Abstract

Introduction

Embryonic stem (ES) cells are pluripotent cells derived from the inner cell mass of the blastocyst that can be maintained in culture for an extended period of time without losing differentiation potential. The successful isolation of human ES cells (hESCs) has raised the hope that these cells may provide a universal tissue source to treat many human diseases. However, directed differentiation of hESCs into specific tissue types poses a formidable challenge. Protocols are currently available for only a few cell types, mostly of neural identity [1–3], and differentiation into many of the cell types derived from the paraxial mesoderm has not been reported, with the exception of a recent study indicating osteoblastic differentiation [4]. Mesenchymal stem cells (MSCs) have been isolated from the adult bone marrow [5], adipose tissue [6], and dermis and other connective tissues [7]. Harvesting MSCs from any of these sources requires invasive procedures and the availability of a suitable donor. The number of MSCs that can be obtained from a single donor is limited, and the capacity of these cells for long-term proliferation is rather poor. In contrast, hESCs could provide an unlimited number of specialized cells. In this study, we present techniques for the generation and purification of mesenchymal precursors from hESCs and their directed differentiation in vitro into various mesenchymal derivatives, including skeletal myoblasts. Our isolation method for mesenchymal precursors is the first example, to our knowledge, of efficiently deriving structures of the paraxial mesoderm from ES cells, and further highlights the potential of hESCs for basic biology and regenerative medicine.

Methods

Results

Mesenchymal differentiation of hESCs (lines H1 [WA-01] and H9 [WA-09]) [9] was induced by plating undifferentiated hESCs on a monolayer of murine OP9 stromal cells [10], in the presence of 20% heat-inactivated FBS in alpha MEM medium. OP9 cells have been previously shown to induce blood cell differentiation from mouse ES cells [11]. After 40 d of coculture, cells were harvested and sorted by FACS for CD73, a surface marker expressed in adult MSCs [5] (Figure 1A). An average of 5% CD73+ cells was obtained from the mixed culture of OP9 and differentiated hESC progeny. CD73+ cells were replated in the absence of stromal feeders on tissue culture plates and expanded in alpha MEM medium with 20% FBS for 7–14 d. We next established the membrane antigen profile of the resulting population of flat spindle-like cells. The H1- and H9-derived CD73+ cells expressed a comprehensive set of markers that are considered to define adult MSCs, including CD105(SH2), STRO-1, VCAM (CD106), CD29(integrin β1), CD44, ICAM -1(CD54), ALCAM(CD166), vimentin, and alpha smooth muscle actin (Figure 1B and ​and1C).1C). The cells were negative for hematopoietic markers such as CD34, CD45, and CD14. They were also negative for neuroectodermal, epithelial, and muscle cell markers including nestin, pancytokeratin, and desmin (data not shown). The human identity of these presumed mesenchymal cells (termed hESMPC-H1.1, -H1.2, -H1.3, -H1.4, and -H9.1) was confirmed for all experiments by immunocytochemistry for human nuclear antigen to rule out the possibility of contamination with OP9 cells (Figure S1).

Open in a separate window

Figure 1

Isolation and Characterization of hESMPCs

(A) FACS (MoFlo, Cytomation) for the isolation of CD73+ precursors (right) and isotype control (left).

(B) Flow cytometry analysis of the CD73+ hESMPC population for various markers characteristic of MSCs, including CD44, CD73, CD105, CD166, VCAM, ICAM-1, CD29, and STRO-1.

(C) Immunocytochemistry of hESMPCs for MSC markers (VCAM, STRO-1, CD73, and CD105). The cells also express vimentin and alpha smooth muscle actin. Scale bar = 50 μm.

(D) Venn diagram presenting the overlap among transcripts selectively expressed in hESMPC-H1.2, hESMPC-H9.1, and primary adult human MSCs.

To further characterize hESMPCs, we performed genome-wide expression analysis using oligonucleotide arrays (Affymetrix U133A). The expression profiles of hESMPC-H1.2 and hESMPC-H9.1 were compared with that of human primary adult MSCs. Housekeeping genes for each of the mesenchymal cell populations were eliminated by subtracting those transcripts also expressed in at least one of three independent samples of undifferentiated hESCs. Based on this analysis, 1,280 transcripts were selectively expressed in hESMPC-H1.2, 932 transcripts in hESMPC-H9.1, and 1,218 transcripts in primary adult MSCs. A remarkable overlap of 579 transcripts shared among the three mesenchymal populations was observed (Figure 1D). Using the genes that were selected in the initial filter, we performed a statistical analysis on the expression levels to determine whether the genes were expressed significantly differently in the two cell types. We used a Bayesian extension to the standard t-test [12] to assess this difference. Of the 579 genes, 412 of them were significantly different, at a false discovery rate cutoff of 0.05. The relative fold changes were also extremely large in many of the cases. We also looked at the variance of the expression levels within the cell types. For the MSCs, 94% had a coefficient of variation less than 20% for the expression (log transformed); for the ES-derived cells, 72% had a coefficient of variation less than 20%. Numerous known MSC markers were included in the list of 412 genes, such as the mesenchymal stem cell protein DSC54 (13.9-fold increase, p < 0.001), neuropilin 1 (30.4-fold increase, p < 0.001), hepatocyte growth factor (48.1-fold increase, p < 0.001), forkhead box D1 (14.8-fold increase, p < 0.001), and notch homolog 2 (2.9-fold increase, p < 0.001) . Table S2 lists the p-values from the test, the mean and standard deviation of the expression levels, and the relative fold change of all 412 genes between the two types.

Known markers of MSCs, such as mesenchymal stem cell protein DSC54, were all included within the 579 shared transcripts. These findings support the immunocytochemical data and suggest that hESMPCs and primary MSCs are highly related.

MSCs are characterized functionally by their ability to differentiate into mesenchymal tissues, such as fat, cartilage, and bone. Therefore, we tested whether hESMPCs have the same potential (Figure 2).

Open in a separate window

Figure 2

Selective Differentiation of hESMPCs into Various Mesenchymal Derivatives

(A) Adipocytic differentiation in the presence of dexamethasone, insulin, and isobutylxanthine. Adipocytic characterization by Oil Red O staining and RT-PCR analysis for PPARγ.

(B) Chondrocytic differentiation in the presence of TGF-β3 and ascorbic acid. Chondrocytic characterization by Alcian Blue staining and RT-PCR for aggrecan and collagen II.

(C) Osteogenic differentiation in the presence of β-glycerolphosphate, dexamethasone, and ascorbic acid. Osteocytic characterization by von Kossa staining and RT-PCR for bone-specific alkaline phosphatase (ALP) and bone sialoprotein (BSP).

(D) Phase-contrast image of hESMPCs and RT-PCR for the ES cell markers Nanog and Oct-4 in hESMPC-H1.1 and -H9.1 compared with undifferentiated H1 hESCs.

Scale bar = 50 μm for all panels.

Adipocytic differentiation of hESMPCs was induced under conditions described previously for primary adult MSCs [5]. Appearance of cells harboring fat granules was observed after 10–14 d in culture. After 3 wk of induction, more than 70% of the cells displayed Oil Red O+ fat granules, and PPARγ, a marker of adipocytic differentiation, was detected by RT-PCR. (Figure 2A).

Chondrocytic differentiation was achieved using the pellet culture system [5]. After 28 d in culture, more than 50% of all cells exhibited robust staining for Alcian Blue, a marker specific for extracellular matrix proteoglycans. Chondrocytic differentiation was confirmed by the gene expression of collagen II and aggrecan, two components of extracellular matrix selectively expressed by chondrocytes, using RT-PCR (Figure 2B).

Osteogenic differentiation was induced in the presence of β-glycerolphosphate [5]. Osteogenesis was demonstrated by specific staining for calcium deposition in the matrix (von Kossa, Figure 2C; or Alizarin Red, Figure S2A) and increased expression of bone-specific alkaline phosphatase and bone sialoprotein at day 28 of treatment (Figures 2C and S2B). At day 28, Alizarin Red staining was detected in approximately 70% of all cells. Throughout these studies, human adult MSCs and foreskin fibroblasts were used as positive and negative controls, respectively.

In addition to adipocytic, chondrocytic, and osteogenic differentiation, reports suggested that adult MSCs can form skeletal muscle [13]. Although generation of skeletal muscle cells from adult MSCs remains controversial, we tested whether hESMPCs exhibit this potential. Under the conditions previously described [13], hESMPC-H1.1 and -H9.1 did not yield significant numbers of MyoD+ cells after 15–20 d in culture. However, when confluent cells were maintained in culture in the presence or absence of 5-AzaC without passage for more than 21 d, expression of specific skeletal muscle markers such as MyoD and fast-switch myosin was observed (Figure 3A). More rapid myogenic differentiation was obtained in the presence of 24-h-conditioned medium from the murine myoblastic cell line C2C12 previously induced to form myotubes [14]. Direct coculture of hESMPCs with C2C12 cells led to the formation of hESMPC-derived myotubes, as visualized by expression of human nuclear antigen (Figure 3B), similar to those formed by host C2C12 cells. After 1 wk of coculture, myotubes composed of human nuclei accounted for more than 10% of the total number of human cells present, and each human myotube was composed of up to ten human nuclei. Human cell contribution to myotubes in coculture was confirmed by expression of human muscle-specific transcripts such as MyoD, myosin heavy chain IIa, and myogenin (data not shown). These data demonstrate that hESMPCs can give rise to mesenchymal derivatives typically obtained from primary adult MSCs, as well as to cells expressing markers of skeletal muscle.

Open in a separate window

Figure 3

Myogenic Differentiation of hESMPCs

(A) Immunocytochemistry for MyoD (red) and fast-switch myosin (green). RT-PCR for MyoD in human skeletal muscle as a positive control (hSM), and in hESMPC-H9.1 cells differentiated for 10 d in the presence of C2C12-conditioned medium (hESMPC).

(B) Myotube formation induced at high cell densities in the presence of C2C12 cells. Myotube characterization by immunocytochemistry for MF20 against sarcomeric myosin (green) and human nuclear antigen (hNA, red). Left panel: Control undifferentiated hESCs (H9) do not fuse with C2C12. Right panel: Under identical culture conditions, hESMPCs (line 9.1) efficiently fuse with C2C12 cells, forming myotubes containing human nuclei. RT-PCR for human specific muscle transcripts myosin heavy chain IIa (MYHC-2) and MyoD in C2C12 cells, in human skeletal muscle as positive control (huSM), and in hESMPC-H9.1 cells cocultured with C2C12 cells.

One concern for the clinical application of hESC-derived progeny in regenerative medicine is the risk of teratoma formation due to the presence of residual undifferentiated ES cells among the differentiated progeny. We did not detect markers of undifferentiated hESCs, such as Nanog [15] or Oct-4 [16], in any of the hESMPCs by RT-PCR (see Figure 2D) and immunocytochemistry (data not shown), suggesting the lack of any undifferentiated ES cells in hESMPC cultures. However, future in vivo studies are required to rule out the potential of these cells for teratoma formation.

Discussion

Previous studies have demonstrated the derivation of neural cells [1–3], hematopoietic [17] and endothelial lineages [18], and cardiomyocytes [19] from hESCs. This study presents the induction of paraxial mesoderm with the generation of multipotent mesenchymal precursors. We calculate that under these conditions a single undifferentiated hESC yields an average of one CD73+ cell at day 40 of differentiation, suggesting a balance between cell proliferation and cell selection. There were no obvious differences in marker and gene-expression profile or in differentiation behavior among the five hESMPC lines generated. However, some of the lines (e.g., hESMPC9.1) exhibited a tendency of spontaneous osteogenic differentiation after long-term propagation. Directed differentiation of hESCs into somatic stem-cell-like precursors represents a substantial advancement in harnessing the developmental potential of hESCs. The high purity, unlimited availability, and multipotentiality of hESMPCs will provide the basis for future therapeutic efforts using these cells in preclinical animal models of disease. Such in vivo studies will also be required to properly assess the safety profile of these cells. Furthermore, our system also offers a novel platform to study basic mechanisms of mesodermal induction and differentiation during early human development.

Supporting Information

Acknowledgments

Abbreviations

Footnotes

References