The importance of knowing your identity: Sources of confusion in stem cell biology


  • Markus Grompe

    Corresponding author
    1. Department of Molecular and Medical Genetics, Oregon Health & Sciences University, Portland, OR
    • Dept. of Molecular and Medical Genetics, L103, Oregon Health & Sciences University, 3181 SW Sam Jackson Pk Rd., Portland, OR 97239
    Search for more papers by this author
    • fax: 503-494-6886.

  • See Article on Page 258

Recent progress in stem cell biology has created much enthusiasm and hope about the therapeutic potential of such cells. While this resurgence in interest applies to multiple tissues, some of the most promising advances pertain to our favorite organ, the liver. Human clinical trials have demonstrated some clinical benefits of hepatocyte transplantation.1 Furthermore, the liver is similar to the hematopoietic system in that extensive tissue replacement (repopulation) can be achieved when the transplanted cells have a selective growth advantage.2 Spontaneous liver repopulation occurs in humans and multiple preclinical animal models of therapeutic liver reconstitution exist.3–9 The procurement of human hepatocytes for transplantation is difficult,10 and, for this reason, much effort has been placed into the isolation and characterization of liver stem cells.

Recently, another compelling reason for the interest in hepatic stem cells has emerged: liver stem cells may be useful for the treatment of type 1 diabetes. Many observations suggest that there is a tight relationship between pancreatic progenitors/stem cells and hepatic progenitors. During embryonic development, both organs originate from the same region of the ventral foregut, suggesting the existence of a common hepatopancreatic precursor cell in this location. In addition, multiple examples of pancreatic-hepatic cell fate switches in adult mammals exist.11 Now, several studies have indicated that hepatic progenitors can give rise to insulin-producing cells in vitro12 and in vivo.13 Liver-derived insulin-producing cells have even been effective in reducing experimental diabetes.13, 14 These hopeful preclinical results have already resulted in company spin-offs. Type 1 diabetes is an important medical problem, and currently, most patients are treated by life-long insulin replacement therapy. Cell therapy, which restores the missing β-cell mass, can replace insulin15 and, therefore, a very large (and lucrative) commercial market exists for β-cell transplantation.

There are 2 basic techniques to demonstrate the differentiation of putative progenitors/stem cells into their differentiated progeny. The first is in vitro differentiation in a tissue culture system. Progenitor cells (preferably at single cell density) are exposed to conditions that promote their conversion to the cell type of interest, which is then ascertained by the presence of markers, such as insulin production. In many cases, the in vitro conditions involve co-culture with other cells or culture on feeder layers. The second method is transplantation of the stem cell in vivo. The donor derived cells are tracked in the organ of interest to determine whether they have differentiated into the desired functional progeny.

The purpose of this editorial is to highlight and discuss the potential pitfalls that exist when these techniques are used to study stem cells. The lessons to be learned do not only apply to liver and pancreatic stem cells, but to stem cell biology in general.

The first artifact to be cognizant of is cell fusion in tissue culture. Co-culture or culture on feeder-layers is a frequently used technique to provide stem cells with growth factors and other signals that influence their cellular phenotype. For example, embryonic stem cells are usually cultured on feeder layers of embryonic fibroblasts.16 Some groups have attempted to differentiate embryonic stem (ES) cells into desired cell fates by co-culture with other cell types. During the course of such experiments, 2 groups independently discovered that ES cells could actually fuse with the cells with whom they are being co-cultured and thereby become “differentiated.”17, 18 The “differentiated” ES cells contained genetic markers from both the ES cells and the cells with which they had been mixed. Without this careful analysis, the authors of these studies could have concluded that the ES had actually differentiated and acquired a phenotype similar to that of the co-culture population.

The second artifact is cell fusion in vivo. Several studies have demonstrated the unexpected plasticity of adult bone marrow derived stem cells to become epithelial cell types in vivo.19–22 In general, these studies interpreted their findings as evidence for transdifferentiation of bone marrow stem cells. Now, however, it has been shown that stem cell plasticity is not the only explanation for such results. Bone marrow derived hepatocytes can be produced by in vivo cell fusion23, 24 by a process very similar to that seen in vitro before.17, 18 Again, careful analysis of markers from both the donor and the host was necessary to uncover the mechanism of the apparent “plasticity.”

The third potential problem results from the sole use of immunocytochemical markers to demonstrate the acquisition of a differentiated phenotype. Several studies have suggested that embryonic stem cells can become pancreatic endocrine cells in vitro and produce insulin.25, 26 Recently, however, this conclusion has been called into question by the discovery that insulin can be taken up from the tissue culture media.27 Thus, cells may stain positive for insulin without activation of the respective messenger RNA and, hence, without being actual insulin producing cells.

Finally, co-culture lends itself to artifact even without cell fusion. An example for this problem is provided by an article published in HEPATOLOGY,12 whose authors publish a letter in this issue. In this paper, several cell lines were derived from the livers of rats treated with an oval cell regimen (allyl alcohol). The authors established these liver progenitor cells on feeder layers of γ-irradiated mouse embryonic fibroblasts (STO cells). After the cell lines had been established, they were then weaned from the feeders and propagated independently. Interestingly, several of the liver progenitor cell lines could be propagated indefinitely and then differentiated into hepatocyte-like cells or ducts after many cell divisions. The markers analyzed were classic liver epithelial genes, including α-fetoprotein, albumin, glucose-6-phosphatase, etc. The authors concluded that they had generated rat liver epithelial progenitor cell lines that could be used to study their developmental potential, including their pancreatic abilities. Now, however, some of the authors of this original study realized that the expressed markers were murine in origin and not from the rat.28 They summarize their findings as follows28: “The combined results suggest that the previously reported “rat” liver progenitor cell lines were most likely generated during early derivation in cell culture from γ-radiation-resistant or ineffectively irradiated mouse STO cells used as the feeder layers.”

Together, these stories tell a cautionary tale that every stem cell researcher should recognize. It is very easy to draw erroneous conclusions about stem cell fates when the analysis is focused only on genetic markers of the putative progenitor/stem cell. To rule out cell fusion or feeder cell contamination, it is crucially important to use positive markers for the stem cell itself, as well as the other cells in the system. This is quite straightforward when species differences can be exploited, such as rat versus mouse, but becomes more difficult when all cells are from the same species. In those settings, the experiments should be carefully designed to make sure that readily identifiable markers are available. These can be cytogenetic (sex chromosomes), marker genes (green fluorescent protein, β-galactosidase), cell surface antigens (Ly 5.1/5.2), or other immunocytochemical markers. However, as illustrated by insulin uptake from the media, it is also very important to not rely on only protein markers, but to also include actual proof of marker gene expression at the messenger RNA level.

It is particularly important to be meticulous and careful about these kinds of experiments when the results suggest potential future clinical benefits for patients and/or commercial pay-offs. It is all too easy to get carried away by the enthusiasm generated by positive findings and ignore the less attractive, alternative explanations.

Given the problems and set-backs mentioned above, it is important to emphasize that stem cell research, particularly liver stem cell biology still has a bright future. Many different potential sources for the production of transplantable hepatocytes have been defined, including multipotent adult progenitor cells,29, 30 embryonic stem cells,31 fetal hepatoblasts,32 salivary gland progenitors,33 pancreatic liver stem cells,34 and, most importantly, the intrahepatic progenitors themselves.35 These cell sources need to be carefully evaluated bearing in mind the potential pitfalls described here. The most important criterion for an actual hepatic stem cell is functional restoration of liver function in a repopulation assay. As mentioned earlier, several animal models now exist to evaluate putative progenitors in this way.3–9 It is noteworthy that functional liver reconstitution has been achieved to date only by hepatocytes themselves or closely related cells, such as fetal hepatoblasts, intrahepatic stem cells, or pancreatic liver progenitors. Thus, it is clear that much needs to be learned about the conditions needed to convert immortally growing progenitors into “the real thing”: functional hepatocytes that can cure disease.