Collagen IV Induces Trophoectoderm Differentiation of Mouse Embryonic Stem Cells


  • Katja Schenke-Layland,

    1. Department of Medicine and Physiology, Cardiovascular Research Laboratory, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Ekaterini Angelis,

    1. Department of Medicine and Physiology, Cardiovascular Research Laboratory, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Katrin E. Rhodes,

    1. Department of Molecular, Cellular and Developmental Biology, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Sepideh Heydarkhan-Hagvall,

    1. Department of Surgery, Regenerative Bioengineering and Repair Laboratory, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Hanna K. Mikkola,

    1. Department of Molecular, Cellular and Developmental Biology, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • W. Robb MacLellan M.D.

    Corresponding author
    1. Department of Medicine and Physiology, Cardiovascular Research Laboratory, University of California Los Angeles, Los Angeles, California, USA
    • Cardiovascular Research Laboratory, UCLA School of Medicine, 675 C.E. Young Dr., MRL 3-645, Los Angeles, California 90095-1760, USA. Telephone: 310-825-2556; Fax: 310-206-5777
    Search for more papers by this author


The earliest segregation of lineages in the developing embryo is the commitment of cells to the inner cell mass or the trophoectoderm in preimplantation blastocysts. The exogenous signals that control commitment to a particular cell lineage are poorly understood; however, it has been suggested that extracellular “niche” and extracellular matrix, in particular, play an important role in determining the developmental fate of stem cells. Collagen IV (ColIV) has been reported to direct embryonic stem (ES) cell differentiation to mesodermal lineages in both mouse and human ES cells. To define the effects of ColIV on ES cell differentiation and to identify the resulting heterogeneous cell types, we performed microarray analyses and determined global gene expression. We observed that ColIV induced the expression of mesodermal genes specific to hematopoietic, endothelial, and smooth muscle cells and, surprisingly, also a panel of trophoectoderm-restricted markers. This effect was specific to collagen IV, as no trophoblast differentiation was seen on collagen I, laminin, or fibronectin. Stimulation with basic fibroblast growth factor (FGF) or FGF4 increased the number of trophoectodermal cells. These cells were isolated under clonal conditions and successfully differentiated into a variety of trophoblast derivatives. Interestingly, differentiation of ES cells to trophoblastic lineages was only seen in ES cell lines maintained on embryonic feeder layers and was caudal-type homeobox protein 2 (Cdx2)-dependent, consistent with Cdx2's postulated role in trophoectoderm commitment. Our data suggest that, given the appropriate extracellular stimuli, mouse embryonic stem cells can differentiate into trophoectoderm.

Disclosure of potential conflicts of interest is found at the end of this article.


In the early mouse preimplantation embryo, two distinct cell fates are generated that give rise to two morphologically and functionally distinct cell lineages. One is the inner cell mass (ICM), and the other is the trophoectoderm (TE) of the blastocyst. The ICM goes on to differentiate into the epiblast and overlying primitive endoderm. The epiblast gives rise to the embryo proper as well as extraembryonic mesoderm, whereas the primitive endoderm gives rise to the entire endoderm layer of the yolk sac. The remaining extraembryonic tissues, including the trophoblast layers of the placenta, are derived from the trophoectoderm [1]. Mouse embryonic stem (mES) cells derived from the ICM have virtually unlimited self-renewal potential in vitro and can give rise to all three germ layers of the embryo and to the extraembryonic mesoderm. However, when reintroduced into early embryos to form chimeras, mES cells contribute poorly to the extraembryonic endoderm and rarely, if ever, to trophoblast layers of the placenta [2, 3]. This has led to the widespread belief that mES cells have irreversibly committed to an epiblast lineage and rarely spontaneously differentiate into TE derivatives [4, 5]. Only with manipulation of master regulatory factors, such as octamer-4 (Oct4) and caudal-type homeobox protein 2 (Cdx2), could mES cells be efficiently diverted to trophoectoderm lineages. Deletion of Oct4 expression in mES cells triggers trophoblast-like differentiation [6], whereas forced expression of the earliest trophoblast-specific regulator Cdx2 can also promote differentiation of mES cells into trophoblast-like cells [7]. In contrast, numerous reports have documented the ability of human embryonic stem (hES) cells to differentiate into cells of trophoblast lineage upon appropriate stimulation without genetic manipulation [8, 9]. The reduced trophoectodermal differentiation potential of mES cells has been cited as one of the prime differences between mES and hES cells [10].

Although the genetic regulation of the ICM/TE lineage decision has received much attention [1], the exogenous cues that regulate this determination or control stem cell fate during in vitro differentiation, more generally, have not been well defined. The role that the extracellular “niche” plays on the developmental fate of pluripotent stem cells is determined not only by soluble factors and cell-cell contacts but also by the extracellular matrix (ECM). Thus, we sought to determine the effects of ECM modulation on mES cell differentiation. We were specifically interested in collagen IV (ColIV), as it has been reported to direct embryonic stem (ES) cell differentiation to mesoderm lineages, including hematopoietic (HC), endothelial (EC), and smooth muscle cells (SMC), in both mouse [11, [12]–13] and human [14] cells. It has been proposed that this represents a model of lateral mesoderm [15]; however, other mesodermal compartments, such as the allantoic mesoderm that forms parts of the developing placenta, can also give rise to these three cell types [16, 17].

In this study, we sought to identify the effects of ColIV on mES cell differentiation. We observed that ColIV-differentiated mES cells expressed mesodermal genes specific to hematopoietic, endothelial, and smooth muscle cells and, surprisingly, a panel of genes normally restricted to cells of the trophoectodermal lineage. The ability to induce TE differentiation was specific to ColIV, Cdx2-dependent, and only seen in ES cell lines maintained on embryonic feeder layers. Stimulation with basic fibroblast growth factor (bFGF) or fibroblast growth factor 4 (FGF4) increased the number of trophoectodermal cells, which were isolated under clonal conditions and successfully differentiated into a variety of trophoblast derivatives. Taken together, our results suggest that mES cells retain the ability to differentiate into trophoectoderm when given the appropriate extracellular stimuli.

Materials and Methods

Mouse ES Cells and Cell Cultures

The murine ES cell lines D3 (CRL-1934; American Type Culture Collection [ATCC], Manassas, VA,, R1 (SCRC-1011; ATCC), v6.5 C57BL/6 × 129/sv (MES1402; Open Biosystems, Huntsville, AL,, and Cdx2-deficient v6.5 C57BL/6 × 129/sv [18] (a kind gift from Rudolph Jaenisch) were cultured on mitomycin-C-treated (M0503; Sigma-Aldrich, St. Louis, primary mouse embryonic fibroblasts (MEF) in mES cell medium (knockout Dulbecco's modified Eagle's medium [Invitrogen, Carlsbad, CA,] supplemented with 15% ES cell-qualified fetal calf serum [ES-FCS; Invitrogen], 0.1 mM β-mercaptoethanol [Sigma], 2 mM glutamine [Invitrogen], 0.1 mM nonessential amino acids [Invitrogen], and 1,000 U/ml recombinant leukemia inhibitory factor [LIF; Chemicon, Temecula, CA,]) at 37°C, 5% CO2. CCE-ES cells [19] (00300; Stem Cell Technologies, Vancouver, BC, Canada, were maintained MEF-free on 0.1% gelatin (Sigma) in mES cell culture medium at 37°C, 5% CO2. The medium was changed on a daily basis. Cells were passaged every second day using 0.05% trypsin-EDTA (Invitrogen).

Differentiation Assays

For differentiation assays, the MEF-dependent mES cells were initially plated for 2 × 30 minutes on 0.1% gelatin-coated plastic flasks at 37°C, 5% CO2 to remove fibroblasts. All mES cells, including CCE-ES cells, were then transferred to either collagen type I-, collagen type IV-, laminin-, or fibronectin-coated flasks (BD BioCoat; BD Biosciences, San Diego, and cultured in α-minimum essential medium (α-MEM; Invitrogen) (supplemented with 10% ES-FCS, 0.1 mM β-mercaptoethanol, 2 mM glutamine, and 0.1 mM nonessential amino acids) without LIF at 37°C, 5% CO2. After 4 days the cells were either harvested for analysis, or trypsinized and cultivated for an additional 7 days on fibronectin-coated flasks and culture slides (BD Biosciences), in either α-MEM; vascular endothelial growth factor (VEGF) medium (endothelial growth medium-2; Cambrex, Walkersville, MD,, supplemented with 50 ng/ml VEGF [R&D Systems Inc., Minneapolis,]); or platelet-derived growth factor-BB medium (PDGF-BB) (smooth muscle growth medium-2; Cambrex, supplemented with 10 ng/ml PDGF-BB [R&D Systems Inc.]), at 37°C, 5% CO2.

Isolation, Maintenance, and Differentiation of TE Cells Under Clonal Conditions

To determine whether TE cells respond to growth factors, D3-ES cells were additionally cultured on collagen type I and type IV in either α-MEM, α-MEM supplemented with 10 ng/ml bFGF (Invitrogen) [5], or trophoblast stem cell (TS) medium, containing 25 ng/ml FGF4 (Peprotech, Rocky Hill, NJ, and 1 μg/ml heparin (Sigma), with 70% of the medium being MEF-preconditioned [20]. After 1, 2, 3, or 4 days, the cells were harvested for analysis. Cloning of TE cells was carried out by limited dilution. Briefly, D3-ES cells were cultured for 2 days on ColIV in TS medium, trypsinized, resuspended to a density of 10 cells per milliliter, and recultured in TS medium on 0.1% gelatin-coated 96-well plates (BD Biosciences). Cell clonality was confirmed by phase-contrast microscopy using a Zeiss Axiovert 200 microscope (Carl Zeiss, Jena, Germany, The cloned cells were subcultured in triplicate, and, 4 days after subculturing, immunofluorescence analysis with an anti-Cdx2 antibody was carried out (refer to Antibodies and Immunocytochemistry for more detailed information). The cloned Cdx2-expressing cells were expanded in TS medium and either analyzed as undifferentiated TE cells or cultured for an additional 6 days in differentiation medium (TS medium without supplement of FGF4, heparin, and MEF-conditioned medium) prior to analysis.


Primary antibodies used include (a) rabbit polyclonal antibodies: anti-α-smooth muscle actin (SMA) (ab5694 [1:400]; Abcam, Cambridge, U.K.,, anti-vimentin (ab7783 [1:500]; Abcam), anti-mouse connexin 31 (Cx31) (CX31-A [1:100]; Alpha Diagnostics, San Antonio,, and anti-cow cytokeratin (Z0622, 1:1,000; Dako North America Inc., Carpinteria, CA,; (b) mouse monoclonal antibodies anti-Cdx2 (MU392-UC [1:20]; BioGenex, San Ramon, CA, and anti-Cadherin 3/P-Cadherin (clone 56C1 [1:20]; Lab Vision, Fremont, CA,; as well as (c) a rat monoclonal antibody anti-mouse CD31 (PECAM-1, catalog number 550274 [1:50]; BD Pharmingen, San Diego, Secondary antibodies included Alexa Fluor 488- and Alexa Fluor 594-conjugated goat-anti-mouse IgG (H+L); Alexa Fluor 488-conjugated goat-anti-rabbit IgG (H+L); Alexa Fluor 488-conjugated goat-anti-rat IgG (H+L) (1:250; all from Molecular Probes, Carlsbad, CA, To visualize the F-actin cytoskeleton, cells were stained using Alexa Fluor 594 phalloidin (Molecular Probes). For counterstaining of cell nuclei, 4,6-diamidino-2-phenylindole (DAPI) (Sigma) was added to the final phosphate-buffered saline (PBS) washing. Cell staining without primary antibodies served as controls. Bright-field images were acquired using the Zeiss Axiovert 200 microscope.


Prior to immunocytochemical staining, all cells were washed with 1× PBS (Invitrogen), fixed for 10 minutes in 4% paraformaldehyde (Sigma), and rinsed twice in PBS. Cells were then permeabilized using 1% Triton X-100 (Sigma) for 30 minutes and subsequently incubated in serum blocking buffer (2% goat serum [Chemicon], 1% bovine serum albumin [Sigma], 0.1% cold fish skin gelatin [Sigma], 0.1% Triton X-100, and 0.05% Tween 20 [Sigma] in PBS). After 3× rinsing with 0.05% Tween 20, cells were incubated with the primary antibody (diluted as specified earlier) overnight at 4°C followed by 2× washing with 0.05% Tween 20. For secondary antibody detection, the appropriate Alexa Fluor-conjugated antibodies were incubated at a 1:250 dilution in PBS for 30 minutes at room temperature in the dark, either alone or in combination with Alexa Fluor 594 phalloidin (1:40). After secondary antibody incubation, the cells were washed 3× in PBS and incubated in a DAPI-PBS solution followed by 3× washes with PBS. In a final step, the cell culture chambers were removed from the slides (according to the manufacturer's protocol), and slides and coverslips were mounted using ProLong Gold antifade mounting medium (Molecular Probes). The slides were then stored in the dark overnight at 4°C prior to imaging. Images were acquired using a confocal TCS SP2 AOBS laser-scanning microscope system (Leica Microsystems Inc., Exton, PA, with 40× (1.3 numerical aperture [NA]) and 63× (1.4 NA) oil-immersion objectives. Images were processed with Adobe Photoshop 7.0 (Adobe Systems Inc., San Jose, CA,

RNA Extraction, cDNA Synthesis, and Semiquantitative Reverse Transcription-Polymerase Chain Reaction

Total RNA was extracted from placenta tissues at day 12.5 of gestation (positive control) as well as from harvested cells by a modification of the acid-guanidinium-phenol-chloroform method (TRIzol Reagent; Sigma) as per manufacturer's instructions. Precipitated RNA was resuspended in RNase-free water and subjected to an additional RNA purification step to remove possible genomic DNA contamination (RNeasy Plus Mini Kit; Qiagen, Valencia, CA, before final storage at −80°C. First strand cDNA was generated from 2 μg of total RNA by using the Omniscript Reverse Transcriptase (RT) Kit (Qiagen) as per manufacturer's instructions. All samples, along with the corresponding “no- RT” control (RNA) to confirm the absence of contaminating genomic DNA, were subjected to polymerase chain reaction (PCR) and carried out using 2.5 units of Taq DNA polymerase, 10× PCR buffer, 2.5 mM MgCl2, 200 μM dNTP, Q-solution (Qiagen), 0.2 μM gene-specific forward, and 0.2 μM reverse PCR primers. The sequences of each specific primer set, including their annealing temperatures and cycles, are listed in the supplemental online Table 1. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA was used as an internal control. PCR reactions were performed under the following conditions: 94°C denaturation for 30 seconds, specific primer annealing temperature (Supplemental Table 1) for 45 seconds, and primer extension at 72°C for 45 seconds (all except Eomesodermin [Eomes]); 94°C denaturation for 30 seconds, 50°C for 30 seconds, 68°C for 2 minutes, and primer extension at 68°C for 7 minutes (Eomes). The Oct4, Acta2, Cald1, Cdh5, and Vwf RT-PCR primer sets were obtained from SuperArray Bioscience Corporation (Frederick, MD, and used with the ReactionReady HotStart PCR master mix (including an internal normalizer) following the manufacturer's instructions. The resultant PCR products were resolved through 2% agarose gels stained with ethidium bromide.

Real-Time RT-PCR

Real-time PCR was conducted using the ABI PRISM 7700 Sequence Detection System, TaqMan (Applied Biosystems, Foster City, CA, The Cdx2 and GAPDH primer sets utilized for real-time quantification were obtained from Qiagen (QuantiTect Primer Assay) and used following the manufacturer's instructions. PCR amplicons were detected by fluorescent detection of SYBR Green (QuantiTect SYBR Green PCR Kit, Qiagen). Cycling conditions were as follows: 95°C for 15 minutes followed by 40 cycles at 94°C for 15 seconds, 55°C for 30 seconds, and 72°C for 30 seconds. For statistical analysis, all data are presented as mean ± SD. Significant differences between the samples were assessed by analysis of variance with Tukey's multiple comparison test. We defined p values less than .05 as statistically significant.

Gene Expression Analysis

RNA samples from three 75-cm2 flasks (each flask 3–5 × 106 cells) of (a) undifferentiated D3-ES cells and (b) D3-ES cells, cultured for 4 days in α-MEM on collagen type IV, were analyzed at the UCLA Illumina Microarray Laboratory. Briefly, biotinylated cRNA was prepared using the Illumina RNA Amplification Kit (Ambion Inc., Austin, TX, starting with 100 ng of total RNA. Samples were purified and used for hybridization on a Sentrix MouseRef-8 Expression BeadChip System (Illumina Inc., San Diego, containing approximately 24,000 reference-sequence-based probe sequences per array. Scanning was performed according to the Illumina BeadStation 500× manual. Microarray raw data were analyzed using BeadStudio version Software was provided by the manufacturer. Differential expression analysis was selected to quantify gene expression intensity values as well as to determine changes of the gene expression levels between undifferentiated ES cells (reference group) and ColIV-differentiated ES cells. To filter out nonspecific signal intensities, local background subtraction was performed. Only genes with intensities >0.99 were selected for analysis. A differential score of >13.0 demonstrated that gene expression from ColIV-differentiated ES cells had changed significantly when compared with genes of undifferentiated mES cells. A summary of the upregulated cardiovascular-, HC-, and TE-specific genes in D3-ES cells cultured on ColIV, with a more than 1.5-fold difference, is presented in Table 1. The full microarray analysis is provided as a supplemental data Excel (Microsoft Corporation, Redmond, WA, spreadsheet.

Table Table 1.. Summary of upregulated (>1.5-fold) cardiovascular-, hematopoietic-, and trophoectoderm-specific genes in D3-embryonic stem cells cultured on collagen IV
original image

Fluorescence-Activated Cell Sorter Analysis

Cells were detached using 0.05% trypsin-EDTA, pelleted by centrifugation, washed in PBS, and fixed using the 4% paraformaldehyde-containing BD Cytofix solution (BD Pharmingen). For double labeling, cells were first processed for surface staining using an anti-Cadherin 3 antibody (see Antibodies), followed by permeabilization using BD Perm/Wash buffer (BD Pharmingen) according to the manufacturer's instructions and staining with the anti-Cdx2 antibody as described earlier (see Antibodies). Nonspecific isotype-matched IgGs (Santa Cruz Biotechnology Inc., Santa Cruz, CA, served as controls. Secondary detection was done using appropriate Alexa Fluor 488-, and Alexa Fluor 647-conjugated antibodies. All analyses were performed using a BD LSR2 flow cytometer (BD Biosciences). FCS files were exported and analyzed using the FlowJo 8.3.3 software (Tree Star Inc., Ashland, OR,


Induction of Hematopoietic, Endothelial, and Smooth Muscle Markers in ColIV-Differentiated mES Cells

It has been reported that mES cells cultured on ColIV and stimulated with growth factors differentiate into cultures of EC, SMC, and HC [11, [12]–13]. To confirm these results, we plated mES cells on ColIV for 4 days and then replated the disaggregated cells in the presence or absence of VEGF and PDGF-BB. After 48 hours, ColIV-differentiated mES cells developed a characteristic colony-like morphology that was visible in all cultures, independent of the culture medium (Figs. 1, 2). Clusters of cells with a colony-like structure were surrounded by cuboidal cells with a boundary layer of spindle-shaped cells (Fig. 2A). To better characterize the identity of the cells within each of these three areas, we performed immunostaining with a panel of cell-type-specific antibodies. Cells within the clusters were predominantly positive for HC (data not shown) and EC markers including CD31 (Fig. 2Bg–2Bi); however, some of the inner, more compact cells continued to express markers such as Nanog and Oct4 (data not shown), likely representing a small fraction of undifferentiated mES cells. The spindle-shaped cells at the periphery expressed vimentin (Fig. 2Bd–2Bf) as well as SMC markers including basic calponin, h-caldesmon, smooth muscle-myosin (data not shown), and α-SMA (Fig. 2Ba–2Bc), indicating a SMC phenotype. The flattened cuboidal-shaped cells in the border zone did not express markers consistent with EC, SMC, or HC, and some of them showed an enlarged morphology (Fig. 1B–1D) with multiple nuclei (Fig. 2A).

Figure Figure 1..

Phase-contrast images show D3-embryonic stem cells cultured for 4 days on collagen IV in α-minimal essential medium (MEM) (A) followed by culture for 7 days on fibronectin in α-MEM (B), vascular endothelial growth factor-supplemented endothelial growth medium (C), or platelet-derived growth factor-BB-supplemented smooth muscle growth medium (D). White arrows point to cells with giant cell characteristics.

Figure Figure 2..

ColIV-differentiated embryonic stem cells express hematopoietic, endothelial, and smooth muscle cell markers. (A): Extracellular matrix-induced differentiation of mouse embryonic stem cells leads to a highly heterogeneous cell population. F-actin staining (red) shows clearly the different cytoskeleton dimensions of a variety of cells, which grow in a characteristic colony-like pattern: large cuboidal cells surround central cell clusters of round cells, and in the adjacent regions of the cuboidal cells elongated, spindle-shaped cells can be found. Within the cuboidal cells some large flattened cells with polyploid nuclei (∗) are visible. (B): Spindle-shaped, fibroblast-like cells are positive when stained with antibodies against α-SMA (green) (Ba–Bc) and vimentin (green) (Bd–Bf), indicating a smooth muscle cell phenotype, whereas the majority of cells situated within the cell clusters are positive for the endothelial cell marker CD31 (green) (Bg–Bi). Cell nuclei are stained with 4,6-diamidino-2-phenylindole (blue). Abbreviations: cc, cuboidal cells; MEM, minimal essential medium; PDGF-BB, platelet-derived growth factor-BB; rc, round cells; sc, spindle-shaped cells; SMA, smooth muscle actin; VEGF, vascular endothelial growth factor.

Mouse ES Cells Cultured on ColIV Differentiate into Trophoblast Lineages

To ascertain the identity of the cuboidal cells induced by ColIV, microarrays were used to characterize global gene expression. In addition to the expected expression of cardiovascular- and HC-specific genes, ColIV-differentiated D3-ES cells expressed a panel of genes normally restricted to trophoblast cells (Table 1). Although mES cells have been thought to be incapable of differentiating into trophoblasts, the large flattened cuboidal-shaped cells with enlarged and multiple cell nuclei (Fig. 1B–1D; Fig. 2A) demonstrated morphological characteristics that are classic for trophoblast giant cells and were reminiscent of those previously reported in studies using trophoblast stem cells [20] or forced Cdx2 expression in mES cells [7].

To confirm that the cuboidal cells within the highly heterogeneous population of differentiating D3-ES cells were indeed trophoblast-like cells, we surveyed them for expression of known TE markers, including cytokeratin [21], Cx31 [22], P-Cadherin (Cadherin 3), and Cdx2 [23] by immunofluorescence staining (Figs. 3, 4A). Double immunocytochemical labeling using these markers demonstrated cells positive for cytokeratin and Cadherin 3, cytokeratin and Cdx2, and Cx31 and Cdx2 within the zone of cuboidal cells and at the outer edge of the cell clusters (Fig. 4A). It is possible that trophoblast cells closer to the compact cell clusters represent the more primitive trophoblast stem cells, since TE cells with a more differentiated phenotype such as giant cells appeared to grow out and were further away from the cell clusters, a phenomenon also seen in previous studies using trophoblast stem cells [20].

Figure Figure 3..

Collagen IV-differentiated embryonic stem cells cultured in α-MEM (A–E), VEGF-supplemented endothelial growth medium (F–J), or PDGF-BB-supplemented smooth muscle growth medium (K–O) express markers of trophoectodermal cells. Images show cytoskeleton-staining (red) and positive staining (green) for cytokeratin (B, G, L), Cx31 (C, H, M), Cadherin 3 (D, I, N), and Cdx2 (E, J, O). Staining using the secondary antibody alone (A, F, K) serves as control. Staining with 4,6-diamidino-2-phenylindole was performed to show cell nuclei (blue). Abbreviations: MEM, minimal essential medium; PDGF-BB, platelet-derived growth factor-BB; VEGF, vascular endothelial growth factor.

Figure Figure 4..

ColIV-differentiated cells express markers of trophoblast lineages. (A): Immunolabeling of differentiating mES cells identifies double-positive cells for cytokeratin (green)/Cadherin 3 (red) (Aa, Ab); cytokeratin (green)/Cdx2 (red) (Ac, Ad); and Cx31 (green)/Cdx2 (red) (Ae, Af). (B): D3-embryonic stem cells express a panel of trophoectoderm-related genes during the course of extracellular matrix-induced differentiation. Abbreviations: ColIV, collagen IV; d, day; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MEF, mouse embryonic fibroblast; MEM, minimal essential medium; mES, mouse embryonic stem; PDGF-BB, platelet-derived growth factor-BB; VEGF, vascular endothelial growth factor.

Several groups have proposed detailed gene profiles that allow identification of specific trophoblast subtype lineages [24, 25]. Thus, we sought to identify the expression of trophoblast-restricted genes in ColIV-differentiated mES cells using semiquantitative RT-PCR (Fig. 4B). Murine placenta at day 12.5 of gestation was included as positive control for most of the markers along with MEF and undifferentiated D3-ES cells as negative controls. mES cells cultured for 4 days on ColIV showed a downregulated expression of the pluripotency-associated genes Nanog and Oct4 (data not shown). Expression of the brachyury gene, a transcription factor essential for the genesis and maintenance of mesoderm and notochord [26], as well as Nodal, a growth factor implicated in regulating trophoblast differentiation and placental development, were increased. Expression of a panel of TE-restricted genes that identify trophoblast subtypes was upregulated in ColIV-differentiated ES cells. These included markers for trophoblast stem cells (Cdx2, Eomes, estrogen-related receptor [Esrrβ, Err2], spongiotrophoblast cells (mammalian achaete-scute homologous protein-2 [Mash2], trophoblast specific protein alpha [Tpbp/-4311]), trophoblast giant cells (placental lactogen [PL-1], Homo sapiens heart and neural crest derivatives expressed 1 [Hand1]), and labyrinthine trophoblasts (extraembryonic spermatogenesis homeobox 1 [Esx1], glial cells missing homolog 1 [Gcm1], and distal-less homeobox 3 [Dlx3]) (Fig. 4B).

When the ColIV-differentiated mES cells were replated in the presence of VEGF- or PDGF-BB-supplemented media, a similar expression pattern was observed, although relative differences were seen (Fig. 3; Fig. 4B). Expression levels of the mesodermal marker brachyury and early trophoblast markers Cdx2 and PL-1 declined, accompanied by a corresponding increase in expression of the spongiotrophoblast-specific marker Tpbp/-4311 and labyrinthine trophoblast marker Dlx3. Furthermore, cells that were cultured in PDGF-BB-supplemented medium showed enhanced expression of Gcm1 and Mash2, which was not detectable in cells cultured in VEGF-supplemented medium. Nodal was expressed in all cells but seemed to be upregulated in cells cultured in VEGF- and PDGF-BB-supplemented medium. The expression levels of Eomes, Esrrβ, Hand1, and Esx1 were constant across the different culture conditions.

TE Differentiation Is Specific to ColIV and Is Restricted to MEF-Dependent mES Cells

To determine if TE differentiation of mES cells is specific to ColIV, we compared trophoblast gene expression in ColIV-differentiated cells with D3-ES cells cultured on various ECM proteins including collagen type I, laminin, and fibronectin (Fig. 5A, 5B). Expression of Cdx2, Eomes, PL-1, and Tpbp/-4311 was only seen in ColIV-differentiated mES cells. Expression of Cdx2 was over sixfold higher on ColIV when compared with undifferentiated mES cells or ES cells differentiated on other ECM proteins (p < .05; Fig. 5B). In contrast, expression of less specific markers such as Esrrβ, Esx1, and Gcm1 was seen in all samples, although Esx1 and Gcm1 were decreased on collagen I (Fig. 5A). These results indicate that TE differentiation of mES cells is indeed specific to collagen type IV.

Figure Figure 5..

Trophoectoderm differentiation is specific to ColIV and is restricted to embryonic stem (ES) cells that were maintained on MEF feeder layers. (A): Gene expression analysis of D3-ES cells cultured on a panel of extracellular matrix (ECM)-proteins. The expression of trophoblast lineage genes including Cdx2, Eomes, PL-1, and Tpbp/4311 is exclusively seen in cells cultured for 4 days on ColIV. (B): Real-time reverse transcription-polymerase chain reaction analysis of Cdx2 expression in D3-ES cells cultured on ColIV compared with undifferentiated ES cells as well as ES cells cultured on other ECM proteins (∗, p < .05 vs. mouse ES cells, MEF, Coll I, laminin, and Fibro.). (C): Expression of TE-restricted markers is observed in both MEF-dependent ES cell lines D3 and R1, but is not seen in the MEF-free CCE-ES cells. Abbreviations: ColIV, collagen IV; Coll I, collagen I; d, day; Fibro., fibronectin; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MEF, mouse embryonic fibroblast; mESC, mouse embryonic stem cell.

One potential explanation for the lack of previous reports of mES cells cultured on ColIV differentiating into TE cells may be that feeder-free mES cell lines were used, which may have a more limited differentiation potential [11, 12]. Data from a previous study showed that the heterogeneous population of differentiated J1 and 210-58 ES cells that were maintained on MEFs contained a small population of cells with TE potential [5]. To determine whether the ability of mES cells to differentiate into cells of the TE lineage is a general phenomenon or indeed restricted to specific cell lines, we examined the ability of a second MEF-dependent cell line, R1-ES cells, as well as MEF-free CCE-ES cells, to differentiate into trophoblast cells (Fig. 5C). As shown, only mES cells that had been maintained on MEF layers (D3 and R1) could be induced to differentiate into TE cell types. CCE-ES cells, differentiated on ColIV and further on fibronectin, appeared to be morphologically distinct when compared with ColIV-differentiated D3-ES and R1-ES cells (data not shown), showing no expression of TE markers including Cdx2 or Tpbp/-4311 (Fig. 5C). In contrast, markers for endothelial and smooth muscle differentiation were unaffected (supplemental online Fig. 1).

ColIV-Induced Trophoblast Differentiation Is Cdx2-Dependent

To determine if TE differentiation of mES cells on ColIV is Cdx2-dependent, we cultured Cdx2-deficient ES cells [18] and the wild-type (WT) parental ES cell line for 4 days on ColIV and analyzed trophoblast gene expression and cell morphology (Fig. 6). No expression of Cdx2, Eomes, or Tpbp/4311 was observed in Cdx2-deficient cells, and PL-1 was only weakly expressed (Fig. 6A). The morphological characteristics of WT cells cultured on ColIV were distinct from those of Cdx2-deficient ES cells (Fig. 6B). ColIV-differentiated WT ES cells showed a more heterogeneous cell population (Fig. 6Ba) when compared with ColIV-differentiated Cdx2-deficient ES cells (Fig. 6Bb). Double immunocytochemical labeling of WT ES cells revealed the presence of cells positive for cytokeratin and Cadherin 3, cytokeratin and Cdx2, and Cx31 and Cdx2 (Fig. 6Ca, 6Cc, 6Ce). Although a very limited number of cells positive for Cadherin 3, cytokeratin, or Cx31 were detectable within the ColIV-differentiated Cdx2-deficient ES cells (Fig. 6Cb, 6Cd, 6Cf), these cells showed a very different morphology when compared with the WT cells. Cells positive for Cdx2 were never observed within ColIV-differentiated Cdx2-deficient ES cells (Fig. 6Cd, 6Cf).

Figure Figure 6..

ColIV-induced differentiation into trophoectoderm cells is Cdx2-dependent. (A): Semiquantitative reverse transcription-polymerase chain reaction showing that trophoblast-restricted markers are exclusively expressed in WT cells when cultured on ColIV. No expression of trophoectoderm-specific markers could be detected in Cdx2-deficient embryonic stem (ES) cells. (B): Bright-field microscopy reveals morphological differences between WT (A) and Cdx2-deficient (B) ES cells cultured for 4 days on ColIV. White arrows point to flattened cuboidal-shaped cells seen in the WT but not in the Cdx2-deficient ES cell cultures. (C): Immunocytochemical staining of ColIV-differentiated WT (Ca, Cc, Ce) and Cdx2-deficient (Cb, Cd, Cf) ES cells. WT ES cells show double-positive cells for cytokeratin (green) and Cadherin 3 (red) (Ca); cytokeratin (green) and Cdx2 (red) (Cc); as well as Cx31 (green) and Cdx2 (red) (Ce). In contrast, Cdx2-deficient ES cells are only positive for Cadherin 3 (red) (Cb). Some single cells are positive for cytokeratin (green) (Cb, Cd) and Cx31 (green) (Cf). No Cdx2-positive cells (red) (Cd, Cf) were detectable. Abbreviations: ColIV, collagen IV; d, day; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; WT, wild-type.

ColIV-Differentiated TE Cells Are FGF-Responsive and Can Be Grown Under Clonal Conditions

To determine if the TE differentiation on ColIV is modulated by growth factors known to promote trophoblast cells, cells were cultured for 1, 2, 3, and 4 days on collagen type I and type IV in either α-MEM, bFGF-supplemented α-MEM, or TS medium containing FGF4 and heparin. Fluorescence-activated cell sorter (FACS) analysis showed significantly higher numbers of Cdx2-expressing cells in the ColIV-differentiated cultures, grown in either bFGF-supplemented α-MEM or in TS medium, when compared with α-MEM cultures without supplements (Fig. 7A). In ColIV-α-MEM cultures, the Cdx2-positive population represented 1.07% of cells on day 1 and increased to 2.62% by day 4. Addition of bFGF to the medium significantly increased the number of Cdx2-positive cells from 3.01% to 6.51% (day 1 to day 4), whereas the FGF4-/heparin-supplemented TS medium resulted in a further increase (3.83% to 15.71% [day 1 to day 4]) of Cdx2-expressing cells (Fig. 7A). No Cdx2-positive cells and no FGF-responsiveness was observed in cultures grown on collagen type I (Fig. 7A). Analysis of gene expression in ColIV-/TS medium-differentiated mES cells revealed a progressive downregulation of Oct4 accompanied by an upregulation of trophoblast stem cell markers including Cdx2, Eomes, and Esrrβ (Fig. 7B).

Figure Figure 7..

ColIV-differentiated TE cells are FGF-responsive and can be grown under clonal conditions. (A): Fluorescence-activated cell sorting (FACS) analysis of Cdx2-expressing cells within the ColIV- and Coll I-differentiated cultures (∗, p < .05 vs. α-MEM; †, p < .05 vs. bFGF). (B): Gene expression pattern of the heterogeneous population of ColIV-differentiated D3-embryonic stem cells, cultured for 1, 2, 3, and 4 days in TS medium. (C): Histogram showing the FACS profile of clonally derived TE cells that were cultured in TS medium. (D): Bright-field and immunofluorescence images of clonally derived, undifferentiated TE cells in TS medium (Da–Dc) and after 6 days of culture in differentiation medium (Dd–Df). Immunolabeling demonstrates that the majority of the undifferentiated TE cells are positive (red) for Cdx2 (Db) and Cadherin 3 (Dc), whereas only a few Cdx2-positive cells were detectable within the differentiated cells (De). Cadherin 3 was expressed by most of the differentiated cells (Df). Cell nuclei are stained with 4,6-diamidino-2-phenylindole (blue). (E): Reverse transcription-polymerase chain reaction analysis of three isolated TE cell clones at day 0 of differentiation in TS medium and after a 6-day exposure to differentiation medium. Abbreviations: APC, allophycocyanin; bFGF, basic fibroblast growth factor; ColIV, collagen IV; Coll I, collagen I; d, day; Diff., differentiation; FITC, fluorescein isothiocyanate; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MEF, mouse embryonic fibroblast; MEM, minimal essential medium; mES, mouse embryonic stem; TE, trophoectoderm; TS, trophoblast stem cell.

To ensure that we had committed but undifferentiated TE cells, Cdx2-positive clones were isolated after 2 days in culture on ColIV by limited dilution. Isolated clones were expanded and subsequently analyzed (Fig. 7C–7E). FACS analysis demonstrated that 99.7% and 98.7% of the clonally expanded populations retained Cdx2 and Cadherin 3 expression, respectively, suggesting that these clones most likely represent trophoblast stem cells (Fig. 7C). These FACS results were confirmed by immunocytochemical staining (Fig. 7Da–7Dc). Gene expression analysis of the clonally derived, undifferentiated Cdx2 cells showed expression of trophoblast stem cell genes including Cdx2, Eomes, and Esrrβ, whereas markers for differentiated TE cells were absent (Fig. 7E). However, when these clones were cultured in differentiation-promoting conditions by withdrawal of FGF4 and heparin, they differentiated into a variety of TE derivatives, as showed by increased expression of markers for giant cells, spongiotrophoblast, and labyrinth and a concomitant decrease in expression of trophoblast stem cell genes (Fig. 7E). Further analysis of these cultures demonstrated the presence of cells with trophoblast giant cell morphology (Fig. 7Dd). Immunocytochemistry showed a downregulated expression of Cdx2; however, differentiated TE cells were still positive for cytokeratin (data not shown) and Cadherin 3 (Fig. 7Dd–7Df). Expression of markers for endothelial or smooth muscle differentiation was not observed in these cultures at any time point, suggesting that the progenitors that give rise to TE derivatives do not produce vascular cells (supplemental online Fig. 2).


Our data suggest that mES cells retain the capacity to differentiate into cells of the TE lineage given the correct extracellular signals. Previous dogma held that mES cells were incapable of differentiating into TE derivatives [1, 27]; however, this information was based on in vivo data, where it was found that mES cells rarely give rise to the trophoblast layers of the placenta [2, 3]. This lack of contribution to TE cells in the developing embryo may simply represent a preferred or default differentiation pathway of mES cells. Consistent with this hypothesis, recent reports have documented that genetic manipulation can divert mES cells to TE lineages, which suggests that there might be some plasticity to the commitment of mES cells to epiblast-restricted lineages [5, [6]–7]. Thus, reported differences in the differentiation potential among species of mES cells may be more related to limitations in determining the correct extracellular microenvironment than true biological distinctions.

Although relative differences have been reported between mouse and human ES cells, including morphology, signaling pathways mediating self-renewal, and expression of certain genes [27, [28]–29], the major purported difference of developmental significance between human and mouse ES cells was that hES cells can give rise to TE-like cells spontaneously[8, 29] or with bone morphogenetic protein 4 (BMP-4) treatment [9]. This has led to the belief that the regulatory pathways for differentiation of early embryonic structures such as the placenta and extraembryonic membranes were distinct in mouse and human embryos. Although the specific role of Oct4, Cdx2, and other lineage-associated genes in regulating TE differentiation in human ES cells remains to be determined, similarities to mouse developmental programs have already been seen, as downregulation of Oct4 in human ES cells leads to upregulation of Cdx2 and trophoblast markers [30]. It will be important to determine if hES cells spontaneously differentiate into TE cells on ColIV. If so, this would provide an in vitro model to directly compare mouse and human ES-differentiation potential and resolve whether the differences reported in developmental usage of some factors in early lineage specification [31] represent true biological differences versus specious differences related to technical disparities.

Cells of the trophoblast lineage are essential for the establishment and maintenance of the placenta, and a number of specialized trophoblast subtypes have evolved to address specific physiological needs. Using a lineage-specific gene profile, we were able to identify markers of all lineages of placental trophoblast cells including trophoblast stem cells (Cdx2, Id2, Eomes, Esrrβ), trophoblast giant cells (PL-1, Hand1), spongiotrophoblasts (Tpbp/-4311), and cytotrophoblasts (Esx1, Gcm1, and Dlx3) [24, 25]. A similar potential to differentiate into multiple trophoblast lineages was also seen in human ES cells [9]. Interestingly, our results and others have documented the presence of HC, as well as EC and SMC, in mES cells differentiated on ColIV [11, [12]–13]. These cell types are all present in the developing placenta along with ColIV [32, 33]. The importance of the placenta in hematopoietic development has been recently appreciated, as placental hematopoietic stem cell (HSC) localization appears in parallel with the aorta-gonad-mesonephros region before HSCs are found in circulation or have colonized the fetal liver [34]. This placental HS cell colonization culminates in a rapid expansion of a definitive HSC pool. The ColIV-based ES cell differentiation model presented here could provide a unique opportunity to study the HSC-promoting properties of the placental niche and the role of trophoblasts in this process specifically. Furthermore, if trophoblast cells indeed support self-renewal of HSC, and if hES cells demonstrate a similar response on ColIV, this in vitro differentiation model might represent a novel culture system to generate HSCs for stem cell-based therapies in the future.

ColIV is detected in the mouse embryo as early as the 32–64-cell stage [35] and is highly expressed in the developing placenta in both fetal and maternal layers [32]. ColIV likely plays a dual role in embryonic development both as an inducer of differentiation and as a structural protein. ColIV is essential for basement membrane stability and provides a scaffold that integrates other components such as laminins, nidogens, and perlecan into a highly organized supramolecular structure. Consistent with this role, ColIV-null mutant mice die at midgestation with placental defects [36]. The mechanisms whereby ColIV might be directing ES cells to differentiate into trophoblast are speculative; however, integrin-dependent mechanisms are known to regulate BMP-4 expression in ES cells. BMP-4, a member of the transforming growth factor-beta superfamily, was capable of inducing the differentiation of human ES cells to trophoblast [9].


The data we have provided challenge the widely held notion that mouse ES cells are incapable of spontaneously differentiating into TE without genetic manipulation. We could identify the exact extracellular stimuli and the appropriate microenvironment that are necessary for mES cells to successfully differentiate into TE cells. This in vitro model of ColIV-induced TE differentiation should prove useful, both as a tool for studying the differentiation and function of early trophoblasts as well as further elucidating the specific developmental role of the large number of factors on TE differentiation that, when genetically deleted, result in abnormal placental development and early embryonic lethality [25].

Disclosures of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


The authors would like to thank Y. Bukshpun and E. Butylkova for their technical assistance and Rudolph Jaenisch (Whitehead Institute, Cambridge, MA) for providing the Cdx2-deficient ES cells. This work was supported by the Laubisch and Glazer Funds as well as the Deutsche Forschungsgemeinschaft (Sche701/2-1, 3-1 [K.S.-L.]), NIH R21DK069659 (H.K.M.), R01HL70748, and AHA 0340087N Grants (W.R.M.).