SEARCH

SEARCH BY CITATION

Keywords:

  • Human embryonic stem cell;
  • Embryoid body;
  • Differentiation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Like other cell populations, undifferentiated human embryonic stem cells (hESCs) express a characteristic set of proteins and mRNA that is unique to the cells regardless of culture conditions, number of passages, and methods of propagation. We sought to identify a small set of markers that would serve as a reliable indicator of the balance of undifferentiated and differentiated cells in hESC populations. Markers of undifferentiated cells should be rapidly downregulated as the cells differentiate to form embryoid bodies (EBs), whereas markers that are absent or low during the undifferentiated state but that are induced as hESCs differentiate could be used to assess the presence of differentiated cells in the cultures. In this paper, we describe a list of markers that reliably distinguish undifferentiated and differentiated cells. An initial list of approximately 150 genes was generated by scanning published massively parallel signature sequencing, expressed sequence tag scan, and microarray datasets. From this list, a subset of 109 genes was selected that included 55 candidate markers of undifferentiated cells, 46 markers of hESC derivatives, four germ cell markers, and four trophoblast markers. Expression of these candidate marker genes was analyzed in undifferentiated hESCs and differentiating EB populations in four different lines by immunocytochemistry, reverse transcription–polymer-ase chain reaction (RT-PCR), microarray analysis, and quantitative RT-PCR (qPCR). We show that qPCR, with as few as 12 selected genes, can reliably distinguish differentiated cells from undifferentiated hESC populations.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Currently, more than 100 distinct human embryonic stem cell (hESC) lines have been derived, and efforts at new derivations are ongoing. Approximately 20 lines from the 78 derivations undertaken before August 9, 2001, are available in sufficient quantities for general research use (National Institutes of Health [NIH] stem cell registry, http://stemcells.nih.gov/research/registry). Of these, only a small subset of lines is available for detailed characterization [18]. As expected, various hESC lines have a number of similarities. For example, undifferentiated hESCs are similar in expressing surface antigens and markers characteristic of the undifferentiated ESC state, including Oct4 (POU5F1), Nanog, UTF1, DPPA5, TERT, gap junction proteins, and SSEA and TRA antigens [18]. hESCs are also similar in their ability to proliferate and differentiate into cell types of the three germ layers in vitro and in vivo [916]. Properties of hESCs have also been compared using microarray, expressed sequence tag (EST) scan, serial analysis of gene expression (SAGE), and massively parallel signature sequencing (MPSS) [4, 1727]. These studies suggest that it is likely that markers shared by hESC lines, but absent in other cell populations, exist.

Although these studies have highlighted similarities among hESC lines and markers that distinguish them from mouse ESCs, it is likely that differences also exist. These include potential differences in methylation patterns [28, 29], likely HLA differences [25], allelic differences, variability of X-inactivation, and adaptation of cells to different culture conditions [2, 17, 30, 31]. Indeed, important differences among hESC lines in growth rates, methods of propagation, and karyotype have been reported using a variety of different techniques, suggesting that although shared markers may exist, care will be needed to identify them. Identifying such shared markers, however, remains an easier task than the technically challenging, direct comparisons of hESCs under identical culture conditions—experiments that are being undertaken at the Stem Cell Center at the NIH (Dr. Ronald McKay) and at the International Stem Cell Initiative (Dr. Peter Andrews) to identify fundamental differences among cell lines. Such experiments are beyond the scope of our laboratories. The available data, however, indicate that identifying a common pattern of gene expression that is conserved independent of culture conditions and reflects the fundamental properties of hESC is possible.

Several other experiments suggest that a unique molecular signature can be defined to distinguish undifferentiated hESCs from their differentiated progeny and that this signature can be used to define the states of hESCs [1719]. These experiments used MPSS, EST scan, and microarray data to suggest that a large pool of potential markers that could distinguish embryoid bodies (EBs) and other differentiated cells from hESCs exists. We have reasoned, therefore, that at the current level of resolution of techniques, it is possible to identify a core set of genes that are conserved and/or required to maintain hESC identity. These genes should be expressed irrespective of the conditions of culture, numbers of passages, and methods of propagation as long as undifferentiated hESCs are present. Moreover, these core markers should be present regardless of the methods of hESC derivation and ethnic phenotypes of the blastocysts used. If present at lower levels, they should be detectable by reverse transcription–polymerase chain reaction (RT-PCR) as well as by SAGE/MPSS, and if robust, by SAGE and microarray. Furthermore, if the expression of such genes is examined in EBs, then a subset of markers that are rapidly downregulated or rapidly induced as cells differentiate can be identified [19]. A combination of such markers can serve to reliably assess the states of ESCs and EBs.

To test this hypothesis, we performed a meta-analysis of published reports examining hESCs and EBs and identified a list of potential markers. We tested a substantial number of these markers by quantitative RT-PCR (qPCR) and immunocytochemistry and identified a combination of markers to distinguish hESCs from EBs.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

hESC Culture and In Vitro Differentiation via EB

hESC lines BG01, BG02, BG03, and I6 were maintained either on inactivated mouse embryonic fibroblast (MEF) feeder cells in medium comprised of Dulbecco's modified Eagle's medium/Ham's F-12 supplemented with 20% knockout serum replacement, 2 mM nonessential amino acids, 2 mM L-glutamine, 50 μg/ml Penn-Strep (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com), 0.1 mM β-mercaptoethanol (Specialty Media, Phillipsburg, NJ, http://www.specialtymedia.com), and 4 ng/ml of basic fibroblast growth factor (bFGF; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), or on dishes coated with fibronectin (BD Biosciences, Bedford, MA, http://www.bdbiosciences.com) in medium conditioned with MEF for 24 hours as previously described [4, 25].

Differentiation via EB formation was performed as previously described [4]. Briefly, hESCs were dissociated into small clumps by collagenase IV (Sigma-Aldrich) and grown as floating spheres in hESC medium without bFGF for up to 2 weeks, with a medium change every second day.

RT-PCR and qPCR Analysis

Total RNA was extracted from undifferentiated hESCs or (7-day and 14-day) EBs using RNA STAT-60 (Tel-Test Inc., Friendswood, TX, http://www.isotexdiagnostics.com). cDNA was synthesized by using a reverse transcription kit (RETRO-script; Ambion, Austin, TX, http://www.ambion.com) according to the manufacturer's recommendations. The PCR primers are listed online in supplemental Table 1.

Real-time qPCR was used to quantify the levels of mRNA expression of 12 selected genes (Oct4, Nanog, Sox2, UTF1, DPPA5, Lin41, Sox1, DCN, H19, IGF2, GATA4, and Hand1) in hESCs or EBs at different times of differentiation. PCR reactions were carried out by DNA Engine Opticon Fluorescence Detection System (MJ Research, Waltham, MA, http://www.mjr.com) using a DyNAmo SYBR Green qPCR kit according to the manufacturer's instructions. The content of selected genes was normalized to the content of 18S-rRNA, and standard curves were generated using 10 to 1000 ng cDNA per 20 μl reaction volume. All PCR products were checked by melting curve analysis to exclude the possibility of multiple products or incorrect product size. PCR analyses were conducted in triplicate for each sample.

Immunocytochemistry

Immunocytochemistry and staining procedures were as described previously [32]. Briefly, hESCs grown either on MEF feeder cells or under feeder-free conditions and 7-day and 14-day EBs either attached or floated and were fixed with 2% paraformaldehyde for 30 minutes. Parts of EBs were embedded in OCT blocks and were cut on a cryostat to obtain 8-μm sections. Fixed cells and sections were blocked in blocking buffer (5% goat serum, 1% bovine serum albumin, and 0.1% Triton X-100) for 1 hour followed by incubation with the primary antibody at 4°C overnight. Appropriately coupled secondary antibodies (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com) were used for single and double labeling. All secondary antibodies were tested for cross-reactivity and nonspecific immunoreactivity.

The following antibodies were used: Nanog (1:1000, AF1997; R&D Systems Inc., Minneapolis, http://www.rndsystems.com), CD29/Integrin β-1 (ITGB1, 1:1000, MAB1951; Chemicon, Temecula, CA, http://www.chemicon.com), E-Cadherin (CDH1, 1:2500, MAB1838; R&D Systems Inc.), Podocalyxin (PODXL, 1:500, MAB1658; R&D Systems Inc.), Sox2 (1:1000, MAB2018; R&D Systems Inc.), Oct4 (1:1000, AF1759; R&D Systems Inc.), Brachyury (1: 1000, AF2085; R&D Systems Inc.), cardiac actin (ACTC, 1:50, PRO61075; Research Diagnostics, Inc., Concord, MA, http://www.researchd.com), GATA4 (1:100, sc-25310; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), GATA6 (1:100, sc-9055; Santa Cruz Biotechnology Inc.), alpha-fetoprotein (AFP) (1:500, A8452; Sigma-Al-drich), and β-III-tubulin (TUBB3, 1:2000, T8660; Sigma-Aldrich). Bis-benzamide (Dapi, 1:1000; Sigma-Aldrich) was used to identify the nuclei. Images were captured on a fluorescence microscope (Olympus, Tokyo, http://www.olympus-global.com).

Microarray Analysis Using BeadArray Platform

RNAs from undifferentiated BG01, BG02, and BG03 cells and the matched 14-day EBs were hybridized to prototype Illumina RefSeq-8 BeadChips (San Diego, http://www.illumina.com), which contained more than 24,000 genes [33]. A detailed description of the BeadChip system has been provided elsewhere [33]. Samples were coded and run in duplicates, and the results were analyzed using standard bioinformatics tools and the Bead Studio, a tool kit developed by Illumina. A detailed analysis of these and other samples will be reported elsewhere (J.L., personal communication).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Meta-Analysis Procedure

A large dataset on gene expression of undifferentiated hESCs and EBs that differentiated from them has been generated using large-scale array experiments, including MPSS, EST enumeration, and microarrays, and described by a variety of investigators [1724]. To generate a list of genes characteristic of undifferentiated hESCs and differentiated EBs, we examined three published reports on gene expression in hESCs and EBs: (a) a list of 92 genes identified as “stemness” genes that are expressed at high levels in six hESC lines as assessed by a 16,695 70-bp oligonucleotide microarray [18], (b) a comprehensive list of genes common to undifferentiated hESCs obtained by MPSS analysis using pooled hESCs samples [17], and (c) a large list of genes that are highly expressed in differentiated EBs detected by both MPSS and EST scan [19]. (A complete list of EST scan data is available on http://www.ncbi.nlm.nih.gov/UniGene/library.cgi?all=yes&ORG=Hs&LID=14183 and http://www.ncbi.nlm.nih.gov/UniGene/library.cgi?all=yes&ORG=Hs&LID=14184.)

The criteria for selection were that the levels of expression should differ between hESCs and EBs by at least threefold by array data or by fivefold for EST scanning and MPSS, the levels of expression should be high (>10 transcript per million for MPSS, >5 by EST scan), and the genes must be detected by at least one independent method and in more than one hESC line. Some exceptions to the select ion criteria included: genes that were not detected by MPSS or EST scanning but that are used as an ESC marker by convention; genes that were highly expressed in both hESCs and EBs or expressed at higher levels in EBs but that were reported as ESC-specific genes either by microarray analysis when compared with human universal RNA [18] (Table 1) or because they were potential cell surface markers. A list of several hundred genes generated after this selection was pruned to a list of approximately 100 by examining published literature; our goal was to include known genes, with approximately half of them being candidate genes highly expressed in undifferentiated hESCs (but not in EBs or down-regulated in EBs) and the other half being those that are expressed in EBs (but lower or absent in hESCs). We did not include any unknown genes even though many of them were significantly differentially expressed in hESCs and EBs [19].

The final list of 109 potential markers included 55 stem cell markers, most of which were highly expressed in undifferentiated hESCs. For potential markers of the EB state, we selected 46 markers, including representative genes from each germ layer: 12 ectoderm, 15 mesoderm, and 19 endoderm markers that are likely overexpressed in EBs. In addition, four trophoblast markers and four germ cell markers were included (Tables 1, 2; supplemental online Table 1).

Markers of the ESC State

Forty-four genes highly expressed in undifferentiated hESCs but downregulated in EBs were selected as markers of stem cells to represent the undifferentiated ESC state (Table 1). Several of these genes, such as Oct4, Nanog, and TERT, are generally accepted markers of pluripotency. Many genes in the list, including UTF1, Sox2, Lin28, Lin 41, PODXL, LeftyB, GJA1, FoxD3, and Rex1 (ZFP42), have recently been reported by several research groups to be expressed in undifferentiated hESCs [4, 1719, 34]. Other undifferentiated markers included genes that encoded transcriptional factors, growth factors, signal transducers, and cell surface antigens and receptors. In addition, 11 genes were included as ESC markers although they did not meet our selection criteria. Of them, eight genes, NOG, TFCP2L1, CommD3, TERT, NR5A2, DPPA5, NODAL, and ITGB1BP3, were selected by convention. Two genes, GJA1 and IMP2, were selected because they expressed at significant higher levels in hESCs when compared with human universal RNA by microarray analysis, although their expression was equally high in EBs. Finally, despite higher level expression in EBs, ITGB1 was included because it is a cell surface receptor that may bind fibronectin and has been reported to be a substrate capable of supporting hESC growth [35, 36].

Markers of the EB State

To represent the complexity of EBs, we included as many types of early markers of differentiation as possible and selected the following: (a) 12 ectoderm markers, including markers for neural precursors such as nestin and Sox1 and for terminal differentiated neural cells like Tuj1 (TUBB3), TH, and GFAP; (b) 19 endoderm markers, including pancreatic marker insulin, imprinted maternally expressed gene H19, HNF, and AFP; (c) 15 mesoderm markers, including collagen, Brachyury, and ACTC; (d) four trophoblast markers, KRT1, EOMES, GCM1, and CDX2; and (e) four germ cell markers, SYCP3, DDX4, IFITM1, and IFITM2 (Table 2).

Expression of Candidate Markers by RT-PCR

Our selection criteria indicated that candidate markers would be expressed in the appropriate stage of development and should be detected by RT-PCR. We therefore generated a PCR primer list for all 109 genes (supplemental online Table 1) and tested 35 genes (Fig. 1 and data not shown; supplemental online Table 1) using three different cell lines. Representative results for a subset of markers for the I6 line are shown. Of the seven ESC markers shown, all were expressed in undifferentiated hESCs. A subset (Dppa5, UTF1, and ZFP42) was undetectable in 14-day EBs, whereas the others were downregulated (GAL, Lin28, Lin41, and TDGF1) in EBs (Fig. 1, left panel). Likewise, the seven EB-specific markers (AFP, FoxA2, Hand1, HGF, IGF2, Msx1, and MSI1) were expressed in EBs but absent (AFP, HNF3b, IGF2, and Msx1) in undifferentiated hESCs or only slightly expressed (Hand1, HGF, and MSI1) in undifferentiated hESCs (Fig. 1, right panel). These results showed the relatively specific pattern of expression of candidate ESC and EB markers and indicate the suitability of using some or all of these to assess the overall state of cultured cells.

Antibodies to Test the Expression of Candidate Markers

Of the 109 genes, we were able to locate commercially available antibodies for about two thirds (76) with reactivity against human antigens. These antibodies and their sources are provided in supplemental online Table 1. We tested a subset of these commercially available antibodies by immunocytochemistry, and at least 15 of them worked well (supplemental online Table 1). Representative staining of ESC- or EB-specific genes in 7-day and 14-day differentiated EBs and in undifferentiated hESCs is shown in Figures 2 and 3. In general, all the ESC markers (Nanog, Oct4, ITGB1, CDH1, and PODXL) were strongly positive in undifferentiated hESCs but weakly expressed in 7-day EBs and not expressed in 14-day EBs (Fig. 2A–L). The one exception was Sox2, which was expressed in both undifferentiated hESCs and the two stages of EBs (Fig. 2M–O). All the markers of differentiation examined (Brachyury, ACTC, AFP, GATA4, GATA6, and TUBB3) were strongly expressed in both 7-day and 14-day EBs but negative or only weakly expressed in undifferentiated hESCs (Fig. 3). In addition to these markers, known pluripotency markers SSEA (SSEA3 and 4) and TRA (TRA-1–60 and -1–81) were downregulated in differentiated EBs (Table 3 summarizes the results of immunocytochemistry). Thus, antibodies to most of the candidate markers exist, and a significant subset can be used for immunocytochemistry. We notice, however, that ITGB1 was strongly positively stained in undifferentiated hESC but downregulated in EBs, which is in conflict with the MPSS and array data. This suggests that not all genes could be used in all methodologies.

Monitor Differentiation by Microarray

Although offering sufficient resolution, RT-PCR and immunocytochemistry are difficult to perform for a large number of genes and cannot be easily automated. To test whether differences in gene expression were of a magnitude sufficient to be detected by a more global and less quantitative measurement, we assessed the expression of candidate ESC and EB markers by analyzing their expression in three hESC lines derived by BresaGen (three undifferentiated hESC samples of BG01, 02, and 03 and EBs derived from them; Thebarton, South Australia, Australia, http://www.bresagen.com.au) using the Illumina Bead Array containing about 48,000 unique features. All samples were examined in duplicate, and only data from duplicate samples that showed 99% or greater correlation were used. The present results were focused on expression of the genes that were selected as candidate markers of the ESC and EB state.

Global pairwise comparisons among different hESC lines (Fig. 4A; supplemental online Tables 2– 4) or different EBs (Fig. 4B) showed similar levels of gene expression, and approximately 90% of the genes detected at a greater than 99% confidence limit were expressed at approximately similar levels (within the 2.5-fold range) (Fig. 4F). Pairwise comparisons of hESCs with hESCs or of EBs with EBs showed a high degree of similarity of samples (correlation coefficient, >0.90). Most of the differential expression seen in Figure 4A and B is the result of biological differences between the cultures; technical replicates have correlation coefficients greater than 0.90 (J.F. Loring et al., in preparation). Comparisons of hESCs with EBs showed a much lower degree of similarity (correlation coefficient, <0.8; Fig. 4C). This suggests that different hESCs are similar to each other and that this similarity is greater than that between hESCs and EBs derived from the same line. The entire comparison is presented in supplemental online Table 2, and a restricted list of genes that were selected as ESC or EB markers showing a greater than 2.5-fold difference in expression is shown in Figure 4G. The large difference between hESCs and EBs detected by this global comparison indicates that arrays can readily distinguish hESCs from the EBs derived from them.

To further test whether the similarity in gene expression among hESC lines can be generalized, we analyzed an additional hESC line, H9, obtained from WiCell Research Institute (Madison, WI, http://www.wicell.org) rather than from Bresa-Gen. As shown in Figure 4D, gene expression profiles of H9 were remarkably similar to those of the BG lines, with a correlation coefficient of 0.93 when compared with BG01. This suggests that gene expression profiles in hESC lines derived from different laboratories are similar.

qPCR to Monitor Differentiation

Our results suggested that a global assessment by a relatively nonquantitative method such as RT-PCR or microarray could be used to detect differentiation. Given the dramatic differences in gene expression, we reasoned that assessing a smaller number of markers using a more quantitative measurement could be sufficient in monitoring the overall state of hESCs. To test this hypothesis, we selected a small number of genes from the 109-gene list and tested their expression in undifferentiated hESCs and in two differentiating stages of EBs (7-day and 14-day) using BG03. These included six undifferentiated ESC markers (Oct4, Nanog, UTF1, DPPA5, Lin41, and Sox2) and six markers of differentiation with at least one gene from each germ layer (Sox1, DCN, H19, IGF2, GATA4, and Hand1). The expression level of these genes was determined as the ratio to the level of 18S RNA, and differential expression of these genes in undifferentiated hESCs and EBs is shown as the ratio of expression in hESCs to EBs (ESC markers) or EBs to hESCs (EB markers).

As expected, the expression of ESC markers Oct4, Nanog, UTF1, DPPA5, and Lin41 was higher in undifferentiated hESCs than in EBs, whereas the expression of EB markers Sox1, DCN, H19, IGF2, GATA4, and Hand1 was upregulated in EBs compared with undifferentiated hESCs (Fig. 5). In particular, expression of UTF1 and Nanog was rapidly downregulated upon differentiation, with more than a 10-fold decrease in 7-day EBs and 200-fold decrease in 14-day EBs. Downregulation of Oct4, an important gene for the maintenance of pluripotency in both hESCs and mESCs, was less marked, with only a fivefold decrease in 14-day EBs. Sox2 was expressed in both undifferentiated hESCs and differentiated EBs (threefold higher in 14-day EBs and fivefold higher in 7-day EBs), which is expected because Sox2 is known to express in neural stem cells that are present in EBs. Interestingly, expression of Lin41 was rapidly decreased in 7-day EBs, but the expression level increased in 14-day EBs (Fig. 5).

All of the differentiation markers, except for Decorin, were rapidly induced as the cells differentiated. The most dramatic changes were seen for an imprinted gene, IGF2, the levels of which were several thousand–fold higher in EBs than in undifferentiated hESCs. Expression of the imprinted gene H19, as well as of Hand1 and GATA4, was also rapidly increased as the cells underwent differentiation.

qPCR Detects Changes That May Be Missed by Immunocytochemistry

To test whether our qPCR assay can detect more subtle changes in hESC cultures that affect the undifferentiated state, we used qPCR and immunocytochemistry to examine hESC cultures maintained with bFGF and cultures in which bFGF was withdrawn for a period of 72 hours. For qPCR assay, we chose to analyze two markers of the ESC (UTF1 and Nanog) and EB (IGF2 and Hand1) state, as expression of these four genes changed most significantly upon differentiation in our qPCR analysis (Fig. 4). No change in expression of either ESC (SSEA4 and Oct4) or EB markers (AFP) could be detected in this time period by immunocytochemistry (Fig. 6). However, qPCR readily detected a significant change in cultures maintained without bFGF for 72 hours: IGF2 and Hand1 were expressed 3.6-fold and 2.3-fold higher, respectively, in hESC cultures without bFGF, whereas no significant changes were observed for the two most differentially expressed ESC markers (UTF, 1.4-fold; Nanog, 1.5-fold) detected by our qPCR analysis. These changes, despite being smaller than those seen in 7-day and 14-day EBs, were similar in profile to changes when cells undergo differentiation.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Since the first derivation of hESC lines in 1998, information on gene expression in hESCs and other human stem cells has been accumulated rapidly using a variety of techniques, including microarray, SAGE, EST scan, and MPSS [1721, 24]. A large number of genes have been identified that are expressed at high levels in undifferentiated hESCs, and “stemness” genes that define a stem cell state have been proposed [18, 37]. Expression of many of these genes is downregulated as hESCs differentiate, and parallel to this, many genes are induced during differentiation. Nevertheless, there is no defined set of markers that can be routinely used for assessing the different states of hESCs (i.e., the undifferentiated ESC stage and different stages of differentiating EBs). In the present study, we compared published reports on gene expression in ESCs and EBs and selected a set of 109 known genes, including 55 stem cell, 12 ectoderm, 19 endoderm, 15 mesoderm, four trophoblast, and four germ cell markers as potential ESC- or EB-specific markers. We show that this set of genes can serve as an indicator of the states of hESCs, using four independent methods, qPCR, immunocytochemistry, RT-PCR, and microarray, and at least three different hESC lines for each method.

It is clear that no single marker is sufficient to define the state of hESCs. Several surface antigens, including SSEA and TRA, are useful markers for undifferentiated hESCs because their level of expression is downregulated as hESCs differentiate. The genes encoding them, however, have yet to be identified. Other pluripotency genes, including two well-characterized transcription factors, Oct4 and Nanog, are good markers to assess the presence of undifferentiated hESCs. Oct4 and Nanog are essential for the maintenance of pluripotency in both hESCs and mESCs, and knockout or knockdown of either gene causes differentiation [38, 39]. Oct4 and Nanog are, however, not uniquely expressed in undifferentiated hESCs: Oct4 is expressed in germ cells, and Nanog has recently been reported to be expressed in mature tissues [40, 41]. Moreover, the expression of Oct4 declines slowly as cells differentiate, and the change in levels is small. For example, we have detected Oct4 expression in hESCs that underwent differentiation for a week or underwent neuronal differentiation on PA6 cells for 2 weeks (X. Zeng, unpublished results). Our qPCR results likewise showed only a moderate decrease in Oct4 expression in 7-day EBs and only a fivefold decrease in 14-day EBs. Taken together, we believe that although SSEA, TRA, Oct4, and Nanog are useful for distinguishing undifferentiated hESCs from their differentiated progeny, expression of these markers alone or in combination is not enough to define the undifferentiated hESC populations. Likewise, expression of a single marker is not a definitive indicator of the differentiated EB state. Indeed, some of the early differentiation genes, including keratin, actin, and tubulin, were expressed at low levels in undifferentiated hESCs; however, their expression was strongly upregulated as hESCs formed EBs [19]. Because the differentiated progeny of hESCs include a number of cell types, it is important to assess EBs using a combination of endoderm, mesoderm, and ectoderm markers.

Our assessment of RT-PCR as a method of examining hESC cultures showed that even though it is not quantitative, it is quite robust provided appropriate genes are selected for assessment. Primers were designed for all 109 genes, and the expression of 35 candidate ESC and EB genes was confirmed by RT-PCR. Our data showed that undifferentiated hESCs could be readily distinguished from differentiating EBs by assessing 10–20 markers using semiquantitative RT-PCR. The relatively specific expression of ESC and EB markers in undifferentiated hESCs and in EBs provides a simple method to assess the quality of RNA samples for different purposes and to estimate the level of differentiation in hESC culture.

In addition to the confirmation of differential expression in hESCs and EBs by RT-PCR, we examined the expression of many of these genes by immunocytochemistry. These included markers that have not previously been analyzed by antibody staining in hESCs: PODXL, ITGB1, and Nanog. Nanog, PODXL, ITGB1, and CDH1 were downregulated in 7-day EBs and further decreased in 14-day EBs. Similarly, expression of differentiation markers such as AFP, GATA4, and GATA6 was strongly upregulated in 7-day EBs. These markers, together with the SSEA and TRA surface markers, can reliably detect the differentiation of hESCs and can be used for routine examination of differentiation in hESC cultures. Although immunostaining is more time consuming, it offers unprecedented resolution, allowing rapid assessment of the degree of contamination or the extent of differentiation.

RT-PCR and immunocytochemistry, however, are not suitable for scaling-up or processing of a large number of markers. We therefore examined whether the genes identified as candidate markers could be used to assess differentiation using a microarray platform. Our results showed that many, though not all, genes show detectable changes in gene expression, even in a relatively poor quantitative method such as microarray. For example, Oct4, Lin28, TDGF1, and GDF3 were present at significantly higher levels in hESCs than in EBs, whereas Col2A1, Col1A1, and SerpinA1 were present at higher levels in EBs using the Illumina BeadArray. The differences in gene expression were present in five hESC lines tested in this study and in six other cell lines evaluated (J.L., personal communication), indicating that these genes may serve as a standard measure of changes irrespective of the cell line being used. Other genes that were expected to serve as useful markers and showed utility in immunocytochemistry and RT-PCR were not as useful in this microarray format. Such genes included Nanog, Sox2, and Sox1 (supplemental online Table 2), indicating that candidate genes will have to be assessed in each individual platform to determine whether they are adequate within the limitations of that particular technology.

Among the 109 genes, six of each of the undifferentiated and differentiated markers (at least one marker of each germ layer) were further examined by qPCR in undifferentiated hESCs and two stages of EBs (7-day and 14-day). We reasoned that careful quantitation may allow one to use only a small subset of markers. Indeed, our results showed that as few as six markers may be sufficient provided both positive and negative markers were used. Whereas downregulation of Oct4 was gradual during differentiation, expression of Nanog and UTF1 declined more than 200-fold in 14-day EBs, suggesting that these markers are good indicators of the undifferentiated ESC state. Dramatic upregulation of expression in EBs was also found for an imprinted gene, IGF2, and for Hand1. Expression of IGF2 and Hand1 was quickly upregulated in 7-day EBs by several thousand folds, and by day 14, the expression levels were 3 million–fold higher for IGF2. It therefore seems that undifferentiated hESCs and their derivatives can be discriminated by examining a few genes using a quantitative method if the genes are appropriately selected. UTF1 and Nanog are excellent candidates for markers of the undifferentiated state, whereas IGF2 and Hand1 are good markers for differentiated EBs. The dramatic changes in expression level of these four genes upon differentiation can be reliably used for assessing the undifferentiated ESC and differentiated states. Moreover, negative markers (differentiated EB markers) are more sensitive than positive markers (undifferentiated ESC markers) in detecting differentiation in hESCs. These conclusions are supported by experiments designed to detect smaller changes of differentiation in hESC cultures in which bFGF was removed for a period of 3 days. Whereas immunocytochemistry could not detect any difference in bFGF-treated and bFGF-withdrawn cultures using ESC or EB markers, bFGF-withdrawn hESC cultures could be readily distinguished from their bFGF-maintained sister culture by qPCR using negative markers (IGF2 and Hand1).

As more data on hESCs are collected, additional markers for undifferentiated and differentiated cells will undoubtedly be identified. For example, a large number of novel genes or genes of unknown function that show a similar robust alteration in expression levels as hESCs differentiate [18] may be included in future arrays or qPCR sets to provide an additional level of sensitivity and allow a finer dissection of the state of differentiation. The present lists, however, provide useful information for evaluating the states of hESC populations and the extent of differentiation or for quality control of hESC cultures. Our results suggest that any of the four methods we describe here can be used to monitor the transition of undifferentiated hESCs to differentiated EBs when a combination of ESC and EB markers from our list is tested.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Our strategy of including a combination of genes that are downregulated and upregulated during differentiation, which includes genes that represent different cell types (undifferentiated cells and cell types of the three germ layers, as well as trophoblast and germ cells) allows one to identify a set of markers that can be readily assessed by routine molecular or cellular biology methods. We believe that any of the methods we tested is sufficient to monitor the state of hESC, but each method has its advantages and disadvantages. If qPCR is used, a small number of genes is sufficient provided both positive and negative markers are used (present results and [42]). However, the most cost-effective method for the wealth of information obtained may be a focused array that includes many markers such as the genes we have described. Efforts to generate such an array are in progress ([43], Ian Lyons, Invitrogen, personal communication). Alternatively, microfluidic plates allow the custom design of markers and have the advantages of being able to be adapted to a very small number of cells.

Table Table 1.. Human embryonic stem cell–specific markers
Thumbnail image of
Thumbnail image of
Table Table 2.. Embryoid body specific markers
Thumbnail image of
Thumbnail image of
Table Table 3.. Expression levels of embryonic stem cells and embryoid body markers in human embryonic stem cells
Thumbnail image of
thumbnail image

Figure Figure 1.. Differential expression of ESC- or EB-specific genes by RT-PCR. Expression of selected markers of ESCs (Dppa5, GAL, LIN28, LIN41, TDGF1, UTF1, and ZFP42) and EBs (AFP, FoxA2, HAND1, HGF, IGF2, Msx1, MSI1) was examined by RT-PCR in undifferentiated hESCs and in 7-day and 14-day EBs. Consistent with other independent analyses, all the ESC markers are highly expressed in hESCs but are quickly downregulated in the two EB populations, whereas all the EB markers are detected in EB samples but are slightly expressed or not expressed in undifferentiated hESCs. Abbreviations: EB, embryoid body; ESC, embryonic stem cell; hESC, human embryonic stem cell; RT-PCR, reverse transcription–polymerase chain reaction.

Download figure to PowerPoint

thumbnail image

Figure Figure 2.. Staining of ESC markers on undifferentiated hESCs and on 7-day and 14-day EBs. ESC markers Nanog, ITGB1, CDH1, PODXL, Oct4, and Sox2 are expressed by most of the undifferentiated hESCs (A, D, G, J, M), whereas their expression was downregulated in both 7-day and 14-day EBs ([B, C, E, F, H, I, K, L], red in [N, O]) except Sox2 (green in [N, O]), which is also a neural stem cell marker. Abbreviations: EB, embryoid body; ESC, embryonic stem cell; hESC, human embryonic stem cell.

Download figure to PowerPoint

thumbnail image

Figure Figure 3.. Staining of differentiated markers on 7-day and 14-day EBs and on undifferentiated hESCs. Mesoderm markers Brachyury and ACTC, endoderm markers AFP, GATA4, and GATA6, and ectoderm marker TUBB3 were detectable in both 7-day and 14-day EBs (B, C, E, F, H, I, K, L, N, O), but their expression was not detected in undifferentiated hESCs (A, D, G, J, M). Spontaneously differentiated hESCs also expressed Brachyury (C), GATA4, and GATA6(I). Abbreviations: EB, embryoid body; ESC, embryonic stem cell; hESC, human embryonic stem cell.

Download figure to PowerPoint

thumbnail image

Figure Figure 4.. Assessment of potential hESC and EB markers using microarrays. Four different ESC lines (BG01, 02, 03, and H9) and 14-day EBs that differentiated from them were compared using an Illumina BeadArray. (A, B): Comparisons of gene expression among three hESC lines or three EBs show similar levels, with a correlation coefficient (r2) greater than .93. (C): Pairwise comparisons of undifferentiated hESCs with their matched EBs reveal that approximately 50% of the genes are expressed with at least a 2.5-fold difference. (D): Comparisons of gene expression between two hESC lines derived from different laboratories (BG-01 and H9) show similar levels, with a correlation coefficient (r2) greater than .92. (E): Summary of the numbers of genes detected in this array. (F–G): Selected genes that are differentially expressed in three hESC lines (BG01, 02, and 03) and in their matched EBs. Note that only the genes that are detected at >0.99 confidence (blue dots) are considered valid for further analysis. Dots that fall between the thin red lines represent genes that are commonly expressed in hESCs and EBs, whereas dots outside the red lines correspond to differentially expressed genes at >2.5-fold. Abbreviations: EB, embryoid body; ESC, embryonic stem cell; hEB, human embryoid body; hESC, human embryonic stem cell.

Download figure to PowerPoint

thumbnail image

Figure Figure 5.. qPCR analysis of 12 genes during hESC differentiation. Expression of six ESC markers and six EB markers was quantified by qPCR. (A): Amplification curves corresponding to IGF2 and 18S RNA (standard curve) are shown from left to right: IGF2 of 14-day EB (blue and red); 18S RNA of 14-day EB (red and blue); 18S RNA of undifferentiated hESCs (green and red); and IGF2 of undifferentiated hESCs (blue and yellow). (B): The expression level of these genes was determined as the ratio to the level of 18S RNA, and differential expression of these genes in undifferentiated hESCs and EBs was shown as the ratio of expression in hESCs to EBs (ESC markers) or EBs to hESCs (EB markers). Abbreviations: EB, embryoid body; ESC, embryonic stem cell; hESC, human embryonic stem cell; qPCR, quantitative reverse transcription–polymerase chain reaction.

Download figure to PowerPoint

thumbnail image

Figure Figure 6.. No difference in ESC and EB marker expression by immunocytochemistry between hESC cultures maintained with bFGF and cultures in which bFGF was withdrawn for a period of 72 hours. (A, B): Immunostaining of SSEA4 (red) shows similar expression levels in these two cell populations (with and without supplement bFGF). (C, D): Immunostaining of Oct4 (red) shows that most of the cells are positive, whereas only occasional AFP-positive cells (green) are seen outside the colonies in both hESC populations. Scale bar = 100 μm. Abbreviations: AFP, alpha-fetoprotein; bFGF, basic fibroblast growth factor; EB, embryoid body; ESC, embryonic stem cell; FGF, fibroblast growth factor; hESC, human embryonic stem cell.

Download figure to PowerPoint

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

We acknowledge the support of Children's Neurobiological Solutions Foundation, National Institute on Aging, National Institute on Drug Abuse, and the Packard Center. We thank Drs. Tim McDaniel and David L. Barker for sharing the Illumina BeadArray analysis and Rose Amable for technical assistance. J. Cai and J. Chen contributed equally to this work.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
SC050143SuppFile1.pdf1513KSupplemental Data

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.