Clonal Analysis of Human Embryonic Stem Cell Differentiation into Teratomas


  • Barak Blum,

    1. Department of Genetics, Silberman Institute of Life Sciences, The Hebrew University, Jerusalem, Israel
    Search for more papers by this author
  • Nissim Benvenisty M.D., Ph.D.

    Corresponding author
    1. Department of Genetics, Silberman Institute of Life Sciences, The Hebrew University, Jerusalem, Israel
    • Department of Genetics, Silberman Institute of Life Sciences, The Hebrew University, Jerusalem 91904, Israel. Telephone: 972-2-6586774; Fax: 972-2-6584972
    Search for more papers by this author


Differentiation of human embryonic stem cells (HESCs) can be studied in vivo through the induction of teratomas in immune-deficient mice. Cells within the teratomas differentiate into all three embryonic germ layers. However, the exact nature of the proliferation and differentiation of HESCs within the teratoma is not fully characterized, and it is not clear whether the differentiation is cell autonomous or affected by neighboring cells. Here, we establish a genetic approach to study the clonality of differentiation in teratomas using a mixture of HESC lines. We first demonstrate, by means of 5-bromo-2′-deoxyuridine incorporation, that cell proliferation occurs throughout the teratoma, and that there are no clusters of undifferentiated-proliferating cells. Using a combination of laser capture microdissection and DNA fingerprinting analysis, we show that different cell lines contribute mutually to the same distinctive tissue structures. Further support for the nonclonal differentiation within the teratoma was achieved by fluorescence in situ hybridization analysis of sex chromosomes. We therefore suggest that in vivo differentiation of HESCs is polyclonal and, thus, may not be cell autonomous, stressing the need for a three-dimensional growth in order to achieve complex differentiation of HESCs.

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


Human embryonic stem cell (HESC) lines are derived from the inner cell mass of blastocyst stage human embryos. They are defined by their ability to proliferate indefinitely while retaining an undifferentiated state (self-renewal) and by their ability to differentiate into cells of the three embryonic germ layers (pluripotency). Since their isolation in 1998 [1], HESCs have attracted the attention of scientists and physicians worldwide as a potential source of cells for replacement of malfunctioning tissues in various diseases. HESCs can also serve as a tool for studying human developmental events. This is important, since many such events remain obscure due to the difficulties of studying human embryos at early stages.

HESCs can differentiate in vitro by allowing the cells to aggregate in suspension, to form embryoid bodies (EBs) [2, [3]–4]. Through this method, the HESCs develop in a way that somewhat resembles early embryonic development [5, 6], and a variety of cell types representing the three embryonic germ layers can be observed. In vitro cell differentiation into specific lineages can be induced by adding growth factors to the medium, by coculturing the cells with more mature cell types, or by genetic manipulations of the cells [7]. HESCs can also differentiate in vivo. Such an in vivo environment is created when HESCs are transplanted into an immune-deficient mouse. The cells then proliferate and differentiate to form tumors called teratomas. In these tumors, the cells spontaneously differentiate to cells representing all three embryonic germ layers and even, to some extent, form complex, organ-like structures [1, 7, [8], [9]–10].

To date, teratomas from human embryonic stem cells are used mainly as a model for the assessment of the pluripotency of newly derived HESC lines [1, 8, 9]. This, in fact, remains a standard proof for the differentiation capacity of virtually every new HESC line that is created. Therefore, there is a vast documentation regarding the range of tissue types arising within these tumors. These include ectoderm derivatives such as skin cells, neural rosettes or retinal epithelia, mesoderm derivatives such as cartilage, bone, or muscle, and endoderm derivatives such as gut epithelia and liver cells (reviewed in [7]). The exact nature of the tumor and the mechanisms by which HESCs differentiate within it, however, were largely ignored [11]. For instance, it is still not determined whether this tumor represents a bona fide model for embryonic development or an artifact produced by a subset of tumorigenic cells within the transplanted HESC population. Only if the former is correct can this model be applied to investigate developmental events.

In this study, we set out to analyze the clonality of HESC-induced teratomas. Using microsatellite analysis of teratomas from a mixture of three different HESC lines, we show that this tumor is polyclonal, derived from all cells initially transplanted, and is not formed by an expansion of a dominant clone. We also show that proliferation takes place throughout the tumor and not in proliferating foci within quiescent tissue, and that the differentiated structures themselves are proliferating. We then choose to investigate the clonality of differentiated structures within the tumor. Using a combination of laser-capture microdissection and analysis of genomic markers and sex-specific genomic in situ hybridization, we show that differentiation within the tumor is polyclonal, indicating that the HESC-induced teratomas may imitate developmental niches occurring in the developing embryo.

Materials and Methods

Cell Culture and Induction of Teratomas

HESC lines H9, HUES12, and HUES13 were cultured as previously described [2, 3]. For teratoma induction, approximately 2–3 × 106 cells were trypsinized, washed with phosphate-buffered saline (PBS), and suspended in 100 μl of PBS. Cells were transplanted under the kidney capsule of male BALB/c-nude or severe combined immunodeficient (SCID)/beige mice (Harlan Israel, Rehovot, Israel, as described [12]. Mice were kept in the specific pathogen-free unit of the Institute of Life Sciences at the Hebrew University. After 4–8 weeks, mice were sacrificed, and the tumors were excised with the adjacent kidney. All experiments were performed according to approval of the Committee for Animal Care and Use of the Faculty of Sciences at the Hebrew University of Jerusalem. Approximately half of the total mass of the tumor was processed immediately for DNA preparation, and the rest, including the kidney, was fixed with 4% buffered formalin (Bio-Lab Ltd., Jerusalem, and embedded in paraffin for laser-capture microdissection and fluorescence in situ hybridization (FISH).

5-Bromo-2′-Deoxyuridine Incorporation Assays and Immunofluorescence

For 5-bromo-2′-deoxyuridine (BrdU) incorporation assays, mice carrying HESC-induced teratomas were injected intraperitoneally with 1 ml of 1 mg/ml BrdU in PBS (approximately 0.03 mg/gr body mass). Mice were sacrificed 6–24 hours later, and teratomas were collected, fixed with 4% buffered formalin, and embedded in paraffin. For each tumor, 6-μm serial sections were placed on glass slides. Every third section was stained with hematoxylin and eosin for histology. Unstained sections were examined immunohistochemically for BrdU incorporation to proliferating nuclei using BrdU In Situ Detection Kit (BD Pharmingen, San Diego, according to manufacturer instructions. For immunofluorescent detection of Oct4 expression, cultured HESCs were washed twice with PBS, and teratomas were embedded in OCT (Sakura, Torrance, CA,, snap frozen in liquid nitrogen, and cut into 6-μm sections. HESCs and teratoma samples were then fixed for 15 minutes in 4% buffered formalin. Blocking and permeabilization were performed with 3% bovine serum albumin, 10% low-fat milk, and 0.1% Triton X in PBS. Samples were stained for Oct4 expression using monoclonal mouse anti-Oct4 antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA, at 1:100 dilution and Cy3-conjugated goat anti-mouse polyclonal secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA, at 1:200 dilution. Nuclear staining was performed using Hoechst 33258 (Sigma-Aldrich, St. Louis,

Laser-Capture Microdissection

Formalin-fixed paraffin embedded samples were serially sectioned into 12–6-μm-thick slices on MMI membrane slides (MMI Molecular Machines & Industries, Glattbrugg, Switzerland, Slides were deparaffinized with two 5-minute washes of xylene followed by 10-second successive immersions with 70%, 95%, and 100% ethanol. Slides were then stained briefly (5 seconds) with eosin, dehydrated with 10-second successive immersions in 70%, 95%, and 100% ethanol, 20 seconds in xylene, and air-dried. Microdissection was performed using SL μCut laser-capture microdissection system (MMI). The sampling of each structure was performed as follows. Each teratoma was serially sectioned six times, and a collection of the same dissected structure from all six slides was combined to give one structure. To ensure that no structure is sampled twice, all structures from the same teratoma were selected from the same plane.

Microsatellite DNA Analysis

Genomic DNA was extracted from culture cells using EZ-DNA kit (Biological Industries, Kibbutz Beit Haemek, Israel, according to the manufacturer's instructions. For extractions of genomic DNA from teratomas, the tumors were digested overnight at 55°C with 0.36 mg/ml proteinase K in teratoma digestion buffer (10 mM Tris-HCl pH 8.0, 100 mM NaCl, 10 mM EDTA, 0.5% sodium dodecyl sulfate), and DNA was extracted using phenol/chloroform. To obtain DNA from laser-capture microdissection (LCM) dissected samples, LCM tubes were incubated overnight at 37°C with 45 μl of 0.04% proteinase K in LCM digestion buffer (10 mM Tris-HCl, 1 mM EDTA, 1% Tween 20) and then heated to 95°C for 8 minutes to inactivate the proteinase K. Genotyping was performed at the Center for Genomic Technologies at the Hebrew University of Jerusalem, Israel, using markers from the Genethon human linkage map (ABI PRISM Linkage Mapping Set MD10; Applied BioSystems, Foster City, CA, Polymerase chain reaction (PCR) amplification of individual markers (fluorescence-dye-labeled forward primer and unlabeled reverse primer) was performed in a PTC 225 DNA Engine (Bio-Rad, Hercules, CA, using 25–30 ng of genomic DNA, 6 pmoles of each primer, 1.5 mM MgCl2, 0.14 mM deoxynucleoside-5′-triphosphate, 1× PCR Gold Buffer (15 mM Tris-HCl, pH 8.0, 50 mM KCl), and 0.4 units of AmpliTaq Gold DNA Polymerase (both from Applied Biosystems) in a total volume of 10 μl. PCR conditions were as follows: an initial 12 minutes denaturation at 95°C, 10 cycles of 15 seconds at 94°C, 15 seconds at 55°C and 30 seconds at 72°C, 10 minutes at 72°C, and hold forever at 10°C. After amplification, 1–2 μl were sampled into 9 μl of loading buffer (formamide with GeneScan-400HD [ROX] size standard; Applied Biosystems). PCR product electrophoresis and detection were performed using the 3700 Automated DNA Analyzer (Applied Biosystems). Sizing and genotyping were performed using GENSCAN and GENOTYPER software (Applied Biosystems).

FISH Analysis

Formalin-fixed paraffin embedded samples were serially sectioned into 6-μm slices on glass slides. Slides were deparaffinized by three 10-minute washes of xylene rehydrated with 5-minute washes of 100%, 85%, and 70% ethanol and 5-minute immersion in double-distilled water. Slides were baked for 20 minutes in 0.1 M citric acid in a microwave oven and were let to cool down to room temperature. Slides were then washed briefly in 2× standard saline citrate (SSC), denaturized in 70% formamide in 2× SSC for 3 minutes at 75°C, dipped for 3 minutes in 70% cold ethanol, dehydrated with 85% and 100% ethanol, and air-dried. FISH was carried out according to the manufacturer's recommendations and as described previously [13] using probe mixtures specific for chromosomes X and Y (CEP X Spectrum Green and CEP Y Spectrum Orange [Vysis, Downers Grove, IL,] or fluorescein isothiocyanate-labeled chromosome X and rhodamine-labeled chromosome Y satellite [catalog number PSAT2324; Qbiogene, Irvine, CA,]). The criterion for signal scoring was that signals had to be at a minimum of a signal's width apart to be scored as two separate signals. Confocal imaging was done using an MRC-1024 Bio-Rad confocal scan head coupled to a Zeiss Axiovert 135M inverted microscope (Carl Zeiss, Jena, Germany,


Our aim is to assess whether differentiated structures derived from HESCs originate in a mono- or polyclonal manner. Therefore, we took a genetic approach in which we induced teratoma formation from a mixture of three different HESC lines. The relative contribution of each of the cell lines to the teratomas was analyzed by the presence of their genomic microsatellite signature. The advantage of this method is that it enables analysis of native HESCs rather than genetically modified subclones.

Figure 1 illustrates the possible fate for a pool of three HESCs injected in vivo in generating differentiated structures. One possibility is that the teratoma itself may be clonal, and one of the cell lines undergoes positive selection and takes over the teratoma. The tumor will then be derived solely from this line (Fig. 1A). In this case, all differentiated structures will appear monoclonal whether or not polyclonality of differentiation is possible, and it will not be possible to trace the mono- or polyclonal origin of the differentiated structures. Another possibility is that the teratoma may be polyclonal (Fig. 1B). In this case, the mono- or polyclonal manner of differentiation could be determined. If differentiation is clonal, that is, the structures originate by the expansion of a single cell within the pool that was autonomously committed to a particular fate (Fig. 1C), structures within the teratomas are expected to be composed of cells from only one origin. If, on the other hand, the differentiation is not clonal, that is, cells from different origins react to the same microenvironmental signals and unite to form a differentiated structure (Fig. 1D), some of the structures are expected to be composed of cells from several origins.

Figure Figure 1..

A scheme describing the possibilities for the origin of complex structures within a teratoma. Injection of a mixture of three different human embryonic stem cell lines into a severe combined immunodeficient mouse (green, orange, and blue dots at the top of the figure). (A): The tumor may be clonally derived from cells that underwent selection. In this case, the composition of differentiated structures, in terms of cell-of-origin, is untraceable. (B): The tumor may instead be composed of all lines injected. In this case, the monoclonal or polyclonal origin of differentiation can be traced. A differentiated structure within the tumor, such as the one on the bottom left of the figure, could then be clonally derived from a prepatterned cell of origin (C) or composed of cells from different origins, which react to signals from the microenvironment and unite to form complex structures (D).

Establishment of Teratomas from Several HESC Lines

To study clonal differentiation of HESCs, we induced teratomas from a mixed pool of H9, HUES12, and HUES13 lines. We first screened for the composition of the polymorphic sites of these lines at different genomic loci (Table 1). The genomic locus that was chosen as the most informative for discrimination between these HESC lines was the D3S3681 locus. In this locus, H9 cells are heterozygous for 150 and 156 nucleotide repeat alleles, HUES13 cells are heterozygous for 152 and 156 nucleotide repeat alleles, and HUES12 cells are homozygous for a 124-nucleotide repeat allele (Table 1 and Fig. 2A–2C). All alleles could be detected in a mixture of cells from all three lines (Fig. 2D). We next created teratomas from a mixed pool of these HESC lines. Accordingly, H9, HUES12, and HUES13 cells were mixed at equal amounts and injected under the kidney capsule of SCID and nude mice. Additional SCID mice were injected, each with cells from only one of the HESC lines. All mice developed teratomas, which were extracted 4 weeks after transplantation. Among the teratomas induced by a single HESC line, the HUES12 and HUES13 tumors displayed a somewhat more cystic morphology compared with that of the H9 teratoma. Nevertheless, the three mixed teratomas were fairly similar, and all tumors demonstrated differentiation to all three germ layers. Each teratoma was dissected, and approximately half of the total mass was submitted to DNA microsatellite analysis of the D3S3681 locus. In two of the mixed teratomas, we could clearly find the characteristic microsatellite signature of all three HESC lines (Fig. 2E–2G), demonstrating that cells from all three HESC lines exist in these tumors. Moreover, the ratios among the three HESC lines prior to injection were maintained in the teratoma, as was evident by the resemblance of the relevant microsatellite signatures. In the third teratoma, the intensity (peak height) of the 150-nucleotide repeat allele, characteristic of H9, was slightly lower than the intensity of the 152-nucleotide repeat allele (Fig. 2H). We were thus unable to be certain that this peak is a true allele and not a “shadow” of the 152-nucleotide allele. However, the presence of alleles characteristic of both HUES12 and HUES13 was clearly evident. To confirm the suspected contribution of the H9 cell line to the third teratoma, we hence analyzed it for another microsatellite marker, namely that of the D20889 locus. In this locus, H9 cells are homozygous for a 92-nucleotide repeat allele; HUES12 are homozygous for a 103-nucleotide repeat allele, whereas HUES13 are heterozygous for 97- and 103-nucleotide repeat alleles (Table 1). The results obtained with this microsatellite marker (Fig. 2I) evidently showed the presence of these three alleles. Taken together, the analysis of the two microsatellite markers (D3S3681 and D20S889; Fig. 2H, 2I) indicate that this tumor was indeed a mixture of all three HESC lines.

Table Table 1.. Microsatellite composition of H9, HUES12, and HUES13 on different genomic loci
original image
Figure Figure 2..

Microsatellite analysis of polyclonality in human embryonic stem cell (HESC)-induced teratomas. Three HESC lines (H9, HUES12, and HUES13) were independently injected or mixed and injected under the kidney capsule of severe combined immunodeficient mice. One month later, teratomas were harvested, and genomic DNA was prepared and subjected to microsatellite analysis. (A–H): Results for the D3S3681 locus. (A): DNA from H9 teratoma. (B): DNA from HUES12 teratoma. (C): DNA from HUES13 teratoma. (D): DNA from a pool of the three cell lines at equivalent ratios, before injection. (E): DNA from a 4-week-old teratoma induced from that same mix. (F): DNA from a different part of the same teratoma as in (E). (G, H): DNA from two other teratomas induced from a mix of the three HESC lines. The inset in (G) is an enlargement of the 150–156 allele section. (I): Results for the D20S889 locus of the same teratoma shown in (H).

Proliferation of Differentiated Cells Within HESC-Induced Teratomas

Before the examination of clonality of differentiated structures, we wanted to learn about the manner of proliferation within HESC-induced teratomas. First, we asked whether the increase in tumor volume comes from foci of proliferating undifferentiated cells surrounded by differentiated quiescent tissue or whether the whole tissue and differentiated structures within it are proliferating. To examine the proliferation of the different cell types within the teratomas, we performed a BrdU incorporation assay on developing teratomas. We transplanted H9 or HUES12 cells under the kidney capsules of five SCID mice. Three of the mice were sacrificed after 4 weeks, and two were left to develop the tumors for 8 weeks. Each mouse was injected intraperitoneally with BrdU 6 or 24 hours prior to the collection of the teratoma. Proliferating nuclei were detected immunohistochemically using anti-BrdU antibodies (Fig. 3A–3E). To score the proliferation index of the tumor, we counted the percentage of BrdU incorporated nuclei in a randomly aligned grid. The average proliferating index of the teratoma 6 hours after injection of the BrdU was 28%. The average proliferating index of the teratoma 24 hours after injection of the BrdU was 43%. No BrdU incorporation was identified in the spleen of any of the mice. Interestingly, extensive BrdU incorporation was observed even in 8-week-old teratomas. We next scored the proliferation index of four histologically distinguishable tissues within the teratoma, namely cartilage, epithelial tubes resembling neural rosettes, and two types of mesenchyme: a mucoidal component and a mesenchymal tissue surrounding the tubular epithelial component (Fig. 3F). All tissues showed extensive BrdU incorporation 24 hours after BrdU injection with labeling indexes of 41%, 63%, 26%, and 45%, respectively. These relative proliferation ratios resemble the relative proliferation ratios observed with in vivo BrdU labeling assays in mature teratomas in humans [14]. Although some tissue types within the teratoma were proliferating slightly more rapidly than others, proliferation was scattered across the teratoma, and there were no isolated foci of proliferation in the tumor. Also, and in agreement with other reports [10], there was no indication of undifferentiated cells expressing Oct4 (Fig. 3G–3L).

Figure Figure 3..

Analysis of cell proliferation in human embryonic stem cell (HESC)-induced teratomas. (A–F): Mice bearing HESC-induced teratomas were injected with BrdU. Teratomas were harvested 24 hours later and analyzed immunohistochemically for BrdU incorporation in proliferating cells (brown) and counterstained with hematoxylin (blue). (A): Analysis of 4-week-old teratoma induced from H9 cells 24 hours after BrdU injection. Proliferation is scattered throughout the tumor. No foci of proliferating cells are observed. Cartilage and epithelial tube are marked. (B): Analysis of 4-week-old teratoma induced from H9 cells 6 hours after BrdU injection. Two types of mesenchyme (mes1 and mes2) are shown. (C): The spleen of the same mouse as in (A). (D, E): Different magnifications of 8-week-old teratoma induced from HUES12 cells. (F): A histogram showing the relative proliferation indexes of several tissues within the teratoma. (G–I): Immunofluorescence analysis of Oct4 expression in an undifferentiated H9 HESC colony. Cells were immunostained for Oct4 (red). Nuclear DNA was stained with Hoechst (blue). (J–L): Immunofluorescence analysis of Oct4 expression in a 4-week-old teratoma induced from H9 cells. Cells were immunostained for Oct4 (red). Nuclear DNA was stained with Hoechst (blue). No Oct4 expression is observed. Scale bars = 50 μm. Abbreviations: BrdU, 5-bromo-2′-deoxyuridine; cart., cartilage; epi., epithelial tube; mes1, mucoidal mesenchyme; mes2, mesenchyme surrounding the neural tissue.

Differentiation in the Teratoma Is Not Clonal

Since HESC-induced teratomas were found to contain cells from various HESCs and the proliferation was found to be scattered throughout the teratoma, we could now try and elucidate some of the mechanisms underlying the complex differentiation within these tumors. As discussed above (Fig. 1C, 1D), differentiation within the tumor may be a cell-autonomous process or an inductive process by which cell fate is determined by niches forming as the teratoma develops. We hypothesized that, if the differentiated structures originated from single cells that were randomly committed to a certain fate (Fig. 1C), we would not find any polyclonal structures within the tumor. Alternatively, if differentiation in the tumor is not entirely cell-autonomous and certain local signals within the teratoma regulate its differentiation (Fig. 1D), then at least some of the differentiated structures would appear polyclonal. We therefore isolated differentiated structures from the mixed teratomas using LCM and examined their clonality by the composition of microsatellite alleles. We dissected 20 structures of distinctive nature from two mixed teratomas and could obtain DNA of good quality from 15 of them. Of these, two structures displayed microsatellite signatures characteristic of H9 only, two were HUES13 only, four were HES12 only, and five structures displayed microsatellite signatures of two or more cell lines. Another two structures displayed only the 156-nucleotide peak and could thus be composed of H9 cells, HUES13 cells, or both. Figure 4 shows representative results form one teratoma. Whereas the four upper structures are composed of only one cell line, the two bottom ones are clearly showing the contribution of two different lines. To exclude the possibility of false-positive results generated by a contamination of cells from outside of the selected structures, we assessed the sensitivity of the microsatellite method. For this, we mixed fixed ratios of DNA from two cell lines and submitted those to microsatellite analysis (data not shown). Since the sensitivity of the method used to detect the presence of various alleles within a mixture was limited to a 10-fold difference between the alleles, and since we dissected all structures within a “safety zone” from inside the structure's borders, we ruled out the possibility that our data result from a contamination of cells outside the observed structure. To confirm the above results, slices of the same teratomas were also examined using FISH analysis. This was made possible due to the fact that H9 and HUES12 lines are of female origin, and HUES13 is of male origin (Fig. 5). Indeed, in some structures, we detected the presence of nuclei with two X chromosomes and, in the same structures, the presence of nuclei with one X and one Y chromosome. Taking into account the results of these two complementary methodologies, we concluded that at least some of the differentiated structures originated from the combination of different cell lines. Differentiation of HESCs in vivo as HESC-induced teratoma is thus polyclonal and not created by autonomously committed cells.

Figure Figure 4..

Microsatellite analysis of the polyclonality of differentiation within the teratoma. Differentiated structures were excised from the teratoma using LCM and subjected to microsatellite analysis as described. In order to obtain a sufficient amount of DNA, structures were collected from as many as six successive serial sections. Results of the D3S3681 locus from representative structures are shown. A magnification of each structure before dissection (before LCM) is shown together with the total amount of tissue dissected from that structure (after LCM). (A–D): Structures composed of single-origin cells. (E, F): Structures composed of cells from two origins. Abbreviation: LCM, laser-capture microdissection.

Figure Figure 5..

X-Y chromosome fluorescence in situ hybridization (FISH) analysis of differentiated structures within the teratoma. (A, B): Fluorescent microscopy images of X-Y chromosomes FISH of control female and male human embryonic stem cells, respectively. X chromosome probe: green; Y chromosome probe: orange. (C, E): Confocal microscopy images of X-Y chromosomes FISH of two epithelial tubes from teratomas. X chromosome probe: green; Y chromosome probe: red. Arrowheads represent X chromosomes and arrows represent Y chromosomes of XX and XY cells within the same structure. (D, F): Enlargement of the insets in (C) and (E), respectively. Dashed lines represent the borders of the selected cells. Scale bars = 10 μm.


In this work, we have studied the clonality of differentiation of teratomas induced by the in vivo transplantation of HESCs. Many tumors are thought to be monoclonal in origin, derived from single precursor cells. Likewise, in human germ cell tumor (GCT), components were shown to display similar loss of heterozygosity patterns throughout the entire tumor [15, [16], [17]–18]. In the present study, we have shown that HESC-derived teratomas are polyclonal, given that the microsatellite signature of the entire tumor included the alleles of all HESC lines initially transplanted. Moreover, even discrete structures within the tumor were found to be polyclonal. This implies that these tumors are initiated by more than one precursor cell.

HESCs are characterized by their self-renewal capability and, as such, have the potential to proliferate indefinitely. However, their in vitro differentiated counterparts (EBs) display a much lower proliferation capacity [2]. Likewise, after sorting of differentiated hepatic cells from HESC-induced teratomas, the cells fail to proliferate for extended periods in culture [19]. This raises the question of whether the proliferation of differentiated cells, or that of foci of undifferentiated cells, contributes to tumor growth. As was raised in a recent review [11], the capacity of the HESC-induced tumor growth can be limited by the initial number of HESCs transplanted or, instead, be based on a subpopulation of constantly proliferating cells and thus be infinite. Using BrdU incorporation assays, we show a high ratio of BrdU containing nuclei as early as 6 hours after BrdU injection, and extensive incorporation of BrdU was evident in teratomas as old as 8 weeks. These results indicate that, in contrast to EBs, the teratoma is rapidly proliferating. We also show that BrdU incorporated nuclei are scattered throughout the tumor, and the differentiated structures themselves are proliferating. Moreover, no foci of cells expressing markers of undifferentiated cells could be detected in HESC-induced teratomas (Fig. 3, [10]). This is in contrast to naturally occurring malignant human testicular GCTs and teratocarcinomas induced from mouse embryonic stem cells, where highly proliferating foci expressing both Oct4 and Ki-67 were detected [20]. We suggest that the explanation to this intriguing difference between HESC-induced teratomas and naturally occurring GCTs is that, in teratomas induced from HESCs, clonal selection is not as fundamental a part of the tumorigenic process as in other tumors. However, this issue requires further in-depth investigation.

The question of the clonality of organs and tissue development in humans has been mainly addressed by extrapolations from studies of model animals such as the mouse. Experiments using mouse chimeras to trace the lineages of tissues and organs were initiated in the late 1970s [21, 22]. Recently, this method was enhanced by using chimeric mice carrying fluorescent transgenes of three different colors [23]. However, in humans, the analysis of clonality of normal tissues was restricted to the crude patterns of X-inactivation patches, due to the obvious lack of human chimeras [24]. We suggest that our system may provide an attractive model for the study of the clonality of human tissues. HESCs spontaneously differentiate, forming clusters of specific cell types. This can be seen in vitro, as mature EBs display specific regional gene expression in RNA in situ hybridizations with specific lineage markers [2] or express a reporter fluorescent protein under the control of a tissue specific promoter [19]. It is also evident in vivo, in HESC-induced teratomas, where markers for different tissue types were shown to be expressed in a localized fashion [10, 19]. However, such clustering was never shown to be regulated and could be simply attributed to proliferation of specific cell types. Using two independent methods, we demonstrate that the differentiation of structures within the teratomas is not clonal. Our laser-capture microdissection coupled to DNA microsatellite analysis shows that there is incorporation of cells from different origins into the same tissue structure. This was striking, since as much as a third of the dissected structures showed polyclonality. Given that some structures were only partially sampled, it could also be that structures that appeared monoclonal were actually comprised from more than one cell line, stressing this ratio even further. FISH analysis with sex-specific probes indicated, by a complementary methodology, that indeed the differentiated structures within the teratoma contain cells of distinct origins. However, since in our assay not all differentiated structures showed polyclonality, it is still plausible that some tissue types will consistently differentiate in a polyclonal manner, whereas others will always be monoclonal.

In spontaneous differentiation of EBs in vitro, the first event generating variability in the homogeneous HESC pool is thought to take place as cells in the outer layer of the aggregate are exposed to a different environment than the inner cells. This in turn creates two distinct cell populations that may now influence each other. In the teratoma, the cells are exposed to a more complex environment formed by the much variable atmosphere of the host. Indeed, it was recently demonstrated that HESCs transplanted into the liver of nude mice formed teratomas that were strikingly dissimilar to those formed by subcutaneous injection of the same cells, arguing that differentiation of HESC-induced teratomas is affected by their external environment [25]. As a consequence, the intertissue influences within the teratoma are thought to be more complex. Our results demonstrate that adjacent cells within the teratoma acquire the same cellular fate, suggesting the presence of an inductive niche. But do these interactions truly imitate the differentiation cascades occurring during normal embryonic development? Lavon et al. [19] showed that cells expressing α-cardiac actin colocalized with albumin expressing cells in a teratoma and suggested that, as in the rodent embryo [26], the former create a niche that induces the differentiation of the latter. Similarly, tissue organization suggesting simple organogenesis was speculated in teratomas induced from HESCs [10]. However, in these studies, the evidence for the occurrence of an embryo-like microenvironment in HESC-induced teratomas was indirect.

To date, the in vitro differentiation of HESCs as EBs or as monolayers generated only simple cells and tissues [7]. The formation of many tissues needed in the clinic, such as pancreatic β-cells or kidney tissue, requires the complex interaction of many cell types in an orderly fashion. Indeed, insulin-secreting cells were elusive using in vitro attempts [7] but were reported to arise after the cotransplantation of HESCs with dorsal pancreatic fragments from rodent embryos [27], and human mesenchymal stem cells created complex kidney structures after they were cultured in the environment of rodent embryonic kidney in an organ culture system [28]. Our finding that cells within the teratoma are influenced by the microenvironmental differentiation signals implies that HESC-induced teratomas may truly represent events in the developing embryo. The teratoma system, providing inductive environments that are not possible for in vitro imitation, provides a complex environment to investigate cell-cell and cell-tissue interactions during tissue differentiation and early organogenesis.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank Tsvia Frumkin for assistance with FISH analysis, Dr. Shmuel Wolf and Hanita Zemach for help with LCM, and Daphne Benvenisty for assistance with BrdU analysis. We also thank Yoav Mayshar, Nadav Sharon, and Ori Bar-Nur for critically reading the manuscript. This work was partially supported by funds from Bereshit Consortium, the Israeli Ministry of Trade and Industry (Grant number 37675), and by the European Community (ESTOOLS, Grant number 018739).