The formation of the proamniotic cavity in the mammalian embryo is the earliest of many instances throughout development in which programmed cell death and the formation of epithelia play fundamental roles (Coucouvanis, E., and G.R. Martin. 1995. Cell. 83:279–287). To determine the role of the basement membrane (BM) in cavitation, we use embryoid bodies derived from mouse embryonic stem cells in which the LAMC1 genes have been inactivated to prevent BM deposition (Smyth, N., H.S. Vatansever, P. Murray, M. Meyer, C. Frie, M. Paulsson, and D. Edgar. 1999. J. Cell Biol. 144:151–610). We demonstrate here that LAMC1−/− embryoid bodies are unable to cavitate, and do not form an epiblast epithelium in the absence of a BM, although both embryonic ectodermal cells and extraembryonic endodermal cells do differentiate, as evidenced by the expression of cell-specific markers. Acceleration or rescue of BM deposition by exogenous laminin in wild-type or LAMC1−/− embryoid bodies, respectively, results in cavitation that is temporally and spatially associated with restoration of epiblast epithelial development. We conclude that the BM not only directly regulates development of epiblast epithelial cells, but also indirectly regulates the programmed cell death necessary for cavity formation.
The formation of cavities in solid blocks of cells is a widespread event in organogenesis throughout embryonic development. Over the last decade, it has become apparent that programmed cell death (PCD) plays a fundamental role in cavity formation in many tissues (Coles et al. 1993; Coucouvanis and Martin 1995; Humphreys et al. 1996; Jacobson et al. 1997). Although it is known that basement membranes (BM) are necessary for the survival and differentiation of epithelial cells surrounding the cavities (Ekblom et al. 1980; Coucouvanis and Martin 1995; Streuli 1996), any involvement of BMs in the regulation of PCD and the mechanisms coordinating epithelialization with PCD during cavitation remain unknown.
Formation of the proamniotic cavity is the first instance of cavitation during mammalian development. Shortly before implantation, the inner cell mass (ICM) of the mouse embryo consists of a small group of cells separated from an outer layer of primitive endoderm by a BM (Salamat et al. 1995). Subsequently, the primitive endoderm cells remaining in contact with this BM differentiate to become visceral endoderm (VE), while the remaining ICM cells differentiate to become the epiblast, or primitive ectoderm (see Fig. 1). Initially, the differentiation of epiblast cells is reflected by an alteration in the profile of expressed genes, and is not accompanied by any obvious morphological differentiation (Kaufman 1992; Rathjen et al. 1999). However, a few hours later, the epiblast cells in contact with the BM become polarized to form the columnar epiblast epithelium (CEE), while cells at the center of the ICM undergo PCD, thereby giving rise to the proamniotic cavity (Coucouvanis and Martin 1995).
Embryoid bodies (EBs), which are derived from differentiating mouse embryonic stem (ES) or embryonal carcinoma (EC) cells, are widely used model systems suitable for analysis of such events, cell differentiation in EBs closely reflecting that of the ICM during the peri-implantation period (Robertson 1987). Thus, cavitating and noncavitating EBs, which are derived from PSA1 and S2 EC cell lines, respectively, have been used to show that VE cells secrete a diffusible factor that induces PCD of the majority of epiblast cells (Coucouvanis and Martin 1995). However, the BM that lies between the VE and the epiblast supports the survival of the CEE cells in contact with it (Coucouvanis and Martin 1995). More recent studies using these EC cells have implicated bone morphogenetic proteins 2 and/or 4 in the apoptosis of epiblast cells by showing BMP4 only to be expressed by cavitating PSA1 EBs and being able to induce cell death in normally noncavitating S2 EBs (Coucouvanis and Martin 1999).
To investigate the role of BMs during embryonic development in vivo and EB development in vitro, we previously used homologous recombination in mouse ES cells to knockout one or both copies of the LAMC1 gene encoding the laminin γ1 subunit (Smyth et al. 1999). This defect renders the cells incapable of assembling a laminin type-1 trimer, which is necessary for BM deposition; hence, both LAMC1−/− preimplantation embryos and EBs lack BMs (Smyth et al. 1999). In vivo, the consequence of the lack of BMs is that the embryo dies during the peri-implantation period around the time when cavitation occurs, although the cause of this remains to be established (Smyth et al. 1999). However, the LAMC1−/− ES cells offer a unique system to help delineate the reasons for this lethality by establishing the role of the BM in proamniotic cavitation (Coucouvanis and Martin 1995). By being able to experimentally manipulate BM deposition in LAMC1−/− EBs by addition of exogenous laminin, we are able to demonstrate that the BM is not only necessary for formation of the columnar epiblast epithelium, but is also necessary for the cell death leading to cavitation. Involvement of the BM in both these processes indicates that this extracellular matrix structure plays a key role in the coordination of events necessary for cavitation in developing tissues.
Materials and Methods
ES Cell and EB Culture
The production of R1 mouse LAMC1+/− and LAMC1−/− ES cells has been described in detail previously (Smyth et al. 1999). The LAMC1+/− ES cells, used here as controls, were from the clone previously used to produce healthy heterozygous germline animals (Smyth et al. 1999). The absence of clonal artefacts in the LAMC1−/− cells used here was confirmed by rescue of the phenotype by adding laminin type-1 (Sigma Chemical Co.) to developing LAMC1−/− EBs (see Fig. 4). ES cells were cultured on mitomycin-treated STO feeder cells in gelatinized 3.5-cm tissue culture dishes. The culture medium was DME (GIBCO BRL) supplemented with 15% (vol/vol) ES grade FBS (GIBCO BRL), 0.1 mM β mercaptoethanol, 1 mM l-glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, and 1,000 U/ml of LIF (ESGRO; GIBCO BRL). ES cells were subcultured every 2 d. Before EB formation, ES cells were passaged once on gelatinized tissue culture dishes and incubated in the above medium for 2 d to eliminate STO cells. To make EBs, ES cells were trypsinized, triturated, and split 1:10 by replating into bacterial petri dishes, under which conditions the ES cells remained in suspension and formed aggregates. The EB culture medium was as above, except that LIF was omitted and the FBS content was reduced to 10% (vol/vol). After 2 d of suspension culture, the EB population of each 10-cm petri dish was divided into two and supplemented with fresh EB culture medium, after which the medium was changed on every second day. For the rescue experiment, 20 μg/ml of laminin type-1 was added to the culture medium immediately after replating of ES cells into petri dishes. For toluidine blue staining, immunostaining and terminal transferase–mediated biotinylated-dNTP end labeling (TUNEL) analysis, EBs were fixed for 1 h with 4% (wt/vol) paraformaldehyde and gelatin-embedded for preparation of frozen cryostat sections. For transmission EM, EBs were fixed for 1 h in 2:4% glutaraldehyde/paraformaldehyde (wt/vol) and processed as previously described (Fleming et al. 1984).
Immuno- and Fragmented DNA Staining
The primary antibodies used were rabbit anti-EHS laminin that recognizes all three subunits of laminin type-1 (Kücherer-Ehret et al. 1990), and so could be used to localize α1 and β1 subunits in the absence of γ1 in LAMC1−/− EBs. Rabbit antiperlecan antibodies were raised against recombinant perlecan domain III3 (Schulze et al. 1995). Incubations with primary antibodies were carried out overnight in 1% (vol/vol) goat serum in PBS at room temperature, and the sections were washed three times in PBS. The secondary antibody was TRITC-conjugated swine anti-rabbit IgG (Dako), which was applied in 1% (vol/vol) goat serum in PBS at room temperature for 2 h, and the sections were washed three times in PBS. For the detection of fragmented DNA on EB frozen sections, the TUNEL method was used as described previously (Smyth et al. 1999). Sections were mounted in fluorescent mounting medium (Dako) and photographed using a Leitz fluorescence microscope. All digital images were prepared with Adobe Photoshop 5.
Reverse Transcription–PCR (RT-PCR) Analysis of mRNA
Total RNA was extracted from LAMC1+/− and LAMC1−/− ES cells or EBs using guanidinium isothiocyanate (Chomczynski and Sacchi 1987), and reverse-transcribed using SuperscriptTM II RT (GIBCO BRL). For undifferentiated ES cells and day 2 EBs, whole populations were used, but for day 10 EBs, 10–15 cavitated LAMC1+/− EBs and an equal number of LAMC1−/− EBs were selected using phase-contrast microscopy. α-Fetoprotein (AFP) primers were those used for riboprobe synthesis (see below), and BMP4 and FGF-5 primers were as described previously (Johansson and Wiles 1995). GAPDH primers were as follows: forward (5′-GGTGAAGGTCGGAGTCAACGG-3′) and reverse (5′-GGTCATGAGTCCTTCCACGAT-3′; product size, 520 bp). Semi-quantitative RT-PCR was performed as previously described to determine mRNA levels relative to that of GAPDH (Squitti et al. 1999).
Whole-mount In Situ Hybridization
A sequence containing nucleotides 309–770 of mouse AFP cDNA (Tilghman et al. 1979) was amplified by PCR with forward primer (5′-ACATCAGTGTCTGCTGGCAC-3′) and reverse primer (5′-AGCGAGTTTCCTTGGCAACAC-3′), from cDNA reverse-transcribed from total RNA extracted from day 10 EBs. The PCR fragment was cloned into the T-Easy™ Vector (Promega) and transcribed with T7 or SP6 and digoxigenin-UTP for sense or antisense probes. Whole-mount in situ hybridization was performed as previously described (Leibl et al. 1999).
Results and Discussion
Fig. 1 shows a schematic diagram of the organization of cells and BM during the periimplantation stages of mouse development.
Cavitation Fails in LAMC1−/− EBs Despite the Presence of VE Cells
Histological analysis of LAMC1+/− control EBs, which are able to synthesize BMs and have a wild-type phenotype (Smyth et al. 1999), showed that they cavitated in suspension culture as expected (Fig. 2 a). In contrast, the LAMC1−/− EBs failed to form a cavity (Fig. 2 b). Furthermore, EM revealed that in addition to the lack of deposition of a BM (Smyth et al. 1999), a columnar ectodermal epithelium (CEE) failed to form in the LAMC1−/− EBs (see Fig. 3 e). Despite these differences, EM also showed that cells with the morphological characteristics of VE, namely apical vacuoles and microvilli, were present at the periphery of both LAMC1−/− and LAMC1+/− EBs (Fig. 2c and Fig. d). Additionally, whole-mount in situ hybridization showed that the VE marker AFP (Dziadek and Adamson 1978) was expressed in some of the peripheral cells of both LAMC1−/− and LAMC1+/− control EBs (Fig. 2e and Fig. f). Semi-quantitative RT-PCR was used to demonstrate that the relative levels of AFP mRNA were similar or even somewhat higher in the LAMC1−/− EBs than in controls while being absent in undifferentiated EBs (Fig. 2 g). Taken together, these results indicate that the BM, while having no apparent effect on VE cell differentiation, is necessary for the previously reported regulation of PCD by endodermal cells (Coucouvanis and Martin 1995).
Epiblast Cell Differentiation in LAMC1−/− EBs
Several lines of evidence have been presented suggesting that VE regulates epiblast cell differentiation. For example, disruption of the VE-specific gene Evx1 inhibits epiblast cell differentiation (Spyropoulos and Capecchi 1994), and factors expressed by an endodermal cell line can induce the differentiation of ES cells to an epiblast-like cell population in vitro (Rathjen et al. 1999). Given the need for a BM to obtain cavitation of the epiblast, we wished to determine whether the BM was necessary for all aspects of epiblast differentiation, or, alternatively, if the role of the BM was more restricted to being a requirement for polarization of CEE cells. Therefore, RT-PCR was used to determine the relative mRNA levels of FGF-5, a gene that is not expressed in the undifferentiated ICM cells before implantation, but subsequently is turned on in epiblast cells just before they become polarized to form the CEE (Haub and Goldfarb 1991). The results show that the profile of FGF-5 expression in EBs mimics that seen in vivo; only trace amounts of FGF-5 mRNA were present in undifferentiated LAMC1+/− ES cells and at early time points during EB differentiation, whereas the levels increased at later time points (Fig. 2 h). BMP4 mRNA levels were also investigated, as this signaling molecule is normally expressed in early epiblast cells before cavitation, but not in the CEE, and has been implicated in the PCD observed in EC cell-derived EBs (Coucouvanis and Martin 1999). We found that while BMP4 mRNA levels were initially similar in LAMC1−/− and control LAMC1+/− EBs, the levels were maintained in the LAMC1−/− EBs, whereas they were markedly reduced in the control EBs that had cavitated by this time (Fig. 2 i). The maintenance of BMP4 mRNA expression in the LAMC1−/− EBs, together with the FGF-5 data, indicates that the BM has no apparent effect on initial epiblast cell differentiation. The development of epiblast cells is considered to be an obligatory intermediate step in the differentiation of ES cells into embryonic cell lineages (Rathjen et al. 1999). Thus, the fact that myoblasts, endothelial and neuronal cells can differentiate from LAMC1−/− ES cells (Smyth et al. 1999) also indicates that the BM is not required for at least some aspects of epiblast cell differentiation, although it is necessary for the polarization of cells to form the CEE.
Relationship between PCD, Epiblast Cell Polarization and BMs
In control LAMC1+/− EBs, we found that the first stage of cavitation involved a loss of cell–cell contact between the polarized CEE cells and the cells positioned at their apical surface (Fig. 3 b). Subsequently, small pockets of cell debris could be identified at the apical surface of the CEE (Fig. 3 c), and, finally, a cavity became evident as the debris was phagocytosed by the cells of the CEE (Fig. 3 d). During cavitation, cell debris was restricted to the apical surface of the CEE, and was never observed in the vicinity of the unpolarized epiblast cells (Fig. 3 a). TUNEL analysis of control and age-matched LAMC1−/− EBs showed that whereas only randomly scattered TUNEL-positive cells were present in the LAMC1−/− EBs (Fig. 3 h), clusters of TUNEL-positive cells were observed exclusively at the apical surface of the CEE in control EBs (Fig. 3 f). Thus, there is a precise correlation between the development of the CEE and PCD.
To demonstrate that the BM was responsible for both the PCD and polarization of the CEE cells, the mutant phenotype of the LAMC1−/− EBs was rescued by the addition of exogenous laminin type-1. This resulted in the deposition of a BM-like sheet defined by anti–laminin type-1 and antiperlecan immunoreactivity between the outer endoderm and inner core cells of LAMC1−/− EBs (Fig. 4, a–d). In addition, CEE cells were found aligned on this sheet, and cells at the apical surface of the CEE cells had either detached or had undergone PCD, thereby forming a cavity (Fig. 4 e). The indirect effect of the BM on the PCD in the epiblast indicates either that the CEE is responsible for inducing PCD of those cells positioned at its apical surface, or, alternatively, the PCD was induced by a VE cell–derived molecule with restricted diffusion and whose synthesis was dependent upon contact of VE cells with the BM. To decide between these alternative hypotheses, we made use of the observation that a complete BM is not observed in wild-type LAMC1+/− EBs until after day 4 of differentiation (results not shown). However, by adding exogenous laminin at the start of differentiation, the rate of BM deposition was accelerated so that a BM was evident by day 2 of differentiation (Fig. 5, a–d). This observation supports our previous conclusion that laminin expression is the rate-limiting step in BM deposition (Smyth et al. 1999). In addition, histological analysis of these EBs showed that the early deposition of a BM-like sheet was accompanied by premature CEE formation and the initiation of cavitation (Fig. 5 f). However, the expression of AFP was unaffected by the absence of a BM (Fig. 2 f) and did not appear prematurely in laminin-treated EBs (results not shown). Thus, although the differentiation of the mature VE phenotype occurs independently of the BM, the PCD of epiblast cells is closely linked to the differentiation of CEE cells, which in turn is dependent upon a BM.
Our results demonstrate a novel BM-dependent mechanism for the coordination of cellular events leading to cavitation. First, extra-embryonic endodermal cells deposit a BM and also induce undifferentiated ICM cells to become epiblast cells (Spyropoulos and Capecchi 1994; Rathjen et al. 1999). This induction is independent of the BM as it also occurs in LAMC1−/− EBs. Second, the epiblast cells in contact with the BM become polarized to form the CEE, the survival of which is then dependent upon contact with the BM (Coucouvanis and Martin 1995). Finally, unpolarized epiblast cells that lie at the apical surface of the CEE undergo PCD, thereby forming the cavity. Our data show for the first time that laminin (and consequently the BM) could be both a death signal (acting indirectly) and a survival signal (acting directly). While the mechanism responsible for this indirect BM-dependent PCD remains to be determined, the fact that it is observed directly at the apical surfaces of newly polarized CEE cells is consistent with the involvement of a CEE-derived factor whose diffusion is highly restricted or, alternatively, a cell–cell contact phenomenon within the epiblast.
During organogenesis in later development, cavity formation occurs in many tissues including the exocrine glands (Hieda and Nakanishi 1997), lungs (Schuger et al. 1995), mammary glands (Humphreys et al. 1996) and kidneys (Coles et al. 1993). In the submandibular gland and lung, both initially derived from solid masses of cells, there is a strong association between development of a continuous BM, epithelialization, and cavity formation (Schuger et al. 1995; Hieda and Nakanishi 1997). Although the mechanism of cavity formation has not been investigated in most cases, PCD followed by phagocytosis of cell debris by the epithelial cells recently has been demonstrated in mammary gland (Humphreys et al. 1996) and kidney development (Coles et al. 1993). Therefore, it is likely that the ability of BMs to coordinate both epithelialization and cell death is used throughout development whenever a lumen or cavity is to be created from a solid structure.
We are grateful to Drs. Neil Smyth (University of Cologne, Germany), Marie Dziadek (University of Melbourne, Melbourne, Australia), and David Garrod (University of Manchester, Manchester, UK) for their helpful suggestions and comments on the manuscript, and we are indebted to Ms. Anne Currie, Ms. Marion Pope and Dr. Brian Boothroyd (all from the University of Liverpool) for invaluable technical assistance. We thank Dr. Rupert Timpl (Max-Planck-Institute for Biochemistry, Martinsried, Germany) for his generous gift of the antiperlecan antibodies.
This work was supported by a Postgraduate Research Studentship awarded by the Medical Research Council and a scholarship from the British Federation of Women Graduates to P. Murray.
Abbreviations used in this paper: AFP, α-fetoprotein; BM, basement membrane; CEE, columnar epiblast epithelium; EB, embryoid body; EC, embryonal carcinoma; ES, embryonic stem; ICM, inner cell mass; PCD, programmed cell death; RT-PCR, reverse transcription–PCR; TUNEL, terminal transferase–mediated biotinylated-dNTP end labeling; VE, visceral endoderm.