p45NF-E2 represses Gcm1 in trophoblast cells to regulate syncytium formation, placental vascularization and embryonic growth

Intrauterine growth restriction (IUGR) is a frequent but incompletely understood pregnancy complication that occurs in ~8% of pregnancies (McIntire et al., 1999). Knockout studies in mice have identified a number of genes required for normal placentation (Cross, 2005), including genes that regulate the activity of the coagulation system (Brenner and Kupferminc, 2003; Sood and Weiler, 2003). The developmental function of the coagulation system is at least partially independent of its well-known hemostatic function in the regulation of the fibrin-platelet interaction, as embryos simultaneously lacking fibrinogen and carrying a severe platelet defect develop normally (Camerer et al., 2004; Palumbo et al., 2004). The notable exception is an IUGR and impaired placental vascularization observed in p45NF-E2 (Nfe2 – Mouse Genome Informatics) null mice lacking platelets (Palumbo et al., 2004). A similar developmental defect is not detectable in embryos with normal platelet numbers but disturbed platelet function [Par4 or Gαq (F2rl3 or Gnaq – Mouse Genome Informatics) null mice] (Camerer et al., 2004; Palumbo et al., 2004). A developmental role for platelets is not necessarily precluded by the normal development of Par4^−/− or Gαq^−/− embryos, as platelets lacking Par4 or Gαq can still be activated (Ohlmann et al., 2000; Sambrano et al., 2001). This raises the question of whether embryonic platelets have an as yet undefined developmental function or whether p45NF-E2 has a developmental function that is independent of platelets.

NF-E2 belongs to the basic region leucine zipper (bZip) superfamily of transcription factors. It consists of two subunits: a 45 kDa (p45) subunit thought to be restricted to megakaryocytes, erythrocytes and mast cells; and a widely expressed 18 kDa (p18) subunit. Mice lacking p45NF-E2 display severe thrombocytopenia caused by a megakaryocyte maturation defect (Shivdasani et al., 1995). Megakaryocytes can be detected in the embryonic compartment as early as 7.5 days post-coitum (dpc) (Xu et al., 2001), and platelets may alter placental development through various mechanisms (Ohlsson et al., 1999; Sato et al., 2005). Hence, it is generally assumed, but remains to be shown, that the IUGR of p45NF-E2-deficient embryos results from the absence of embryonic platelets.

However, a role of the bZip transcription factor p45NF-E2 in placental development independent of platelets or megakaryocytes cannot currently be ruled out. Other bZip transcription factors, such as JunB, Fra1 (Fosl1), Tfeb (Tcfeb), Creb and Oasis (Creb3l1), regulate trophoblast differentiation or placental development (Schorpp-Kistner et al., 1999; Schreiber et al., 2000a; Schreiber et al., 2000b; Schubert et al., 2008; Steingrimsson et al., 1998). Creb and Oasis regulate the expression of Gcm1, a key regulator of placental development that is required for chorioallantoic branching and syncytiotrophoblast formation (Anson-Cartwright et al., 2000). The stability and binding activity of Gcm1 are furthermore post-translationally regulated by acetylation. Hence, p45NF-E2 might regulate Gcm1 expression by interacting with other bZip transcription factors or by modulating acetylation.

In the current study we identify a novel function of p45NF-E2 in non-hematopoietic cells. In trophoblast cells, p45NF-E2 acts as a transcriptional repressor, reducing Gcm1-dependent syncytiotrophoblast differentiation by inhibiting acetylation.

Cell culture media, fetal bovine serum, glutamine, penicillin/streptomycin mix, neuromycin, non-essential amino acids, sodium pyruvate, PCR4-TOPO cloning vector and pcDNA3.1/V5-His-TOPO vector were obtained from Invitrogen (Karlsruhe, Germany). ESGRO (LIF), thrombopoietin (Thpo), M2 medium and KSOM medium were purchased from R&D Systems (Wiesbaden-Nordstadt, Germany). P24 tag antigen ELISA and ChIP assay kits were from Biocat (Heidelberg, Germany). ECL reagent was from Amersham Biosciences (Little Chalfont, UK). Murine p45NF-E2 and Gcm1 shRNA constructs (pLKO1-NF-E2, pLKO1-Gcm1) and the murine Gcm1 full-length cDNA construct (pCMV-SPORT6-Gcm1) were from Open Biosystems (Huntsville, AL, USA). GoTaq DNA polymerase and the AP-1 consensus oligo sequence for electrophoretic mobility shift assay (EMSA) were from Promega (Mannheim, Germany). P45NF-E2 consensus oligo sequence for EMSA and rabbit polyclonal IgGs (NF-E2, CBP) and goat polyclonal IgG (Gcm1) were from Santa Cruz Biotechnology (Heidelberg, Germany). Rabbit polyclonal anti-acetylated lysine IgG antibody was from Cell Signaling (Boston, MA, USA). All primers were synthesized by Thermo Scientific (Langenselbold, Germany). All other materials were obtained from Sigma-Aldrich (Taufkirchen, Germany).

p45NF-E2^−/− mice were kindly provided by R. Shivdasani (Harvard Medical School, Cambridge, MA, USA). CD-1 mice were from Charles River Laboratories (Sulzfeld, Germany). Aggregation of wild-type ES cells with tetraploid p45NF-E2 embryos was achieved essentially as described previously (Isermann et al., 2001; Nagy et al., 1993). In brief, p45NF-E2-deficient mice were backcrossed for one generation onto the CD-1 background and then heterozygous p45NF-E2 offspring (F1) were bred to generate p45NF-E2^−/− embryos with a mixed C57Bl/6–CD-1 genetic background. These embryos were used to generate tetraploid embryos as previously described (Isermann et al., 2001). Tetraploid embryos were aggregated with clusters of 10-15 R1 wild-type ES cells (Isermann et al., 2001).

All animals were housed and the experiments were performed at the interdisciplinary biomedical research institution (IBF) at the University of Heidelberg, Germany, following standards and procedures approved by the local Animal Care and Use Committee (Regierungspräsidium Karlsruhe, Germany).

Placental and embryonic tissues were obtained from timed matings of p45NF-E2^+/− females and p45NF-E2^−/− males on a C57Bl/6 background. Embryonic tissue was used for genotyping. Half of the placental tissue was fixed in 4% buffered formalin while the other half was further divided and stored either in RNAlater (Ambion) or snap frozen in liquid nitrogen.

For analyses of gross placental morphology, placental sections were stained with Bandeiraea simplicifolia BS-I B4 (40 ng/μl) as previously described (Hemberger et al., 1999). Nuclear counterstaining was carried out with Hematoxylin and the spongiotrophoblast and labyrinthine layer area was measured by a blinded investigator using UTHSCSA ImageTool software on three placental sections per placenta. Hematoxylin and Eosin-stained placental sections were analyzed for vascular space quantification. In each section, at least ten randomly selected microscopic fields in the labyrinthine region from three nonconsecutive placental sections (magnification 20×) were acquired at 1280×960 pixel resolution using a digital camera (Nikon Digital Sight DS-U1) connected to a light microscope (Nikon Eclipse TE 2000). We implemented an image analysis routine using NIS Elements AR2.30 imaging software. Briefly, after acquisition, the images underwent an automated light analysis procedure and were given a color threshold to cover the area corresponding to the blood spaces lumen. The coverage percentage was calculated as the ratio between the number of pixels covered by the area defined by the threshold and the overall number of pixels in the image. Analyses were performed by a blinded investigator. Analysis of placental perfusion was performed essentially as previously described (Sood et al., 2007). Briefly, for analyses of embryonic perfusion, pregnant females were injected with 100 μl 25 mg/ml FITC-labeled dextran (molecular weight 2×10⁶ Da) via the tail vein at 14.5 dpc. For analysis of embryonic perfusion, embryos with an intact yolk sac vasculature and placenta were isolated at 17.5 dpc and ~20 μl of FITC-labeled dextran was injected intracardially. The placenta was isolated after 15 minutes, embedded in tissue-freezing medium (Tissue-Tek) and flash frozen in a dry ice-ethanol bath. Micrographs of serial tissue sections were obtained using the same microscope, camera and software as above. The perfused area was quantified following a routine automated light analysis procedure with a color threshold to cover the FITC-positive, perfused area, similar to the procedure described above for the vascular space analyses.

Regular RT-PCR was performed using the following primers (5′ to 3′): actin, CTAGACTTCGAGCAGGAGATGG and GCTAGGAGCCAGAGCAGTAATC; MafF, GTGCCTGAGTTCCTTCTGCT and CACAGTTGGGAGGTCTGCTT; MafG, GAGGAGCTGGAGAAGC AGAA and GAGCATCCGTCTTGGACTTT; and MafK, GCACACTCAAGAACAGAGGC and CTGAGAAGGGTACAGAGGTGG.

Quantitative RT-PCR (qRT-PCR) was performed as previously described (Rudofsky et al., 2007). Initially, RT-PCR was performed (57°C for 1 minute, 72°C for 1 minute, 30 cycles) using the following primers (5′ to 3′): actin, CTAGACTTCGAGCAGGAGATGG and GCTAGGAGCCAGAGCAGTAATC; Cebpa, CGCTGGTGATCA AACAAGAG and GTCACTGGTCAACTCCAGCA; Esx1, GTACCCGGACCTGTTTGCTA and CTTGCTGCATTGGGTATGTG; Gcm1 coding sequence, GACTGCTCCACAGAGGAAGG and GGAGAGCCATAGGTGAGCAG; p45NF-E2, CACCAAATACTC CCAGGTGATATG and AACTTGCCGGTAGATGACTTTAAT; PL-II, CTCCTTCTCTGGGCAACTC and GGACTTGCTCGCTGTTTTC; Synb, TGACCTTGAAGTGGGTAGGG and TGACCTTGAAGTGGGTAGGG; and 4311, CAGGTACTTGAGACATGACTC and AGAGATTTCTTAGACAATG. PCR fragments were cloned using the TOPO TA Cloning Kit (Invitrogen) and confirmed by sequencing. Generation of standard curves and evaluation of final data were performed as described (Rudofsky et al., 2007). All RT-PCRs were performed in triplicate.

AP-1 or p45NF-E2 binding activity was determined by EMSA as described previously (Bierhaus et al., 2001). Immunoblotting was performed as described (Isermann et al., 2007).

ChIP was performed on placental tissue extracts using the EpiQuik ChIP Assay Kit for tissue following the manufacturer's instructions (Epigentek, Brooklyn, NY, USA). Gcm1 (third intron) primers were 5′-ATTCGTGAATAAGTAGGGAGGA-3′ and 5′-GCTGATTCTGAGTCTGAGGTGTC-3′. Co-immunoprecipitation was performed as described (Qiu et al., 2004).

Placental tissues obtained at 16.5 dpc were fixed with 2.5% glutaraldehyde, 2.5% polyvidone 25, 0.1 M sodium cacodylate pH 7.4. After washing with 0.1 M sodium cacodylate buffer (pH 7.4), samples were post-fixed in the same buffer containing 2% osmium tetroxide and 1.5% potassium ferrocyanide for 1 hour, washed in water, contrasted en bloc with uranyl acetate, dehydrated using an ascending series of ethanol and embedded in glycidyl ether 100-based resin. Ultrathin sections were cut with a Reichert Ultracut S ultramicrotome (Leica Microsystems, Wetzlar, Germany), contrasted with uranyl acetate and lead citrate, and were viewed with an EM 10 CR electron microscope (Carl Zeiss NTS, Oberkochen, Germany).

Mouse trophoblast stem (TS) cells were obtained from J. Rossant (Hospital for Sick Children, Toronto, ON, Canada) and were maintained as stem cells or induced to differentiate as previously described (Isermann et al., 2003; Tanaka et al., 1998). Syncytiotrophoblast formation was monitored in nearly confluent TS cells differentiated for 6 days. In some cases, TS cells were differentiated in the presence of HDAC inhibitor [100 nM trichostatin A (TSA) or 2.5 mM sodium butyrate] (Maltepe et al., 2005) or HAT inhibitor (30 nM curcumin or 5 μM EGCG) (Balasubramanyam et al., 2004; Tachibana et al., 2004).

For quantification of syncytiotrophoblast formation, cells were incubated with 2.5 μm CellMask Deep Red (Invitrogen) for 5 minutes at 37°C and then fixed for 3 minutes in 3.75% formaldehyde at room temperature. Cells were counterstained with DAPI (0.5 μg/ml in PBS) for 10 minutes at room temperature in the dark. For quantification of syncytiotrophoblasts, at least 30 random images per slide were captured (UTHSCSA ImageTool software) and the frequency of syncytial cells observed in these random images was determined by a blinded investigator.

Murine p45NF-E2 (pLKO1-NF-E2) and Gcm1 (pLKO1-Gcm1) knockdown constructs containing shRNA for p45NF-E2 or Gcm1 cloned into the lentiviral plasmid vector pLKO1 were obtained from Open Biosystems. Gcm1 expression plasmid pLV-Gcm1 was generated by replacing the EGFP fragment (EcoRI/XhoI) of the plasmid pLV-EGFP [kindly provided by Masahito Ikawa, Osaka, Japan (Okada et al., 2007)] with full-length Gcm1 cDNA (EcoRI/XhoI fragment; Open Biosystems). The p45NF-E2-expressing plasmid pLV-NF-E2 was generated by replacing EGFP in the plasmid pLV-EGFP with PCR-derived murine p45NF-E2 cDNA. In detail, a p45NF-E2-coding PCR fragment was generated from wild-type 14.5 dpc placental cDNA using the forward primer 5′-TCAGCTGGCACAGTAGGATG-3′ and the reverse primer 5′-CATCAGAAGGGAATGAGGGA-3′ (57°C annealing temperature, 35 cycles). The fragment was cloned into pcDNA3.1/V5-His-TOPO, yielding the plasmid pTOPO-NF-E2. Subsequently, the EGFP coding sequence of pLV-EGFP was replaced by the BamHI-XhoI fragment of pTOPO-NF-E2, yielding pLV-NF-E2, as confirmed by sequencing.

Gcm1 mutants lacking Lys387 and Lys426 were generated by site-directed mutagenesis, replacing lysine with arginine residues. Using pLV-Gcm1 as a template, a two-step PCR was conducted with the following primers (5′ to 3′): Gcm1 387K2Rf, ATGCACTGCCCGGCAGGAGCAACAGGTGG; Gcm1 387K2Rr, CCACCTGTTGCTCCTGCCGGGCAGTGCAT;Gcm1426K2Rf,GAGCCTTACGAAGAGAAGGTATCTGTGGATCTGAGC; and Gcm1 426K2Rr, GCTCAGATCCACAGATACCTTCTCTTCGTAAGGCAC. pLV-Gcm1^ΔKK was confirmed by sequencing.

VSV-G pseudotyped lentiviral particles were generated as described (Follenzi et al., 2000). Briefly, HEK 293T cells were transfected with pLV-NF-E2, pLV-Gcm1, pLV-Gcm1^ΔKK, pLKO1-NF-E2 or pLKO1-Gcm1 together with the packaging plasmid (pCMV-dR8.91) and VSV-G-expressing plasmid (pVSV-G) using the standard calcium phosphate method. Lentiviral particles were harvested 2 and 3 days after transfection and concentrated by ultracentrifugation (50,000 g, 2 hours, twice). After resuspension in Hank's Buffered Salt Solution, the viral particle concentration was determined by measuring p24 tag antigen by ELISA. Viral particles were stored at −80°C.

Two-cell stage embryos were collected at 1.5 dpc and incubated for 2 days to obtain blastocysts. The zona pellucida was removed with acidic Tyrode's solution. Zona pellucida-free embryos were incubated individually in 50 μl KSOM medium containing lentiviral vector (1 μg p24/ml) for 4 hours (Okada et al., 2007). Transduced blastocysts were transferred into pseudopregnant females.

Knockdown of p45NF-E2 or of Gcm1 and knock-in of pLV-Gcm1 in mouse TS cells was performed by lentiviral transduction. pLKO1-NF-E2, pLKO1-Gcm1 or pLV-Gcm1 viral particles were generated as described above. Stable knockdown (p45NF-E2^kd) TS cell lines were generated by infecting TS cells with viral particles [5×10⁵ transducing units (TU)/ml] in the presence of 8 μg/ml polybrene and 1 μg/ml puromycin. Efficiency of knockdown was assayed by RT-PCR and EMSA in undifferentiated and differentiated TS cells. To generate p45NF-E2/Gcm1 double-knockdown TS cells, p45NF-E2^kd cells were transiently infected with pLKO1-Gcm1 shRNA lentiviral particles, yielding p45NF-E2^kd Gcm1^kd cells. For ectopic expression of Gcm1, TS cells were transiently infected with pLV-Gcm1 lentiviral particles. The efficiency of the Gcm1 knockdown or knock-in was determined after 3 days of differentiation by RT-PCR. In all experiments, cells transduced with non-specific shRNA or pLV-EGFP served as controls.

To inhibit HDAC or HAT activity in vivo, pregnant females (p45NF-E2^+/+ for HDAC inhibition, p45NF-E2^−/− for HAT inhibition) were injected intraperitoneally with the HDAC inhibitor TSA [15 μg/mouse (Nervi et al., 2001)] or valproic acid [sodium salt, 155 mM, 25 ml/kg body weight (Gottlicher et al., 2001)] or the HAT inhibitor EGCG [50 mg/kg in saline (Giunta et al., 2006)] or curcumin [0.5 mg/kg (Pan et al., 2008)] in 150 μl vehicle [10% (v/v) DMSO in PBS/kg] at 7.5 dpc. The injections were repeated every 48 hours until analysis.

In vitro experiments were performed at least in triplicate. Data are expressed as mean ± s.e.m. Statistical analyses were performed using Student's t-test or ANOVA. Post-hoc comparisons of ANOVA were corrected using the method of Tukey. Excel (Microsoft, Munich, Germany) and StatistiXL software were used for statistical analyses. Statistical significance was set at P<0.05.

We performed timed matings (p45NF-E2^+/− females × p45NF-E2^−/− males) to determine the onset of embryonic growth retardation and impaired placental vascularization in p45NF-E2^−/− embryos (Palumbo et al., 2004; Shivdasani et al., 1995). The growth retardation of p45NF-E2^−/− embryos first became evident at 16.5 dpc (15.9 mm embryonic length versus 16.6 mm in p45NF-E2^+/−, P<0.05) and was readily detectable by 17.5 dpc (16.5 mm versus 19.8 mm, P<0.001) (Fig. 1A and see Table S1 in the supplementary material). Impaired vascularization of the placental labyrinthine layer could already be detected at 14.5 dpc (75% vascularization in p45NF-E2^−/− embryos as compared with 100% in p45NF-E2^+/− embryos; P<0.001; Fig. 1B). The reduction of the vascularized area, as detected by digital image analyses, was associated with reduced placental perfusion (see Fig. S1 in the supplementary material). The manifestation of the IUGR subsequent to the reduced placental vascularization indicates that the IUGR is caused by the placental defect.

We next determined whether the placental defect is the consequence of p45NF-E2 deficiency within the embryo proper or within cells of the trophectodermal cell lineage. To achieve p45NF-E2 expression only in the embryo proper and not in the placenta we performed tetraploid aggregation (TA) experiments to generate chimeric embryos derived from p45NF-E2 wild-type (+/+) embryonic stem (ES) cells and p45NF-E2^−/− tetraploid embryos. Consistent with p45NF-E2 expression in the embryo proper, p45NF-E2 mRNA and platelets were readily detectable within the embryo proper (see Fig. S2A in the supplementary material; data not shown). However, vascularization of the labyrinthine layer remained significantly lower than that of control chimeric embryos obtained by fusion of p45NF-E2^+/+ ES cells with p45NF-E2^+/+ tetraploid embryos (76% versus 98%, P=0.006; Fig. 1C,D).

We next restored p45NF-E2 expression specifically within the cells of the trophectodermal lineage by transducing p45NF-E2^−/− blastocysts with a p45NF-E2-expressing lentivirus (Okada et al., 2007). Following viral transduction (VT) of p45NF-E2^−/− blastocysts, expression of p45NF-E2 was readily detectable within the placenta, but not within the embryo proper (see Fig. S2B in the supplementary material). Expression of p45NF-E2 within the placenta rescued the vascularization of the labyrinthine layer (99% versus 75%, P=0.004; Fig. 1C,D) and resulted in the normal growth of embryos (weight 98% and length 99% of that of control embryos; Fig. 1E,F). Therefore, expression of p45NF-E2 within trophoblast cells is required for normal placental vascularization and embryonic growth.

Expression of p45NF-E2 was readily detectable in the placental tissue of p45NF-E2^+/− embryos (Fig. 2A,B). Placental expression of the smaller p18 units of NF-E2 (MafF, MafG, MafK) was independent of embryonic p45NF-E2 expression (Fig. 2A).

The binding activity of p45NF-E2 in 14.5 dpc placental extracts from p45NF-E2^+/− embryos was readily detectable by EMSA (Fig. 2C, lanes 1 and 2). Placental extracts were co-incubated with radiolabeled probes for activator protein 1 (AP-1) and NF-E2, which share a similar DNA-binding sequence. A slower migrating band, corresponding to the larger AP-1 complex, was clearly observed, demonstrating the specificity of the observed p45NF-E2 signal (Fig. 2C, lanes 3 and 4).

In vitro, p45NF-E2 was expressed in murine trophoblast stem (TS) cells but not in endothelial cells (ECs; SVEC4-10 cells) (Fig. 2D). In differentiating trophoblast cells, expression of p45NF-E2 was transiently induced at day 3 (183% versus 100% at day 0, P=0.002; Fig. 2D,E), at which time p45NF-E2 expression was similar to that in megakaryocytes (Fig. 2D). Thus, p45NF-E2 is expressed in a differentiation-dependent manner in trophoblast cells.

The differential expression of p45NF-E2 during trophoblast differentiation raises the question of whether p45NF-E2 itself modulates trophoblast differentiation. Morphometric analyses revealed a small but significant reduction of the spongiotrophoblast layer (88% versus 100% in p45NF-E2^+/− placentae, P=0.0028; Fig. 3A,B) and in the expression of the spongiotrophoblast marker 4311 (Tpbpa – Mouse Genome Informatics) (86%, P=0.007; Fig. 3C) in p45NF-E2^−/− placentae. The area of the labyrinthine layer did not differ between p45NF-E2^−/− and p45NF-E2^+/− embryos (101% versus 100%, P=0.84; Fig. 3A,B). Interestingly, expression of the labyrinthine layer marker Esx1 was slightly reduced (76%, P=0.004; Fig. 3C), whereas expression of the syncytiotrophoblast marker Gcm1 was markedly increased (193%, P=0.001; Fig. 3C,D). Expression of the giant trophoblast marker placenta lactogen II (Pl-ll; Prl3b1 – Mouse Genome Informatics) did not differ between p45NF-E2^−/− and p45NF-E2^+/− embryos (Fig. 3C). Expression of p45NF-E2 colocalized with that of Gcm1 but not with the EC marker Pecam1 in the labyrinthine layer of the placenta (Fig. 3E and see Fig. S3 in the supplementary material).

To determine the consequences of p45NF-E2 deficiency for syncytiotrophoblast morphology in vivo, we performed ultrastructural analyses. Evaluation of transmission electron microscopy images showed an increased width of the total maternal-embryonic barrier in p45NF-E2^−/− embryos as compared with p45NF-E2^+/− littermates at 16.5 dpc (121% versus 100% in p45NF-E2^+/−, P=0.004; Fig. 4A,B). Detailed morphological analyses revealed thickening of syncytiotrophoblast layer 2 (SynT-II) in p45NF-E2^−/− embryos (136%, P<0.001; Fig. 4A,B). No obvious alteration of ECs or mononuclear sinusoidal trophoblast giant cells was detected. Consistent with the increased width of SynT-II, the expression of Gcm1 and of the Gcm1-dependent genes Cebpa and Synb (D930020E02Rik – Mouse Genome Informatics) (Fig. 4C), which are markers of SynT-II (Simmons et al., 2008a), was significantly increased. Expression of Syna (Gm52 – Mouse Genome Informatics), which is not expressed in SynT-II, or of Ctsq, a marker of sinusoidal giant trophoblast cells, was not altered (Fig. 4C). Lentiviral-mediated reconstitution of p45NF-E2 expression in trophectodermal cells of p45NF-E2^−/− embryos (p45NF-E2^−/−,+) rescued the width of the syncytiotrophoblast layers and the expression of Gcm1, Cepba and Synb (Fig. 4A-C). The absence of the transcription factor p45NF-E2 from trophoblast cells therefore enhances the width of SynT-II.

To determine whether p45NF-E2 modulates syncytium formation cell-autonomously, we performed in vitro experiments employing p45NF-E2 knockdown (p45NF-E2^kd) trophoblast cells. The expression and binding activity of p45NF-E2 mRNA were significantly reduced in p45NF-E2^kd cells (see Fig. S4 in the supplementary material). Differentiation of p45NF-E2^kd cells for 6 days resulted in enhanced syncytiotrophoblast formation [average number of trophoblast syncytia per random image: 2.7 versus 0.5 in control (p45NF-E2^c) cells, P=0.005; Fig. 5A,B]. Consistent with the enhanced syncytium formation, the expression of the syncytiotrophoblast markers Gcm1 (193%, P=0.002), Cebpa (163%, P=0.008) and Synb (176%, P=0.004) (Fig. 5C,D) was markedly increased.

We next determined whether p45NF-E2-dependent regulation of Gcm1 expression is mechanistically linked with altered syncytiotrophoblast formation. Lentiviral-mediated overexpression of Gcm1 in control TS cells increased syncytiotrophoblast formation (2.2 in p45NF-E2^c Gcm1⁺ cells versus 0.5 in p45NF-E2^c cells, P=0.007; Fig. 5A,B) and expression of the Gcm1-dependent genes Cebpa (175%, P=0.004) and Synb (184%, P=0.004) (Fig. 5C,D) after 6 days of differentiation. Thus, increased expression of Gcm1 is sufficient to enhance syncytiotrophoblast formation in vitro.

To determine whether increased Gcm1 expression is required for the enhanced syncytiotrophoblast formation in p45NF-E2-deficient trophoblast cells, we reduced the expression of Gcm1 in p45NF-E2^kd TS cells. This rescued syncytiotrophoblast formation (0.4 in p45NF-E2^kd Gcm1^kd cells versus 2.7 in p45NF-E2^kd control cells, P<0.001; Fig. 5A,B) and reduced the expression of Cebpa (27%, P<0.001) and Synb (25%, P<0.001) (Fig. 5C,D) after 6 days of differentiation. These results establish that p45NF-E2 cell-autonomously represses Gcm1, preventing excess syncytiotrophoblast formation.

The transcriptional activity of Gcm1 is increased by CBP (Crebbp – Mouse Genome Informatics)-dependent acetylation (Chang et al., 2005). Since p45NF-E2 interacts with CBP, which regulates acetylation in mouse erythroleukemia (MEL) cells (Kiekhaefer et al., 2004), we explored whether p45NF-E2 regulates Gcm1 via acetylation.

First, we evaluated whether p45NF-E2 modulates the CBP-Gcm1 interaction. Using immunoprecipitation we detected only a weak CBP-Gcm1 interaction in the presence of p45NF-E2, whereas a strong CBP-Gcm1 interaction was observed in the absence of p45NF-E2 (Fig. 6A). As expected, no p45NF-E2 could be detected in the immunoprecipitates of p45NF-E2-deficient tissues or cells (Fig. 6A). However, in tissue extracts from p45NF-E2^+/− placentae or p45NF-E2 control trophoblast cells, p45NF-E2 strongly co-immunoprecipitated with CBP (Fig. 6A). Thus, in the absence of p45NF-E2, the interaction of Gcm1 with CBP, which possesses an intrinsic acetyltransferase activity, is increased.

Immunoprecipitation experiments revealed increased acetylation of Gcm1 in p45NF-E2^−/− placentae and in p45NF-E2^kd trophoblast cells [149% compared with p45NF-E2^+/− placentae (P=0.002) and 137% compared with p45NF-E2-expressing control cells (P=0.005), respectively; Fig. 6B,C]. Hence, p45NF-E2 modifies Gcm1 post-translationally via acetylation. Furthermore, acetylation of histone H4 (ac-H4) within the Gcm1 promoter was enhanced in p45NF-E2^−/− placentae or p45NF-E2^kd trophoblast cells [187% (P=0.006) and 169% (P=0.008), respectively; Fig. 6D,E], implying that p45NF-E2 in addition epigenetically regulates Gcm1 expression via acetylation.

Compared with p45NF-E2-expressing trophoblast control cells (p45NF-E2^c), histone deacetylase (HDAC) activity was reduced to 68% in p45NF-E2^kd cells (P=0.03; Fig. 7A). To determine the mechanistic relevance of altered acetylation we modulated acetylation in p45NF-E2^c and p45NF-E2^kd trophoblast cells in vitro. Treatment of p45NF-E2^c trophoblast cells with the HDAC inhibitor TSA reduced HDAC activity to 62% (versus untreated p45NF-E2^c cells, P=0.01; Fig. 7A). HDAC inhibition with TSA enhanced syncytium formation (2.7 versus 0.5 in p45NF-E2^c cells, P=0.004; Fig. 7B,C) and the expression of syncytiotrophoblast markers (Gcm1, Cebpa, Synb) significantly (Fig. 7D).

Conversely, the histone acetyltransferase (HAT) inhibitor curcumin rescued HDAC activity in p45NF-E2^kd trophoblast cells (107% versus P45NF-E2^kd, P=0.006; Fig. 7A). HAT inhibition using curcumin reduced syncytium formation (0.7 versus 2.7 in untreated p45NF-E2^kd cells, P=0.009; Fig. 7B,C) and decreased the expression of syncytiotrophoblast markers (Gcm1, Cebpa, Synb) in p45NF-E2^kd cells (Fig. 7D). Similar results were obtained using the HAT inhibitor EGCG (Giunta et al., 2006) (data not shown). Thus, enhanced syncytiotrophoblast formation in p45NF-E2-deficient trophoblast cells is acetylation dependent.

Previous data established that the transcriptional activity of human GCM1 is enhanced by acetylation of lysine residues (Chang et al., 2005). To determine whether post-translational modification of Gcm1 via acetylation contributes to enhanced syncytiotrophoblast formation we replaced two lysine residues of murine Gcm1 with arginine (Gcm1^ΔKK). Following lentiviral-mediated transient transfection of p45NF-E2^kd cells with Gcm1^ΔKK, acetylation of Gcm1 was markedly reduced (as determined by immunoprecipitation, data not shown). In p45NF-E2^kd Gcm1^ΔKK trophoblast cells, syncytiotrophoblast formation was markedly reduced (1.1 versus 2.7 in p45NF-E2^kd cells, P<0.001; Fig. 7B,C). However, syncytiotrophoblast formation remained slightly higher than in curcumin-treated p45NF-E2^kd cells (1.1 versus 0.7 in p45NF-E2^kd cells, P=0.010; Fig. 7B,C). Concomitantly, we observed a reduction in expression of the Gcm1-dependent genes Cebpa (36%, P<0.001) and Synb (25%, P<0.001) in Gcm1^ΔKK-transfected p45NF-E2^kd cells (Fig. 7D). These data establish that the increased syncytiotrophoblast formation in p45NF-E2-deficient trophoblast cells depends largely on post-translational modification of Gcm1 by acetylation. However, other effects, such as the acetylation of other proteins and epigenetic gene regulation via acetylation, might contribute to the enhanced syncytiotrophoblast formation in p45NF-E2-deficient trophoblast cells.

Injection of HDAC inhibitors (TSA or valproic acid) into pregnant p45NF-E2^+/+ mice mated with p45NF-E2^+/+ males reduced HDAC activity and increased acetylation within the placenta (53%, P=0.004; Fig. 8A and data not shown). In utero exposure of p45NF-E2^+/+ embryos to HDAC inhibitors impaired placental vascularization [73% in p45NF-E2^+/+ TSA-treated embryos versus 100% in untreated p45NF-E2^+/+ control (p45NF-E2^+/+c) embryos, P=0.007; Fig. 8B,C], impaired embryonic growth [weight, 72% (P<0.017); length, 93% (P<0.021) compared with p45NF-E2^+/+c embryos] and increased the width of SynT-II (145%, P<0.001; Fig. 8D,E). The expression of the syncytiotrophoblast markers Gcm1 (172%), Cebpa (145%) and Synb (163%) was induced to a comparable extent as in p45NF-E2^−/− control (p45NF-E2^−/−c) placental tissues (Fig. 8F).

Conversely, treatment of pregnant mice carrying p45NF-E2^−/− embryos with HAT inhibitors (curcumin or EGCG) increased HDAC activity and reduced protein acetylation (Fig. 8A and data not shown). Following HAT inhibition, placental vascularization (98% versus 100% in p45NF-E2^+/+c mice, P=0.96; Fig. 8B,C), embryonic growth [weight, 97% (P=0.86); length, 98% (P=0.79)] and the width of SynT-II (98%, P=0.58; Fig. 8D,E) were rescued in p45NF-E2^−/− embryos (p45NF-E2^−/− Cur). Furthermore, the expression of the syncytiotrophoblast markers Gcm1 (103%), Cebpa (97%) and Synb (109%) was rescued (Fig. 8F).

The reversibility of reduced placental vascularization, impaired embryonic growth, enhanced syncytiotrophoblast formation and increased expression of Gcm1, Cebpa and Synb in p45NF-E2^−/− placentae or trophoblast cells upon inhibition of acetylation establishes that the placental and trophoblast differentiation defects in p45NF-E2^−/− embryos are mechanistically linked to altered acetylation activity.

The present study identifies a novel mechanism regulating syncytium formation of trophoblast cells. The transcription factor p45NF-E2, which hitherto was thought to be specific to hematopoietic cells, cell-autonomously represses Gcm1-dependent syncytium formation in trophoblast cells. This interaction of p45NF-E2 and Gcm1 depends, in vitro and in vivo, on acetylation, establishing that acetylation is crucial for syncytiotrophoblast formation, placental function and embryonic growth.

Hitherto, p45NF-E2^−/− mice were best known for their megakaryocyte maturation defect, which results in thrombocytopenia (Shivdasani et al., 1995). We now demonstrate that the placental defect and the IUGR of p45NF-E2^−/− embryos are independent of thrombocytopenia. This establishes that embryonic platelets and platelet-released mediators are dispensable for embryonic and placental development. This provides conclusive evidence that the developmental function of the embryonic hemostatic system is independent of the primary endpoint of hemostasis (the interaction of fibrin and platelets). The lethal embryonic phenotypes of embryos lacking some coagulation factors (e.g. tissue factor, prothrombin, factor X) must therefore be caused by mechanisms that are independent of the fibrin-platelet interaction and instead most likely related to the developmental function of PAR proteins in vascular and non-vascular tissues (Bugge et al., 1996; Camerer et al., 2010; Carmeliet et al., 1996; Cui et al., 1996; Griffin et al., 2001; Sun et al., 1998; Toomey et al., 1996; Xue et al., 1998).

The current study identifies a novel developmental function of p45NF-E2 in non-hematopoietic cells. Unlike in hematopoietic cells, where p45NF-E2 is perceived as a positive regulator (Andrews, 1998), p45NF-E2 must act as a repressor during syncytiotrophoblast development as the absence of p45NF-E2 enhances the expression of Gcm1 and of Gcm1-dependent genes and promotes syncytiotrophoblast formation. Considering the distinct functions of p45NF-E2 during the development of hematopoietic cells (transcriptional activation) and trophoblast cells (transcriptional repression), we propose that p45NF-E2 modulates hematopoietic and trophoblast cell differentiation through at least partially diverse mechanisms.

The mechanism through which p45NF-E2 governs megakaryocyte maturation remains incompletely understood. Of note, treatment of patients with HDAC inhibitors results in thrombocytopenia (Prince et al., 2009) and combined Hdac1 and Hdac2 deficiency results in megakaryocyte apoptosis and thrombocytopenia in mice (Wilting et al., 2010). Although it remains unknown whether the thrombocytopenia observed in p45NF-E2-deficient mice is related to altered acetylation, we establish that the placental phenotype and IUGR of p45NF-E2^−/− embryos are mechanistically linked to acetylation-dependent syncytiotrophoblast differentiation. This role of acetylation in trophoblast cells is in agreement with the acetylation-dependent syncytiotrophoblast formation in aryl hydrocarbon receptor nuclear translocator (Arnt, or Hif1β)-deficient trophoblast cells in vitro (Maltepe et al., 2005). The activity of HDACs, which are also referred to as lysine deacetylases (KDACs) or protein deacetylases (PDACs), was reduced in the absence of p45NF-E2. Enhanced acetylation was not specific for Gcm1, as we consistently observed a general increase in protein acetylation (data not shown). Indeed, in addition to the post-translational modification of Gcm1, we observed enhanced histone H4 acetylation within the Gcm1 promoter and increased Gcm1 mRNA levels in p45NF-E2-deficient trophoblast cells, indicating that p45NF-E2 regulates Gcm1 both epigenetically and post-translationally. Furthermore, it is possible that the post-translational modification of proteins other than Gcm1 contributes to the p45NF-E2-dependent placental phenotype. However, given the well-established pivotal role of Gcm1 in labyrinthine layer formation (Anson-Cartwright et al., 2000) and the current data showing that overexpression of Gcm1 is sufficient to increase syncytiotrophoblast formation, we conclude that enhanced expression of Gcm1 is crucial for the increased syncytiotrophoblast formation and the placental defect observed in p45NF-E2^−/− embryos.

The enhanced Gcm1 acetylation observed in the absence of p45NF-E2 is associated with the increased expression of Gcm1-dependent genes (Cebpa and Synb) relevant to murine syncytiotrophoblast formation (Simmons et al., 2008b). This is in agreement with findings in human trophoblast cells showing that acetylation increases the stability of GCM1 and the binding of GCM1 to the promoter of the syncytin (ERVWE1 – Human Gene Nomenclature Committee) gene, which encodes a placental fusogenic membrane protein (Chang et al., 2005; Chuang et al., 2006).

The transcription factor Gcm1 has two functions during placentation. It is required for (1) the initial folding of the chorion into simple branches and the subsequent branching morphogenesis within the placental labyrinthine layer and (2) syncytiotrophoblast fusion (Anson-Cartwright et al., 2000). Although, in the absence of Gcm1, both branching and syncytiotrophoblast fusion are impaired resulting in embryonic lethality at E10.5 (Anson-Cartwright et al., 2000; Schreiber et al., 2000a), overexpression of Gcm1 in p45NF-E2^−/− placenta has no apparent impact on EC or gross branching morphology. Thus, chorioallantoic fusion, embryonic EC invasion and chorioallantoic branching appear to be unaffected by high levels of Gcm1 expression in p45NF-E2^−/− placentae. The observed thickening of SynT-II in p45NF-E2^−/− placentae with high-level Gcm1 expression reflects a terminal syncytiotrophoblast differentiation defect during placental labyrinthine formation, resulting in a milder, non-lethal phenotype at a later stage. The different phenotypes that result from reduced or increased Gcm1 expression indicate that the two Gcm1-dependent effects are controlled through partially distinct mechanisms during placentation.

The enhanced Gcm1 expression and syncytiotrophoblast formation observed in murine p45NF-E2-deficient placentae and trophoblast cells might be of pathophysiological relevance to placental dysfunction in humans, as we have observed p45NF-E2 expression in human placental tissues (M.K. and B.I., unpublished) and because altering GCM1 expression triggers parallel changes in human trophoblast cells (Baczyk et al., 2009) to those observed in mice. Interestingly, both increased and decreased GCM1 expression have been reported in human placenta obtained from complicated pregnancies (Chen et al., 2004; Fujito et al., 2006; McCaig and Lyall, 2009). The placental defects associated with either the loss or gain of Gcm1 function observed in mice (Anson-Cartwright et al., 2000) (this paper) provide a rationale for the apparently paradoxical findings of the aforementioned clinical studies. The current results might aid in deciphering the mechanisms that result in placental dysfunction associated with low or high levels of GCM1 expression in humans.

P45NF-E2-deficient mice were kindly provided by R. Shivdasani (Harvard Medical School, Cambridge, MA, USA). We thank S. Schmidt and F. Zimmermann for technical support; and Hilmar Bading, Department of Neurobiology, Interdisciplinary Center for Neurosciences, University of Heidelberg, for providing the opportunity to carry out the electron microscopy work in his laboratory. This work was supported by grants of the Deutsche Forschungsgemeinschaft (IS 67/4-1, IS 67/4-2) to B.I. and a grant of the German Israeli foundation to B.I. and P.P.N. K.S. has a scholarship from the DAAD.