Human Molecular Genetics Advance Access originally published online on April 27, 2006
Human Molecular Genetics 2006 15(11):1894-1913; doi:10.1093/hmg/ddl112
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The role of PI3K/AKT, MAPK/ERK and NF
ß signalling in the maintenance of human embryonic stem cell pluripotency and viability highlighted by transcriptional profiling and functional analysis
1Centre for Stem Cell Biology and 2Developmental Genetics, Institute of Human Genetics, University of Newcastle, Newcastle upon Tyne, UK
* To whom correspondence should be addressed at: Centre for Stem Cell Biology and Developmental Genetics, Institute of Human Genetics, University of Newcastle upon Tyne, International Centre for Life, Central Parkways, Newcastle upon Tyne NE1 3BZ, UK. Tel: +44 1912418688; Fax: +44 1912418666; Email: majlinda.lako{at}ncl.ac.uk
Received January 16, 2006; Revised March 11, 2006; Accepted April 11, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSMAPK/ERK, PI3/AKT and...DISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Understanding the molecular mechanism by which pluripotency is maintained in human embryonic stem cells (hESC) is important for the development of improved methods to derive, culture and differentiate these into cells of potential therapeutic use. Large-scale transcriptional comparison of the hES-NCL1 line derived from a day 8 embryo with H1 line derived from a day 5 embryo (WiCell Inc.) showed that only 0.52% of the transcripts analysed varied significantly between the two cell lines. This is within the variability range that has been reported when hESC derived from days 5–6 embryos have been compared with each other. This implies that transcriptional differences between the cell lines are likely to reflect their genetic profile rather than the embryonic stage from which they were derived. Bioinformatic analysis of expression changes observed when these cells were induced to differentiate as embryoid bodies suggested that quite a few of the downregulated genes were components of signal transduction networks. Subsequent analysis using western blotting, flow cytometry and antibody arrays implicated components of the PI3K/AKT kinase, MAPK/ERK and NF
ß pathways and confirmed that these components are decreased upon differentiation. Disruption of these pathways in isolation using specific inhibitors resulted in loss of pluripotency and/or loss of viability suggesting the importance of such signalling pathways in embryonic stem cell maintenance. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSMAPK/ERK, PI3/AKT and...DISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Human embryonic stem cells (hESC) hold great promise for understanding early human embryonic development and generating clinically useful cell lines (1
,2). In view of this a great deal of effort is focused on experiments aimed at improving culture conditions (3), generating new hESC lines suitable for clinical purposes (4), genetic manipulation (5,6) and differentiation regimes to produce human cells suitable for transplantation and drug testing (7,8). To date, more than 300 hESC lines have been derived worldwide; however, only a handful of these have been extensively characterized using large-scale transcriptomic approaches. These techniques have shown that similar to murine ESC (mESC), hESC express some of the classical markers of pluripotent stem cell lines such as OCT4, NANOG, alkaline phosphatase (AP) and high levels of telomerase activity (1). However, there are many notable differences between human and mouse ESC in both morphological and behavioral characteristics such as colony thickness, shape and population doubling time (9). The expression of cell surface marker proteins differs considerably (10) and large-scale transcriptional profiling studies undertaken by a number of groups reveal that 
10% of the genes showing significant expression changes between mouse and human ESC are involved in oxidative phosphorylation, suggesting that mESC have a higher ability to generate ATP and higher metabolic rate powered by mitochondrial oxidation (11). Although leukaemia inhibitory factor (LIF) modulated through gp130 signalling and the JAK/STAT pathway is sufficient for the maintenance of undifferentiated mESC, it cannot prevent the spontaneous differentiation of hESC (12). In addition, hESC unlike mESC can spontaneously differentiate to trophectoderm (13). These differences between mouse and human cells suggest that although the same pluripotentiality genes are expressed in both mouse and human ESC, their function and downstream signalling pathways may differ between the two species. This highlights the importance and necessity of studying signalling pathways that underline the maintenance of pluripotency and viability in hESC given the potential importance of stem cell therapies in therapeutic medicine.
Recent publications detailing transcriptional profiling of several hESC lines (14–23) have shown that there are common genes expressed in several cell lines. However, each cell line possesses a unique expression signature which may reflect either the genetic profile of the embryo from which they have been derived or the differences in conditions used either for the derivation, culture or cryopreservation of cell lines in different laboratories (22,23). The differences in gene expression might also be a function of the developmental stage of the embryo from which the hESC lines are derived; however, this hypothesis has been difficult to test because the majority of the hESC lines were derived from days 5–6 blastocysts. More recently, derivation of cell lines from morulae has been described (24), and 2 years ago, our group was able to isolate a new hESC line, hES-NCL1 from day 8 human embryos using a novel three-step culture condition approach (25). This cell line gives us a unique ability to assess whether the developmental stage of the embryo affects the transcriptional profile of the hESC lines. In this manuscript, we report for the first time that the developmental stage of the embryo does not enhance the transcriptional differences already reported between different cell lines and therefore these are more likely to be due to genetic profile of the embryo from which they are derived. We also highlight three signalling pathways, namely PI3K/AKT, RAS/MAPK and NFß, and indicate that they are crucial for the maintenance of pluripotency and/or viability in hESC.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSMAPK/ERK, PI3/AKT and...DISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
hES-NCL1 cell line shows similar phenotypic and transcriptional characteristics to H1 cells
Recently, we reported the derivation of a new hESC line, hES-NCL1, from day 8 embryos, using novel three-step culture conditions (25
). The majority of the hESC lines reported to date have been derived from days 5 and 6 embryos. In view of this, we set out to compare our cell line hES-NCL1 with the H1 line available from WiCell Inc. (Madison, USA) and to determine whether these cells derived from day 8 embryos showed any phenotypic or significant transcriptional difference to cells derived from earlier embryos (such as the H1 line). Both of these cell lines have been extensively characterized in vitro and shown to contribute to teratoma formation in vivo (1,25). We cultured the two cell lines under the same conditions (see Materials and Methods) and by the same investigator continuously up to passage 70. We first examined the expression of cell surface markers (TRA-1-60, TRA-1-81, SSEA-3, 4, 1, CD117 and CD135) by flow cytometry at passages 38 (Fig. 1A), 56 (data not shown) and 70. In all cases, we observed very similar marker expression for both cell lines. No gross karyotypic abnormalities were found up to 70 continuous passages in culture (Fig. 1B). Both of the cell lines expressed markers such as OCT4, NANOG, REX1 and TERT at both passages 38 (data not shown) and 70 (Fig. 1C) as shown by the reverse transcription assay. As this method is only semi-quantitative and does not reflect properly the levels of gene expression between the two cell lines, we investigated their relative expression using microarray technology as outlined in Expression profiling of hES-NCL1 and H1: comparison with other hESC lines. This indicated that there was no significant difference in the expression levels of NANOG, OCT4, TERT and REX1 between the two cell lines (Fig. 1D; in all cases, P>0.5). This result was confirmed by real-time reverse transcription (RT)–PCR analysis (data not shown).
|
To assess the growth rate of the two cell lines, hESC colonies were disrupted to single cells by digestion with accutase, which results in much higher cell viability compared with trypsin digestion (our own unpublished data). Fifty thousand cells were plated in triplicate on fresh mitotically inactivated mouse embryonic fibroblasts and allowed to grow for 5 days before being subjected to the digestion procedure and counted again. This was repeated for five passages and the results indicate that the average population doubling time for the hES-NCL1 line was 43.5 h and that for the H1 line was 44.2 h.
To assess the degree of variability in differentiation capacity, we removed colonies of both hESC lines from the feeder cells and cultured them in suspension as embryoid bodies (EBs) for up to 15 days. Differentiation was confirmed by flow cytometry for the hESC-specific cell surface marker, TRA-1-60 which showed significant downregulation during the differentiation of hESC (74% of the hESC stained with TRA-1-60) to EBs (6.05 and 4.11% of the cells remained TRA-160-positive after 7 and 14 days of differentiation, respectively; data are presented as an average of both cell lines). Samples of cells were removed at days 0, 5, 10 and 15 and subjected to quantitative real-time RT–PCR for markers of pluripotency (GDF3), endoderm (AFP), mesoderm (KDR) and ectoderm (PAX6; Fig. 1E). As reported for other cell lines (14), both lines used in the study differentiated into cells that expressed markers indicative of the three germ layers. However, there were some differences in the expression levels of the same gene between the two cell lines studies at identical time points which might indicate their preference for differentiation towards a specific lineage. Most importantly, changes in marker expression can reflect the level of spontaneous differentiation occurring at different rate in different hESC lines. Although we have tried to minimize such differences between the two cell lines (Fig. 1A), the real-time RT–PCR is a sensitive technique, and hence minor changes in the level of spontaneous differentiation can result in rather major changes in gene expression during the differentiation of these two cell lines.
Expression profiling of hES-NCL1 and H1: comparison with other hESC lines
The gene expression pattern in both cell lines was assessed by hybridization of their respective mRNA output at passages 38–45 to the human U133 Plus 2.0 array which represents a coverage of 47 000 transcripts and variants including 38 500 well-characterized human genes and ESTs. We only considered a gene to be present in each cell line if that gene received a present call in three independent samples and the detection P-value was less than 0.000244. Twenty-two thousand two hundred and thirty-two transcripts (47.3%) were found to be present in the H1 cell line and 21 683 transcripts (46.13%) were detected in the hES-NCL1 line. To identify genes that were present in both cell lines, a Venn diagram was drawn using Genespring (Fig. 2A) and this showed that 20 382 transcripts were present in both cell lines (43.3%), whereas 1850 (3.9%) transcripts were present in H1 but not in hES-NCL1 and 1301 (2.7%) transcripts were present in hES-NCL1 but not in H1. Abeyta et al. (14) reported that 52% of the total transcripts were shared by all the three cell lines they analysed (H9, HSF-1 and HSF-6) and this is slightly higher than that we have observed (43.3%); however, the differences might relate to stringency of criteria chosen because ‘present and marginal’ expression was included in their list, whereas ‘present only’ was the selection criterion in our study. Most importantly, this can be dependent on the number of cell lines being compared. To verify this, we extended our analysis to the four hESC cell lines of University of Helsinki (FES21, FES22, FES29 and FES30) and three hESC cell lines of Karolinska University (HS181, HS 235, HS237) kindly provided to us by Dr Heli Skottman (23). A comparative analysis of all nine cell lines indicated that less transcripts were shared between all of them (27.8% of total transcripts, Fig. 2B). This highlights the need to compare larger number of cell lines, especially if new ESC biomarkers are sought.
|
A number of groups have tried to identify genes that can be used as hESC markers (22
,23). This approach has resulted in several gene lists which show increasing discrepancies from one publication to the next (22,23). The variability in the results probably reflects the differences in the technical approach used for such large-scale expression studies (22), differences in culture conditions (23), the number of replicates and the stringency of statistical analysis as well as the availability of raw data from a large number of hESC lines (26). Because the number of transcripts shared between different cell lines is partly dependent on the number of hESC lines being compared, we hypothesized that more specific hESC biomarkers can be identified by increasing the number of cell lines being compared. To investigate this, we initially compared the H1 and hES-NCL1 lines with the differentiated progeny derived from each line to identify hESC-enriched genes. The comparison level analysis of each hESC line to non-lineage differentiated reference samples (these were differentiated progeny from hESC lines obtained after 3 weeks) defined a gene as significantly enriched in hESC if the average signal fold change (FC) between the hESC and the reference sample was larger than two and the gene in hESC sample was present. This analysis identified 1232 transcripts that were enriched in both H1 and hES-NCL1 compared with more differentiated progeny (Fig. 2C). Addition of another seven hESC lines (FES21, FES22, FES29, FES30, HS181, HS235 and HS237) to this comparison reduced the list of ESC-specific transcripts to 235 (representing 206 genes and ESTs; Fig. 2C, and for a complete list, see Supplementary Material, Table S1), thereby highlighting the validity of our hypothesis. Most importantly, the list of hESC-specific genes comprised several well-known hESC markers already identified in other studies such as OCT4, NANOG, CD24, GDF3, TDGF1, TERF1, GABRB3, POLR3G, LECT1, DPPA4, GALANIN, DNMT3B and LIN28 (14–24,34). In addition to the hESC-enriched transcripts, we were also interested to identify highly expressed genes in all nine cell lines and any potential overlaps between the two populations. To select highly expressed transcripts in both cell lines, the Genespring feature called ‘filter on expression level’ was used by applying a minimum cut-off point of 110 to normalized data from the nine cell lines. A list of 117 transcripts representing 85 genes and ESTs (Supplementary Material, Table S2) was identified. From this analysis, we observed that the most abundant class of transcripts expressed in both cell lines were ribosomal proteins followed by transcription factors and nucleic acid binding proteins and protein processing and trafficking systems, which is perhaps indicative of a high rate of protein turnover in hESC. This is in accordance with other studies which have used SAGE and MPSS as a technique for identification of genes highly expressed in hESC lines (11,20). Comparison of the 85 highly expressed genes and ESTs with the 206 hESC-enriched genes and ESTs showed that only two genes were shared between them (HDCMA18P and CACNA2D3). Investigation of several databases including NetAffix, NCBI, Unigene, etc. revealed that HDCMA18P is likely to be involved in RNA binding and processing, whereas the CACNA2D3 (calcium channel voltage-dependent alpha2/delta3) is likely to mediate the influx of calcium ions into the cell upon membrane polarization. Both of the genes are novel, therefore further detailed investigations are needed before any conclusion can be reached regarding their function in hESC. Other key pluripotency factors such as OCT4, REX1, NANOG, LIN28, CD24 and TERT are not represented among the highly expressed transcripts. This shows that highly expressed genes, although being part of the molecular signature for the hESC, will not necessarily be part of the functional machinery of hESC that can be exploited further for directing their differentiation.
Identification of most changed transcripts between the two cell lines
Because the hES-NCL1 line was derived from day 8 blastocysts, we were interested to determine whether it showed significant changes in gene expression from the H1 line which is derived from day 5 embryos. Using the Genespring software, we identified genes that were unique to H1, unique to hES-NCL1 or differentially expressed between the two cell lines. To identify the genes unique to each cell line, we followed the criteria that the gene had to be present in three out of the three replicates for the target cell line and absent in three out of the three replicates for the reference cell line. This identified 61 genes present in H1 (0.27% of all transcripts present) but completely absent in hES-NCL1 (Supplementary Material, Table S3) and 49 genes which were present only in hES-NCL1 (0.22% of all transcripts present; Supplementary Material, Table S3). The differential expression of six of the uniquely expressed candidates was confirmed by real-time PCR analysis (Fig. 3B). The most differentially expressed transcripts were of various natures: about one-third is as yet uncharacterized on a functional level and more than 10% are involved in DNA transcription (Supplementary Material, Table S3). The H1 and hES-NCL1 lines have different karyotype (i.e. H1 is male, NCL-1 is female; Fig. 1B), and this might contribute to their unique expression pattern. However, this does not seem to be a major factor, because only 3/61 genes that are unique to H1 are located on the Y chromosome (Fig. 3A). It is highly likely that these genes do not represent important regulators of pluripotency or differentiation because both cell lines are capable of differentiating into multiple cell lineages. It is interesting, however, that the nine genes we investigated by real-time RT–PCR not only show changes in expression between the two hESC lines, but also between each line and the embryonic carcinoma cell line (NTera2-SP.12) which was used as a positive control (Fig. 3B). Although the embryonic carcinoma cells show chromosomal abnormalities, they do retain the capacity to differentiate into several lineages and express most of the genes normally found in hESC lines albeit not necessarily at exactly the same level. This again suggests that variability in the expression of the candidate genes we have selected through the microarray analysis is unlikely to affect the pluripotency or differentiation capacity of these cells; however, genes which differentially expressed between the two cell lines might be useful as a ‘fingerprint’ for each cell line.
|
Using the Genespring software, we also investigated genes that consistently show more than 2-fold expression changes between the two cell lines in three replica experiments. A stringent P-value was chosen to ensure significant differences between the two cell lines. Surprisingly, we found that only six genes which comprise 0.03% of the transcripts present in the two cell lines (Fig. 3A) fell into this category. The differential expression of three of these candidates (SLC1A6, PDCD6 and MNS1) was also confirmed by real-time RT–PCR (Fig. 3B). These results are very similar to a study published by Carpenter et al. (21
), which showed that only three genes out of 2802 cDNA clones showed differential expression between four hESC lines (H1, H7, H9 and H14). The number of the unique genes that we found for each hESC line together with the genes that are differentially expressed between them is significantly lower than what has been previously reported for other hESC lines such as HSF-1, HSF-6 and H9 (14) or HS181, HS235, HS237, FES21, FES22, FES29 and FES30 (23). These differences could be attributed to changes in number of replicates used in the gene expression studies [two replicates were used by Skottman et al. (23)], the number of hESC lines being compared or the stringency of gene selection criteria [present and marginal genes were used by Abeyta et al. (14)]. Our own unpublished data suggested that both the increase in number of replicates and selection of only genes deemed to be present reduced significantly the number of genes considered unique or differentially expressed for each cell line (for example, comparison of the two cell lines using the criteria of present or marginal in one out of three samples for each hESC line increased the number of genes that showed two or more FC to 64).
In summary, the percentage of genes that are differentially expressed between the hES-NCL1 cell line derived from day 8 embryos to H1 derived from day 5 embryos is only 0.52%. This is within the variability limits described for ESC derived from days 5–6 embryos such as HSF-1, HSF-6 and H9 (2.4–4.5%) (14) and HS181, HS235, HS237, FES21, FES22, FES29 and FES30 (2.1–3.6%) (23). Large-scale gene expression studies in lymphoblastoid cells lines derived from unrelated individuals have shown that up to 5% of the genes can vary from one individual to another (27). In view of this, we would like to suggest that changes in the gene expression between the two cell lines we have analysed are likely to be due to the changes in the genetic profile of the embryo rather than the developmental stage of the blastocyst from where they were derived. To verify this further, one will have to compare directly gene expression patterns between ICMs dissected from different stages of the blastocyst's development and hESC lines derived from earlier embryos such as morulae. This work is currently progressing within our group.
Identification of new hESC markers
One of our aims in this investigation was to identify new hESC markers; therefore, we collected RNA samples from days 6, 10, 14 and 21 EBs from the hES-NCL1 line and hybridized these samples to the human U133 Plus 2.0 array. Using the Genespring software and a stringent selection criterion of at least 4-fold downregulation between the hES-NCL1 line and each group of EBs collected at different days (days 6, 10, 14 and 21, respectively), four groups of transcripts were selected (Fig. 4A). A Venn diagram was drawn to identify the set of transcripts that were common to all the four subsets and indicated genes that were expressed in the hES-NCL1 but were downregulated 4-fold or more at all differentiation time points examined. One hundred and eighteen transcripts which represented 105 genes were identified and among those the major category comprised uncharacterized transcripts and genes involved in signalling pathways (Fig. 4B, and for a complete list, see Supplementary Material, Table S4).
|
It is interesting to note that 12.38% of the identified transcripts have been already associated with maintenance of pluripotency in human and mouse ESC previously (OCT4, NANOG, TDGF1, GDF3; Supplementary Material, Table S4) or selected on the basis of higher expression in ESC versus EBs (GAL, SEMA6A, LECT1, FLJ10884, POLR3G, LGALS1, GABRB3, TFRC, TERF1, DPPA4, DNMT3B, NODAL, HESX, PRDM14) by other groups (14
–23). Preliminary studies in murine and human ESC have already revealed a role for DNMT3B, NODAL, POLR3G and GAL in cell growth, viability and maintenance of pluripotency (28–35). In view of the published evidence and availability of commercial antibodies for some of these factors, we decided to investigate in further detail the expression and localization of LGALS1, TFRC and GABRB3 in hESC (Fig. 4C). LGALS1 also known as galectin1 can restrict carcinoma cell growth via suppression of RAS/MEK/ERK signalling pathway and induction of p21 and p27 transcription (36). This evidence seems to suggest that the high and selective expression of LGALS1 in hESC is required for maintaining ESC homeostasis and perhaps preventing a tumorigeneic expansion; however, further work is needed to verify this suggestion. The transferrin receptor, TFRC, acts normally as a gatekeeper for maintaining iron homeostasis by mediating endocytosis of its ligand, iron-loaded transferrin, and high expression of this receptor has been observed in human mesenchymal stem cells and mESC (37,38). However, ablation of this gene in murine embryos suggests that transferrin and its receptor TFRC are mostly required for development of haematopoietic cells such as erythrocytes and lymphocytes but they are not essential for development of other tissues (38,39). The human GABA A receptor subunit beta 3 (GABRB3) was found to be an imprinted gene and a possible candidate for Prader–Willi syndrome (40), and the downregulation of its paternal allele in murine embryos results in a neurodevelopmental disorder which mimics Angelman syndrome (41); however, no functional role in stem cells has been described so far. A recent paper described the successful use of 
-aminobutyric acid and pipecolic acid which enhances GABA A receptor responses together with FGF2, LiCl2 and TGFß in defined medium to grow and derive new hESC lines (42). This was based on previous reports which indicated that GABA can stimulate proliferation of both neural and non-neural tissue (43) and microarray results similar to the ones presented here (15). Expression differences for these three genes were examined by flow cytometry (both intra and extracellular, data not shown) and immunocytochemistry (Fig. 4C). This indicated that TFRC is the only marker with cell surface expression; therefore, we compared TFRC expression with that of TRA-1-60 in hESC and day 14 EBs by flow (Fig. 4D). This analysis showed that 4% of the cells maintained the expression of TFRC after 14 days of differentiation; however, this was also the case with TRA-1-60 staining, suggesting that TFRC may serve as a new marker to distinguish hESC from cells present in EBs.
Signalling pathways in hESC
The second largest category of transcripts enriched in hESC similar comprised the genes involved in signalling pathways (Figs 4B and 5A). Interestingly, a significant number of those candidates fell within three major signalling categories belonging to the RAS/MAPK/ERK signalling pathway (G3BP, RASAL2, SOS1 and ITGB1BIP3), the PI3K/AKT (PI3KCB and PTEN) and NFß signalling (LCK, PELLINO1 and TNSF11).
|
The RAS/MAPK/ERK, PI3K/AKT and NF
ß signalling pathways identified above represent novel pathways that have not been thoroughly investigated in hESC. These pathways are very complex and involve a large number of components, and to facilitate the unfamiliar reader, we have outlined them schematically in Figure 5B.
RAS is a small GTPase responsible for transducing signals from a vast array of receptor tyrosine kinases and other receptors. RAS itself is regulated by GTPase-activating proteins (GAPs), which alternate signalling. RAS GAP transduces signals downstream of RAS in a MAPK-dependent manner (Fig. 5B, top panel). RAF serine/threonine protein kinases are considered to be the primary RAS effectors and are normally localized in the cytosol in inactive form. However, c-RAF (also known as RAF1) associates with RAS at the plasma membrane after growth factor-induced RAS guanine nucleotide exchange and its activation, which links a series of phosphorylation and changes in subcellular localization, links RAS to mitogen and extracellular-regulated kinase (MEK1/2) and MAPK. RAF directly phosphorylates and activates the protein kinase MEK1/2 which in turn phosphorylates and activates MAPK/ERK1 and MAPK/ERK2 (Fig. 5B, top panel). Activated MAPK/ERKs are then translocated from the cytoplasm to the nucleus where they activate nuclear transcription factors leading to changes in gene expression (Fig. 5B, top panel). Pi3K enzymes are normally regulated by growth factors and serve to phosphorylate phospholipids at the plasma membrane (Fig. 5B, bottom panel). Activated PI3K phosphorylates PIP2 and generates PIP3 which acts as a second messenger. AKT interacts with PIP3 and subsequently translocates to the plasma membrane. AKT also referred as PKB plays a critical role in controlling the balance between survival and apoptosis (Fig. 5B, bottom panel) and is activated by phospholipid binding and activation loop phosphorylation at threonine 308 by PDK1 and phosphorylation at the C-terminus at serine 473. It is inactivated by PTEN phosphatase. AKT promotes cell survival by inhibiting apoptosis through its ability to phosphorylate and inactivate several target genes such as BAD, c-RAF and caspase-9 (Fig. 5B, bottom panel). In addition, it regulates glycogen synthesis through phosphorylation and inactivation of GSK3ß and GSK3
. Activation of AKT survival signalling pathways leads to phosphorylation of Iß, the cytoplasmic inhibitor of nuclear factor ß (NFß). Phosphorylation of Iß initiates proteasome-dependent degradation and allows NFß to enter the nucleus and regulate NFß-dependent gene expression (Fig. 5B, bottom panel).
Ras-GTPase-activating protein, SH3-domain-binding protein (G3BP1), was downregulated more than 4-fold during the differentiation of hESC (Fig. 5A). G3BP1 binds to the SH3 domain of Ras GAP and is shown to act as a Ras effector in breast cancer cell lines (44). Complete ablation of this gene in mouse results in embryonic lethality indicating an important role during embryonic development (45). Other activators of the pathway, such as RASAL2 and SOS1, which are guanine nucleotide exchange factors are also downregulated at the transcriptional level during differentiation, implying a greater role for RAS/MAPK signalling in hESC (Fig. 5A). We observed that one of the integrin ß1 binding proteins, ITGB1BP3, which has been shown to affect laminin deposition and myogenic differentiation (46,47) is expressed at much higher levels in ESC compared with EBs (Fig. 5A). We investigated other members of the RAS/MAPK/ERK signalling pathway (Supplementary Material, Table S5) and noticed that several other components (MAP4K1, MAP2K6, MAP3K7, MAP4K3, RAF1, MAPK13, MAPK9, MAPK6, MAPK1/ERK1, MAP2K1, KRAS, MAP3K4, NRAS, BRAF) were transcriptionally downregulated during the differentiation process, albeit less than 4-fold. Taken together these insights acquired from the transcriptional study suggest a role for the RAS/MAPK/ERK signalling pathway in hESC.
Our transcriptional data also suggested a downregulation of several components of the NFß signalling pathway such as LCK, PELLINO1 and TNFS11/ RANKL (Fig. 5A). LCK, a lymphocyte-specific tyrosine kinase, (Src kinase) required for the activation of the NFß signalling pathway (47) has been shown to be downregulated during mESC differentiation (48). In addition, the product of another candidate gene, PELLINO1, identified in this screen is required for NFß activation, probably through interactions with IRAK and tumour necrosis factor receptor associated factor TRAF6 (49). Other members of these signalling pathways were either absent (TLRs) or slightly upregulated (Supplementary Material, Table S5). However, as for most of the signalling pathways, phosphorylation/translocation of key components as well as abundance of ligands/activators plays the most important role. In view of these, we looked carefully at NAK (NFß activating enzyme; Supplementary Material, Table S5) which showed less than 4-fold downregulation during the differentiation process and the IKBK
(which enhances the stability of Is and therefore downregulates NFß activity) which was absent in both hESC lines and upregulated during the differentiation process (Supplementary Material, Table S5). Most importantly, the receptor activator of nuclear factor (NFß) ligand (RANKL/TNFS11) which is also enriched in hESC compared with EBs has been shown to be translationally regulated by PDK1 via the PI3K/AKT pathway suggesting a potential link between these two pathways (50,51). Interestingly, PDK1 activates NFß signalling by phosphorylating IKKß which leads to its degradation and therefore allowing NFß p65 to enter the nucleus (51,52). AKT has also been shown to be involved in transactivation of p65 and subsequent NFß-dependent transcription (52). These data seem to suggest an intrinsic link between the NFß and PI3K/AKT signalling pathways in other cell lines; however, this has not been explored previously in hESC. Of interest is the finding that PI3K/AKT pathway has been shown to be important for the maintenance of pluripotency in mESC (53), whereas FGF2 which is necessary for hESC growth activates ERK signalling (54–56). In view of this, it is important to investigate whether FGF2 signalling activates PI3K/AKT pathway in addition to MAPK/ERK signalling in hESC.
Although our transcriptional data suggest that the three selected pathways, PI3K/AKT, NFß and MAPK signalling, might be important for proliferation and maintenance of pluripotency in hESC, in-depth studies to investigate the role of these pathways in hESC and potential links between them have not been carried out as yet. We investigated the expression of some crucial components of these three pathways in hESC and changes during their differentiation to EBs. Immunocytochemistry on both H1 and hES-NCL1 on various components of these three signalling pathways (see Fig. 5C, D and E for RAS/MAPK, PI3K/AKT and NFß signalling pathways, respectively) demonstrated cytoplasmic/cell membrane localization for the phosphorylated form FRS2 and nuclear localization for phospho-ELK1 and c-MYC (Fig. 5C). The phosphorylated form of c-RAF serine 338 as well as GAB1, SOS1 and GRB2 showed both nuclear and cytoplasmic/cell membrane localization (Fig. 5C). Most of the components of the PI3K/AKT pathway (phospho-GSK3ß, phospho-c-RAF serine 259, phospho-PDK1 and phospho-AKT threonine 308) showed cytoplasmic/cell membrane localization and a few (unphosphorylated AKT, phospho-AKT serine 308 and phospho-PTEN) displayed both nuclear and cytoplasmic/cell membrane localization (Fig. 5D). In contrast, all the tested components of the NFß pathway showed nuclear localization with the exception of TNSF11 which showed some additional spotty expression in the cytoplasm (Fig. 5E). We were unable to detect the expression of phospho-MEK1/2, phospho-ERK1/2 (also known as p44/42) and phospho-90 RSK (Fig. 5B) which suggests that they might be expressed at lower levels. To improve the detection sensitivity and assess changes between hESC and EBs, we carried out flow cytometry analysis for all the components involved in these pathways for which commercial antibodies were available (Fig. 5F–H). This showed that the key components of the RAS/MAPK/ERK pathway (phospho-c-RAF serine 338, phospho-MEK1/2, phospho-ERK1/2, phospho-ELK1, phospho-c-MYC and phospho-p90; Fig. 5F), PI3K/AKT pathway (phospho-PDK1, phospho-PTEN, phospho-AKT serine 473 and phospho-AKT threonine 308, phospho-c-RAF serine 259; Fig. 5G) and NFß pathway (phospho-65 and TNFS11; Fig. 5H) were all downregulated during differentiation. For some of the MAPK/ERK signalling pathway, this downregulation was confirmed by western blotting (Fig. 5F, lower panel). To confirm our initial flow cytometry results, we also took advantage of the panorama Ab microarray for cell signalling components where a good percentage of the antibodies of interest are spotted in duplicate on nitrocellulose filters. We hybridized hESC and day 14 EB cell extract to those arrays and calculated the ratio of hESC/EB for each protein of interest (Fig. 5I). This analysis confirmed our flow cytometry results and suggested that hESC expressed higher levels of key components of the MAPK/ERK signalling pathways (phospho-ERK1/2, phospho-c-RAF and phospho-c-MYC) as well as the PI3K/AKT (phospho-AKT serine 473 and phospho-AKT threonine 308) and NFß pathways (p65 and NAK- NFß-activating kinase). These results were confirmed by dye reversal experiment (Supplementary Material, Fig. S1). In both experiments, we observed that hESC/EB ratio for GRB2 was different in each of the cell lines. Investigation of the normalized values for GRB2 in the two cell lines showed very similar expression (Supplementary Material, Table S5); however, different expression was observed in day 14 EBs obtained from the two different hESC lines. This transcriptional difference could account for rather different hESC/EB ratios in the two cell lines.
FGF, BMP/TGFß and WNT signalling have been implicated in the maintenance of pluripotency in human ESC (54–56). In view of these, we investigated our microarray data for the expression of main components of these signalling pathways. A close investigation of the FGF2 signalling pathway indicated an abrupt downregulation of FGF2 during the differentiation process as well as two of the FGF target genes, c-FOS and c-KIT (Supplementary Material, Table S5). This is in accordance with other published findings which demonstrate a crucial role for FGF2 in maintaining the pluripotency of hESC (54–56). Investigation of the BMP/TGFß signalling pathways suggested that a large number of the components of this pathway were present in undifferentiated hESC, although some time at minimal levels (Supplementary Material, Table S5). Like other researchers, we found that NODAL and two of its inhibitors LEFTY1 and LEFTY2 were expressed in human ESC, but they were abruptly downregulated upon differentiation suggesting a tight control of this signalling pathway in pluripotency hESC (16,54). Interestingly, ALK5 that has been implicated in the TGFß signalling is the only one among the ALKs that show downregulation upon differentiation, thus providing further support for the role of TGFß in maintaining the hESC pluripotency as suggested by Amit et al. (3). Most interestingly, two inhibitors of Activin A and BMP signalling, FOLLISTATIN and BAMBI (BMP and Activin membrane bound inhibitor) are greatly increased upon differentiation, whereas a third inhibitor, CERBERUS, shows an abrupt decrease upon differentiation. Vallier et al. (54) suggested that Activin/Nodal pathway in co-operation with FGF2 is necessary for the maintenance of pluripotency in hESC. Our data does corroborate these findings. In spite of this, it highlights once more the complexity of these pathways and the fine tune imposed not only by ligands and receptors, but also by specific inhibitors.
|
MAPK/ERK, PI3/AKT and NFß are important for maintenance of pluripotency and viability in human ESC
|
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSMAPK/ERK, PI3/AKT and...DISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
To validate the importance of these signalling pathways for maintenance of pluripotency and viability of hESC, we used three inhibitors: Ly294002 (10 µM) which is a specific PI3 kinase inhibitor already shown to interfere with the maintenance of pluripotency in mESC (57
), U0126 (20 and 30 µM) a specific MEK1/2 inhibitor and sodium pyrrolidinethiocarbamate (PTDC) (50 µM) a specific NFß inhibitor which prevents p65 entry into the nucleus. Inhibitors concentration was chosen carefully based on the criteria of inhibiting the chosen pathway without causing other non specific effects. In the case of U0126, lower concentrations (10 µM) did not result in inhibition of MAPK/ERK even after 5–30 min of application. This could be due to the presence of FGF2 in the medium which activates ERK signalling and counteracts the effects of U0126; therefore, higher concentrations of this inhibitor (20–60 µM) were tested, 50 and 60 µM caused extensive cell death. Therefore, we selected the lowest concentrations (20 and 30 µM) at which inhibition of the MAPK/ERK signalling pathway was evident. The same logic was applied for the NFß inhibitor, although published literature recommends a much higher effective concentration (100 µM) compared with the other two inhibitors. We noticed that at 50 µM concentration, the entry of p65 into the nucleus was inhibited (Fig. 6H) and the effects on cell differentiation and death were the same as when 100 µM PTDC was applied; therefore, 50 µM PTDC was used in all the following experiments.
|
Two to three days after the application of PTDC and LY294002, cell differentiation was prominent in the middle of hESC colonies compared with the control which was vehicle alone (Fig. 6A). Cell differentiation was not evident in the lower (20 µM) U0126 concentration (Fig. 6A); however, this was observed as soon as the U0126 concentration was increased to 30 µM (data not shown). In addition, significant cell death was present when cells were treated with U0126 and PTDC as assessed by Annexin V staining (Fig. 6B). The increase in U0126 concentration also caused an increase in total cell death which reached 62.6% at 30 µM. The amount of differentiation was assessed by alkaline phosphatise staining (Fig. 6C), real-time RT–PCR analysis for OCT4, NANOG and SOX2 (Fig. 6D) and flow cytometry for SSEA-4 (Fig. 6E). In the case of LY294002 and PTDC, a significant reductions in OCT4, NANOG and SOX2 expression and AP staining and the percentage of SSEA-4-positive cells (Fig. 6C–E) were observed, suggesting loss of pluripotency upon inactivation of PI3K/AKT and NF
ß signalling pathways. This was also evident at the higher concentration of U0126 (30 µM; Supplementary Material, Fig. S2A and B) but not at the 20 µM concentration (Fig. 6B–E).
PI3K inactivation by LY294002 resulted as expected in downregulation of the phosphorylated AKT kinase (both serine 473 and threonine 308) and the downstream targets of AKT kinase such as phosphorylated form of GSK3ß and c-RAF (serine 338; Fig. 6F and I). In addition, it resulted in downregulation of several components of the MAPK/ERK pathway such as the phosphorylated form of MEK1/2, ERK1/2 and one of the intracellular effectors, phosphorylated form of c-MYC (Fig. 6F and I). The active form of AKT has been shown to directly phosphorylate and activate c-RAF (58) (Fig. 5B), and from our results, it is obvious that inhibition of the PI3/KAKT pathway results in downregulation of the phosphorylated form of c-RAF (serine 338). Because the active form of c-RAF phosphorylates MEK1/2, we are inclined to speculate that the observed downregulation of the other components of the MAPK/ERK signalling pathway does not simply reflect the process of differentiation ongoing in these cells but rather reflects a direct link between the PI3K/AKT and MAPK/ERK pathway occurring at the level of c-RAF (Fig. 7). Application of LY294002 also resulted in reduction of phosphorylated form of p65; however, inhibition of the NFß pathway did not result in inactivation of the PI3K/AKT pathway (Figs 6F, H, I and 7) suggesting that the PI3K/AKT pathway acts upstream of the NFß pathway. It is widely reported in the literature that activated AKT kinase phosphorylates Iks, which serve as cytoplasmic sequestrators of p65 and p50. Phosphorylation of Iks results in their ubiquitination and subsequent degradation (Fig. 5B) and translocation of the activated p65/p50 complex into the nucleus. Our initial results suggest that this is also likely to be the case for hESC; however, further investigation into the Iks and their expression and function in hESC is required.
|
In several tumour cell lines, p65 expression has been observed in both cytoplasm and nucleus, and activation of the NF
ß pathway is marked by translocation of the cytoplasmic form of p65 into the nucleus where together with p50 it activates transcription of several genes. Immunocytochemistry in both hESC cell lines indicated that p65 and p50 are observed only in the nucleus (Fig. 5E). Application of PTDC, a known inhibitor of the NFß pathway, inhibits the translocation of p65 to the nucleus and this is evident following addition of PTDC to hESC cultures where several nuclei without p65 expression are duly observed (Fig. 6H). As expected, no changes in p50 expression or localization were observed. Downregulation of nuclear p65 expression caused loss of viability as well as differentiation of hESC highlighting the importance of NFß pathway in the maintenance of viability and pluripotency (Fig. 6A–E).
Inhibition of the MAPK/ERK signalling pathway by U0126 resulted as expected in decrease in the phosphorylated form of ERK1/2 and one of its downstream effectors, the phosphorylated form of p90, RSK (Fig. 6G). It is worth mentioning that no changes were observed in the expression of most of the components of the PI3K/AKT pathway (phospho-AKT serine 473 and threonine 308, phospho-PDK1, phospho-GSK3ß and phospho-c-RAF), whereas inhibition of PI3K/AKT pathway resulted in downregulation of most of the components of MAPK/ERK signalling cascade (Fig. 6F). These results imply that MAPK/ERK signalling is downstream of the PI3K/AKT kinase. Inhibition of the MAPK/ERK signalling at the level of MEK1/2 does not seem to affect two of the components of the NFß pathway (Fig. 6G); however, inhibition of the NFß pathway results in downregulation of the activated form of ERK1/2 and its downstream effector, p90 RSK (Fig. 6H and I), implying that NFß pathway acts upstream of the MAPK/ERK signalling pathway. These results are summarized in a table (Fig. 6I) and schematically outlined in Figure 7.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSMAPK/ERK, PI3/AKT and...DISCUSSIONMATERIALS AND METHODSSUPPLEMENTARY MATERIALREFERENCES |
|---|
Elucidation of the signalling events that govern the self-renewal and pluripotency of hESC is important for understanding their physiology and devising optimal culture conditions for their large-scale growth. It is of equal importance to understand the changes which take place in such signalling networks as a cause or consequence of hESC differentiation, as this may help us to devise better strategies to direct differentiation into particular developmental pathways. This will be of paramount importance to our efforts to produce defined populations of cells for therapeutic use. Transcription factors such as OCT4 (59
) and NANOG (60) are indispensable for the maintenance of pluripotency in hESC but it has recently become evident that activation of FGF2 signalling (61) or ERK activation (55) and its co-operation with the Activin/Nodal signalling pathway (54) are necessary to maintain the pluripotency of hESC. Substantial differences exist in a number of key signalling pathways between human and mouse ESC; for example, LIF/Stat3 and Bmp4 signalling which are important for maintenance of pluripotency in mESC do not play the same role in hESC (12,13,61,62). In striking contrast to mESC which display high ERK1/2 activity when they are stimulated to undergo differentiation, hESC display high basal activity of ERK1/2 in the undifferentiated state which correlates with large amounts of endogenous, high molecular mass FGF2 (63). These fundamental differences highlight the need for new and thorough signalling studies in hESC per se.
We undertook a large transcriptional study which was focused on the molecular characterization of a new cell line derived by our group from a day 8 blastocyst (hES-NCL1) and highlighted important changes in gene expression during differentiation to EBs. We identified an interesting phenomenon in that the second largest group of transcripts downregulated during differentiation of hESC were components of several important signalling pathways, namely PI3K/AKT, RAS/MAPK/ERK and NFß signalling. Using a combination of flow cytometry, western blotting and an antibody signalling array, we confirmed that the majority of factors involved in these signalling pathways and identified by our large-scale transcriptional study were downregulated during the differentiation of hESC, indicating a possible role in the maintenance of pluripotency and viability in hESC. This hypothesis was further confirmed by using specific inhibitors for each of the signalling pathways (LY294002 for PI3K/AKT, U0126 for MAPK/ERK and PTDC for NFß) which caused loss of pluripotency and viability at specific concentrations. These results are not entirely surprising given the contribution of the aforementioned signalling pathways in a number of human cancers (64) and during embryogenesis (65–67).
A number of studies conducted in mouse embryos have shown that most of the MAPK/ERK pathway proteins were present in trophoblast and the inner cell mass at embryonic day 3.5, where they may mediate mitogenic signals within the embryo or coming from the uterus (65). Most importantly, knockouts of FRS2, Gab1, Grb2, Sos-1, Raf-B and Raf1 result in peri-implantation or placental lethality (66) and this could well be due to defects in formation of placental trophoblasts given the role of Ras in extraembryonic endoderm differentiation (67). In this context, it worth mentioning two main differences between the human and the murine system that can help with further interpretation of our results. First, in mouse, FGF/MAPK signalling seems to be involved in trophoblast and thus placental development (68,69), whereas in the humans, relatively high and stable concentration of FGF2 is needed to maintain the proliferation and pluripotency of hESC (61). Secondly, in the murine system, the suppression of MAPK/ERK signalling by BMP4 promotes the self-renewal of mESC and aids their derivation (70–72), whereas in the human, addition of BMP4 causes differentiation of hESC to trophoectoderm (13). This suggests a different role for MAPK/ERK signalling in hESC and our results show that this is indeed the case. We observed significant cell death in hESC cultures treated with 20 µM of MEK1/2 inhibitor (U0126) and this increased with increasing concentrations of inhibitor. No prominent loss of hESC morphology and pluripotency was observed at the 20 µM concentration of U1026, whereas an increase in U0126 concentration to 30 µM caused cell differentiation as described in