complex (aPKC/PAR) regulates the orientation of cell cleavage planes and cell polarity (Alarcón, 2010; Dard et al., 2009; Plusa et al, 2005). Thus, the aPKC/PAR complex also influences the localization of blastomeres to an outer or inner position in the blastocyst as well as blastocoel formation (Alarcón, 2010; Plusa et al, 2005). However, the connection between these two processes is unclear.
Planar cell polarity (PCP) is manifested as the coordinated, polarized orientation of cells within epithelial sheets, or as the directional cell migration and intercalation during convergent extension (Axelrod, 2009; Goodrich and Strutt, 2011; Simons and Mlodzik, 2008; Zallen, 2007). The signaling pathway that controls PCP consists of Celsr/Flamingo, Frizzled (Fzd), Dishevelled (Dvl/Dsh), Van Gogh/strabisumus (Vang/stbm), Diego, and Prickle (Pk). The PCP pathway is proposed to modulate the cytoskeleton and influence cell morphology rather than cell fates in many cases (Wansleeben and Meijlink, 2011). Nevertheless, some PCP components are essential for asymmetric cell division and cell fate determination during mouse neurogenesis (Goodrich, 2008; Lake and Sokol, 2009). Furthermore, some PCP components appear to participate in multiple pathways and to create crosstalk with other pathways such as that for AB determination, suggesting that some PCP genes have acquired new functions that exploit their fundamental properties in PCP signaling (Goodrich, 2008; Goodrich and Strutt; 2011; Nishita et al., 2010; Wansleeben amd Meijlink, 2011).
Two prickle (pk) genes, Pk1 and Pk2, have been identified in the mouse (Katoh and Katoh, 2003), and the deletion of Pk1 revealed its early developmental role of Pk1 in establishing the AB polarity of the epiblast (Tao et al., 2009). In addition, the mouse PCP core component Vangl2 interacts genetically with Scribble (Scrib) (whose Drosophila orthologue regulates AB polarity) in determining the planar polarity of inner cell cilia (Montcouquiol et al., 2003, 2006). Together, these results highlight a functional link between the PCP and AB polarity systems (Nishita et al., 2010). On the other hand, the apical determinants aPKC and PARD6b regulate cell lineage during mouse preimplantation development (Alarcón, 2010; Dard et al., 2009; Plusa et al., 2005), although their roles in controlling TE differentiation and blastocyst formation are still unclear. The functional relationship between PCP components and aPKC/PAR complex proteins, the key players in AB polarity formation, are also unclear. However, evidence suggests that aPKC phosphorylates and inhibits the activity of Fzd to regulate PCP in the Drosophila eye (Djiane et al., 2005) and Dvl is involved in establishment of AB polarity by binding to and regulating the activity of Lgl, a target of aPKC in Xenopus (Dollar et al., 2005).
In this article, we show that Pk2 is expressed throughout mouse preimplantation development and regulates AB cell polarity during compaction, thus also regulating the first cell fate decision. The Pk2−/− embryos showed arrested development at the late morula stage, failed to form the blastocyst cavity, and at the late 8-cell stage, showed defects in microtubule network redistribution and Na+/K+ ATPase α1 subunit accumulation on the basolateral membrane. Furthermore, nuclear-targeted but not cytoplasmically localized Pk2 rescued the polarity defect phenotypes, indicating that nuclear retention of Pk2 is essential for its function. We provide the first evidence that Pk2 plays a crucial role in mouse preimplantation development and highlight the functional link between AB polarity establishment and PCP pathway.
Pk2 mutant mice were generated as described (Tao et al., 2011). Two Pk2 mutant mouse strains (line1, #64 and line 2, #134) were established from independent homologous recombinant ES cells, and no difference in preimplantation phenotype (defective formation of blastocyst cavity) was apparent between them (Table S1A). Unless stated otherwise, all of experiments were carried out with line 2 mice. For pre-implantation embryos, genotyping was performed on individually isolated embryos directly or after observation in culture of following antibody staining. Vangl2Lp/+ mice (strain LPT/LeJ) were obtained from The Jackson Laboratory and maintained by backcrossing to C57BL/6J. Inbred C57BL/6J were purchased from LARGE (RIKEN CDB) or CLEA CO., LTD., Japan. Inbred CBA strains were purchased from Oriental Yeast CO., LTD. and Charles River CO., LTD. Mice were housed in environmentally controlled rooms at the NIBB and/or RIKEN CDB, under the institutional guidelines for animal and recombinant DNA experiments, respectively.
The preimplantation embryos were collected from timed natural matings or in vitro fertilization pre-Embryo transfer (IVF-ET). The preimplantation embryos were collected by flushing dissected oviducts or uteri with M2 medium (M7167, Sigma) with HEPES (M7167, Sigma). Embryos were cultured in 80 μl drops of KSOM (ARK Resource Co., Ltd.) under mineral oil (M8410, Sigma) at 37°C in a 5% CO2 incubator. The inhibitor treatment was performed in 3.5 cm dish (Nunc). Embryos with or without the zona pellucida were cultured in 80–160ul drops containing several dose of inhibitors and covered with mineral oil. Details of antibodies and inhibitors are shown in Table S2.
Embryos were fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) for 15–30 min at room temperature, and then embryos were treated with acidified Tyrode’s solution (T1788, SIGMA) to remove the zona-pellucida. Embryos were subsequently permeabilized with 0.2% Triton X-100 in PBS for 20 min at room temperature, washed in PBSS for 5 min, blocked with 2% goat serum in PBS (blocking solution), and incubated overnight with primary antibodies diluted in blocking solution at 4°C. The Primary antibodies lists were provided in Table S2. After washing in PBSS for 15 min, 3 times, embryos were incubated with the following secondary antibodies diluted in PBSS for 1 hr at room temperature: Alexa Fluor 488 goat anti-rabbit, goat anti-mouse, donkey anti-goat (A11034, A11029, A11055; 1:1000–4000, Molecular Probes) and/or Alexa Fluor 555 goat anti-rabbit, goat anti-mouse, goat anti-rat (A21428, A21422, A21434; 1:1000–4000, Molecular Probes). Embryos were washed in PBSS and postfix in 4% PFA/PBS. Then, embryos were placed in a drop of 40% glycerol/PBS or PermaFluor aqueous mounting medium (TA-030-FM, Thermo Scientific) on a glass-bottom dish (D110400, Matsunami).
Laser scanning microscopy was performed using LSM 510 Meta, LSM 710 ZEN and LSM 780 ZEN (Zeiss), with optical sections obtained approximately every 3–5μm. The relative fluorescence of intensity in each picture was measured by ImageJ, and accounts between value 0 and 255. These indices were based on a pixel-based quantitative analysis of confocal images. The obtained relative fluorescence intensity profiles were transferred into Microsoft EXCEL for further quantification. Same blastomeres with more intense nuclear than cytoplasmic staining were considered positive. The images shown in some of the figures were modified using contrast enhancement or and brightness (Adobe Photoshop CS4). Statistically significant differences (* p < 0.05, **p < 0.01) are indicated by asterisks.
Embryos were treated with acidified Tyrode’s solution (T1788, SIGMA) to remove the zona-pellucida. Three sets of 100–200 embryo pools were used for preparation of protein extracts separately. Embryos were collected in a sample buffer (x4 ME, Wako) (62.5mM Tris-HCl at pH6.8, 0.5x PBS, 2% SDS, 10% Glycerol, 5% 2-mercaptothanol), and then sonicated to cleave genomic DNA. The nuclear and cytoplasmic extracts were separated by using Nuclear/cytosol fractionation kit (K266-25, BIOVISON), following the manufacturer’s instructions. These proteins were denatured by heating at 95°C for 5 min, separated by electrophoresis on SDS-5% polyacrylamide gels, and transfered onto Immobilon (IPVH20200, Millipore). Blots were blocked with 5% skimmed milk, incubated overnight with primary antibodies at 4°C. After several washes, blots were incubated with a 1:10,00 dilution of HRP-conjugated anti-rabbit IgG antibody, and detected by using ECL Advance Western Blotting Detection Kit (RPN2106, GE Healthcare) and LAS3000 lumino-image analyzer (Fuji).
The full-length coding sequences of mouse Pk1 (ID A630056F08) and Pk2 (ID M5C1081L03) were amplified from RIKEN FANTOM cDNA library. The Myc-tagged Prickle2 and various myc tagged Prickle2 mutant expression vectors were constructed with 6xMyc pCAGGS or 6xMyc pcDNA3.1 poly (Yamagata et al., 2005) vectors, respectively. For mutagenesis, the specific primer used with the KOD -Plus- mutagenesis kit (SMK-101, TOYOBO) are shown in Table S3A. Theses oligo cDNAs were performed by Operon Biotechnologies, Japan.
The total RNAs were isolated from individual embryos using TRIzol plus RNA purification kit (#12183-555, Invitrogen) following the manufacturer’s instruction. All of total RNA was used for cDNA synthesis using Superscript III reverse transcriptase (#18080-044, Invitrogen) following the manufacturer’s instruction. The resultant cDNA was diluted appropriately for semi-quantitative PCR. Primers used are shown in Table S3B.
Poly(A)-tailed RNA was synthesized from several cDNAs cloned into the pcDNA3.1- poly(A)83 plasmid (Yamagata et al., 2005) and in vitro transcription was performed using the mMESSAGE mMACHINE T7 and Sp6 kit (AM1344 and AM1340, Ambion). Synthesized RNAs were purified by NucAway Spin Columns (AM10070, Ambion) and dissolved in water and −20°C before microinjection. The purified RNAs were injected into both blastomeres of 2-cell-stage embryos according to standard protocols (Nagy et al., 2003). The DNAs were injected in pronuclei of 1-cell stage embryos according to standard protocols (Nagy et al., 2003).
Analysis of active Rho protein was conducted as previously described (Xie et al., 2008). Briefly, embryos were fixed with freshly prepared 2% PFA in PBS for 20 minutes at room temperature. After washing, embryos were permeabilized in PBS containing 3% BSA, 0.1 M glycine and 0.05% Triton X-100, followed with blocking in the same buffer containing 2% donkey serum, then incubated with 50 μg/ml GST tagged Rhotekin-RBD (RT01A, Cytoskelton) for 2 hours at 4°C. Anti-GST primary (1:200, B-14, Santa Cruz) and secondary antibodies were used to visualize the GST-GTP-bound active Rho proteins.
To investigate whether Pk2 is involved in preimplantation development, we first examined the expression and subcellular localization of the Pk2 protein in MII oocytes and preimplantation embryos by immunostaining with an anti-Pk2-specific antibody (Figs. 1A-F). Pk2 protein was undetectable in MII oocytes (n = 23/23) (Fig. 1A), and was first detected in the nucleus of 2-cell-stage embryos (n = 17/18) (Figs. 1B and B’). The nuclear signal became stronger until the 8–16-cell stage (n > 15 embryos, respectively) (Figs. 1B-D’; Fig. S1A), suggesting that Pk2 is not maternally expressed but produced only in the embryo. To distinguish whether maternal or zygotic transcripts produce Pk2, we treated embryos from the late 1-cell to the 4-cell stage with 0.1 mg/ml α-amanitine (α-AM), an inhibitor of RNA polymerase II, and therefore of zygotic gene activation (Hamatani et al., 2004) (Fig. S2). In most of the α-AM-treated embryos, nuclear Pk2 signals were absent (2-cell stage; n = 21/21, 4-cell stage, n = 19/22) (Figs. S2D, D’, F and F’), suggesting that the Pk2 protein is not maternally expressed. From the 16-cell stage onwards, the nuclear Pk2 signal decreased as the cytoplasmic Pk2 signal increased (n > 20 blastomeres, n = 5 embryos, respectively) (Figs. 1E and E’; Figs. S1A, C and D), an observation that was confirmed by western blotting of nuclear and cytoplasmic extracts from compacted 8-cell stage and blastocyst stage embryos (Fig. S1B). Pk1, the other prickle family gene in the mouse, was also expressed in the nucleus of all blastomeres from the 2-cell to the blastocyst stage, as shown by immunostaining with an anti-Pk1-specific antibody (n > 20 embryos, respectively) (Fig. S3A). Notably, Pk2 was mostly localized to the nuclei of embryonic cells up to the 16-cell stage, and primarily before blastocyst formation, but Pk1 remained in the nuclei at the blastocyst stage.
To study the in vivo function of Pk2, we generated knock-out mice lacking PET/LIM, one of the functional domains of Pk2 (Tao et al., 2011). In the Pk2−/− embryos, no Pk2 protein was observed at the 8-cell stage (n = 8/8) (Fig. S3B), while the localization and expression level of Pk1 was not affected (n = 5/5), compared with Pk2+/+ and Pk2+/− (control) embryos (n = 23/23) (Fig. S3C). Two independent Pk2 homozygous mutant alleles displayed similar defective formation of blastocyst cavity on the C57BL/6 x CBA mixed background (Table S1A). After crossing pairs of Pk2+/− mice, we did not find any Pk2−/− mice on the CBA background (Tables S1B and C), suggesting the homozygous mutation was lethal in early embryonic development. However, when the Pk2+/− mice were backcrossed with C57BL/6 over 6 generations, the Pk2−/− progenies were viable and fertile (Table S1D), although they had a slightly reduced body size and an epilepsy-like neuronal disorder as we recently reported (Tao et al., 2011). We assume that the presence of modifier genes for Pk function underlies these background-dependent phenotypes.
To reveal abnormalities in the Pk2−/− embryos, we first examined their preimplantation development. The Pk2−/− embryos underwent compaction at the late 8-cell stage and were morphologically indistinguishable from controls up to the 25–30-cells stage (Figs. 1G and J). Subsequently, the Pk2−/− embryos failed to form a definitive blastocyst cavity, and instead maintained a morula-like morphology (Figs. 1H and K). Up to the late blastocyst stage, Pk2−/− embryos started to exhibit small cell fragments on the surface (Figs. 1I and L). The number of nuclei in Pk2−/− embryos was significantly decreased compared with that in controls at the early blastocyst stage (Fig. 1M), indicating that the cell proliferation of Pk2−/− embryos was arrested or cell death occurred. To test for cell death, we immunocytochemically examined cleavage-stage embryos for cleaved caspase-3, which faithfully represents the execution of apoptosis. The results showed that the Pk2−/− embryos progressed beyond the 16-cell stage, but then the number of cleaved caspase-3 positive cells gradually increased throughout the embryo (n = 8/8) (Figs. S4A-E).
Next, we asked if the differentiation of TE/ICM is affected in Pk2−/− embryos. Cdx2 is the key factor in TE-fate specification (Plusa et al., 2005; Strumpf et al., 2005). Tead4, the TEA domain transcription factor, was recently identified as an essential factor for TE development acting upstream of Cdx2 (Yagi et al., 2007; Nishioka et al., 2008, 2009). Tead4−/− embryos do not express Cdx2, fail to form a blastocyst cavity, and all the cells follow the ICM fate (Yagi et al., 2007; Nishioka et al., 2008). To address whether Pk2 is involved in this cell-differentiation process, we first examined the expression of YAP1, the Tead4 co-activator protein, and Cdx2 in Pk2−/− embryos (Figs. 2A, B, D and E; Figs. S5A-D). In the Pk2−/− embryos, the expression of YAP1 and Cdx2 was observed at the 14–16-cell stage (n = 6/6, respectively) (Figs. S5A-D). After the next cell division, however, the nuclear Yap1 and Cdx2 signals disappeared from most cells in the Pk2−/− embryos (n = 9/9, respectively) (Figs. 2A-F).
Although the outer cells of the Pk2−/− embryos were morphologically distinct (flattened) from the inner cells at 13–15-cell stage, it was still possible that the outer cells have adopted an ICM fate, as was reported for Cdx2−/− blastomeres that inappropriately express Nanog and Oct3/4 (Strumpf et al., 2005). Therefore, we examined the expression of Nanog and Oct3/4 in Pk2−/− embryos (Figs. S5E-H). At 13–15-cell stage, expression levels of Nanog and Oct3/4 in the Pk2−/− embryos were indistinguishable from those in controls (n = 9/10, n = 4/4) (Figs. S5E-H). Cdx2−/− mutants develop into early blastocysts, but the blastocoel eventually collapses (Strumpf et al., 2005). In this mutant, the expression of TE-specific genes significantly decreases, and all the cells are positive for Nanog in early blastocyst (Strumpf et al., 2005). Tead4−/− mutants, and Pard6b siRNA-injected embryos (Alarcón, 2010; Nishioka et al., 2008) show similar phenotypes to Cdx2−/−. Likewise, in Pk2−/− embryos at 26–29-cell stage, the ratio of Nanog-positive blastomeres/total blastomeres was significantly increased compared with controls (n = 11/14) (Figs. 2A’, B’, D’, E’ and G).
Moreover, RT-PCR analysis at the late morula stage revealed the absence of transcripts of Yap1 and Cdx2, and upregulated Nanog expression in the Pk2−/− embryos (Fig. 2H), indicating that Pk2 is essential for the specification and development of TE possibly through the proper establishment of cell polarity and maintenance of gene expression of Yap1 and Cdx2.
We next examined whether Pk2 regulates the asymmetric reorganization of the MT network and/or the accumulation of cell-cell contact via E-cadherin during compaction as markers for epithelial polarity formation. In control and Pk2−/− non compacted-8-cell blastomeres, the MT meshwork was found mainly around the nucleus and at the cell cortex (n = 8) (Figs. S6A-B’), whereas pericentriolar material 1 (PCM1), a major component of the centriole, was diffused in the cytoplasm (n = 4) (Figs. S6C and D). During compaction, the MTs accumulated in the apical part of cell and decreased in the basal domain (Figs. 3B, D and F). In contrast, the Pk2−/− blastomeres exhibited an irregular α-tubulin-staining pattern, with a dispersed MT network, and no distinct organization (n = 8/9) (Figs. 3C, E and G), although E-cadherin localized fairly normally to the AJs (n = 12/12) (Fig. 3H; Figs. S6E and F). In most of the Pk2−/− blastomeres, α-tubulin staining showed an irregular local deposition at the apical cell surface, and accumulated at intercellular contact points, as compared with controls (Figs. 3B-H). These findings revealed that assembly of MT meshwork at the cell cortex was perturbed in the Pk2−/− blastomeres, although phalloidin staining that marks F-actin were normal in Pk2−/− embryos at the 8–12-cell stage (Figs. 3I-J2).
During compaction, the majority of PCM1 puncta accumulate at the apical domain of the controls (Figs. S6G and G’), where PCM1acts as a microtubule-organizing center (MTOC) in blastomeres (Houliston et al., 1987). By contrast, aberrant aggregates of PCM1 were often observed in the apical part of Pk2−/− blastomeres (27/32 blastomeres, n = 4) (Figs. S6H and H’). Interestingly, tyrosinated tubulin, a microvillus structural component, was dispersed, in part, on the apical membrane of Pk2−/− blastomeres (19/24 blastomeres, n = 3) (Figs. S6I and J). Together these findings indicated that the loss of epithelial polarity was accompanied by a delocalization of the microtubular network.
We next examined the possible defects in signaling pathways of Pk2−/− embryos during compaction. Recent studies showed that Wnt-c-jun N-terminal kinase (JNK) signaling and Ras-MAPK signaling are required for cavity formation (Lu et al., 2008; Xie et al., 2008). In Pk2−/− embryos, there were no significant effects on the apical membrane localization of Erk2, which is the downstream effector of Ras-MAPK signaling, β-catenin, or the nuclear localization of phosphorylated c-Jun, a JNK substrate protein (n = 3/3 in each case) (Figs. S7A-F).
Vangl2/stbm is a homolog of the Drosophila PCP gene Vangl/stbm, and its function is closely related to pk’s in zebrafish gastrulation cell movements (Carreira-Barbosa et al., 2003). At the compacted 8-cell stage, the localization of Vangl2 was not affected in Pk2−/− embryos (n = 4/5) (Figs. S7G and H). To investigate whether Pk2 and Vangl2/stbm act cooperatively in the preimplantation embryo, we crossed Pk2+/− female mice to Looptail (Vangl2Lp/+) males, Vangl2 hypomorphic mutants. The double heterozygous mutant embryos formed a normal blastocyst cavity indistinguishable from wild-type embryos (Fig. S7I). Although the penetrance of the phenotype appeared to vary with the genetic background, the result was in contrast to the interaction of these genes in the establishment of cell polarity in the inner ear (Deans et al., 2007).
Rho proteins are required for the maintenance of microtubule orientation (Clayton et al., 1999). We thus performed an in situ Rho-GTP affinity assay (Xie et al., 2008) to examine the RhoA activity in the Pk2−/− embryo. The nuclear signals of total and GTP-bound (active) RhoA GTPase were dramatically decreased at the 8–16-cell stage in Pk2−/− embryos (n = 9/10, for both) (Figs. 3K-N), suggesting that Pk2 regulates microtubules orientation via the activation of Rho family GTPase in preimplantation embryos.
The abnormal microtubule orientation during compaction suggested that Pk2 acts as an AB polarity determinant in the early mouse embryo. To test this hypothesis, we investigated the localization of several polarity components in the blastomeres during compaction, focusing on the aPKC/PAR complex, which plays a role in establishing cell polarity and cell division orientation in many other systems (Suzuki and Ohno, 2006). We first examined the expression and localization of PARD6b, a Par6 homolog, and an atypical PKC, PKCζ. Both aPKC and Par complex adopt a polarized localization from the 8-cell stage onwards; manipulating their function re-directs cell positioning and consequently influences cell fate (Alarcón 2010; Dard et al., 2009; Plusa et al., 2005). We found that, at compacted 8-cell stage of control embryos, significant portions of PARD6b and phosphorylated PKCζ (p-PKCζ) were observed at the apical pole of newly polarized cells (n > 30 embryos, respectively) (Figs. 4A, C and E; Fig. S8A). Notably, on the apical cell surface in Pk2−/− embryos, this distribution of PARD6b, p-PKCζ and PKCζ was, respectively, lost and partially reduced (n = 5 for both) (Figs. 4B, D and E; Figs. S8A and B). The fluorescence intensity of nuclear PARD6b was also significantly reduced in the Pk2−/− blastomeres (22/24 blastomeres, n = 3 embryos) (Fig. 4F). As the loss of Pk2 affect both apical membrane and nuclear localization of PARD6b, Pk2 may functionally interact with PARD6b.
In a variety of polarized cells, the aPKC/PAR complex and PAR-1/EMK1 exhibit complementary localizations along the polarity axis of each cell (Suzuki and Ohno, 2006). At the compacted 8–12-cell stage, EMK1 was mostly found in the nucleus and weakly distributed at the membrane (Vinot et al., 2005) (Figs. 4G and I; Fig. S8C). In the Pk2−/− embryos, much of the EMK1 was reduced from cell-cell contacts (n = 3/5) (Figs. 4H, J and K; Fig. S8C).
Since the Scribble complex (Scribble and Lethal giant larvae) has an essential but context-dependent role in regulating the directed cell migration and establishment of AB polarity in epithelial cells of other systems (Nelson, 2009; Suzuki and Ohno, 2006), we investigated the expression and localization of the Scribble (Scrib) and Lethal giant larvae homolog 1 (Lgl1) proteins. At compacted 8–12-cell stage, Scrib was dispersed from the basolateral membrane in the Pk2−/− embryos (n = 3/4), while it remained at regions of cell-cell contact in controls (Figs. 4L and M; Fig. S8D). In contrast, Lgl1 retained its normal basolateral distribution in most Pk2−/− blastomeres (n = 3/4) (Figs. 4P and Q; Fig. S8E). Interestingly, Scrib and Lgl1 were co-localized with Na+/K+ ATPase α1 subunit whose function is essential for blastocyst cavity formation, along the basolateral membranes at the compacted 8-cell stage (n > 15 embryos, respectively) (Figs. 4L, N, P and R; Figs. S8D and E). Similar to Scrib, Na+/K+ ATPase α1 was also disrupted at the basolateral membrane in Pk2−/− embryos (n = 8/8) (Figs. 4O and S; Figs. S8D and E). Together, these results suggested that Pk2 is required to establish the AB polarity of outer cells during compaction, which is prerequisite for blastocyst cavity formation.
We next examined which functional domains of Pk2 are required for preimplantation development. The prenylation motif at the C-terminus of Pk is known to be necessary for its function (Mapp et al., 2011; Shimojo and Hersh, 2003, 2006). Prenylation mediated by farnesyltransferase (FT) or geranylgeranyltransferase (GGT) attaches isoprenyl anchors to C-terminal motifs in substrate proteins (Maurer-Stroh et al., 2007). In addition, preimplantation embryos treated with mevastatin, an HMG CoA reductase inhibitor that blocks prenylation, show arrested development around the late morula stage (Surani et al., 1983).
To address how the nuclear localization of Pk2 affects cell polarity, we treated late 4-cell-stage mouse embryos with an FT inhibitor, B581, or GGT inhibitor, GGTI-298, by adding it to the culture medium (Fig. S9A). The FT inhibitor-treated embryos showed disruptions in ZO-1 and PARD3 localization (n = 4/6, n = 12/14, respectively) (Fig. S9B). After 18 hrs, many of the FT inhibitor-treated embryos became arrested and maintained a morula-like morphology, while the DMSO-treated embryos formed a blastocyst cavity (n = 88/96) (Fig. S9C). These data suggested that FT activity is required to maintain epithelial integrity during preimplantation development. Late 4-cell-stage embryos treated with a GGT inhibitor were morphologically indistinguishable from DMSO-treated control embryos (n = 32/42) (Fig. S9C).
Interestingly, the nuclear signal of Pk2 was decreased in the 8-cell embryos treated with 10 μM FT inhibitor B581 for 8 hrs (n = 12/12) (Fig. S9D). To confirm this finding biochemically, we performed western blots of nuclear and cytoplasmic fractions obtained from DMSO- and FT inhibitor-treated embryos at the late 8-cell stage (Fig. S9E). These data suggested that farnesylation is essential for its nuclear localization and thus for preimplantation development. Next, we examined the subcellular localization of myc-tagged Pk2 ΔCIIS, in which the farnesylation motif (CIIS) was deleted, and myc-tagged full-length Pk2 by injecting their respective mRNAs into each blastomere of wild-type 2-cell-stage embryos (Figs. S10A and C). The injection of 5 ng/μl full-length Pk2 mRNA did not affect cell polarity or cause developmental arrest, compared with EGFP-injected embryos (Figs. S10C and D). However, an excess amount (50 ng/μl) of full-length Pk2 mRNA caused developmental retardation or arrest (Fig. S10C), whereas the injection of 10 ng/μl or 50 ng/μl full-length Pk1 mRNA did not affect cell polarity (n = 30/30, n = 25/25) (Fig. S11A). Myc-tagged-Pk2 was localized to the nucleus (Fig. S10D), and myc tagged-Pk2 ΔCIIS was localized to the cytoplasm (n = 12/12) (Fig. S10D). Notably, Prickle2 ΔCIIS RNA caused developmental arrest even at a low dose (5 ng/μl) (Fig. S10C), suggesting that the excess cytoplasmic Pk2 interrupts normal development. These data demonstrated that the CIIS motif of Pk2 is required for its nuclear localization, and together with the result of FT inhibitor-treatment, suggested that loss of the nuclear Pk2 signal prevents cell polarization during compaction.
To further test the hypothesis that farnesylation of Pk2 is critical for cell polarization and for the nuclear localization of Pk2 during preimplantation development, we replaced the 15 C-terminal residues of Pk2 with the sequence for farnesylation identified in Lamin B1 (Pk2ΔFT-Lamin B1) or Kras2b (Pk2ΔFT-Kras2b), each targeting Lamin B1 and KRAS2B to nucleus and cytoplasm/plasma membrane, respectively (Fig. S10A) (Maurer-Stroh et al., 2007). At the 8-cell stage, Lamin B1 was present at the nuclear envelope and KRAS2b was distributed in a punctate pattern throughout the cytoplasm (Fig. S10B). The localizations of endogenous Pk2 and Lamin B1 or KRAS2b were not altered in embryos injected with RNA encoding Pk2ΔFT-LaminB1 or Pk2ΔFT-Kras2b, respectively (n > 10 in each case) (Fig. S10B). To correlate cellular distribution of Pk2 with its function, RNA encoding myc-tagged forms of the above constructs were injected into each blastomere of wild-type 2-cell stage embryos and analyzed by immunofluorescence confocal microscopy with an anti-myc antibody (Fig. S10E). Pk2ΔFT-Lamin B1-injected embryos showed the nuclear localization of myc and disruption of the apical PKCζ localization at 14–18-cell stage (n = 19/21) (Fig. S10E). In Pk2ΔFT-Kras2b-injected embryos, myc staining showing a cytoplasmic localization, and the apical PKCζ localization was disrupted at 14–18-cell stage (n = 17/18) (Fig. S10E). The injection of 5 ng/μl Pk2ΔFT-Lamin B1 or Pk2ΔFT-Kras2b caused developmental arrest, compared with 5 ng/μl full-length Pk2-injected embryos (Fig. S10F). These data suggested that the overexpression of Pk2 in either the nucleus or cytoplasm disrupts cell polarity during compaction and proper development.
We next tested cellular localization-function relationship of Pk2 and asked whether Pk2ΔFT-Lamin B1 or Pk2ΔFT-Kras2b could rescue the cellular phenotypes of Pk2−/− embryos (Fig. 5). The myc-tagged full-length Pk2 RNA-injected embryos and myc-tagged Pk2ΔFT-Lamin B1 RNA-injected embryos showed the nuclear localization of myc at the compacted 8-cell stage in Pk2−/− embryos (n = 4/4, n = 3/3) (Figs. 5A and B). In myc-tagged Pk2ΔFT-Kras2b RNA-injected embryos, myc staining showed a cytoplasmic localization (n = 3/3) (Fig. 5C). In embryos injected with 5 ng/μl full-length Pk2 RNA and Pk2ΔFT-Lamin B1 RNA, PKCζ, PARD6b and Scrib were located in membrane at the compacted 8–12-cell stage in the Pk2−/− embryos (Pk2; n = 4/4, 6/6 and 4/5, Pk2ΔFT-Lamin B1; n = 3/3, 4/5 and 5/6) (Figs. 5D, E, G, H, J-K’), suggesting that the cell polarity defect had been rescued as we expected. In contrast, cytoplasmically retained Pk2 (Pk2ΔFT-Kras2b) failed to rescue (PKCζ, PARD6b and Scrib; n = 8/8, 6/7 and 6/8) (Figs. 5F, I, L and L’). In addition, the injection of 10 ng/μl or 50 ng/μl full-length Pk1 cDNA into the pronuclei never restored apical localization of PKCζ and formed a blastocyst cavity in Pk2−/− embryos (Fig. S11B and C), suggesting again that Pk2, but not Pk1, can affect the polarization of blastomeres. Importantly, the MT network of the apical part of Pk2−/− blastomeres rescued with the full-length Pk2 or nuclear-targeted Pk2 was indistinguishable from that of wild-type blastomeres (Pk2; n = 5/5, Pk2ΔFT-Lamin B1; n = 5/6) (Figs. 5M and N), although cytoplasmically retained Pk2 failed to restore the disruption of α-tubulin staining (Pk2ΔFT-Kras2b; n = 8/8) (Fig. 5O). Together, these results suggest that nuclear translocation of Pk2 is prerequisite to its function in establishing the cell polarity of early blastomeres and that the loss of epithelial polarity was accompanied by a delocalization of the MT network.
Here we demonstrated new roles of Pk2 in controlling the redistribution of the MT network during compaction and of functional proteins that are normally asymmetrically localized, both of which depend on the establishment of AB polarity. Misregulation of these events by Pk2 gene knockout leads to aberrant positional information and thus to inappropriate cell fate decisions. These results provide the first evidence that Pk2 plays a crucial role in mouse preimplantation development.
Intriguingly, these phenotypes are only seen predominantly CBA background on which almost all of Pk2−/− embryos die before implantation due to the loss of zygotic Pk2. However, when the chimeras were backcrossed with C57BL/6 mice over a few generations, Pk2−/− mutant mice started to survive. Although less commonly observed as a general situation, it is possible that modifier gene(s) are involved in controlling embryo survival and thus in increasing viability. These results also rule out the possibility that the phenotypic diversity of Pk2 is due to a difference between the C57BL/6 and CBA strains in either the Pk2 gene itself or a closely linked gene. We would reason that interactions of Pk with various interacting proteins could be differently affected in C57BL/6 and CBA strains. The presence as well as the number of modifier genes involved in this variation of preimplantation phenotype will be unveiled by a precise genetic mapping.
Models for cell fate determination in the early mouse embryo include the “Inside-Outside” model, in which topological differences dictate cell fates, and the polarity model, in which a differential inheritance of information along the AB axis dictates both cell position and fate (Marikawa and Alarcón, 2009; Sasaki, 2010; Yamanaka et al., 2006). Consistent with the polarity model, the presence of polarity promotes adoption of the TE fate (Alarcón, 2010; Jedrusik et al., 2008; Nishioka et al., 2008, 2009; Plusa et al., 2005). In the present study, Pk2−/− embryos showed aberrant AB polarity during compaction, and then, at the late morula stage, nuclear signals of YAP1 and/or Cdx2 were almost completely lost. Interestingly, nuclear localization of YAP1 seems to be disappeared altogether rather than relocating to the cytoplasm in outer cells of Pk2−/− embryos at 26–28-cell stage. These results led us to speculate how Pk2 regulates YAP1 transcription. Previous reports (Alarcón, 2010; Nishioka, 2008, 2009) together with our results suggest that Pk2 regulates a Tead4-Cdx2-dependent pathway for TE development, supporting the polarity model. However, a significant difference between Pk2−/− and Tead4−/− embryos is that the former impairs PKCζ localization (our study) and latter does not (Nishioka et al., 2008). The difference may be the result of additional roles of Pk2 in polarization of TE. Therefore, our result also suggests the possibility that Pk2 regulates this process either upstream or independently from Tead4.
Obviously, it is important to clarify how the cell polarization and transcriptional factors-based mechanisms are connected. Thus, we can assume two possibilities about this concern. First, Tead4 activity is regulated by cell-cell contact through a Hippo signaling pathway components (Nishioka et al., 2009), and the Crumbs polarity complex interacts with YAP (Varelas et al., 2010). These reports presented a model in which cell-cell contact controls transcriptional factor-mediated TE determination. In our study, the EMK1, Na+/K+ ATPase α1, and Scrib localizations on the basolateral membrane were disrupted in Pk2−/− embryos before the first cell fate decision, suggesting a possible functional relationship between these polarity components and Pk2. Second, it has been reported a model in which Cdx2 is involved in cell polarization process (Jedrusik et al., 2008, 2010), and there is a positive feedback loop between Cdx2 and cell polarization (Jedrusik et al., 2008, 2010; Ralston and Rossant, 2008). It is noteworthy that Cdx2 mRNA and cell polarity molecules (e.g. aPKC) become localized apically at 8- and 16-cell stage (Plusa et al., 2005; Ralston and Rossant, 2008; Vinot et al., 2005). Importantly, Cdx2 depletion by injection of Cdx2 RNAi at 2-cell stage caused strongly down-regulation of PKCζ expression at the 8- and 16-cell stage (Jedrusik et al., 2010), reminiscent of Pk2−/− phenotype. Moreover, in Pk2−/− embryos, apical localization of PKCζ does not rescue by injection of cytoplasmic Pk2. These results reveal that Pk2 may participate as the molecular machinery, which is relationship of Cdx2 and aPKC localization. Interestingly, in Pk2−/− embryos, nuclear Cdx2 is heterogeneity at 14–16-cell stage, and a similar distribution was also observed in the wild-type embryos. Taken together, these observations suggested that although loss of Pk2 impaired apical-basal polarity during compaction, which indirectly led to the failure of up-regulation in Cdx2 expression during first cell fate decision process.
Next, in this study, we mainly focused on cellular distribution at 8–16-cell stage, which is compaction is taking place. Compaction is mediated by at least two pathways, cell-cell contact mediated by E-cadherin-dependent adherence, and MT-mediated interactions between the nuclei and the cell surface, which result in the formation of the polarized (flattened) epithelium (Fleming and Johnson, 1988; Johnson et al., 1986; Maro et al., 1990). Here we showed that cell-cell adhesion, indicated by localization of E-cadherin and β-catenin, was not hardly affected in Pk2−/− embryos. Moreover, Pk2−/− embryos did not show obvious differences in Lgl1 localization. These results are supported by a previous report that Lgl is dispensable for the basolateral and tight junctional localization of β-catenin and β1-integrin (Dollar et al., 2005). Thus, the loss of Pk2 influenced neither the Lgl1 localization nor the cell adhesion via E-cadherin and/or β-catenin during cell polarization. On the other hand, our observation that compaction occurs with disorganized/misoriented MTs and PCM suggested that rather than being the driving force for polarization of the 8-cell blastomere, MTs might coordinate the compaction process and reinforce the asymmetry already established at the cell periphery (Our study, (Houliston and Maro, 1989)). Nevertheless, our data show that the loss of Pk2 may influence the stabilization of centrosomes and MTs, with effects on epithelial cell organization during early mouse development. In contrast, a recent report showed that the MT cytoskeleton is needed for the establishment, but not the maintenance, of the anterior Pk localization of the mesodermal cells during zebrafish gastrulation (Sepich et al., 2011). This report suggested that MT dynamics is tightly linked to Pk function. In any case, we speculate that Pk2 may be an essential component of vesicular membrane transport and membrane polarization through MT-based motility.
Rho GTPases play a role in establishing polarity in mouse blastomeres during compaction (Clayton et al., 1999). In the Pk2−/− embryos, the total and active RhoA were reduced around the time of compaction, and the phenotypes were similar to those of RhoA-injected embryos, i.e., disrupted MTs and relatively normal E-cadherin (Clayton et al., 1999). Such similarities of gain- and loss-of-function phenotypes are often observed for both RhoA and PCP components (Takeuchi et al., 2003; Veeman et al., 2003), and they suggest that the tuning of these activities to the proper levels is crucial. The actual role of PCP signaling is somewhat unclear, but these results reveal that a functional correlation between Pk2 and RhoA signaling for the establishment of cell polarity.
In this study, we showed that Pk2’s C-terminal prenylation by farnesyltransferase is required for nuclear localization of Pk2. Consistent with this, HMG-CoA reductase, which acts upstream of FT and GGT and whose gene is activated from morula to the blastocyst stage (Hamatani et al., 2004) was reported to be essential for mouse preimplantation development (Surani et al., 1983). These observations strongly suggest that in early mouse development, the Pk2 protein is farnesylated and its activity is tightly regulated. Beside early embryogenesis, farnesylation and nuclear localization of Pk1b which is more similar to mouse Pk1 in zebrafish have recently reported to be essential for facial branchiomotor neuron (FBMN) migration (Mapp et al., 2011), suggesting that the post-translational modification of Pk is an evolutionarily conserved and broadly occurring event.
We also showed that a nucleus- but not a cytoplasm-targeted Pk2 mutant could rescue cell polarization, demonstrating that its protein product was functional in nucleus. This may suggest that cell polarity establishment requires the physical exclusion of Pk2 from the cytoplasm, at least in the early cleavage stages. Although these showed that nuclear localization of Pk2 is necessary and sufficient for the establishment of AB cell polarity, how the polarity information is transmitted from the nucleus to the membrane remains to be investigated. One possibility is that Pk may function in transcriptional regulation. Pk1, which is also known as RILP (REST (RE-1 silencing transcription factor) interacting LIM domain protein), is associated with the transcriptional repressor REST in other systems (Bassuk et al., 2008; Mapp, et al., 2011; Shimojo and Hersh, 2003, 2006). REST also appears to promote pluripotency by repressing multiple components of the Wnt pathway in mouse embryonic stem cells (Johnson et al., 2008). To understand Pk2’s contribution to transcriptional control, global gene expression analyses should be performed.
In conclusion, our results reveal a new role for the polarity molecule Pk2 in the establishment of AB cell polarity and cell fate decision, particularly with respect to proper formation of the blastocyst. We have also been able to demonstrate that nuclear localization of Pk2 is the key process, which is somewhat unexpected from cytoplasmic function of Pks which had been broadly recognized. As the Hippo signaling pathway has recently emerged as a novel pathway regulating cell polarity formation and the first cell fate decision during preimplantation development, it is intriguing to investigate crosstalk between this pathway and the PCP signaling pathway to understand how the epithelial architecture is achieved.
We thank Dr. V.B. Alarcon (University of Hawaii) and Dr. Y. Marikawa (University of Hawaii) for critical discussion. We are grateful to the Laboratory for Animal Resources and Genetic Engineering staff for collecting mouse embryos, housing the mice, and excellent technical assistance, and members of division of Morphogenesis in NIBB and the Laboratory for Vertebrate Body Plan in RIKEN CDB for valuable discussion. This work was supported by a Grant-in-Aid for Scientific Research, MEXT, Japan to S.A. (18GS0320) and N.U. (22116511, 22127007).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.