TGF-beta Superfamily Signaling in ES cells (Mice are Not Men)

Although mouse embryonic stem (ES) cells were first isolated and characterized over 25 years ago, the discovery of their human counterpart in 1998 led to a strong surge of interest in identifying the molecular mechanisms that mediate self-renewal and broad differentiative capacity, the key char­acteristics of ES cells.1, 2, 3

Using model systems to represent human biological processes can be challenging.
View Larger Image
Using model systems to represent human biological processes can be challenging. This holds true in the study of mouse and human stem cells, as 75 million years of divergent evolution have produced differences in the responses to molecules that regulate stem cell fate. These include key interspecies differences in the activities of TGF-beta family members.

Studies of the roles of TGF-beta superfamily members in ES cell self-renewal and differentiation have highlighted the fact that despite their fundamental similarities, there appear to be significant differences between mouse and human ES cells. For example, bone morphogenetic protein (BMP) signals have been shown (in combination with leukemia inhibitory factor, LIF) to maintain mouse ES cells in an undifferentiated, pluripotent state.4 In contrast, human ES cells can only maintain an undifferentiated phenotype by suppression of endogenous BMP signaling.5

Signaling through the activin receptor by activin or nodal has been shown to maintain human ES cells in the undifferentiated state by several groups.6, 7, 8, 9 The role of activin/nodal signaling in mouse ES cells is less clear. One report showed that although this pathway is active in undiff­erentiated mouse ES cells as assessed by phos­phor­ylation and nuclear localization of smad 2/3, inhibition by a synthetic compound that prevents smad 2/3 phosphorylation had no effect on the undifferentiated state of the cells.8 However, re­cent results from another group showed that mouse ES cell proliferation, but not pluripotency, was inhibited by the same synthetic Smad 2/3 inhibitor.10

Another TGF-beta superfamily member that plays apparently opposite roles in mouse vs. human ES cells is growth and differentiation factor 3 (GDF-3). GDF-3 falls in the BMP branch of the superfamily, and has the greatest degree of homology to Vg1, a mesoderm inducer found in Xenopus. However, GDF-3 lacks the 4th canonical cysteine residue found in TGF-beta superfamily members, a char­acteristic shared by GDF-9, BMP-15, and Lefty-A and -B.11, 12 GDF-3 is expressed in undifferentiated mouse and human ES cells, and is downregulated upon differentiation.13 In human ES cells, ectopic expression of GDF-3 results in the maintenance of markers of pluripotency even when the cells are cultured in conditions that typically promote differentiation. Thus, increased GDF-3 levels appear to promote the undifferentiated state. Paradoxically, a similar effect is seen in mouse ES cells when GDF-3 levels are decreased. When GDF-3-deficient cells are cultured in the absence of leukemia inhibitory factor (LIF), markers of pluripotency remain high. These results may be explained by the proposed mechanism by which GDF-3 exerts its functions. Levine and Hemmati-Brivanlou present evidence that GDF-3 acts as a BMP antagonist by direct binding to BMP-4.13 Since, as previously dis­cussed, BMP signals can promote pluripotency in mouse ES cells and can cause differentiation in human ES cells, lower levels of a putative BMP antagonist such as GDF-3 in mouse might enhance pluripo­tency. Similarly, higher levels of GDF-3 might neutralize BMP actions in human ES cells, again favoring pluripotency over differentiation. However, GDF-3 presents yet another paradox: a separate study in which GDF-3 null embryos were analyzed suggested that GDF-3 acts as a nodal agonist.14 In this work, GDF-3 signaling in vitro was shown to be dependent on the nodal co-receptor, cripto, and injection experiments in Xenopus showed similar effects of nodal and GDF-3.

There are a number of potential explanations for the conflicts, perhaps the most likely of which invoke the capacity of TGF-beta superfamily ligands, notably BMPs, to act as morphogens. Because morphogens are well known to exert different effects at different concentrations, variability in activity levels, delivery mechanisms, and experimental systems may be confounding the analyses. Either way, the influence of TGF-beta superfamily members on stem cell phenotype is likely to attract considerable future study.


  1. Martin, G.R. (1981) Proc. Natl. Acad. Sci. USA 78:7634.
  2. Evans, M.J. & M.H. Kaufman (1981) Nature 292:154.
  3. Thomson, J.A. et al. (1998) Science 282:1145.
  4. Ying, Q.-L. et al. (2003) Cell 115:281.
  5. Xu, R.H. et al. (2005) Nature Methods 2:185.
  6. Beattie, G.M. et al. (2005) Stem Cells 23:489.
  7. Vallier, L. et al. (2005) J. Cell Sci. 118:4495.
  8. James, D. et al. (2005) Development 132:1273
  9. Xiao, L. et al. (2006) Stem Cells 24:1476.
  10. Ogawa, K. et al. (2007) J. Cell Sci. 120:55.
  11. McPherron, A.C. & S.-J. Lee (1993) J. Biol. Chem. 268:3444.
  12. Levine, A.J. & A.H. Brivanlou (2006) Cell Cycle 5:1069.
  13. Levine, A.J. & A.H. Brivanlou (2006) Development 133:209.
  14. Chen, C. et al. (2006) Development 133:319.