The nuclear envelope protein MAN1 regulates TGFβ signaling and vasculogenesis in the embryonic yolk sac

The nucleus of mammalian cells is bounded by the nuclear envelope (NE) and underlying nuclear lamina. The NE consists of an inner nuclear membrane (INM)and outer nuclear membrane (ONM). The two membranes are contiguous, being linked at the periphery of each nuclear pore complex. The nuclear lamina,which underlies the INM, primarily consists of the nuclear intermediate filament proteins, the A- and B-type lamins(Burke and Stewart, 2002). A growing body of evidence suggests that the lamins, as well as other proteins associated with the lamina and nuclear envelope, are crucial for interphase nuclear architecture, the structural integrity of the nucleus, gene regulation, DNA replication and chromatin organization. These activities are also highlighted by the fact that at least nine different diseases, ranging from muscular dystrophy to premature aging (the laminopathies), are linked to mutations within the lamin A gene (Burke and Stewart, 2002). In addition to the laminopathies, other diseases are caused by mutations in proteins associated with the NE(Gruenbaum et al., 2005; Worman, 2005).

Some eighty different NE-associated proteins have been identified(Schirmer et al., 2003), and among these are members of the LEM domain family(Lee and Wilson, 2004; Schirmer et al., 2003). LEM domain proteins contain a 43 amino acid motif facing the nucleoplasm, shared by the prototype members of the LEM family, LAP2, emerin and MAN1 (LEMD3)(Cai et al., 2001; Laguri et al., 2001; Lin et al., 2000; Wolff et al., 2001). Although mutated forms of emerin and possibly LAP2 result in muscular dystrophy and cardiomyopathy (Melcon et al.,2006; Morris,2004; Taylor et al.,2005), the functions of LEM domain proteins are still ill-defined. In Xenopus, XMAN1/SANE functions as an embryonic neuralizing factor,by antagonizing bone morphogenetic protein (BMP) signaling(Osada et al., 2003; Raju et al., 2003), whereas in humans, mutations in MAN1 (LEMD3) result in bone and connective tissue disorders (Hellemans et al., 2006; Hellemans et al.,2004).

TGFβ/BMP/activin signaling pathways activate SMAD transcription factors and are important in regulating mouse embryogenesis(Goumans and Mummery, 2000). Previous studies revealed that MAN1 binds to and regulates SMAD transcriptional activity, suggesting a potentially significant role for MAN1 in regulating TGFβ/BMP signaling(Hellemans et al., 2004; Lin et al., 2005; Pan et al., 2005). To determine the role of Man1, we analyzed the expression of Man1 and used a gene-trapped (GT) embryonic stem (ES) cell clone to derive mice with a functional mutation in the Man1 gene. We show that Man1 is essential for embryonic vasculogenesis and that the Man1^GT/GT mutation results in hyperactivation of some components in TGFβ signaling pathways.

The ES cell line XST167, containing a gene-trap insertion mutation(Skarnes, 1990) in the Man1 locus (BayGenomics, San Francisco, CA), was microinjected into C57Bl/6 blastocysts as described (Stewart,1993). The resulting Man1^GT line was maintained as heterozygotes and intercrossed to obtain homozygotes(Man1^GT/GT). All mice were maintained in our facility in accordance with the procedures outlined in the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 86-23, 1985).

Genotypes were determined by Southern blotting or RT-PCR. For Southern analysis, a 300 bp genomic fragment, generated using primers 5′-GCGCTGGGTTACTTTGTGTGCTG-3′ and 5′-GCTTCCCGTTCACCACACTTCTG-3′, was used to probe DNA digested with EcoRV. Embryos were genotyped by RT-PCR using forward primer 1 located within exon 4 (5′-GACCATGAATGTGGCAGTTCTA-3′) and reverse primer 2 located within exon 5 (5′-CGTACATGTGGAATAGGCATGTAAGG-3′)to obtain a 424 bp wild-type PCR product and primer 1 and reverse primer 3 located within the β-geo sequence(5′-TCGTCTGCTCATCCATGACC-3′) to obtain a 1.6 kb mutant PCR product.

Total RNA was extracted using the RNeasy Kit (Qiagen, Valencia, CA). Ten micrograms of RNA were separated on 1% formaldehyde 1% agarose gels and transferred to Hybond membranes (Amersham Biotech, Piscataway, NJ). The mouse cDNA corresponding to nucleotides 1261-1898 bp of the Man1 transcript(I.M.A.G.E. Consortium clone ID 455955) was used as a probe and a Gapdh cDNA as a loading control. Tissue and embryonic blots were obtained from Seegene (Rockville, MD).

Blastocysts from Man1 heterozygous intercrosses were used to derive homozygous Man1^GT/GT ES cells as described(Abbondanzo et al., 1993; Stewart et al., 1995). ES clones were genotyped by northern analysis (see above). Embryoid bodies (EBs)were prepared by growing ES cells in suspension. Two-week-old EBs were plated onto 35 mm culture dishes to allow formation of explants.

Homozygous Man1^GT/GT embryonic stem cells were injected into blastocysts and Man1^GT/GT fibroblasts were derived from chimeric embryos at embryonic day 13 (E13) as described(Escalante-Alcalde et al.,2003). Man1^GT/+ and wild-type fibroblasts were derived from embryos of heterozygous crosses as described(Escalante-Alcalde et al.,2003). Primary fibroblasts were immortalized by retroviral infection with SV40 large T antigen (Zhu et al., 1991). For SMAD translocation experiments, PMEFs were infected with retroviral vector (pBabe-Puro) containing a full-length MAN1 cDNA tagged with a Flag epitope (kind gift of Howard Worman, Columbia University, New York, NY).

Wild-type tetraploid embryos were prepared by incubating fertilized(C57Bl/6×C3H F1) oocytes overnight in KSOM (Specialty Media) with 0.25μg/ml cytochalasin D (Sigma), which suppresses the first cleavage division,making the zygotes tetraploid. The embryos were washed four times in KSOM to remove the cytochalasin D and allowed to progress to the two-cell stage. Two-cell-stage embryos were transferred to the oviducts of pseudopregnant recipients for an additional 24-48 hours for further development. Tetraploid blastocysts were recovered and injected with Man1^GT/GT ES cells. Embryos were transferred to pseudopregnant B6CBAF1 recipients with the embryos being recovered on E9.5-12.

Embryos aged E8.0-10.0 were stained for lacZ expression and photographed as previously described(Escalante-Alcalde et al.,2003). Benzidine staining to detect blood islands was performed as described (Orkin et al.,1975).

Embryos were whole-mount immunostained with antibodies to platelet endothelial cell adhesion molecule (PECAM-1; also known as PECAM1) and FLK-1(KDR - Mouse Genome Informatics) (BD Biosciences, San Diego, CA), as described(Byrd et al., 2002; Schlaeger et al., 1995). For immunohistochemistry, embryos were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned and stained using indicated antibodies. Immunofluorescence using an antibody to fibronectin was performed as described(Zwijsen et al., 1999).

EB explants were immunostained with PECAM-1 as described(Escalante-Alcalde et al.,2003). For fluorescence activated cell sorting (FACS) analysis to quantify expression of PECAM-1 in EB explants, 7-dayold explants were trypsinized, immunostained with an antibody to PECAM-1 conjugated to phycoerythrin (PE, BD Biosciences) and analyzed on an LSR1 flow cytometer using CellQuest software (Becton Dickinson). Unstained cultures were used as negative controls.

Mouse embryonic fibroblasts (MEFs) were fixed with 4% paraformaldehyde,stained with primary antibodies to SMAD1 (Santa Cruz, CA), SMAD2/3 (BD Biosciences) and Flag (Affinity Bioreagents, Golden, CO) and then with secondary antibodies conjugated to Alexa 488 or Alexa 568 (Molecular Probes,Eugene, OR). After immunolabeling, coverslips were mounted in Vectashield Mounting Medium (Vector Laboratories, Burlingame, CA) and visualized using a Zeiss Axiophot inverted microscope.

Whole cell lysates of MEFs were immunoblotted with a rabbit polyclonal antiserum raised against a peptide (aa898-911) at the C-terminus of MAN1. Antibodies used were for fibronectin (1:300, ICN), p21 (1:100, BD Pharmingen),SMAD2 (1:250, Zymed, San Francisco, CA), pSMAD2 (1:250, Upstate,Charlottesville, VA), pSMAD1 (Cell Signaling, St Louis, MO), and actin(1:1000, Santa Cruz). Detection was performed using Chemiluminescence Reagent Plus (PerkinElmer, Boston, MA). Cytoplasmic and nuclear fractions were prepared using the NE-PER kit (Pierce, Rockford, IL) according to the manufacturer's instructions.

The pTP3-Lux construct containing a TGFβ1-responsive Plasminogen Activated Inhibitor I (PAI-1) promoter linked to a luciferase cDNA(Wrana, 2000) (a kind gift of Joan Massague, Memorial Sloan-Kettering Cancer Center, New York, NY) was transfected into MEFs using Fugene 6 (Roche, Indianapolis, IN) according to the manufacturer's instructions. To normalize for transfection efficiency,cells were co-transfected with 0.5 μg pTK-RL (Promega, Madison, WI). After 42 hours following the transfection, cultures were incubated for 8 hours with hrTGFβ1 (R&D Systems) at indicated concentrations. Luciferase activities were measured using the Dual-Luciferase Assay according to the manufacturer's instructions (Promega).

MEFs were seeded onto 96-well culture dishes at a concentration of 5×10³ cells per well and treated with indicated concentrations of hTGFβ1 (R&D Systems). The endothelial cell line SVEC (ATCC #CRL-2181) was previously characterized(O'Connell and Edidin, 1990). After 48 hours, cells were incubated overnight with MTT using the Cell Proliferation Kit I (Roche) according to the manufacturer's instructions. Quantification of the formazan reaction product in active cells was performed using the VersaMax Plate reader (Molecular Devices).

RNA from MEFs and EB explants was extracted using the RNeasy Kit and 1μg was reverse-transcribed using the First-Strand cDNA Synthesis Kit(Roche). Real-time PCR was performed on an ABI Prism 7000 using SYBR Green Master Mix (AB, Warrington, UK).

To determine the function of Man1 during development, we used an ES cell line containing a gene-trap insertion into the Man1 locus. The gene-trap insertion, mapped by Southern analysis(Fig. 1A,B), introduces a novel en2 splice acceptor site and a β-galactosidase (β-gal) cDNA linked to the neomycin resistance cDNA (β-geo) within the Man1 intron between exon 4 and 5(Skarnes, 1990). Northern analysis of RNA from heterozygous fibroblasts revealed a 7.7 kb Man1-β-geo hybrid transcript in addition to the wild-type 4.7 kb transcript (Fig. 1C), indicating that the β-geo sequence is in frame with the Man1 transcript and the first 1702 bp of the endogenous Man1 transcript is retained. No additional transcripts were observed,indicating additional alternative splicing to exons downstream of the gene-trap insertion into Man1 did not occur. This analysis predicts that the gene-trapped Man1^GT allele produces a fusion protein consisting of 567 N-terminal MAN1 amino acids fused to β-geo and lacking the C-terminal sequence downstream of exon 4(Fig. 1D). Western analysis using an antibody derived from the C-terminus of MAN1 confirmed that the C-terminus was no longer recognized (Fig. 1E).

The expression of Man1 was initially determined by northern analysis (not shown). In embryonic tissues, the signal intensity of Man1 was low at stages E4.5 through E8.5 and increased in intensity in later developmental stages from days E9.5 through E18.5. Low levels of the Man1 transcript were also observed in most adult tissues, whereas robust expression was seen in the brain, testes and placenta. We used theβ-galactosidase activity of the gene-trapped fusion protein product to further determine the expression of Man1 during development. Expression of Man1^GT was first detectable in the ectoplacental cone and yolk sac at E8.5(Fig. 2A). At E9.5, lacZ expression increased initially in the gut epithelium and the floor plate (Fig. 2B) and by E11 spread to the skin and throughout the embryo(Fig. 2C), consistent with the widespread expression of Man1 detected by northern analysis. At E13,expression was widespread, with robust staining observed in the central nervous system (Fig. 2E-G),including the midbrain and hindbrain, neural tube, pons and cerebellum. Staining was also observed in the placenta(Fig. 2D,E) and yolk sac(Fig. 2E,H,I) in the visceral endoderm, endothelial cells and mesoderm but excluded from blood cells(Fig. 2H,I).

Intercrossing Man1 heterozygotes produced offspring, of which 38%were wild type, 62% were heterozygotes and none were homozygotes(Man1^GT/GT) (Table 1). Examination of embryos at various stages from Man1heterozygous intercrosses revealed that Man1^GT/GT embryos were indistinguishable from wild-type embryos up to E8.5 (data not shown). By E9.5, the Man1^GT/GT embryos were smaller and developmentally delayed (Fig. 3A,B), did not undergo axial rotation and had an abnormally developed heart with an enlarged pericardium(Fig. 3A,B). Expression of lacZ in the Man1^GT/GT embryos at E9.5 was restricted to the yolk sac, allantois and amnion(Fig. 3C). By E10.5, Man1^GT/GT embryos had ceased growing, with the yolk sac being expanded and the pericardium enlarged and edematous(Fig. 3D). In the Man1^GT/GT yolk sacs, the cell layers were highly disorganized, with lacZ expression being restricted to the endoderm,endothelium and mesoderm but not blood cells(Fig. 3E). Closer examination of individual cells revealed β-gal staining was predominantly localized to the nuclear envelope, and to a lesser degree to the nucleus and cytoplasm,indicating that the targeting of MAN1-β-gal to the NE is apparently unaffected by the gene-trap mutation (Fig. 2F). Some β-gal staining was also observed in the endoplasmic reticulum, which is contiguous with the nuclear envelope.

The first site of vasculogenesis is the embryonic yolk sac. The yolk sacs of Man1^GT/GT embryos were distinctly different from wild-type embryos (Fig. 4). Wild-type yolk sacs contained large branching vessels and a capillary network. Man1^GT/GT yolk sacs were completely devoid of large blood vessels, and the blood vessels remained as a primitive capillary plexus, interspersed with blood pools (Fig. 4A). The presence of erythrocytes in Man1^GT/GTembryos was confirmed by benzidine staining(Fig. 4B). Immunostaining for the endothelial marker PECAM-1 in Man1^GT/GT yolk sacs demonstrated the presence of endothelial cells along primitive capillary networks, indicating that vasculogenesis had commenced(Fig. 4C). Although the allantois frequently fused with the chorion in Man1^GT/GTembryos, it often was structurally abnormal (data not shown).

Vasculogenesis was also defective in Man1^GT/GT embryos proper. The embryonic vasculature was visualized by staining with antibodies to the VEGF receptor FLK-1. Well-formed vascular networks were observed in wild-type embryos (Fig. 4D). By contrast, development of the Man1^GT/GT embryos was delayed some 24-36 hours, with the vasculature of Man1^GT/GTembryos being highly disorganized (Fig. 4E,F). These observations suggest that, whereas the initiation of vasculogenesis proceeds in Man1^GT/GT embryos, formation of large blood vessels and angiogenesis are perturbed.

Abnormal development and lethality of embryos around E8.5-9.5 is frequently due to primary defects in placental development(Rossant and Cross, 2001). To establish whether the Man1^GT/GT phenotype was due to placental defects, we performed a tetraploid rescue experiment. Diploid Man1^GT/GT ES cells were microinjected into wild-type tetraploid blastocysts and transferred to pseudopregnant recipients. Postimplantation embryos were recovered on E9.5 or E12. Of the 36 blastocysts transferred, three embryos were found to resemble the Man1^GT/GT phenotype(Table 2), and their genotype was confirmed by RT-PCR. Of these three, two were delayed approximately 36 hours and blood vessels had not formed from the primitive plexus. The third embryo contained irregularly shaped blood vessels and had undergone embryonic axial rotation but was edematous and small compared to E9.5 wild-type embryos. No embryos were observed to develop beyond the Man1^GT/GTphenotype and none were recovered at E12. As the wild-type tetraploid cells form the placenta and visceral endoderm layer of the yolk sac, our results indicate that embryonic lethality was probably not due to a defect in placental function.

Differentiation and formation of the vascular endothelium occurs in the yolk-sac mesoderm (Baron,2003). In wild-type yolk sacs, the endoderm and mesoderm are tightly apposed and contain many developing blood vessels and discreet blood islands (Fig. 5A). In Man1^GT/GT yolk sacs, the endoderm and mesoderm were poorly attached to each other, resulting in large lacunae containing blood cells(Fig. 5B). Immunostaining with antibodies to PECAM-1 and FLK-1 revealed that although endothelial cells were present in Man1^GT/GT yolk sacs, many remained rounded,were misaligned or had failed to incorporate into the vessel walls(Fig. 5B,C).

Formation of branched blood vessels during vasculogenesis requires the proliferation and recruitment of smooth muscle cells to line the vessel walls. TGFβ1 is a potent inhibitor of vascular smooth muscle cell (VSMC)proliferation (Feinberg et al.,2004). We analyzed sections immunostained with an antibody for the VSMC marker α-smooth muscle actin (ACTA2)(Fig. 5D). Yolk sacs from wild-type embryos contained a large number of VSMCs, whereas VSMCs appeared to be less abundant in the Man1^GT/GT embryos. Quantification of Acta2 transcripts in E9.5 yolk sacs by RT-PCR revealed an approximately twofold reduction in transcript levels in the Man1^GT/GT yolk sacs compared with wild type (data not shown). This reduction in VSMCs is probably due to their decreased proliferation and defective recruitment into developing blood vessels,resulting in the subsequent failure of nascent capillaries to form functional blood vessels.

The vascular defects in the Man1^GT/GT embryos may be due to defective differentiation, proliferation or organization of the endothelial cells (ECs) to form the primitive capillaries. To characterize these defects, we generated embryoid bodies (EBs) from wild-type and Man1^GT/GT ES cells. EBs cultured in suspension for 2 weeks were re-plated and formed explants containing a large number of PECAM-1⁺ ECs (Fig. 6A,B). The explants of both Man1^GT/GT and wild-type EBs contained PECAM-1⁺ cells arranged in capillary networks, indicating that Man1^GT/GT ECs can differentiate from precursors and form capillaries. However, the capillary networks in the Man1^GT/GT explants were less expansive and were surrounded by many PECAM-1⁺ cells that had not integrated into the vascular network (Fig. 6B). Formation of branched blood vessels from primitive capillaries requires endothelial cell proliferation and organization to form the vessel lumen(Risau and Flamme, 1995). Thus, despite the formation of ECs in Man1^GT/GT embryos and EBs, further organization into branched structures is defective. Quantitation of PECAM-1⁺ ECs in the EB explants was performed using FACS analysis. During differentiation, ECs first express low levels of PECAM-1, then expression of PECAM-1 increases as ECs differentiate from precursors (Levenberg et al.,2002). In accordance, two peaks of PECAM-1⁺ cells were observed by FACS analysis. No difference was observed between wild-type and Man1^GT/GT cultures in the fraction of low-expressing PECAM-1⁺ precursor cells, while the fraction of high PECAM-1-expressing EC cells was reduced in Man1^GT/GTcultures compared with wild types (∼30% of total cell population was high PECAM-1⁺ in wild-type EB cultures compared with only 7% of cells in the Man1^GT/GT cultures)(Fig. 6C).

Previous evidence suggested that Man1 regulates cell proliferation. TGFβ1 inhibits fibroblast proliferation, and overexpression of Man1 decreases the anti-proliferative effect of TGFβ1 (Lin et al., 2005). By contrast, decreased Man1 expression enhances the anti-proliferative activity of TGFβ1(Pan et al., 2005). To determine whether disruption of Man1 increased the sensitivity of cells to the anti-proliferative effect of TGFβ1, we compared cell proliferation in response to TGFβ1 treatment in Man1^GT/GT and wild-type MEFs(Fig. 6D). Treatment of wild-type MEFs with 1.6 ng/ml of TGFβ1 resulted in a 10% suppression of cell proliferation. By comparison, we also found that TGFβ1 inhibited proliferation of the immortalized mouse endothelial SVEC cell line to a similar extent. By contrast, treatment of Man1^GT/GT MEFs with TGFβ1 resulted in a significantly increased suppression of cell proliferation by 25%. These results confirm that TGFβ1-mediated inhibition of proliferation is more effective in Man1^GT/GTcells.

To further characterize the reduction in the proliferation of endothelial cells, the expression of endothelial cell markers and cell cycle mediators was quantitated in the EB explants. Real-time PCR analysis showed that expression of endothelial markers Flt-1 and Alk5 (Tgfbr1 -Mouse Genome Informatics) were both reduced in Man1^GT/GTcultures (Fig. 6E). Consistent with the observations in the yolk-sac sections(Fig. 5D), the expression of Acta2 was reduced by 60% relative to wild-type EB explants(Fig. 6E). By contrast,increased expression of the cell cycle inhibitors p15ink4b(Cdkn2b - Mouse Genome Informatics) and p27kip1(Cdkn1b - Mouse Genome Informatics) was observed, consistent with the suppression of cell proliferation in the Man1^GT/GT EB explants. Both p15ink4b and Timp1 are direct transcriptional targets of TGFβ signaling (Edwards et al., 1987; Hannon and Beach,1994), and their increased expression suggests that TGFβsignaling is upregulated in the Man1^GT/GT EB explants.

The binding of TGFβ1 to its receptors results in the phosphorylation and subsequent translocation of the receptor-activated SMADs (R-SMADs) from the cytoplasm to the nucleus (Massague and Wotton, 2000). As MAN1 antagonizes BMP and TGFβ1 signaling by interacting with R-SMADs (Hellemans et al., 2004; Lin et al.,2005; Osada et al.,2003; Pan et al.,2005; Raju et al.,2003), we analyzed whether SMAD activity was disrupted in Man1^GT/GT cells. We examined R-SMAD localization in TGFβ1-treated Man1^GT/GT MEFs using antibodies to R-SMADs (Fig. 7Aa-i). Untreated wild-type MEFs express equivalent levels of SMAD2/3 in the nucleus and cytoplasm (Fig. 7Aa). Treatment of wild-type MEFs with TGFβ1 resulted in an increase in SMAD2/3 signal in their nuclei (Fig. 7Ab). In the Man1^GT/GT MEFs, high levels of nuclear SMAD2/3 immunofluorescence were observed in untreated cells(Fig. 7Ac), and this nuclear signal further increased following treatment with TGFβ1 (not shown). Expression of a full-length Flag-tagged MAN1 in wild-type MEFs prevented the nuclear accumulation of SMAD2/3 (Fig. 7Ab,e insets). However, in Man1^GT/GT MEFs expressing Flag-MAN1, SMAD2/3 was equally distributed between the cytoplasm and nucleus (Fig. 7Ac,f inset),indicating that increased nuclear SMAD accumulation in Man1^GT/GT MEFs can be reduced by the reexpression of wild-type MAN1.

The increased accumulation of nuclear SMAD2 was quantified by immunoblotting cytoplasmic and nuclear fractions from Man1^GT/GT and wild-type MEFs(Fig. 7B). Treatment with TGFβ1 resulted in a more robust nuclear localization of SMAD2 in Man1^GT/GT MEFs (upper panel). By contrast, in Man1^GT/GT MEFs expressing a full-length MAN1, the nuclear localization of SMAD2 in Man1^GT/GT MEFs was significantly reduced (Fig. 7B lower panel).

As TGFβ1 signaling also results in phosphorylation of the R-SMADs, we examined the levels of phosphorylated SMADs1 and 2 (pSMADs1 and 2) in Man1^GT/GT and wild-type MEFs(Fig. 7C). pSMAD2 levels were higher in untreated Man1^GT/GT MEFs, compared with wild-type MEFs, and treating Man1^GT/GT MEFs with TGFβ1 resulted in increased levels of pSMAD2 compared with wild-type MEFs(Fig. 7C lower panel). By contrast, the levels of pSMAD1 were not upregulated in Man1^GT/GT MEFs (Fig. 7C lower panel).

To examine whether the transcription of TGFβ1-regulated genes is affected in Man1^GT/GT cells, we performed luciferase reporter assays in Man1^GT/GT and wild-type MEFs(Fig. 7D) transfected with p3TP-Lux (Wrana, 2000). Treatment of Man1^GT/GT and wild-type MEFs with TGFβ1 for 8 hours resulted in approximately threefold higher levels of normalized luciferase reporter activity than in untreated cultures(Fig. 7D). However, basal luciferase activity was approximately tenfold higher in Man1^GT/GT MEFs compared with wild types, indicating increased basal levels of R-SMAD-dependent transcription in Man1^GT/GT MEFs. Together, these data reveal that R-SMAD levels in Man1^GT/GT cells are not only increased in the nucleus, but also have increased functional activity.

We extended these findings to determine what effect the Man1^GT/GT mutation has on specific gene expression levels in MEFs. Several known targets of TGFβ1 signaling were analyzed using quantitative real-time PCR in Man1^GT/GT and wild-type MEFs treated with TGFβ1 for 48 hours (Fig. 7E). Transcript levels of Pai-1 (Serpine1 -Mouse Genome Informatics) were upregulated in untreated Man1^GT/GT MEFs compared with wild-type MEFs and treatment with TGFβ1 resulted in a further ∼fivefold increase. Similarly transcript levels of Timp1 were upregulated in untreated Man1^GT/GT MEFs compared with wild-type MEFs and were further increased by TGFβ1 treatment. Previous reports using gene profiling of MEFs demonstrated that inhibitor of DNA binding-3 (Id3),a target of the BMP signaling pathway, is negatively regulated by TGFβ1 stimulation (Karlsson et al.,2005). Id3 levels were consistently decreased by the TGFβ1 treatment and showed decreased levels in Man1^GT/GT MEFs.

Proper yolk-sac formation also requires a balanced deposition in extracellular matrix molecules (Baldwin,1996). Analysis of fibronectin (Fn; also known as Fn1) expression in yolk-sac sections by immunofluorescence showed that fibronectin levels are increased in the mesoderm but reduced in the endoderm in the Man1^GT/GT yolk sacs, indicating abnormal extracellular matrix deposition in the Man1^GT/GT yolk sacs(Fig. 5E). Consistently, we found an upregulation of Fn mRNA levels in Man1^GT/GT EBs (Fig. 6E) and Man1^GT/GT MEFs(Fig. 7F) and an increase in Fn protein levels in Man1^GT/GT MEFs(Fig. 7F upper panel). Because the extracellular matrix plays an important role in regulating the proliferation of endothelial cells during lumen formation and blood vessel branching (Baldwin, 1996),increased levels of fibronectin, as well as other extracellular matrix proteins including PAI-1 and TIMP1 in Man1^GT/GT embryos,may have a direct impact on EC proliferation and migration, resulting in defective vasculogenesis. In addition, expression of cell cycle regulator p21waf1 (Cdkn1a - Mouse Genome Informatics) was increased in Man1^GT/GT MEFs (Fig. 7F, lower panel), indicating that increased TGFβ1 signaling and activation of cell cycle regulators may contribute to the decreased proliferation of ECs in Man1^GT/GT embryos.

During mammalian embryogenesis, one of the first tissues formed after embryo implantation is the vascular system. This tissue is essential to sustain the growth and differentiation of the other embryonic tissues, to supply them with nutrients and oxygen, as well as to remove their metabolic waste products. Considerable information already exists as to which growth factors regulate the development of the vascular system in mouse embryos, and among these factors the TGFβ signaling pathway has a central role(Mummery, 2001; Rossant and Howard, 2002).

The TGFβ superfamily of growth factors consists of a multitude of ligands, receptors and transcription factors, as well as other less well-defined proteins regulating the activation of the pathways. Such complexity reflects the many proliferative, differentiative and physiological functions regulated by the TGFβ superfamily(Feng and Derynck, 2005; Massague et al., 2005). One of the more recently identified factors implicated in regulating TGFβsignaling is the nuclear-envelope-associated protein MAN1. Recent studies showed that the nucleoplasmic RRM domain at the C-terminus of MAN1 binds, and seemingly sequesters, SMADs1-3 to the nuclear envelope. Such sequestration regulates SMAD nuclear translocation and transcriptional activity, by an as yet undefined mechanism (Lin et al.,2005; Pan et al.,2005). Here we show that disruption of the Man1 gene leads to the death of homozygous mouse embryos around the time vascularization commences, probably by perturbing TGFβ signaling in the yolk-sac mesoderm.

In the Man1^GT/GT embryos, abnormal yolk-sac vasculogenesis is phenotypically similar to previously reported defects arising from mutations in receptors, ligands and other components of the TGFβ signaling pathway (Agah et al.,2000; Dickson et al.,1995; Larsson et al.,2001; Oshima et al.,1996). Yolk-sac vasculogenesis consists of a proliferative activation state and a final resolution state, which are mediated by the coordinate activation by TGFβ of two parallel receptor-mediated cellular pathways (Oh et al., 2000). TGFβ1 acting through the ALK1 receptor and SMADs1/5/8, mediates the activation state leading to proliferation of endothelial cells as well as recruitment of smooth muscle cells; whereas TGFβ1 acting through the ALK5 receptor and SMADs2/3, inhibits the proliferation of endothelial cells(Bertolino et al., 2005; Goumans et al., 2002; Pepper, 1997). Supportive of this model, functional deletion of Alk5 in mice leads to defects both in hematopoiesis and vasculogenesis and death around day E10.5. Endothelial cells derived from the deficient animals are over-proliferative and show reduced levels of Fn in response to TGFβ1(Larsson et al., 2001). However, mice deficient in Alk1 (Acvrl1 - Mouse Genome Informatics) also develop defects in angiogenesis and show upregulation of the ALK5-regulated Pai-1 (Oh et al.,2000). Comparably, cells overexpressing Alk5 show decreased proliferation, downregulation of inhibitory factor Id1 and upregulation of Pai-1 (Goumans et al., 2002). Finally, disruption of Endoglin, a component of the TGFβ1 receptor complex that antagonizes TGFβ signaling through the ALK5 receptor, also results in defective angiogenesis(Arthur et al., 2000; Li et al., 1999; van den Driesche et al.,2003).

Several lines of evidence from our studies suggest that the defect in vascularization in Man1^GT/GT yolk sacs is preferentially mediated through an overactivated ALK5 pathway. Results from the western analysis showed that ALK5-responsive SMAD2 and not ALK1-responsive SMAD1 is hyperphosphorylated and more concentrated in the nucleus. In Man1^GT/GT MEFs and EBs, proliferation of ECs and VSMCs is reduced, suggesting that vascularization in Man1^GT/GT yolk sacs is compromised due to insufficient numbers of cells being available to remodel the existing capillary plexus. This reduction in cell number correlated with upregulation in the expression of the cell cycle inhibitors p15ink4b, p21waf1 and p27kip1. In addition, increased ECM deposition (Goumans et al.,1999), together with increased PAI-1 inhibit EC migration and angiogenic branching (Ignotz and Massague,1986; McIlroy et al.,2006). Thus, increased Fn and Pai-1 levels in Man1^GT/GT embryos may have also exacerbated the effects on EC proliferation and contributed to defective vasculogenesis in Man1^GT/GT yolk sacs.

During the review process of this paper, a complementary study of the Man1^GT/GT phenotype was published(Ishimura et al., 2006). Consistent with our findings, the authors demonstrated elevated nuclear levels of SMADs2/3, as well as increased ECM deposition in the Man1^GT/GT mice. Interestingly, the authors attribute the increase in ECM due to apoptosis, not cell proliferation, as suggested by an apparent lack of increased phosphohistone 3 signal in the embryo sections. Although not inconsistent with this observation, our conclusion from the cellular proliferation analysis and upregulation of cell cycle inhibitor proteins, suggests that decreased proliferation does play a role in the vascular defect.

R-SMADs are normally in equilibrium between the cytoplasm and the nucleus(Risau and Flamme, 1995; Xiao et al., 2001; Xu et al., 2002). Stimulation by TGFβ1 increases the levels of phosphorylated R-SMADs in the nucleus. Activated R-SMADs are subject to negative regulation by nuclear factors such as c-Ski, SARA and SMURFs (Massague and Chen, 2000). Gating the entry and exit of R-SMADs into the nucleus is another means of regulating a balance of cellular response to TGFβstimulation (Luo, 2004).

The RRM domain of MAN1 physically interacts with R-SMADs, so regulating SMAD activity (Lin et al.,2005; Osada et al.,2003; Pan et al.,2005). We demonstrated that disruption of MAN1 results in enhanced SMAD2/3 nuclear localization and transcriptional activity, leading to the altered expression of genes regulated by TGFβ1. Although the nucleoplasmic localization of pSMAD2 was increased in Man1^GT/GT MEFs in the absence of TGFβ1 addition, this was probably due to stimulation by endogenous TGFβ/BMPs in the serum in which the cells were cultured (Ying et al., 2003). However, treatment of Man1^GT/GTMEFs with TGFβ1 further increased the nuclear concentration and activity of R-SMADs, suggesting that MAN1 regulates the entry/exit of SMADs in the nucleus. Our observations also suggest that MAN1 regulates the nuclear localization and transcriptional activity of SMADs by affecting SMAD phosphorylation. Although MAN1 has not been shown to possess any intrinsic phosphatase activity, the C-terminal domain may act as a scaffold or binding site that brings phosphorylated SMADs into an association with SMAD phosphatases (Chen et al.,2006; Knockaert et al.,2006; Lin et al.,2006). A similar role for the nuclear lamina affecting the phosphorylation state of SMADs was recently proposed, as loss of the A-type lamins may affect TGFβ1-mediated SMAD phosphorylation by regulating protein phosphatase 2A (Van Berlo et al.,2005). Whether the lamin A-mediated pathway is independent of MAN1 in regulating SMAD phosphorylation and activity remains to be demonstrated.

Functional analysis of MAN1 in different species has revealed differing roles, possibly reflecting tissue specificity(Hellemans et al., 2004; Osada et al., 2003). In Xenopus embryos, a reduction in XMAN1 disrupted the formation of anterior neural structures, suggesting that in XMAN1 antagonizes BMP signaling(Osada et al., 2003). Formation of the neural crest and anterior neural structures appeared to be overtly normal in Man1^GT/GT embryos before their death. However, as Man1 transcripts are abundantly expressed in the adult brain, we cannot rule out that Man1 may still play an undefined role in later development of the nervous system. In humans, loss-of-function mutations in MAN1 result in osteopoikilosis, melorheostosis and Buschke-Ollendorf syndrome (Hellemans et al., 2004). Although we performed X-ray scans on the bones from adult Man1^GT/+ mice, we did not detect any evidence of hyperostotic lesions, and the bone density in Man1^GT/+mice was similar to that of wild-type mice (data not shown). It is unclear whether failure to detect bone defects in the heterozygous mice is due to possible differences in the age of onset of lesions between humans and mice,or due to differences between the types of mutation between the human gene and the gene-trap. As the N-terminal sequence of MAN1 is still present in Man1^GT mice, and targeting to the NE is undisturbed, it is possible that the remaining sequence, including the LEM domain, may prevent Man1^GT/GT embryos from developing additional phenotypes.

Our findings and other recent evidence strengthens the notion that the NE/lamina is an important nuclear compartment regulating chromatin organization and gene expression. Studies in yeast revealed the importance of the NE in gene silencing and telomere function(Taddei et al., 2004). Recent studies have shown that the lamina may be important for the sequestration, and to some extent function, of crucial regulatory factors such as the c-Fos and Retinoblastoma (Rb) proto-oncogenes(Ivorra et al., 2006; Markiewicz et al., 2002; Melcon et al., 2006). Also,silencing in the expression of tissue-specific transcription factors such as MyoD, and the activity of the chromatin insulator gypsy, is associated with their localization to the nuclear periphery and lamina,respectively (Capelson and Corces,2005; Lee et al.,2006). In this capacity, MAN1, a transmembrane protein of the INM that is differentially expressed during development and in adult tissues, may act as a scaffold protein regulating SMAD phosphorylation, localization and activity. The NE and lamina may therefore have many additional, as yet undiscovered, roles in regulating cell differentiation and function during development.

This research was supported by the Intramural Research Program of the NIH,National Cancer Institute, Center for Cancer Research. We wish to thank Amanda Boyce and Rocky Tuan for X-ray analysis of the mice; Howard Worman for providing the MAN1 cDNA; Steve Young and Jonathan Cohen for advice; and Richard Frederickson for preparation of the figures.