Correlation of Murine Embryonic Stem Cell Gene Expression Profiles with Functional Measures of Pluripotency
http://www.100md.com
《干细胞学杂志》
a Terry Fox Laboratory, BC Cancer Agency, Vancouver, British Columbia, Canada;
b Michael Smith Laboratories, University of British Columbia, Vancouver, British Columbia, Canada;
c Department of Cancer Endocrinology, BC Cancer Agency, Vancouver, British Columbia, Canada;
d Departments of Chemical and Biological Engineering,
e Medicine, and
f Surgery, University of British Columbia, Vancouver, British Columbia, Canada
Key Words. Embryonic stem cells ? Markers ? Pluripotency ? Leukemia inhibitory factor ? Gene expression ? Microarray
Correspondence: Cheryl D. Helgason, Ph.D., Department of Cancer Endocrinology, BC Cancer Agency, 675 West 10th Avenue, Vancouver, BC, Canada, V5Z 1L3. Telephone: 604-675-8011; Fax: 604-675-8183; e-mail: chelgaso@bccrc.ca
ABSTRACT
Embryonic stem cells (ESCs) are characterized by their ability to both self-renew and differentiate . However, the molecular mechanisms that regulate the decision between these two processes are poorly understood. Mouse ESCs were originally isolated from the inner cell mass (ICM) of preimplantation blastocysts and can be maintained in cell culture indefinitely without loss of their broad pluripotent differentiation capacity as determined by their ability to give rise to all three germ layers both in vitro and in vivo . The more recent establishment of human ESC lines has further increased the interest in ESCs because they raise hope of an unlimited source of cells for tissue engineering and cell therapies in the future. However, realization of this potential requires an increased knowledge of the molecular mechanisms governing self-renewal and pluripotency to guide the development of processes that control the expansion and differentiation of stem cells ex vivo.
Unlike hematopoietic stem cells, for which quantitative assays of stem cell potential have been defined and validated, no such assays currently exist for ESCs. In the murine system, self-renewal is measured by the ability of mouse ESCs to continuously proliferate in culture while maintaining an undifferentiated colony morphology . The most rigorous in vivo assay to establish functionality of cultured mouse ESCs is blastocyst injection and measurement of their ability to give rise to chimeric mice because it requires ESC contribution to all adult tissue, including germ cells . However, injection of 10 to 15 ESCs into a single blastocyst does not provide a quantitative measure of stem cell potential at the single-ESC level. Two in vitro assays have been used extensively as surrogates for chimera formation when testing culture reagents or examining the consequences of genetic manipulation. The colony-forming cell (CFC) assay is used to determine the plating efficiency of ESC populations under various conditions and thus may be considered indicative of self-renewal potential. Formation of embryoid bodies (EBs) can be performed at a clonal level in vitro and reflects multilineage differentiation potential . The correlation between these in vitro assays and chimera generation has not been determined. Assessment of pluripotency has also relied on the expression of selected molecular markers. For murine ESCs, these have included alkaline phosphatase, the POU transcription factor Oct-4, and stage-specific embryonic antigen 1 (SSEA-1) . However, the correlation between marker expression and the various functional assays has not been extensively studied. Knowledge of the intricate mechanisms regulating ESC pluripotency and differentiation potential is currently limited to a few signaling pathways (i.e., leukemia inhibitory factor ) and regulatory factors (i.e., Oct-4 and Nanog). Thus, very little is known about the tolerance limits of different culture conditions for maintaining stem cell function during expansion or how these relate to altered gene expression patterns in ESCs. Identification of molecular markers that correlate with pluripotency would be invaluable to enrich for the desired cells, as well as to monitor their maintenance during expansion protocols.
Achieving the goal of defining the core stem cell regulatory network requires a precise characterization of the functional capacities of the cells for which the transcriptional profile is described. In this study, we established gene expression profiles during early differentiation of the well-defined R1 ESC line and correlated gene expression changes with both phenotypic and functional assessment of the same cells. Functional capacity was determined by blastocyst injections for chimeric mouse formation, EB assays, and CFC counts. Undifferentiated ESCs and ESCs cultured without LIF for 18 or 72 hours were chosen for gene-array analysis. We identified 473 unique genes as significantly differentially expressed during early ESC differentiation, and approximately one third of these have unknown biological function. Among the 275 genes whose expression decreased with ESC differentiation were several factors previously identified as important for, or markers of, ESC pluripotency, including Stat3, Rex1, Sox2, Gbx2, and Bmp4. A significant number of the decreased genes also overlapped with previously published mouse and human ESC data. Reverse transcription–polymerase chain reaction (RT-PCR) validation showed high correlation with the gene-array data, and several genes were also shown to have similar changes after LIF removal in two other murine ESC lines. Expression of the commonly used ESC markers Oct-4 and SSEA-1 was also examined in parallel with the functional assays. However, a close correlation was not observed. Interestingly, among a subset of 48 decreased genes that showed the closest correlation with the functional assays was the stem cell factor (SCF) receptor c-Kit that can be a useful marker of undifferentiated ESCs.
MATERIALS AND METHODS
Pluripotency During Early Differentiation
We hypothesized that loss of ESC pluripotency as defined by measurable functional readouts would correlate with significant alterations in gene expression. Identification of these gene expression changes would thus provide important insights into the genetic regulation of ESC pluripotency and facilitate the identification of new molecular markers of the undifferentiated ESC state. We relied on comparisons of the three measures of ESC potential (chimera formation, EB formation, and CFC assay) to select time points for analysis of gene expression profiles. Undifferentiated ESCs were cultured in medium containing LIF on irradiated primary MEF, which supply several as-yet unidentified factors that enhance the plating efficiency of the ESCs as well as assist in the maintenance of the undifferentiated state. Preplating of ESCs removes the MEF and enriches for ESCs with the capacity to contribute to the developing blastocyst. It has been suggested that those ESCs capable of loosely attaching to the feeder layer in a short (i.e., 1 hour) period of time have the highest likelihood of forming colonies (i.e., self-renewing) and thus may also be the most competent at contributing to the germ-line after blastocyst injection . We thus elected to use only preplated, loosely adherent mouse ESCs for our analyses.
The baseline activities for undifferentiated ESCs in the three different assays were first determined. Undifferentiated, pre-plated R1 ESCs yielded 100% chimeric pups after blastocyst injection (27 blastocysts injected; six born and analyzed in two independent experiments), 6.9% ± 1.0% of plated cells differentiated into EBs in the EB formation assay, and 12.50% ± 2.2% of plated cells gave rise to alkaline phosphatase–positive ESC colonies in the CFC assay. These results were within the normal range for the R1 cell line. ESC differentiation was then initiated by LIF removal and replating without MEF. The morphology of the ESCs before and after LIF removal is shown in Figure 1. At 18 and 24 hours after MEF and LIF removal, the ESCs looked very similar to one another and did not exhibit any clear signs of differentiation. Morphological differentiation was first apparent at 48 hours, and EBs could be seen after 72 hours. Despite the lack of appreciable phenotypic differentiation within the first 24 hours, there were significant changes in the functional properties of the ESCs. Cultured cells were harvested at various time points during this culture period and analyzed in the three different assays. The blastocyst injection assay showed a rapid decrease in the number of chimeras obtained after initiation of ESC differentiation (Fig. 2A). Only 28% of born pups (84 injected, 25 born and analyzed in two independent experiments) were chimeras when ESCs had differentiated for 24 hours, and less than 5% were chimeras after 72 hours of differentiation (66 injected, 29 born and analyzed in two independent experiments). The EB formation assay (Fig. 2B) showed approximately 5.5% ± 1.0%, 3.7% ± 0.3%, and 0.3% ± 0.1% readout at 18, 24, and 72 hours after LIF removal, respectively. In contrast, the frequency of cells replating in the CFC assay increased slightly during the first 24 hours but subsequently declined rapidly such that only 1.3% ± 0.3% retained this activity at 72 hours (Fig. 2C). In summary, all three assays showed clearly that most differentiation potential and self-renewing capacity was gone after 72 hours of differentiation. However, there was a high degree of variation during the first 24 hours amongst the in vitro and in vivo assays, with the EB formation assay correlating most closely with chimeric mouse formation. For the gene expression profiling, because both the EB formation and chimera assays showed a pronounced decrease within the first 18–24 hours, we elected to use the earlier time point, along with 72 hours of differentiation, to compare against the undifferentiated R1 ES.
Figure 1. Embryonic stem cell (ESC) morphology. Morphology of the R1 ESCs grown on mouse embryo fibroblasts (MEFs) with leukemia inhibitory factor (LIF) is shown. ESCs were followed for 72 hours after MEF and LIF removal, and pictures were taken at the indicated time points and with the indicated resolution.
Figure 2. Assays of embryonic stem cell (ESC) pluripotency. Comparison of in vitro and in vivo functional measures of ESC potential with expression of the commonly used ESC markers Oct-4 and stage-specific embryonic antigen 1 (SSEA-1). R1 ESCs were thawed, cultured, and assayed as outlined in Materials and Methods. The frequency of cells capable of (A) generating chimeric mice (summary of two independent experiments), (B) giving rise to embryoid bodies (summary of three independent experiments), or (C) giving rise to colonies in the colony-forming cell assay (summary of three independent experiments) is shown as a function of time after leukemia inhibitory factor removal. (D): Flow cytometric analysis (summary of three independent experiments) was used to determine the percentage of gated cells expressing cell-surface SSEA-1 and intracellular Oct-4 at the same time points. Data are the mean ± standard error of the mean for replicated experiments.
In parallel with the functional assays, the expression patterns of two markers commonly used to identify undifferentiated ESCs, the POU transcription factor Oct-4 and SSEA-1, were also analyzed to establish the level of correspondence in the readouts using each method. Although Oct-4 and SSEA-1 have been extensively used in ESC research, their expression patterns during ESC differentiation have not been studied in detail. Oct-4 is used as a marker for ESCs because of its requirement in ESC self-renewal . The precise expression level of Oct-4 is important for determining ESC fates, and repression of Oct-4 induces loss of pluripotency and dedifferentiation to trophectoderm . However, forced constitutive expression of Oct-4 cannot prevent ESC differentiation, and a less than twofold increase in expression actually causes differentiation into primitive endoderm and mesoderm . Thus, a critical amount of Oct-4 is required to sustain stem cell self-renewal but is not sufficient to prevent differentiation. SSEA-1 is a glycoprotein expressed during early embryonic development and by undifferentiated ESCs. However, the precise role of SSEA-1 in pluripotency and self-renewal has not been defined. ESCs selected for expression of SSEA-1 and platelet endothelial cell adhesion molecule 1 are enriched for cells that differentiate predominantly into epiblast cells in chimeric embryos . In the present study, protein expression of both SSEA-1 and Oct-4 remained relatively unchanged throughout the first 48 hours of differentiation (Fig. 2D). At 120 hours after LIF removal, 18.5% ± 1.1% of the cells retained expression of SSEA-1, whereas 40.6% ± 6.6% of the cells continued to express Oct-4. Thus there was no clear correlation between expression of these two markers and the various ESC functional assays.
Gene-Array Analysis
Cultured R1 ESCs from matched passage numbers were collected in three separate experiments at 0, 18, and 72 hours of differentiation after LIF removal and were analyzed using the Affymetrix GeneChip MG-U74v2 array containing 36,767 different probe sets. Duplicate samples of the MEF feeders were also collected and analyzed to assess any possible contamination of the undifferentiated ESCs. Hybridization, scanning, and production of raw data files were performed according to standard protocols. The MAS 5.0 software was used for the initial scaling and expression analysis. The data were then normalized and further analyzed using the GeneSpring software. To validate the reproducibility and the overall variation of the data, hierarchical clustering analysis was performed. The data were first filtered for genes present in at least one of the three time points (present in at least two out of three replicates), resulting in 13,002 different probe sets. The average number of present genes for each time point was 12,818 (coefficient of variation , 6%) in undifferentiated ESCs, 11,806 (CV, 4%) at 18 hours, and 12,297 (CV, 1%) at 72 hours after LIF removal. The number of expressed genes at the different time points was not statistically significantly different. Hierarchical clustering was applied to the reduced gene set on individual array samples (three replicates for each time point) using Pearson’s correlation and average linkage clustering as implemented in GeneSpring. The individual samples from the three experiments clustered tightly together according to their respective time points, as seen in Figure 3, indicating that the overall interexperimental variation was low. The clustering also reflects the temporal progression of the ESC differentiation. As expected, the first two time points (0 and 18 hours) clustered more closely together, whereas the 72-hour time point showed a more distinct expression pattern, resulting in a greater relative distance from the other two time points, consistent with the functional data (Fig. 2).
Figure 3. Distance tree. Hierarchical clustering of the individual gene-array samples was carried out using Pearson’s correlation and average linking clustering to determine reproducibility and interexperimental variability. Relative distance values are shown.
Genes that were differentially expressed after LIF removal were determined using the following criteria: The difference in expression level between two time points was at least twofold, the gene was present in all three replicates at the time points with the highest abundance, and the extent of difference in expression was statistically significant (p < .05) in a parametric Welsh ANOVA t-test. With this approach, 473 unique genes were identified as significantly differentially expressed (275 decreased, 194 increased, and 4 both decreased and increased) during early ESC differentiation after LIF removal (complete gene lists in the supplemental data, supplementary online Table 2). Unigene and RefSeq IDs were used to exclude redundant probe sets on the array and to get the true number of affected genes.
Only preplated, loosely adherent ESCs were used in our experiments, and visual assessment of MEF contamination of undifferentiated ESCs suggested it was consistently less than 1 in 500. However, even with these measures, we sought to exclude the possibility that contaminating MEF cells may have distorted the data. Duplicate samples of MEFs were therefore also analyzed on the U74v2 GeneChip. The data from both the MEF and the R1 ESC was scaled in MAS 5.0 and normalized in GeneSpring 6.2 before 1% of the raw value (five times more than the estimated contamination) obtained for the gene in MEF was subtracted from the value of the same gene in the undifferentiated R1 ESCs. The analysis steps to find differentially expressed genes were then performed as described above. Only genes detected as decreased during differentiation in the original analysis were reanalyzed. Genes whose expression increased during differentiation would not be affected by MEF contamination in a way that would give false-positive results. This evaluation indicates that none of the genes that showed a significant twofold or greater decrease in the initial analysis lost their significance after MEF subtraction (data not shown) and suggests that the level of MEF contamination was not sufficient to distort the ESC gene expression data.
Gene Expression Validation
The expression patterns of some genes previously implicated in maintaining ESC pluripotency and self-renewal were analyzed to validate the data and the approach to identify differentially expressed genes (Table 1). Mouse ESCs are usually cocultured on a feeder layer of irradiated MEFs. In addition to providing a matrix for attachment, MEFs produce LIF, required for propagation of pluripotent mouse ESCs . LIF-null MEFs cannot support self-renewal . However, LIF also needs to be complemented by fetal calf serum to block differentiation. LIF binds to the gp130 receptor that leads to activation of the transcription factor Stat3 . Stat3 was clearly detected in undifferentiated R1 ESCs, and LIF removal decreased Stat3 expression with the most pronounced effect during the first 18 hours (49% reduction). This was also seen for two other genes involved in the LIF/gp130 pathway, the Oncostatin M receptor, Osmr, and the interleukin 6 signal transducer, Il6st. A known Stat3 target gene, Pim1 , was also decreased significantly at the transcriptional level. Taken together, these observations confirm that the LIF/gp130/Stat3 pathway was rapidly shut down after MEF and LIF removal. The bone morphogenic proteins can act in combination with LIF to sustain self-renewal and preserve multilineage differentiation, chimera contribution, and germ line transmission properties . Bmp4 transcript levels were significantly decreased during differentiation. However, the onset of the decrease was later than for Stat3 (between 18 and 72 hours, 80% reduction). The same was seen for Rex-1/Zfp42, known to be highly expressed in undifferentiated ESCs and downregulated after retinoic acid–induced differentiation . A 90% reduction in Rex-1/Zfp42 expression was seen between 18 and 72 hours after LIF removal. Transcript abundance of the Akp2 gene coding for alkaline phosphatase, commonly used as a marker for undifferentiated ESCs, was also decreased significantly (55% reduction between 18 and 72 hours). The POU transcription factor message, pou5f1, coding for Oct-4 was highly expressed in undifferentiated ESCs but was not changed significantly during the first 72 hours after LIF removal in keeping with the protein expression data (Fig. 2D). Transcript levels of the embryonal stem cell–specific gene 1 (Esg-1 or developmental pluripotency associated gene 5, Dppa5) did not change during differentiation. However, this observation is consistent with the observation that Esg-1 probably is a downstream target of Oct-4 and is downregulated slowly after Oct-4 suppression . The abundance of FoxD3, a gene expressed early in mouse embryonic development, also remained unchanged during the first 72 hours after LIF removal. The SRY-box containing transcription factor Sox2 may act to maintain ESC pluripotency and is expressed in the ICM, epiblast, and germ cells, just like Oct-4 . Sox2 was significantly decreased (60% reduction between 18 and 72 hours, p = .0136). Sox2, together with Oct-4, is involved in the regulation of fibroblast growth factor 4 (Fgf4), another factor confined to the ICM of the blastocyst . Although Fgf4 transcripts were not significantly decreased (p = .09), it was reduced more than twofold (52% reduction between 18 and 72 hours). This was also the case for the newly discovered marker of ESC pluripotency Dppa3 , which decreased 78% between 0 and 72 hours (p = .073). Similarly, the regulatory factor Nanog, which can rescue ESCs from LIF/Stat3 dependence and maintain Oct-4 expression , was not quite significantly decreased during the first 72 hours of differentiation (p = .07), and the reduction in expression was less than twofold (41%). Several genes indicative of ESC differentiation were increased after LIF removal. For example, the mesoderm marker Brachyury increased 15-fold between 18 and 72 hours (p = .003). The ectoderm marker Nestin increased approximately threefold between 18 and 72 hours (p = 0.04), and the epithelial cell marker Prominin 1 increased fivefold between 18 and 72 hours (p = .0010). In conclusion, most of these changes are consistent with the loss of ESC pluripotency as measured in all three assays and indicate that the thresholds used to define differentially expressed genes in this study are reasonable.
Table 1. Genes previously reported as enriched in undifferentiated ESCs or to be markers of ESC differentiation were compared with the genes identified as differentially expressed in the present study
RT-PCR Validation
To further confirm the fidelity of the gene-array data, a set of 28 genes was selected and their transcript levels were tested using quantitative RT-PCR in R1 ESCs before and after LIF removal. The genes selected included some of the genes mentioned above, as well as other genes increased, decreased, or unchanged after LIF removal according to the gene-array results. The fold change was calculated between 0 and 18 hours, 18 and 72 hours, and 0 and 72 hours, respectively. The individual RT-PCR results can be found in the supplemental data (supplementary online Table 3). The analysis revealed a high degree of correlation between the gene-array data and the RT-PCR results (Pearson’s correlation, r = 0.87, Fig. 4A), with a tendency that the PCR results showed greater changes than the array suggested (proportional bias 1.2 in a Deming method comparison analysis, not shown).
Figure 4. Correlation between gene array and quantitative reverse transcription–polymerase chain reaction (RT-PCR). (A): Pearson’s correlation analysis was done on the log ratio values obtained from the quantitative real-time RT-PCR and the gene-array fold changes (n = 84). (B, C, D): Comparison of quantitative real-time RT-PCR results from the three different embryonic stem cell lines are shown. Pearson’s correlation analysis was done on the log ratios of the values obtained with the 2–CT method (n = 45 in each comparison).
To determine if the gene expression changes observed in the R1 ESCs are general and more broadly observed, the expression patterns of 15 genes were followed for 72 hours after LIF removal in two other ESC lines, J1 and EFC , and compared with the R1 ESC line using quantitative RT-PCR (Table 2). Overall, there was a good correlation amongst all three ESC lines (Figs. 4B–4D). Specifically, RT-PCR analysis was able to verify the expression changes of genes such as Lox, Ankrd1, and c-Kit that showed significantly decreased transcript levels after 18 hours in all three ESC lines tested. Similarly, Rex1, Sox2, Leftb, and Mtf2 were all changed in a similar fashion in the R1, J1, and EFC cell lines. The RT-PCR results also confirmed that Oct-4 transcript levels were not decreased significantly during the first 72 hours of differentiation in any of the cell lines, in agreement with the gene-array results and the protein level observed (Fig. 2D). In contrast, although the decrease in Nanog transcripts was not quite significant in R1 ESCs according to the gene array, the RT-PCR results indicated a reduction. Nanog was reduced by 50%–70% after 72 hours in all three ESC lines according to the quantitative RT-PCR analysis but consistent with delayed reduction in Nanog upon differentiation (Table 2). Only two of the tested genes showed different expression patterns amongst the cell lines. The mesoderm marker Brachyury and the homeobox transcription factor Pbx3 were both increased during differentiation in the R1 and J1 ESCs, but not in the EFC cell line. This observation suggests that the various ESC lines might follow slightly altered differentiation pathways upon LIF removal.
Table 2. Comparison of relative expression levels obtained by quantitative real-time RT-PCR in the R1, J1, and EFC ESC lines during differentiation
These validation analyses demonstrate that the early changes during differentiation uncovered by the gene-array analysis could be verified with an independent method and could in most cases be observed in multiple ESC lines. Taken together, they provide confidence in our approach to identify differentially expressed genes with a high likelihood of exhibiting true expression level changes during the loss of ESC pluripotency.
Functional Classification of Differentially Expressed Genes
It is likely that genes showing similar functional properties and expression patterns form interacting networks that contribute to the phenotypic and functional characteristics of the cells of interest. Functional classification of the differentially expressed genes into appropriate biological processes was thus performed using the NetAffx and GeneOntology Express tools as well as the simplified gene ontology tool in the GeneSpring software package. The differentially expressed genes were separated into 13 main categories (Fig. 5A; see supplementary online Table 4 for complete lists). Many differentially expressed genes were, as expected, classified as being involved in development and differentiation (61 genes, 10%) or directly or indirectly involved in cell-cycle control and cell proliferation (32 genes, 5%). Furthermore, at least 81 genes (13%) were classified as being involved in intracellular signal transduction, cell–cell signaling, or response to external stimuli (supplementary online Table 4).
Figure 5. Annotation of differentially expressed genes. All genes differentially expressed during embryonic stem cell differentiation were classified according to (A) biological process and (B) cellular component. Some genes are classified in more than one category, resulting in the total number of genes indicated in the figures being greater than the total number of differentially expressed genes.
A closer look at the signaling pathways affected during ESC differentiation revealed that the Yamaguchi sarcoma viral oncogene gene (Yes) was decreased significantly between 0 and 72 hours (60%). Yes has recently been shown to be regulated by LIF and to be important for ESC self-renewal . Yes is a gene coding for an Src tyrosine kinase expressed in both mouse and human ESCs and is downregulated when these cells differentiate . Another significantly decreased factor was the Notch ligand Jagged-1 (55% reduction between 0 and 18 hours), which has been implicated in hematopoietic stem cell self-renewal . Jagged-1 is involved in embryonic vascular development, and a Jag1 knockout is embryonic lethal .
A signaling pathway that is of particular interest during ESC differentiation is the mitogen-activated protein kinase (MAPK) pathway. The self-renewal of ESCs is influenced by the MAPK pathway, in which expression of Erk and Shp-2 counteracts the proliferative effects of Stat3 and promotes differentiation . According to the array results, the Erk gene (Mapk1 or Mapk3) and most other components in this pathway were already present in undifferentiated ESCs, and their expression levels did not change significantly at the transcription level during the first 72 hours of differentiation (data not shown). However, some genes in the MAPK pathway (e.g., Kras2 and Mapk12) had significantly increased transcript levels during differentiation, indicating an activation of the pathway (supplementary online Table 4). Furthermore, the gene coding for Grb2-associated binder 1 (Gab1) was significantly decreased (58% reduction between 18 and 72 hours, p = .0035). Gab1 binds to Shp-2 and is believed to suppress the MAPK pathway in ESCs; also, increased synthesis of Gab1 together with Oct-4 may suppress induction of differentiation . Although most of the factors involved in the MAPK pathway are primarily regulated at the posttranscriptional level, the fact that transcripts for many pathway members were detected in undifferentiated ESCs suggests that they harbor the required components to quickly respond to signals that promote differentiation.
The Wnt signaling pathway is important for maintenance of pluripotency in undifferentiated ESCs, and a recent study also showed that activation of the Wnt pathway by pharmacological inhibition of Gsk-3 is able to maintain pluripotency in both human and mouse ESCs . Most components of the Wnt signaling pathway could be detected in both undifferentiated and differentiated ESCs, but some factors were also significantly changed during differentiation (e.g., Fzd5, Wnt3, and CyclinD3; supplementary online Table 4). However, overall there was no dramatic change at the transcription level of several important factors belonging to this pathway during the first 72 hours of differentiation (e.g., ?-catenin, Gsk-3, and Axin; data not shown).
The Hedgehog signaling pathway plays a critical role during development and has been extensively studied in different species and tissues . In this pathway, two receptors, Ptch and Ptch2, and two transcription factors, Gli1 and Gli2, were decreased significantly during ESC differentiation. Furthermore, at least one suggested downstream target of this pathway, the Bmp4 gene already discussed above, had significantly decreased transcription, providing additional evidence that this pathway might be involved in maintaining ESC self-renewal or pluripotency.
Importantly, out of 473 differentially expressed genes, the largest group consisted of 173 genes or expressed sequence tags with unknown biological function. Further analyses of the genes within this group whose expression changes closely correlate with loss of pluripotency may reveal novel mechanisms involved in ESC maintenance. Of the 275 genes significantly decreased during ESC differentiation, 48 genes showed a more than twofold decrease in transcript levels within the first 18 hours and continued to decrease or remained at a low level of expression until 72 hours (Table 3). Changes in the transcript levels for these genes correlated well with the functional assays, especially the chimera generation and EB formation assays (Fig. 1). At least half of them have been identified as ESC-enriched or to be downregulated during differentiation in previous studies (see below), but several of these genes have unknown biological function or have not previously been implicated in ESC maintenance (e.g., Lox, Tnc, and Jag1; Table 3).
Table 3. Genes decreasing during ESC differentiation after LIF removal and most closely correlating with the loss of pluripotency
Molecular Markers of Pluripotency
We hypothesized that classification of differentially expressed genes according to cellular component, in combination with functional analyses, may lead to the identification of new markers that more accurately reflect ESC potential. Classification of the significantly changed genes according to cellular component is shown in Figure 5B (complete list provided in supplementary online Table 5). At least 60 gene products (11%) were considered localized to the plasma membrane and might therefore be good candidates as markers for undifferentiated ESCs. Genes increased during differentiation were also included in this list of possible ESC markers because they can be used to detect the onset of differentiation or for negative selection strategies for isolating undifferentiated ESCs (Table 4). Protein expression analyses of some of these have already been conducted in relation to ESC differentiation, i.e., Cd9, Cd44, and Osmr . Additional gene products of interest include the adhesion molecule Vcam1 and the hedgehog signaling pathway receptors Ptch and Ptch2. One gene of special interest encodes the SCF receptor c-Kit. According to the gene-array results, transcript level of the c-Kit gene was one of the most significant changes during differentiation (54% reduction between 0 and 18 hours and 75% reduction between 0 and 72 hours; p = .0016; Table 3), which also correlated well with the functional assays (Fig. 1). Furthermore, this finding was also verified in the R1, J1, and EFC ESC lines by quantitative RT-PCR (Table 2). No c-Kit ligand/SCF was added to the culture medium used in these experiments, and SCF was not detected in the undifferentiated R1 ESCs, although it was indeed present at high levels in the MEF according to the gene-array results (data not shown). Because c-Kit is expressed on several pluripotent cell types, including germ cells and hematopoietic stem cells, and antibodies are commercially available, we were able to compare the results of the array and RT-PCR experiments with protein expression levels measured by flow cytometry (Fig. 6). This analysis revealed that expression of c-Kit on the cell surface of differentiating ESCs closely paralleled the loss of functional potential as measured using the EB formation assay (Fig. 7). At later time points, the correlation was not as close (i.e., EB formation capacity continued to decrease while protein expression remained unchanged), suggesting the emergence of differentiated cell types expressing c-Kit protein.
Table 4. Differentially expressed genes with plasma membrane localization
Figure 6. Fluorescence-activated cell sorting profile. Undifferentiated R1 embryonic stem cells (ESCs) (0 hour) and R1 ESCs differentiated 18 and 72 hours after leukemia inhibitory factor removal were analyzed with flow cytometry using c-Kit antibody. Abbreviations: FSC, forward scatter; SSC, side scatter.
Figure 7. Correlation between c-Kit expression and embryoid body formation. Expression of c-Kit () was monitored at the indicated times during embryonic stem cell differentiation using flow cytometry, and expression levels were plotted as percent gated cells versus time (right axis). Embryoid body (EB)–forming cell frequency () determined on the same cell populations is also shown (percent EB formation indicated on the left axis).
Comparison with Previously Published Gene-Array Data Sets
Gene expression profiling to determine regulatory factors and signaling pathways present in ESCs has been performed extensively in recent years . It has been hypothesized that the undifferentiated state is conferred on various stem cell populations through the use of similar molecular mechanisms. Initial support for this hypothesis came from experiments comparing the gene expression profiles of multiple stem cell populations . However, further analysis of available data indicated minimal overlap between different published stem cell–associated gene sets . The discrepancy observed amongst these studies likely arises in large part from significant differences in the strategies used to identify the stem cell profile. However, true differences in stem cell biology probably also exist among stem cells taken from different tissues (e.g., ESCs, neural stem cells, or hematopoietic stem cells). Some studies have relied on comparison of ESCs to terminally differentiated tissues to establish a list of genes enriched in ESCs. However, this strategy is likely to miss genes involved in pluripotency that are transiently turned off early during the differentiation process but not uniquely expressed in the stem cell population. Another strategy used has been to compare stem cell populations arising from the same tissue source across species. An example is the comparison of mouse with human ESCs. Although mouse and human ESCs are both derived from preimplantation blastocysts, they differ in responsiveness to extrinsic signals and in expression of surface markers; for example, LIF cannot sustain self-renewal of human ESCs, even in the presence of serum, suggesting the existence of other signaling pathways essential for self-renewal in human ESCs . The evidence that human and mouse ESCs share a common core molecular program is also somewhat conflicting . However, the molecular mechanisms that confer pluripotency are likely evolutionary conserved, and thus comparison of mouse and human ESC gene-array data might still be informative.
The gene expression data reported here were compared with nine different gene-array data sets, comprising four studies on murine ESCs , four studies on human ESCs , and one study comparing human and mouse ESCs (summarized in Table 5). Only genes downregulated during differentiation after LIF removal were used in the comparisons because most other studies only report ESC-enriched or ESC-specific genes. Complete gene lists derived from these comparisons can been found in supplementary online Table 6.
Table 5. Summary of the published gene expression studies in mouse and human ESCs that were used for comparison
Overall, there was a large overlap between the downregulated genes from our data and genes identified as either ESC enriched or downregulated during ESC differentiation in the other data sets, despite the wide variety of experimental designs and cell lines used. Of the genes identified as significantly decreased after LIF removal in our study, 60% (164 of 275) were found in at least one other data set, and 28% (78 of 275) were found in at least two other data sets. There was greater similarity with the murine data (129 of 275 or 47% were found in at least one other murine data set) than with the human data (75 of 275 or 27% were found in at least one other human data set). Genes that we identified as differentially expressed between 0 and 18 hours were the least represented in other published data (41% between 0 and 18 hours compared with 67% between 18 and 72 hours and 63% between 0 and 72 hours), possibly because our approach to use early time points during differentiation for defining differential expression has not been widely used. Only two other studies included comparable early time points in their study . Within individual data sets, the degree of overlap depended primarily on the number of genes in the starting list (Table 5). The data set generated by Kelly et al. stood out because of its high degree of overlap (35%), which can potentially be explained by the similarity in the two studies in terms of experimental design and the similarity of the genetic background of the ESCs used. However, this encompasses relatively few genes (6 out of 17 genes). Furthermore, many genes that we report as decreased during the first 72 hours were reported as also being present in human ESCs by Sato et al. . However, they were not identified as differentially expressed in that study. This discrepancy could possibly be due to the later time point (3 weeks) used for determining differential expression in that study (data not shown).
No single gene was found enriched in every human and mouse ESC data set. Jade1, Leftb, and Smarcad1 were the most commonly identified genes in other data sets (i.e., in six of nine other data sets). Leftb is a transforming growth factor-beta family member, which is expressed on the left side of developing mouse embryos and is implicated in left-right determination . Jade1 was recently identified as a gene involved in anteroposterior axis development . Smarcad1 has previously been identified as a marker of preimplantation embryos . The gene coding for Tenascin-C was downregulated within the first 18 hours after LIF removal (Table 3) and was also commonly observed as differentially expressed in other data sets (i.e., in four of eight data sets). Tenascin, also known as hexabrachion and cytotactin, is an extra-cellular matrix protein with a spatially and temporally restricted tissue distribution that is tightly regulated during embryonic development and in adult tissue remodeling . Other commonly identified markers of undifferentiated ESCs were also found in this comparison. Sox2, Rex1, Bmp4, and Gbx2 were all observed in at least three other data sets. This also included the gene Lox, coding for lysyl oxidase, found in three other murine data sets . It was in fact one of the most pronounced and rapidly downregulated genes detected (Table 3). At least one study identified c-Kit as enriched in ESCs, intriguingly in human ESCs . Overall the comparison with previously published data indicated some intriguing similarities but also showed that most genes do not overlap in pairwise comparisons. This underlines the many differences that might arise when different sources of cells, culture conditions, microarray technique platforms, and data analysis tools are used and emphasizes the need to correlate expression changes with rigorous measures of ESC competence and differentiation potential.
DISCUSSION
In this study, we used a novel strategy to identify genes that may play a critical role in regulating the pluripotent potential of murine ESCs. Unlike previous ESC transcriptome analyses, we carried out comparisons of closely related populations both in terms of lineage and time. The correlation of differentially expressed genes with functional measures of ESC pluripotency as well as validation in other murine ESC lines provides added confidence that these genes may be of functional relevance. Our studies also identify candidate novel markers of ESC pluripotency. Taken together, this work provides the foundation for achieving a greater understanding of the molecular mechanisms that govern and reflect the capacity of ESCs for multilineage differentiation and self-renewal.
ACKNOWLEDGMENTS
Smith AG. Embryo-derived stem cells: of mice and men. Annu Rev Cell Dev Biol 2001;17:435–462.
Suda Y, Suzuki M, Ikawa Y et al. Mouse embryonic stem cells exhibit indefinite proliferative potential. J Cell Physiol 1987;133:197–201.
Evans MJ, Kaufman MH. Establishment in culture of pluripotential cells from mouse embryos. Nature 1981;292:154–156.
Martin GR. Isolation of a pluripotent cell line from early mouse embryos cultured in medium conditioned by teratocarcinoma stem cells. Proc Natl Acad Sci U S A 1981;78:7634–7638.
Thomson JA, Itskovitz-Eldor J, Shapiro SS et al. Embryonic stem cell lines derived from human blastocysts. Science 1998;282:1145–1147.
Abbondanzo SJ, Gadi I, Stewart CL. Derivation of embryonic stem cell lines. In: Wasserman P, DePamphilis M, eds.Methods of Enzymology, Guide to Techniques in Mouse Development. Toronto: Academic Press Inc., 1993:803–823.
Keller G, Kennedy M, Papayannopoulou T et al. Hematopoietic commitment during embryonic stem cell differentiation in culture. Mol Cell Biol 1993;13:473–486.
Kirschstein R, Skirboll LR. Stem Cells: Scientific Progress and Future Research Directions. Bethesda, MD: National Institutes of Health, Department of Health and Human Services, 2001:appendix E, 1–11.
Nagy A, Rossant J, Nagy R et al. Derivation of completely cell culture-derived mice from early-passage embryonic stem cells. Proc Natl Acad Sci U S A 1993;90:8424–8428.
Li E, Bestor TH, Jaenisch R. Targeted mutation of the DNA methyltransferase gene results in embryonic lethality. Cell 1992;69:915–926.
Nichols J, Evans EP, Smith AG. Establishment of germ-line-competent embryonic stem (ES) cells using differentiation inhibiting activity. Development 1990;110:1341–1348.
Helgason CD, Damen JE, Rosten P et al. Targeted disruption of SHIP leads to hemopoietic perturbations, lung pathology, and a shortened life span. Genes Dev 1998;12:1610–1620.
Brazma A, Hingamp P, Quackenbush J et al. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet 2001;29:365–371.
Draghici S, Khatri P, Bhavsar P et al. Onto-Tools, the toolkit of the modern biologist: Onto-Express, Onto-Compare, Onto-Design and Onto-Translate. Nucleic Acids Res 2003;31:3775–781.
Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 2001;25:402–408.
Bhattacharya B, Miura T, Brandenberger R et al. Gene expression in human embryonic stem cell lines: unique molecular signature. Blood 2004;103:2956–2964.
Brandenberger R, Wei H, Zhang S et al. Transcriptome characterization elucidates signaling networks that control human ES cell growth and differentiation. Nat Biotechnol 2004;22:707–716.
Ivanova NB, Dimos JT, Schaniel C et al. A stem cell molecular signature. Science 2002;298:601–604.
Ramalho-Santos M, Yoon S Matsuzaki Y et al. "Stemness": transcriptional profiling of embryonic and adult stem cells. Science 2002;298:597–600.
Sato N, Sanjuan IM, Heke M et al. Molecular signature of human embryonic stem cells and its comparison with the mouse. Dev Biol 2003;260:404–413.
Sharov AA, Piao Y, Matoba R et al. Transcriptome analysis of mouse stem cells and early embryos. PLoS Biol 2003;1:E74.
Sperger JM, Chen X, Draper JS et al. Gene expression patterns in human embryonic stem cells and human pluripotent germ cell tumors. Proc Natl Acad Sci U S A 2003;100:13350–13355.
Kelly DL, Rizzino A. DNA microarray analyses of genes regulated during the differentiation of embryonic stem cells. Mol Reprod Dev 2000;56:113–123.
Ginis I, Luo Y, Miura T et al. Differences between human and mouse embryonic stem cells. Dev Biol 2004;269:360–380.
Stewart CL. Production of chimeras between embryonic stem cells and embryos. In: Wassarman PM, DePamphilis ML, eds. Methods in Enzymology: Guide to Techniques in Mouse Development. Toronto: Academic Press, 1993:823–854.
Niwa H, Miyazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet 2000;24:372–376.
Furusawa T, Ohkoshi K, Honda C et al. Embryonic stem cells expressing both platelet endothelial cell adhesion molecule-1 and stage-specific embryonic antigen-1 differentiate predominantly into epiblast cells in a chimeric embryo. Biol Reprod 2004;70:1452–1457.
Williams RL, Hilton DJ, Pease S et al. Myeloid leukaemia inhibitory factor maintains the developmental potential of embryonic stem cells. Nature 1988;336:684–687.
Stewart CL, Kaspar P, Brunet LJ et al. Blastocyst implantation depends on maternal expression of leukaemia inhibitory factor. Nature 1992;359:76–79.
Shirogane T, Fukada T, Muller JM et al. Synergistic roles for Pim-1 and c-Myc in STAT3-mediated cell cycle progression and antiapoptosis. Immunity 1999;11:709–719.
Ying QL, Nichols J, Chambers I et al. BMP induction of Id proteins suppresses differentiation and sustains embryonic stem cell self-renewal in collaboration with STAT3. Cell 2003;115:281–292.
Hosler BA, LaRosa GJ, Grippo JF et al. Expression of REX-1, a gene containing zinc finger motifs, is rapidly reduced by retinoic acid in F9 teratocarcinoma cells. Mol Cell Biol 1989;9:5623–5629.
Rogers MB, Hosler BA, Gudas LJ. Specific expression of a retinoic acid-regulated, zinc-finger gene, Rex-1, in preimplantation embryos, trophoblast and spermatocytes. Development 1991;113:815–824.
Tanaka TS, Kunath T, Kimber WL et al. Gene expression profiling of embryo-derived stem cells reveals candidate genes associated with pluripotency and lineage specificity. Genome Res 2002;12:1921–1928.
Avilion AA, Nicolis SK, Pevny LH et al. Multipotent cell lineages in early mouse development depend on SOX2 function. Genes Dev 2003;17:126–140.
Bowles J, Teasdale RP, James K et al. Dppa3 is a marker of pluripotency and has a human homologue that is expressed in germ cell tumours. Cytogenet Genome Res 2003;101:261–265.
Chambers I, Colby D, Robertson M et al. Functional expression cloning of Nanog, a pluripotency sustaining factor in embryonic stem cells. Cell 2003;113:643–655.
Mitsui K, Tokuzawa Y, Itoh H et al. The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells. Cell 2003;113:631–642.
Wilkinson DG, Bhatt SHerrmann BG. Expression pattern of the mouse T gene and its role in mesoderm formation. Nature 1990;343:657–659.
Wiese C, Rolletschek A, Kania G et al. Nestin expression: a property of multi-lineage progenitor cells? Cell Mol Life Sci 2004;61:2510–2522.
Weigmann A, Corbeil D, Hellwig A et al. Prominin, a novel microvilli-specific polytopic membrane protein of the apical surface of epithelial cells, is targeted to plasmalemmal protrusions of non-epithelial cells. Proc Natl Acad Sci U S A 1997;94:12425–12430.
Anneren C, Cowan CA, Melton DA. The Src family of tyrosine kinases is important for embryonic stem cell self-renewal. J Biol Chem 2004;279:31590–31598.
Vas V, Szilagyi L, Paloczi K et al. Soluble Jagged-1 is able to inhibit the function of its multivalent form to induce hematopoietic stem cell self-renewal in a surrogate in vitro assay. J Leukoc Biol 2004;75:714–720.
Xue Y, Gao X, Lindsell CE et al. Embryonic lethality and vascular defects in mice lacking the Notch ligand Jagged1. Hum Mol Genet 1999;8:723–730.
Burdon T, Chambers I, Stracey C et al. Signaling mechanisms regulating self-renewal and differentiation of pluripotent embryonic stem cells. Cells Tissues Organs 1999;165:131–143.
SatoN, MeijerL, Skaltsounis L et al. Maintenance of pluripotency in human and mouse embryonic stem cells through activation of Wnt signaling by a pharmacological GSK-3-specific inhibitor. Nat Med 2004;10:55–63.
Cohen MM Jr. The hedgehog signaling network. Am J Med Genet 2003;123A:5–28.
Ogden SK, Ascano M Jr, Stegman MA et al. Regulation of Hedgehog signaling: a complex story. Biochem Pharmacol 2004;67:805–814.
Oka M, Tagoku K, Russell TL et al. CD9 is associated with leukemia inhibitory factor-mediated maintenance of embryonic stem cells. Mol Biol Cell 2002;13:1274–1281.
Rose TM, Weiford DM, Gunderson NL et al. Oncostatin M (OSM) inhibits the differentiation of pluripotent embryonic stem cells in vitro. Cytokine 1994;6:48–54.
Wheatley SC, Isacke CM. Induction of a hyaluronan receptor, CD44, during embryonal carcinoma and embryonic stem cell differentiation. Cell Adhes Commun 1995;3:217–230.
Fortunel NO, Otu HH, Ng HH et al. Comment on " ‘Stemness’: transcriptional profiling of embryonic and adult stem cells" and "a stem cell molecular signature." Science 2003;302:393.
Evsikov AV, Solter D. Comment on "‘Stemness’: transcriptional profiling of embryonic and adult stem cells" and "a stem cell molecular signature." Science 2003;302:393.
Meno C, Saijoh Y, Fujii H et al. Left-right asymmetric expression of the TGF beta-family member lefty in mouse embryos. Nature 1996;381:151–155.
Tzouanacou E, Tweedie S, Wilson V. Identification of Jade1, a gene encoding a PHD zinc finger protein, in a gene trap mutagenesis screen for genes involved in anteroposterior axis development. Mol Cell Biol 2003;23:8553–8552.
Schoor M, Schuster-Gossler K, Gossler A. The Etl-1 gene encodes a nuclear protein differentially expressed during early mouse development. Dev Dyn 1993;197:227–237.
Jones PL, Jones FS. Tenascin-C in development and disease: gene regulation and cell function. Matrix Biol 2000;19:581–596.(Lars Palmqvista, Clive H.)
b Michael Smith Laboratories, University of British Columbia, Vancouver, British Columbia, Canada;
c Department of Cancer Endocrinology, BC Cancer Agency, Vancouver, British Columbia, Canada;
d Departments of Chemical and Biological Engineering,
e Medicine, and
f Surgery, University of British Columbia, Vancouver, British Columbia, Canada
Key Words. Embryonic stem cells ? Markers ? Pluripotency ? Leukemia inhibitory factor ? Gene expression ? Microarray
Correspondence: Cheryl D. Helgason, Ph.D., Department of Cancer Endocrinology, BC Cancer Agency, 675 West 10th Avenue, Vancouver, BC, Canada, V5Z 1L3. Telephone: 604-675-8011; Fax: 604-675-8183; e-mail: chelgaso@bccrc.ca
ABSTRACT
Embryonic stem cells (ESCs) are characterized by their ability to both self-renew and differentiate . However, the molecular mechanisms that regulate the decision between these two processes are poorly understood. Mouse ESCs were originally isolated from the inner cell mass (ICM) of preimplantation blastocysts and can be maintained in cell culture indefinitely without loss of their broad pluripotent differentiation capacity as determined by their ability to give rise to all three germ layers both in vitro and in vivo . The more recent establishment of human ESC lines has further increased the interest in ESCs because they raise hope of an unlimited source of cells for tissue engineering and cell therapies in the future. However, realization of this potential requires an increased knowledge of the molecular mechanisms governing self-renewal and pluripotency to guide the development of processes that control the expansion and differentiation of stem cells ex vivo.
Unlike hematopoietic stem cells, for which quantitative assays of stem cell potential have been defined and validated, no such assays currently exist for ESCs. In the murine system, self-renewal is measured by the ability of mouse ESCs to continuously proliferate in culture while maintaining an undifferentiated colony morphology . The most rigorous in vivo assay to establish functionality of cultured mouse ESCs is blastocyst injection and measurement of their ability to give rise to chimeric mice because it requires ESC contribution to all adult tissue, including germ cells . However, injection of 10 to 15 ESCs into a single blastocyst does not provide a quantitative measure of stem cell potential at the single-ESC level. Two in vitro assays have been used extensively as surrogates for chimera formation when testing culture reagents or examining the consequences of genetic manipulation. The colony-forming cell (CFC) assay is used to determine the plating efficiency of ESC populations under various conditions and thus may be considered indicative of self-renewal potential. Formation of embryoid bodies (EBs) can be performed at a clonal level in vitro and reflects multilineage differentiation potential . The correlation between these in vitro assays and chimera generation has not been determined. Assessment of pluripotency has also relied on the expression of selected molecular markers. For murine ESCs, these have included alkaline phosphatase, the POU transcription factor Oct-4, and stage-specific embryonic antigen 1 (SSEA-1) . However, the correlation between marker expression and the various functional assays has not been extensively studied. Knowledge of the intricate mechanisms regulating ESC pluripotency and differentiation potential is currently limited to a few signaling pathways (i.e., leukemia inhibitory factor ) and regulatory factors (i.e., Oct-4 and Nanog). Thus, very little is known about the tolerance limits of different culture conditions for maintaining stem cell function during expansion or how these relate to altered gene expression patterns in ESCs. Identification of molecular markers that correlate with pluripotency would be invaluable to enrich for the desired cells, as well as to monitor their maintenance during expansion protocols.
Achieving the goal of defining the core stem cell regulatory network requires a precise characterization of the functional capacities of the cells for which the transcriptional profile is described. In this study, we established gene expression profiles during early differentiation of the well-defined R1 ESC line and correlated gene expression changes with both phenotypic and functional assessment of the same cells. Functional capacity was determined by blastocyst injections for chimeric mouse formation, EB assays, and CFC counts. Undifferentiated ESCs and ESCs cultured without LIF for 18 or 72 hours were chosen for gene-array analysis. We identified 473 unique genes as significantly differentially expressed during early ESC differentiation, and approximately one third of these have unknown biological function. Among the 275 genes whose expression decreased with ESC differentiation were several factors previously identified as important for, or markers of, ESC pluripotency, including Stat3, Rex1, Sox2, Gbx2, and Bmp4. A significant number of the decreased genes also overlapped with previously published mouse and human ESC data. Reverse transcription–polymerase chain reaction (RT-PCR) validation showed high correlation with the gene-array data, and several genes were also shown to have similar changes after LIF removal in two other murine ESC lines. Expression of the commonly used ESC markers Oct-4 and SSEA-1 was also examined in parallel with the functional assays. However, a close correlation was not observed. Interestingly, among a subset of 48 decreased genes that showed the closest correlation with the functional assays was the stem cell factor (SCF) receptor c-Kit that can be a useful marker of undifferentiated ESCs.
MATERIALS AND METHODS
Pluripotency During Early Differentiation
We hypothesized that loss of ESC pluripotency as defined by measurable functional readouts would correlate with significant alterations in gene expression. Identification of these gene expression changes would thus provide important insights into the genetic regulation of ESC pluripotency and facilitate the identification of new molecular markers of the undifferentiated ESC state. We relied on comparisons of the three measures of ESC potential (chimera formation, EB formation, and CFC assay) to select time points for analysis of gene expression profiles. Undifferentiated ESCs were cultured in medium containing LIF on irradiated primary MEF, which supply several as-yet unidentified factors that enhance the plating efficiency of the ESCs as well as assist in the maintenance of the undifferentiated state. Preplating of ESCs removes the MEF and enriches for ESCs with the capacity to contribute to the developing blastocyst. It has been suggested that those ESCs capable of loosely attaching to the feeder layer in a short (i.e., 1 hour) period of time have the highest likelihood of forming colonies (i.e., self-renewing) and thus may also be the most competent at contributing to the germ-line after blastocyst injection . We thus elected to use only preplated, loosely adherent mouse ESCs for our analyses.
The baseline activities for undifferentiated ESCs in the three different assays were first determined. Undifferentiated, pre-plated R1 ESCs yielded 100% chimeric pups after blastocyst injection (27 blastocysts injected; six born and analyzed in two independent experiments), 6.9% ± 1.0% of plated cells differentiated into EBs in the EB formation assay, and 12.50% ± 2.2% of plated cells gave rise to alkaline phosphatase–positive ESC colonies in the CFC assay. These results were within the normal range for the R1 cell line. ESC differentiation was then initiated by LIF removal and replating without MEF. The morphology of the ESCs before and after LIF removal is shown in Figure 1. At 18 and 24 hours after MEF and LIF removal, the ESCs looked very similar to one another and did not exhibit any clear signs of differentiation. Morphological differentiation was first apparent at 48 hours, and EBs could be seen after 72 hours. Despite the lack of appreciable phenotypic differentiation within the first 24 hours, there were significant changes in the functional properties of the ESCs. Cultured cells were harvested at various time points during this culture period and analyzed in the three different assays. The blastocyst injection assay showed a rapid decrease in the number of chimeras obtained after initiation of ESC differentiation (Fig. 2A). Only 28% of born pups (84 injected, 25 born and analyzed in two independent experiments) were chimeras when ESCs had differentiated for 24 hours, and less than 5% were chimeras after 72 hours of differentiation (66 injected, 29 born and analyzed in two independent experiments). The EB formation assay (Fig. 2B) showed approximately 5.5% ± 1.0%, 3.7% ± 0.3%, and 0.3% ± 0.1% readout at 18, 24, and 72 hours after LIF removal, respectively. In contrast, the frequency of cells replating in the CFC assay increased slightly during the first 24 hours but subsequently declined rapidly such that only 1.3% ± 0.3% retained this activity at 72 hours (Fig. 2C). In summary, all three assays showed clearly that most differentiation potential and self-renewing capacity was gone after 72 hours of differentiation. However, there was a high degree of variation during the first 24 hours amongst the in vitro and in vivo assays, with the EB formation assay correlating most closely with chimeric mouse formation. For the gene expression profiling, because both the EB formation and chimera assays showed a pronounced decrease within the first 18–24 hours, we elected to use the earlier time point, along with 72 hours of differentiation, to compare against the undifferentiated R1 ES.
Figure 1. Embryonic stem cell (ESC) morphology. Morphology of the R1 ESCs grown on mouse embryo fibroblasts (MEFs) with leukemia inhibitory factor (LIF) is shown. ESCs were followed for 72 hours after MEF and LIF removal, and pictures were taken at the indicated time points and with the indicated resolution.
Figure 2. Assays of embryonic stem cell (ESC) pluripotency. Comparison of in vitro and in vivo functional measures of ESC potential with expression of the commonly used ESC markers Oct-4 and stage-specific embryonic antigen 1 (SSEA-1). R1 ESCs were thawed, cultured, and assayed as outlined in Materials and Methods. The frequency of cells capable of (A) generating chimeric mice (summary of two independent experiments), (B) giving rise to embryoid bodies (summary of three independent experiments), or (C) giving rise to colonies in the colony-forming cell assay (summary of three independent experiments) is shown as a function of time after leukemia inhibitory factor removal. (D): Flow cytometric analysis (summary of three independent experiments) was used to determine the percentage of gated cells expressing cell-surface SSEA-1 and intracellular Oct-4 at the same time points. Data are the mean ± standard error of the mean for replicated experiments.
In parallel with the functional assays, the expression patterns of two markers commonly used to identify undifferentiated ESCs, the POU transcription factor Oct-4 and SSEA-1, were also analyzed to establish the level of correspondence in the readouts using each method. Although Oct-4 and SSEA-1 have been extensively used in ESC research, their expression patterns during ESC differentiation have not been studied in detail. Oct-4 is used as a marker for ESCs because of its requirement in ESC self-renewal . The precise expression level of Oct-4 is important for determining ESC fates, and repression of Oct-4 induces loss of pluripotency and dedifferentiation to trophectoderm . However, forced constitutive expression of Oct-4 cannot prevent ESC differentiation, and a less than twofold increase in expression actually causes differentiation into primitive endoderm and mesoderm . Thus, a critical amount of Oct-4 is required to sustain stem cell self-renewal but is not sufficient to prevent differentiation. SSEA-1 is a glycoprotein expressed during early embryonic development and by undifferentiated ESCs. However, the precise role of SSEA-1 in pluripotency and self-renewal has not been defined. ESCs selected for expression of SSEA-1 and platelet endothelial cell adhesion molecule 1 are enriched for cells that differentiate predominantly into epiblast cells in chimeric embryos . In the present study, protein expression of both SSEA-1 and Oct-4 remained relatively unchanged throughout the first 48 hours of differentiation (Fig. 2D). At 120 hours after LIF removal, 18.5% ± 1.1% of the cells retained expression of SSEA-1, whereas 40.6% ± 6.6% of the cells continued to express Oct-4. Thus there was no clear correlation between expression of these two markers and the various ESC functional assays.
Gene-Array Analysis
Cultured R1 ESCs from matched passage numbers were collected in three separate experiments at 0, 18, and 72 hours of differentiation after LIF removal and were analyzed using the Affymetrix GeneChip MG-U74v2 array containing 36,767 different probe sets. Duplicate samples of the MEF feeders were also collected and analyzed to assess any possible contamination of the undifferentiated ESCs. Hybridization, scanning, and production of raw data files were performed according to standard protocols. The MAS 5.0 software was used for the initial scaling and expression analysis. The data were then normalized and further analyzed using the GeneSpring software. To validate the reproducibility and the overall variation of the data, hierarchical clustering analysis was performed. The data were first filtered for genes present in at least one of the three time points (present in at least two out of three replicates), resulting in 13,002 different probe sets. The average number of present genes for each time point was 12,818 (coefficient of variation , 6%) in undifferentiated ESCs, 11,806 (CV, 4%) at 18 hours, and 12,297 (CV, 1%) at 72 hours after LIF removal. The number of expressed genes at the different time points was not statistically significantly different. Hierarchical clustering was applied to the reduced gene set on individual array samples (three replicates for each time point) using Pearson’s correlation and average linkage clustering as implemented in GeneSpring. The individual samples from the three experiments clustered tightly together according to their respective time points, as seen in Figure 3, indicating that the overall interexperimental variation was low. The clustering also reflects the temporal progression of the ESC differentiation. As expected, the first two time points (0 and 18 hours) clustered more closely together, whereas the 72-hour time point showed a more distinct expression pattern, resulting in a greater relative distance from the other two time points, consistent with the functional data (Fig. 2).
Figure 3. Distance tree. Hierarchical clustering of the individual gene-array samples was carried out using Pearson’s correlation and average linking clustering to determine reproducibility and interexperimental variability. Relative distance values are shown.
Genes that were differentially expressed after LIF removal were determined using the following criteria: The difference in expression level between two time points was at least twofold, the gene was present in all three replicates at the time points with the highest abundance, and the extent of difference in expression was statistically significant (p < .05) in a parametric Welsh ANOVA t-test. With this approach, 473 unique genes were identified as significantly differentially expressed (275 decreased, 194 increased, and 4 both decreased and increased) during early ESC differentiation after LIF removal (complete gene lists in the supplemental data, supplementary online Table 2). Unigene and RefSeq IDs were used to exclude redundant probe sets on the array and to get the true number of affected genes.
Only preplated, loosely adherent ESCs were used in our experiments, and visual assessment of MEF contamination of undifferentiated ESCs suggested it was consistently less than 1 in 500. However, even with these measures, we sought to exclude the possibility that contaminating MEF cells may have distorted the data. Duplicate samples of MEFs were therefore also analyzed on the U74v2 GeneChip. The data from both the MEF and the R1 ESC was scaled in MAS 5.0 and normalized in GeneSpring 6.2 before 1% of the raw value (five times more than the estimated contamination) obtained for the gene in MEF was subtracted from the value of the same gene in the undifferentiated R1 ESCs. The analysis steps to find differentially expressed genes were then performed as described above. Only genes detected as decreased during differentiation in the original analysis were reanalyzed. Genes whose expression increased during differentiation would not be affected by MEF contamination in a way that would give false-positive results. This evaluation indicates that none of the genes that showed a significant twofold or greater decrease in the initial analysis lost their significance after MEF subtraction (data not shown) and suggests that the level of MEF contamination was not sufficient to distort the ESC gene expression data.
Gene Expression Validation
The expression patterns of some genes previously implicated in maintaining ESC pluripotency and self-renewal were analyzed to validate the data and the approach to identify differentially expressed genes (Table 1). Mouse ESCs are usually cocultured on a feeder layer of irradiated MEFs. In addition to providing a matrix for attachment, MEFs produce LIF, required for propagation of pluripotent mouse ESCs . LIF-null MEFs cannot support self-renewal . However, LIF also needs to be complemented by fetal calf serum to block differentiation. LIF binds to the gp130 receptor that leads to activation of the transcription factor Stat3 . Stat3 was clearly detected in undifferentiated R1 ESCs, and LIF removal decreased Stat3 expression with the most pronounced effect during the first 18 hours (49% reduction). This was also seen for two other genes involved in the LIF/gp130 pathway, the Oncostatin M receptor, Osmr, and the interleukin 6 signal transducer, Il6st. A known Stat3 target gene, Pim1 , was also decreased significantly at the transcriptional level. Taken together, these observations confirm that the LIF/gp130/Stat3 pathway was rapidly shut down after MEF and LIF removal. The bone morphogenic proteins can act in combination with LIF to sustain self-renewal and preserve multilineage differentiation, chimera contribution, and germ line transmission properties . Bmp4 transcript levels were significantly decreased during differentiation. However, the onset of the decrease was later than for Stat3 (between 18 and 72 hours, 80% reduction). The same was seen for Rex-1/Zfp42, known to be highly expressed in undifferentiated ESCs and downregulated after retinoic acid–induced differentiation . A 90% reduction in Rex-1/Zfp42 expression was seen between 18 and 72 hours after LIF removal. Transcript abundance of the Akp2 gene coding for alkaline phosphatase, commonly used as a marker for undifferentiated ESCs, was also decreased significantly (55% reduction between 18 and 72 hours). The POU transcription factor message, pou5f1, coding for Oct-4 was highly expressed in undifferentiated ESCs but was not changed significantly during the first 72 hours after LIF removal in keeping with the protein expression data (Fig. 2D). Transcript levels of the embryonal stem cell–specific gene 1 (Esg-1 or developmental pluripotency associated gene 5, Dppa5) did not change during differentiation. However, this observation is consistent with the observation that Esg-1 probably is a downstream target of Oct-4 and is downregulated slowly after Oct-4 suppression . The abundance of FoxD3, a gene expressed early in mouse embryonic development, also remained unchanged during the first 72 hours after LIF removal. The SRY-box containing transcription factor Sox2 may act to maintain ESC pluripotency and is expressed in the ICM, epiblast, and germ cells, just like Oct-4 . Sox2 was significantly decreased (60% reduction between 18 and 72 hours, p = .0136). Sox2, together with Oct-4, is involved in the regulation of fibroblast growth factor 4 (Fgf4), another factor confined to the ICM of the blastocyst . Although Fgf4 transcripts were not significantly decreased (p = .09), it was reduced more than twofold (52% reduction between 18 and 72 hours). This was also the case for the newly discovered marker of ESC pluripotency Dppa3 , which decreased 78% between 0 and 72 hours (p = .073). Similarly, the regulatory factor Nanog, which can rescue ESCs from LIF/Stat3 dependence and maintain Oct-4 expression , was not quite significantly decreased during the first 72 hours of differentiation (p = .07), and the reduction in expression was less than twofold (41%). Several genes indicative of ESC differentiation were increased after LIF removal. For example, the mesoderm marker Brachyury increased 15-fold between 18 and 72 hours (p = .003). The ectoderm marker Nestin increased approximately threefold between 18 and 72 hours (p = 0.04), and the epithelial cell marker Prominin 1 increased fivefold between 18 and 72 hours (p = .0010). In conclusion, most of these changes are consistent with the loss of ESC pluripotency as measured in all three assays and indicate that the thresholds used to define differentially expressed genes in this study are reasonable.
Table 1. Genes previously reported as enriched in undifferentiated ESCs or to be markers of ESC differentiation were compared with the genes identified as differentially expressed in the present study
RT-PCR Validation
To further confirm the fidelity of the gene-array data, a set of 28 genes was selected and their transcript levels were tested using quantitative RT-PCR in R1 ESCs before and after LIF removal. The genes selected included some of the genes mentioned above, as well as other genes increased, decreased, or unchanged after LIF removal according to the gene-array results. The fold change was calculated between 0 and 18 hours, 18 and 72 hours, and 0 and 72 hours, respectively. The individual RT-PCR results can be found in the supplemental data (supplementary online Table 3). The analysis revealed a high degree of correlation between the gene-array data and the RT-PCR results (Pearson’s correlation, r = 0.87, Fig. 4A), with a tendency that the PCR results showed greater changes than the array suggested (proportional bias 1.2 in a Deming method comparison analysis, not shown).
Figure 4. Correlation between gene array and quantitative reverse transcription–polymerase chain reaction (RT-PCR). (A): Pearson’s correlation analysis was done on the log ratio values obtained from the quantitative real-time RT-PCR and the gene-array fold changes (n = 84). (B, C, D): Comparison of quantitative real-time RT-PCR results from the three different embryonic stem cell lines are shown. Pearson’s correlation analysis was done on the log ratios of the values obtained with the 2–CT method (n = 45 in each comparison).
To determine if the gene expression changes observed in the R1 ESCs are general and more broadly observed, the expression patterns of 15 genes were followed for 72 hours after LIF removal in two other ESC lines, J1 and EFC , and compared with the R1 ESC line using quantitative RT-PCR (Table 2). Overall, there was a good correlation amongst all three ESC lines (Figs. 4B–4D). Specifically, RT-PCR analysis was able to verify the expression changes of genes such as Lox, Ankrd1, and c-Kit that showed significantly decreased transcript levels after 18 hours in all three ESC lines tested. Similarly, Rex1, Sox2, Leftb, and Mtf2 were all changed in a similar fashion in the R1, J1, and EFC cell lines. The RT-PCR results also confirmed that Oct-4 transcript levels were not decreased significantly during the first 72 hours of differentiation in any of the cell lines, in agreement with the gene-array results and the protein level observed (Fig. 2D). In contrast, although the decrease in Nanog transcripts was not quite significant in R1 ESCs according to the gene array, the RT-PCR results indicated a reduction. Nanog was reduced by 50%–70% after 72 hours in all three ESC lines according to the quantitative RT-PCR analysis but consistent with delayed reduction in Nanog upon differentiation (Table 2). Only two of the tested genes showed different expression patterns amongst the cell lines. The mesoderm marker Brachyury and the homeobox transcription factor Pbx3 were both increased during differentiation in the R1 and J1 ESCs, but not in the EFC cell line. This observation suggests that the various ESC lines might follow slightly altered differentiation pathways upon LIF removal.
Table 2. Comparison of relative expression levels obtained by quantitative real-time RT-PCR in the R1, J1, and EFC ESC lines during differentiation
These validation analyses demonstrate that the early changes during differentiation uncovered by the gene-array analysis could be verified with an independent method and could in most cases be observed in multiple ESC lines. Taken together, they provide confidence in our approach to identify differentially expressed genes with a high likelihood of exhibiting true expression level changes during the loss of ESC pluripotency.
Functional Classification of Differentially Expressed Genes
It is likely that genes showing similar functional properties and expression patterns form interacting networks that contribute to the phenotypic and functional characteristics of the cells of interest. Functional classification of the differentially expressed genes into appropriate biological processes was thus performed using the NetAffx and GeneOntology Express tools as well as the simplified gene ontology tool in the GeneSpring software package. The differentially expressed genes were separated into 13 main categories (Fig. 5A; see supplementary online Table 4 for complete lists). Many differentially expressed genes were, as expected, classified as being involved in development and differentiation (61 genes, 10%) or directly or indirectly involved in cell-cycle control and cell proliferation (32 genes, 5%). Furthermore, at least 81 genes (13%) were classified as being involved in intracellular signal transduction, cell–cell signaling, or response to external stimuli (supplementary online Table 4).
Figure 5. Annotation of differentially expressed genes. All genes differentially expressed during embryonic stem cell differentiation were classified according to (A) biological process and (B) cellular component. Some genes are classified in more than one category, resulting in the total number of genes indicated in the figures being greater than the total number of differentially expressed genes.
A closer look at the signaling pathways affected during ESC differentiation revealed that the Yamaguchi sarcoma viral oncogene gene (Yes) was decreased significantly between 0 and 72 hours (60%). Yes has recently been shown to be regulated by LIF and to be important for ESC self-renewal . Yes is a gene coding for an Src tyrosine kinase expressed in both mouse and human ESCs and is downregulated when these cells differentiate . Another significantly decreased factor was the Notch ligand Jagged-1 (55% reduction between 0 and 18 hours), which has been implicated in hematopoietic stem cell self-renewal . Jagged-1 is involved in embryonic vascular development, and a Jag1 knockout is embryonic lethal .
A signaling pathway that is of particular interest during ESC differentiation is the mitogen-activated protein kinase (MAPK) pathway. The self-renewal of ESCs is influenced by the MAPK pathway, in which expression of Erk and Shp-2 counteracts the proliferative effects of Stat3 and promotes differentiation . According to the array results, the Erk gene (Mapk1 or Mapk3) and most other components in this pathway were already present in undifferentiated ESCs, and their expression levels did not change significantly at the transcription level during the first 72 hours of differentiation (data not shown). However, some genes in the MAPK pathway (e.g., Kras2 and Mapk12) had significantly increased transcript levels during differentiation, indicating an activation of the pathway (supplementary online Table 4). Furthermore, the gene coding for Grb2-associated binder 1 (Gab1) was significantly decreased (58% reduction between 18 and 72 hours, p = .0035). Gab1 binds to Shp-2 and is believed to suppress the MAPK pathway in ESCs; also, increased synthesis of Gab1 together with Oct-4 may suppress induction of differentiation . Although most of the factors involved in the MAPK pathway are primarily regulated at the posttranscriptional level, the fact that transcripts for many pathway members were detected in undifferentiated ESCs suggests that they harbor the required components to quickly respond to signals that promote differentiation.
The Wnt signaling pathway is important for maintenance of pluripotency in undifferentiated ESCs, and a recent study also showed that activation of the Wnt pathway by pharmacological inhibition of Gsk-3 is able to maintain pluripotency in both human and mouse ESCs . Most components of the Wnt signaling pathway could be detected in both undifferentiated and differentiated ESCs, but some factors were also significantly changed during differentiation (e.g., Fzd5, Wnt3, and CyclinD3; supplementary online Table 4). However, overall there was no dramatic change at the transcription level of several important factors belonging to this pathway during the first 72 hours of differentiation (e.g., ?-catenin, Gsk-3, and Axin; data not shown).
The Hedgehog signaling pathway plays a critical role during development and has been extensively studied in different species and tissues . In this pathway, two receptors, Ptch and Ptch2, and two transcription factors, Gli1 and Gli2, were decreased significantly during ESC differentiation. Furthermore, at least one suggested downstream target of this pathway, the Bmp4 gene already discussed above, had significantly decreased transcription, providing additional evidence that this pathway might be involved in maintaining ESC self-renewal or pluripotency.
Importantly, out of 473 differentially expressed genes, the largest group consisted of 173 genes or expressed sequence tags with unknown biological function. Further analyses of the genes within this group whose expression changes closely correlate with loss of pluripotency may reveal novel mechanisms involved in ESC maintenance. Of the 275 genes significantly decreased during ESC differentiation, 48 genes showed a more than twofold decrease in transcript levels within the first 18 hours and continued to decrease or remained at a low level of expression until 72 hours (Table 3). Changes in the transcript levels for these genes correlated well with the functional assays, especially the chimera generation and EB formation assays (Fig. 1). At least half of them have been identified as ESC-enriched or to be downregulated during differentiation in previous studies (see below), but several of these genes have unknown biological function or have not previously been implicated in ESC maintenance (e.g., Lox, Tnc, and Jag1; Table 3).
Table 3. Genes decreasing during ESC differentiation after LIF removal and most closely correlating with the loss of pluripotency
Molecular Markers of Pluripotency
We hypothesized that classification of differentially expressed genes according to cellular component, in combination with functional analyses, may lead to the identification of new markers that more accurately reflect ESC potential. Classification of the significantly changed genes according to cellular component is shown in Figure 5B (complete list provided in supplementary online Table 5). At least 60 gene products (11%) were considered localized to the plasma membrane and might therefore be good candidates as markers for undifferentiated ESCs. Genes increased during differentiation were also included in this list of possible ESC markers because they can be used to detect the onset of differentiation or for negative selection strategies for isolating undifferentiated ESCs (Table 4). Protein expression analyses of some of these have already been conducted in relation to ESC differentiation, i.e., Cd9, Cd44, and Osmr . Additional gene products of interest include the adhesion molecule Vcam1 and the hedgehog signaling pathway receptors Ptch and Ptch2. One gene of special interest encodes the SCF receptor c-Kit. According to the gene-array results, transcript level of the c-Kit gene was one of the most significant changes during differentiation (54% reduction between 0 and 18 hours and 75% reduction between 0 and 72 hours; p = .0016; Table 3), which also correlated well with the functional assays (Fig. 1). Furthermore, this finding was also verified in the R1, J1, and EFC ESC lines by quantitative RT-PCR (Table 2). No c-Kit ligand/SCF was added to the culture medium used in these experiments, and SCF was not detected in the undifferentiated R1 ESCs, although it was indeed present at high levels in the MEF according to the gene-array results (data not shown). Because c-Kit is expressed on several pluripotent cell types, including germ cells and hematopoietic stem cells, and antibodies are commercially available, we were able to compare the results of the array and RT-PCR experiments with protein expression levels measured by flow cytometry (Fig. 6). This analysis revealed that expression of c-Kit on the cell surface of differentiating ESCs closely paralleled the loss of functional potential as measured using the EB formation assay (Fig. 7). At later time points, the correlation was not as close (i.e., EB formation capacity continued to decrease while protein expression remained unchanged), suggesting the emergence of differentiated cell types expressing c-Kit protein.
Table 4. Differentially expressed genes with plasma membrane localization
Figure 6. Fluorescence-activated cell sorting profile. Undifferentiated R1 embryonic stem cells (ESCs) (0 hour) and R1 ESCs differentiated 18 and 72 hours after leukemia inhibitory factor removal were analyzed with flow cytometry using c-Kit antibody. Abbreviations: FSC, forward scatter; SSC, side scatter.
Figure 7. Correlation between c-Kit expression and embryoid body formation. Expression of c-Kit () was monitored at the indicated times during embryonic stem cell differentiation using flow cytometry, and expression levels were plotted as percent gated cells versus time (right axis). Embryoid body (EB)–forming cell frequency () determined on the same cell populations is also shown (percent EB formation indicated on the left axis).
Comparison with Previously Published Gene-Array Data Sets
Gene expression profiling to determine regulatory factors and signaling pathways present in ESCs has been performed extensively in recent years . It has been hypothesized that the undifferentiated state is conferred on various stem cell populations through the use of similar molecular mechanisms. Initial support for this hypothesis came from experiments comparing the gene expression profiles of multiple stem cell populations . However, further analysis of available data indicated minimal overlap between different published stem cell–associated gene sets . The discrepancy observed amongst these studies likely arises in large part from significant differences in the strategies used to identify the stem cell profile. However, true differences in stem cell biology probably also exist among stem cells taken from different tissues (e.g., ESCs, neural stem cells, or hematopoietic stem cells). Some studies have relied on comparison of ESCs to terminally differentiated tissues to establish a list of genes enriched in ESCs. However, this strategy is likely to miss genes involved in pluripotency that are transiently turned off early during the differentiation process but not uniquely expressed in the stem cell population. Another strategy used has been to compare stem cell populations arising from the same tissue source across species. An example is the comparison of mouse with human ESCs. Although mouse and human ESCs are both derived from preimplantation blastocysts, they differ in responsiveness to extrinsic signals and in expression of surface markers; for example, LIF cannot sustain self-renewal of human ESCs, even in the presence of serum, suggesting the existence of other signaling pathways essential for self-renewal in human ESCs . The evidence that human and mouse ESCs share a common core molecular program is also somewhat conflicting . However, the molecular mechanisms that confer pluripotency are likely evolutionary conserved, and thus comparison of mouse and human ESC gene-array data might still be informative.
The gene expression data reported here were compared with nine different gene-array data sets, comprising four studies on murine ESCs , four studies on human ESCs , and one study comparing human and mouse ESCs (summarized in Table 5). Only genes downregulated during differentiation after LIF removal were used in the comparisons because most other studies only report ESC-enriched or ESC-specific genes. Complete gene lists derived from these comparisons can been found in supplementary online Table 6.
Table 5. Summary of the published gene expression studies in mouse and human ESCs that were used for comparison
Overall, there was a large overlap between the downregulated genes from our data and genes identified as either ESC enriched or downregulated during ESC differentiation in the other data sets, despite the wide variety of experimental designs and cell lines used. Of the genes identified as significantly decreased after LIF removal in our study, 60% (164 of 275) were found in at least one other data set, and 28% (78 of 275) were found in at least two other data sets. There was greater similarity with the murine data (129 of 275 or 47% were found in at least one other murine data set) than with the human data (75 of 275 or 27% were found in at least one other human data set). Genes that we identified as differentially expressed between 0 and 18 hours were the least represented in other published data (41% between 0 and 18 hours compared with 67% between 18 and 72 hours and 63% between 0 and 72 hours), possibly because our approach to use early time points during differentiation for defining differential expression has not been widely used. Only two other studies included comparable early time points in their study . Within individual data sets, the degree of overlap depended primarily on the number of genes in the starting list (Table 5). The data set generated by Kelly et al. stood out because of its high degree of overlap (35%), which can potentially be explained by the similarity in the two studies in terms of experimental design and the similarity of the genetic background of the ESCs used. However, this encompasses relatively few genes (6 out of 17 genes). Furthermore, many genes that we report as decreased during the first 72 hours were reported as also being present in human ESCs by Sato et al. . However, they were not identified as differentially expressed in that study. This discrepancy could possibly be due to the later time point (3 weeks) used for determining differential expression in that study (data not shown).
No single gene was found enriched in every human and mouse ESC data set. Jade1, Leftb, and Smarcad1 were the most commonly identified genes in other data sets (i.e., in six of nine other data sets). Leftb is a transforming growth factor-beta family member, which is expressed on the left side of developing mouse embryos and is implicated in left-right determination . Jade1 was recently identified as a gene involved in anteroposterior axis development . Smarcad1 has previously been identified as a marker of preimplantation embryos . The gene coding for Tenascin-C was downregulated within the first 18 hours after LIF removal (Table 3) and was also commonly observed as differentially expressed in other data sets (i.e., in four of eight data sets). Tenascin, also known as hexabrachion and cytotactin, is an extra-cellular matrix protein with a spatially and temporally restricted tissue distribution that is tightly regulated during embryonic development and in adult tissue remodeling . Other commonly identified markers of undifferentiated ESCs were also found in this comparison. Sox2, Rex1, Bmp4, and Gbx2 were all observed in at least three other data sets. This also included the gene Lox, coding for lysyl oxidase, found in three other murine data sets . It was in fact one of the most pronounced and rapidly downregulated genes detected (Table 3). At least one study identified c-Kit as enriched in ESCs, intriguingly in human ESCs . Overall the comparison with previously published data indicated some intriguing similarities but also showed that most genes do not overlap in pairwise comparisons. This underlines the many differences that might arise when different sources of cells, culture conditions, microarray technique platforms, and data analysis tools are used and emphasizes the need to correlate expression changes with rigorous measures of ESC competence and differentiation potential.
DISCUSSION
In this study, we used a novel strategy to identify genes that may play a critical role in regulating the pluripotent potential of murine ESCs. Unlike previous ESC transcriptome analyses, we carried out comparisons of closely related populations both in terms of lineage and time. The correlation of differentially expressed genes with functional measures of ESC pluripotency as well as validation in other murine ESC lines provides added confidence that these genes may be of functional relevance. Our studies also identify candidate novel markers of ESC pluripotency. Taken together, this work provides the foundation for achieving a greater understanding of the molecular mechanisms that govern and reflect the capacity of ESCs for multilineage differentiation and self-renewal.
ACKNOWLEDGMENTS
Smith AG. Embryo-derived stem cells: of mice and men. Annu Rev Cell Dev Biol 2001;17:435–462.
Suda Y, Suzuki M, Ikawa Y et al. Mouse embryonic stem cells exhibit indefinite proliferative potential. J Cell Physiol 1987;133:197–201.
Evans MJ, Kaufman MH. Establishment in culture of pluripotential cells from mouse embryos. Nature 1981;292:154–156.
Martin GR. Isolation of a pluripotent cell line from early mouse embryos cultured in medium conditioned by teratocarcinoma stem cells. Proc Natl Acad Sci U S A 1981;78:7634–7638.
Thomson JA, Itskovitz-Eldor J, Shapiro SS et al. Embryonic stem cell lines derived from human blastocysts. Science 1998;282:1145–1147.
Abbondanzo SJ, Gadi I, Stewart CL. Derivation of embryonic stem cell lines. In: Wasserman P, DePamphilis M, eds.Methods of Enzymology, Guide to Techniques in Mouse Development. Toronto: Academic Press Inc., 1993:803–823.
Keller G, Kennedy M, Papayannopoulou T et al. Hematopoietic commitment during embryonic stem cell differentiation in culture. Mol Cell Biol 1993;13:473–486.
Kirschstein R, Skirboll LR. Stem Cells: Scientific Progress and Future Research Directions. Bethesda, MD: National Institutes of Health, Department of Health and Human Services, 2001:appendix E, 1–11.
Nagy A, Rossant J, Nagy R et al. Derivation of completely cell culture-derived mice from early-passage embryonic stem cells. Proc Natl Acad Sci U S A 1993;90:8424–8428.
Li E, Bestor TH, Jaenisch R. Targeted mutation of the DNA methyltransferase gene results in embryonic lethality. Cell 1992;69:915–926.
Nichols J, Evans EP, Smith AG. Establishment of germ-line-competent embryonic stem (ES) cells using differentiation inhibiting activity. Development 1990;110:1341–1348.
Helgason CD, Damen JE, Rosten P et al. Targeted disruption of SHIP leads to hemopoietic perturbations, lung pathology, and a shortened life span. Genes Dev 1998;12:1610–1620.
Brazma A, Hingamp P, Quackenbush J et al. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet 2001;29:365–371.
Draghici S, Khatri P, Bhavsar P et al. Onto-Tools, the toolkit of the modern biologist: Onto-Express, Onto-Compare, Onto-Design and Onto-Translate. Nucleic Acids Res 2003;31:3775–781.
Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 2001;25:402–408.
Bhattacharya B, Miura T, Brandenberger R et al. Gene expression in human embryonic stem cell lines: unique molecular signature. Blood 2004;103:2956–2964.
Brandenberger R, Wei H, Zhang S et al. Transcriptome characterization elucidates signaling networks that control human ES cell growth and differentiation. Nat Biotechnol 2004;22:707–716.
Ivanova NB, Dimos JT, Schaniel C et al. A stem cell molecular signature. Science 2002;298:601–604.
Ramalho-Santos M, Yoon S Matsuzaki Y et al. "Stemness": transcriptional profiling of embryonic and adult stem cells. Science 2002;298:597–600.
Sato N, Sanjuan IM, Heke M et al. Molecular signature of human embryonic stem cells and its comparison with the mouse. Dev Biol 2003;260:404–413.
Sharov AA, Piao Y, Matoba R et al. Transcriptome analysis of mouse stem cells and early embryos. PLoS Biol 2003;1:E74.
Sperger JM, Chen X, Draper JS et al. Gene expression patterns in human embryonic stem cells and human pluripotent germ cell tumors. Proc Natl Acad Sci U S A 2003;100:13350–13355.
Kelly DL, Rizzino A. DNA microarray analyses of genes regulated during the differentiation of embryonic stem cells. Mol Reprod Dev 2000;56:113–123.
Ginis I, Luo Y, Miura T et al. Differences between human and mouse embryonic stem cells. Dev Biol 2004;269:360–380.
Stewart CL. Production of chimeras between embryonic stem cells and embryos. In: Wassarman PM, DePamphilis ML, eds. Methods in Enzymology: Guide to Techniques in Mouse Development. Toronto: Academic Press, 1993:823–854.
Niwa H, Miyazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet 2000;24:372–376.
Furusawa T, Ohkoshi K, Honda C et al. Embryonic stem cells expressing both platelet endothelial cell adhesion molecule-1 and stage-specific embryonic antigen-1 differentiate predominantly into epiblast cells in a chimeric embryo. Biol Reprod 2004;70:1452–1457.
Williams RL, Hilton DJ, Pease S et al. Myeloid leukaemia inhibitory factor maintains the developmental potential of embryonic stem cells. Nature 1988;336:684–687.
Stewart CL, Kaspar P, Brunet LJ et al. Blastocyst implantation depends on maternal expression of leukaemia inhibitory factor. Nature 1992;359:76–79.
Shirogane T, Fukada T, Muller JM et al. Synergistic roles for Pim-1 and c-Myc in STAT3-mediated cell cycle progression and antiapoptosis. Immunity 1999;11:709–719.
Ying QL, Nichols J, Chambers I et al. BMP induction of Id proteins suppresses differentiation and sustains embryonic stem cell self-renewal in collaboration with STAT3. Cell 2003;115:281–292.
Hosler BA, LaRosa GJ, Grippo JF et al. Expression of REX-1, a gene containing zinc finger motifs, is rapidly reduced by retinoic acid in F9 teratocarcinoma cells. Mol Cell Biol 1989;9:5623–5629.
Rogers MB, Hosler BA, Gudas LJ. Specific expression of a retinoic acid-regulated, zinc-finger gene, Rex-1, in preimplantation embryos, trophoblast and spermatocytes. Development 1991;113:815–824.
Tanaka TS, Kunath T, Kimber WL et al. Gene expression profiling of embryo-derived stem cells reveals candidate genes associated with pluripotency and lineage specificity. Genome Res 2002;12:1921–1928.
Avilion AA, Nicolis SK, Pevny LH et al. Multipotent cell lineages in early mouse development depend on SOX2 function. Genes Dev 2003;17:126–140.
Bowles J, Teasdale RP, James K et al. Dppa3 is a marker of pluripotency and has a human homologue that is expressed in germ cell tumours. Cytogenet Genome Res 2003;101:261–265.
Chambers I, Colby D, Robertson M et al. Functional expression cloning of Nanog, a pluripotency sustaining factor in embryonic stem cells. Cell 2003;113:643–655.
Mitsui K, Tokuzawa Y, Itoh H et al. The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells. Cell 2003;113:631–642.
Wilkinson DG, Bhatt SHerrmann BG. Expression pattern of the mouse T gene and its role in mesoderm formation. Nature 1990;343:657–659.
Wiese C, Rolletschek A, Kania G et al. Nestin expression: a property of multi-lineage progenitor cells? Cell Mol Life Sci 2004;61:2510–2522.
Weigmann A, Corbeil D, Hellwig A et al. Prominin, a novel microvilli-specific polytopic membrane protein of the apical surface of epithelial cells, is targeted to plasmalemmal protrusions of non-epithelial cells. Proc Natl Acad Sci U S A 1997;94:12425–12430.
Anneren C, Cowan CA, Melton DA. The Src family of tyrosine kinases is important for embryonic stem cell self-renewal. J Biol Chem 2004;279:31590–31598.
Vas V, Szilagyi L, Paloczi K et al. Soluble Jagged-1 is able to inhibit the function of its multivalent form to induce hematopoietic stem cell self-renewal in a surrogate in vitro assay. J Leukoc Biol 2004;75:714–720.
Xue Y, Gao X, Lindsell CE et al. Embryonic lethality and vascular defects in mice lacking the Notch ligand Jagged1. Hum Mol Genet 1999;8:723–730.
Burdon T, Chambers I, Stracey C et al. Signaling mechanisms regulating self-renewal and differentiation of pluripotent embryonic stem cells. Cells Tissues Organs 1999;165:131–143.
SatoN, MeijerL, Skaltsounis L et al. Maintenance of pluripotency in human and mouse embryonic stem cells through activation of Wnt signaling by a pharmacological GSK-3-specific inhibitor. Nat Med 2004;10:55–63.
Cohen MM Jr. The hedgehog signaling network. Am J Med Genet 2003;123A:5–28.
Ogden SK, Ascano M Jr, Stegman MA et al. Regulation of Hedgehog signaling: a complex story. Biochem Pharmacol 2004;67:805–814.
Oka M, Tagoku K, Russell TL et al. CD9 is associated with leukemia inhibitory factor-mediated maintenance of embryonic stem cells. Mol Biol Cell 2002;13:1274–1281.
Rose TM, Weiford DM, Gunderson NL et al. Oncostatin M (OSM) inhibits the differentiation of pluripotent embryonic stem cells in vitro. Cytokine 1994;6:48–54.
Wheatley SC, Isacke CM. Induction of a hyaluronan receptor, CD44, during embryonal carcinoma and embryonic stem cell differentiation. Cell Adhes Commun 1995;3:217–230.
Fortunel NO, Otu HH, Ng HH et al. Comment on " ‘Stemness’: transcriptional profiling of embryonic and adult stem cells" and "a stem cell molecular signature." Science 2003;302:393.
Evsikov AV, Solter D. Comment on "‘Stemness’: transcriptional profiling of embryonic and adult stem cells" and "a stem cell molecular signature." Science 2003;302:393.
Meno C, Saijoh Y, Fujii H et al. Left-right asymmetric expression of the TGF beta-family member lefty in mouse embryos. Nature 1996;381:151–155.
Tzouanacou E, Tweedie S, Wilson V. Identification of Jade1, a gene encoding a PHD zinc finger protein, in a gene trap mutagenesis screen for genes involved in anteroposterior axis development. Mol Cell Biol 2003;23:8553–8552.
Schoor M, Schuster-Gossler K, Gossler A. The Etl-1 gene encodes a nuclear protein differentially expressed during early mouse development. Dev Dyn 1993;197:227–237.
Jones PL, Jones FS. Tenascin-C in development and disease: gene regulation and cell function. Matrix Biol 2000;19:581–596.(Lars Palmqvista, Clive H.)