Author Contribution: Y.Y.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; C.H. and T.M.L.V.: collection and/or assembly of data and data analysis and interpretation; X.Z.: data analysis and interpretation; and S.-C.Z.: conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS October 13, 2011.
Fibroblast growth factor (FGF) signaling and PAX6 transcription are required for neuroectoderm specification of human embryonic stem cells (hESCs). In this study, we asked how FGF signaling leads to PAX6 transcription and neuroectoderm specification from hESCs. Under a chemically defined medium, FGF inhibition blocked phosphorylation of extracellular signal-regulated kinase 1/2 (ERK 1/2) with a significant reduction of PAX6-expressing neuroepithelia, indicating that FGF regulates neural induction through ERK1/2 activation. Activation of FGF-ERK1/2 pathway was necessary for the activity of poly(ADP-ribose) polymerase-1 (PARP-1), a conserved nuclear protein catalyzing polymerization of ADP-ribose units. Pharmacological inhibition and genetic ablation of PARP-1 inhibited neural induction from hESCs, suggesting that FGF-ERK1/2 signal pathway regulates neuroectoderm specification through regulating PARP-1 activity. Furthermore, FGF-ERK1/2-PARP-1 cascade regulated the expression of PAX6, a transcription determinant of human neuroectoderm. Together, we propose that FGF regulates hESC neural specification through the ERK1/2-PARP-1 signaling pathway. STEM CELLS2011;29:1975–1982.
Specification of the neuroectoderm from embryonic ectoderm, also known as neural induction, is the first and critical step toward the formation of the complex nervous system. Historical studies using Xenopus and chick embryos have led to the “default model” of neural induction [1–5]. In this model, bone morphogenetic proteins (BMPs), produced by ectoderm cells, prevent them from becoming neural tissues. During gastrulation, inhibitors of BMPs, such as Noggin, Chordin, and follistatin secreted from the embryonic organizer, remove the inhibition. The ectoderm cells hence become neural tissue by default. Nevertheless, BMP inhibition alone is not sufficient for neural induction in ectopic ectoderm regions. Growing evidences indicate that other signaling molecules, especially fibroblast growth factors (FGFs), also play a crucial role in early steps of neural differentiation [2, 6–10]. In chick embryos, the organizer produces FGFs that are required for neural induction . In Xenopus, inhibition of FGF signaling suppresses neural induction even in the presence of BMP antagonists . These findings raised a question of whether the role of FGFs in neural induction is dependent on BMP signaling. Analysis of signaling pathways initiated by BMPs and FGFs during Xenopus development revealed that the main effectors of the FGF pathway, mitogen-activated protein kinases (MAPKs), phosphorylate SMAD family member 1 (SMAD1), a key downstream transcription factor of BMP signaling, at the linker region. Such modification prevents nuclear translocation of SMAD1 and promotes degradation of SMAD1 through proteosomes [8, 12–15]. This discovery brings together FGF and BMP signalings to the default model of neural induction. Whether such a molecular model of neural induction applies to mammals is presently not clear.
Examinations on mammalian neural induction are performed largely using in vitro models of neuroepithelial differentiation from embryonic stem cells (ESCs). These studies confirm the requirement of BMP inhibition and FGF activation in neural differentiation of mouse ESCs. However, FGF signaling appears independent of its inhibitory function on BMP signaling [6, 16, 17]. For example, the negative effect of FGF inhibition on neural specification of mouse ESCs could not be rescued by BMP inhibitors , similar to the phenomenon in early chick embryo development in which FGF works independently of BMP activity in neural induction . Under similar culture conditions, neuroepithelial differentiation of human ESCs also seems to be positively and negatively regulated by FGFs and BMPs, respectively [19–21]. We found that the protein level and subcellular localization of phospho-SMAD1 (p-SMAD1) was not altered in neural differentiating human ESCs in the presence of FGF signaling inhibitors, suggesting that FGFs regulate human embryonic stem cell (hESC) neural differentiation independent of BMP-SMAD signaling . This raised the question: what would be the alternative pathway that mediates FGF effects on neuroectoderm specification in mammals?
FGFs are known to exert their multiple functions, including neural induction, through the MAPK/ERK pathway [5, 22–25]. While there are many downstream effectors of ERK, poly(ADP-ribose) polymerase-1 (PARP-1), which mediates ERK1/2-initiated transcriptional regulation [26, 27], may be one of the candidates. PARP-1 is a highly conserved DNA binding nuclear protein that catalyzes the covalent attachment of poly(ADP-ribose) (PAR) units to itself and other nuclear target proteins. This leads to modification of target proteins, such as protein-DNA or -protein interactions [28–31]. Although it has originally been identified as a DNA repair-associated protein that is activated by DNA breakage [32, 33], PARP-1 has recently been shown to regulate growth, proliferation, and differentiation of a variety of cells [29, 30] through transcriptional regulation [34–37]. Whether it also mediates the effect of FGFs in neural induction, however, remains unexplored.
In the present study, we found that ERK1/2 is a predominant downstream kinase activated by FGF during neural differentiation of hESCs. Importantly, we found that PARP-1 activity is regulated by FGF-ERK1/2 in neural differentiating hESCs. Suppression of PARP-1 activity using chemical inhibitors or genetic ablation blocks neural differentiation of hESCs, suggesting that PARP-1 is a downstream effector of FGF-ERK1/2 pathway in hESC neural differentiation. Furthermore, we found that PARP-1 bound to the promoter of PAX6, a transcription determinant of human neuroectoderm , suggesting that FGF-ERK1/2-PARP-1 cascade regulates neuroectoderm specification of hESCs through modulation of PAX6 transcription.
The hESCs, H9 (p16-35), and H1 (p20-35) were cultured on irradiated mouse embryonic fibroblasts (MEFs) and neural differentiation of hESCs was performed as previously described [39, 40]. Briefly, neural differentiation was initiated by detaching hESCs from MEF with 1 mg/ml dispase (Invitrogen, Carlsbad, CA), and grown in suspension in the ESC growth medium consisting of Dulbecco's modified Eagle's medium (DMEM/F12, 20% knockout replacement serum, 1× nonessential amino acids, 2 mM glutamine, and 100 μM β-mercaptoethanol, all from Invitrogen) to form aggregates for 4 days. The aggregates (day 4) were then transferred to a serum-free minimal (SFM) medium consisting of DMEM/F12, 1× nonessential amino acids, and 2 μg/ml heparin (all from Invitrogen), and grown in suspension. After 3 days in suspension (day 7), they were attached to laminin-coated (Invitrogen) coverslips for immunostaining. At day 10, cells were harvested and used for further analysis. For inhibitor treatment, SU5402, U0126, and PJ34 were added to cultures between day 4 and day 6 of differentiation.
Plasmids and Viral Transduction
PARP-1-specific shRNA and control vectors were obtained from Open Biosystems (RHS4430-99147879 and RHS4346, respectively). We followed the protocol for lentiviral transduction as previously described . Briefly, 20 μg of transfer vector, 15 μg of lentiviral vector pCMVΔ8.91, and 6 μg of vesicular stomatitis virus G protein (VSV-G) were cotransfected to HEK293FT cells (Invitrogen) with calcium phosphate precipitation method. Six hours later, the medium was changed to hESC media, and 2 days later, the virus containing medium was collected and cleared by centrifugation (3000 rpm for 5 min) and filtration (0.45 μM), and immediately used. For transduction of hESCs, the cells were treated with ROCK inhibitor (Calbiochem, NJ) for overnight before viral infection. Trypsinized cells were incubated with viral-containing media for 1 hour at 37°C, and then plated on MEF cells and cultured overnight in the presence of ROCK inhibitor before changing media in the next day. Forty-eight hours after infection, puromycin (5 μg/ml) was treated for selection of drug-resistant clones. Clones-expressing viral marker green fluorescent protein (GFP) were selected and further analyzed.
The experiments were described in previous study . Affymetrix Human Genome U133 plus 2.0 microarrays were used to analyze gene expression levels of key components of the FGF signaling pathway at days 0, 6, and 10 of hESC neural differentiation. hESC expression profiles were used as a reference. cRNA probe synthesis and array hybridizations were carried out at the NIH Neuroscience Microarray Consortium (http://arrayconsortium.tgen.org/np2/home.do).
Preliminary analysis was performed using Affymetrix Microarray Suite 5.0 (MAS 5.0) and Data Mining Tool softwares. The data were deposited at the NIH Neuroscience Microarray Consortium (http://arrayconsortium.tgen.org/np2/home.do) and in the ArrayExpress database (accession number E-MEXP-2426).
Neuroepithelial cells from hESCs were stained as described in previous studies [39, 40].
Harvested cells were filtered through a 70-μM cell strainer, and fixed with 0.1% paraformaldehyde for 15 minutes on ice and permeabilized in 90% methanol for 20 minutes. Cells resuspended in fluorescence-activated cell sorting (FACS) buffer (phosphate-buffered saline with 2% donkey serum) were incubated overnight with primary antibodies or normal IgG (as control) at 4°C, followed by washing and staining with secondary antibodies for 1 hour. Cells were analyzed using a Becton Dickinson FACSCaliber and CellQuest Pro (BD Biosciences, San Diego, http://www.bdbiosciences.com).
Cells were lysed with radio immunoprecipitation assay buffer as described in previous studies . Lysates were cleared by centrifugation for 20 minutes and the concentration was determined by BCA protein assay kit (Bio-Rad, CA). The lysates were resolved by SDS–polyacrylamide gel electrophoresis and Western blotting was carried out using horse-radish peroxidase-conjugated IgG as a secondary antibody and enhanced chemiluminescence system for detection.
Real-Time Polymerase Chain Reaction
Total RNA was isolated using the Trizol kit (Invitrogen) by manufacturer's manual. One microgram per milliliter of total RNA was used for reverse transcription, followed by real-time polymerase chain reaction (PCR) using Power SYBR kit (Applied Biosystem, UK). Primers used for real-time PCR were listed in Supporting Information Table S1.
Chromatin Immunoprecipitation (ChIP) Analysis
After cross-linking with 1% formaldehyde at 37°C for 10 minutes, cells were harvested and washed, followed by lysis with EZ-Chip kit according to the manufacturer's manual (Millipore, CA). The chromatin was sheared by sonication and immunoprecipitated with anti-PARP-1 antibody (Abcam). The immune complex was then washed five times and reversed. Chromatin DNA was subjected to quantitative PCR (qPCR) using primers for control and PAX6/SOX2 promoters (Supporting Information Table S1).
ERK1/2 Is a Downstream Effector of FGF Signaling in Neuroectoderm Specification from hESCs
We have previously shown that under a chemically defined system, hESCs differentiate to a nearly uniform population of neuroepithelial cells at around days 8-10 by expression of neuroectoderm transcription factors PAX6 and SOX2 . This efficient neural differentiation correlates with expression of FGF signaling components by the differentiating cells (Supporting Information Fig. S1) and is not effectively blocked by BMP . To delineate how FGF regulates neuroectoderm specification of hESCs, we first investigated which downstream pathways are involved in the FGF-mediated neural differentiation. At day 6 of neural differentiation, right before the appearance of PAX6-expressing neuroepithelial cells, 30 minutes of treatment of cultures with SU5402, an inhibitor of FGF receptors, at the concentration that effectively blocks neural differentiation without causing apparent cell death [18, 21], blocked the phosphorylation of ERK1/2 (Fig. 1A). However, the phosphorylation of other related serine/threonine kinases, p38 and c-Jun N-terminal kinase, were not affected. Phospho-AKT level in inhibitor-treated cells was also weakly reduced. These results indicate that ERK1/2 is a predominant downstream kinase activated by FGF signaling in neural differentiating hESCs and suggests that ERK1/2 activation is responsible for the FGF-mediated regulation of neural differentiation.
To confirm whether FGF-ERK1/2 signal cascade underlies neural differentiation of hESCs, differentiating cells were treated with an ERK1/2-specific inhibitor, U0126 between day 4 and day 6 of differentiation and neural differentiation was measured at day 10. FACS analysis, verified by immunostaining on coverslip cultures, revealed that only 30% of the cells were PAX6+ neuroepithelial cells in the presence of U0126 when compared with more than 80% in the control group without U0126 (Fig. 1B). Real-time PCR showed that PAX6 and SOX2 expression was decreased in U0126-treated cells (Fig. 1C). Immunoblotting assay showed that the expression of PAX6, and another neural marker, SOX2 was significantly decreased in the cells treated with U0126, compared with control cells (Fig. 1D). Collectively, these data suggest that ERK1/2 works as a downstream effector of FGF signaling to regulate neural induction from hESCs.
FGF-ERK Regulates PARP-1 Activity During hESC Neural Differentiation
How FGF-ERK regulates neural induction is not clear. Using the specific antibody targeting the MAPK phosphorylation sites of SMAD, developed by Dr. DeRobertis, we found that activation or inhibition of FGF-ERK did not alter SMAD activity . This suggests that alternative pathways may mediate the effect of FGF-ERK in neural differentiation. PARP-1 protein, a conserved nuclear enzyme catalyzing polymerization of ADP-ribose units to target proteins, has been shown to be regulated by ERK2 [26, 27]. In nerve growth factor-induced differentiation of rat pheochromocytoma (PC12) cells, which involves ERK activation, PARP-1 activity is increased [42, 43]. We therefore hypothesized that PARP-1 may mediate the FGF effect on hESC neural differentiation. Immunoblotting analysis indicated that the PARP-1 activity, revealed by an antibody against PAR, was increased during neural differentiation, whereas the level of PARP-1 protein was not altered (Fig. 2A). Immunostaining of differentiating cells also indicated that PAR formation was increased during neural differentiation (Fig. 2B). This phenomenon was observed with another hES cell line, H1. In addition, the temporal change of PARP-1 activity correlated with that of p-ERK1/2 during neural differentiation of hESCs (Supporting Information Fig. S2). This temporal course of PAR activity along hESC neural differentiation corresponds well with expression of FGF and ERK1/2, suggesting that PARP-1 may be a downstream of FGF-ERK pathway.
To determine whether PARP-1 activity is regulated by FGF-ERK1/2 pathway in the neural differentiating hESCs, we measured the level of enzymatic product of PARP-1, PAR, in the presence of inhibitors of FGF and ERK1/2. Differentiating cells at day 6 were treated with SU5402, U0126, or a PARP-1-specific inhibitor, PJ34 and the cell lysates were subjected to immunoblotting assay using an anti-PAR antibody. Treatment with PJ34 removed most of the auto-PARylating activity from PARP-1. Interestingly, SU5402 and U0126 treatment significantly reduced the level of PAR formation (Fig. 2C). However, the amount of PARP-1 protein was not altered by the above treatments. Taken together, these results suggest that PARP-1 activity is regulated by FGF-ERK1/2 signal pathway during hESC neural differentiation.
PARP-1 Activity Is Required for hESC Neural Differentiation
As FGF-ERK1/2 signaling is critical for hESC neural differentiation and PARP-1 is regulated by ERK, we then asked whether PARP-1 is required for hESC neural induction. Neural differentiating cells were treated with PJ34 between day 4 and day 6 of differentiation and neural differentiation was analyzed at day 10. FACS analysis, as well as immunostaining of coverslip cultures, showed that PJ34 treatment significantly reduced the PAX6-expressing neuroepithelial population (Fig. 3A, 3B). Immunoblotting assay using lysates from the same cell preparations indicated that the expression of PAX6 and SOX2 was decreased in the cells treated with PJ34 in a dose-dependent manner (Fig. 3C). Similarly, real-time PCR indicated that PARP-1 inhibition resulted in the decrease of PAX6 and SOX2 expression (Fig. 3D). These findings strongly suggest that PARP-1 acts as a downstream effector of FGF-ERK1/2 signal pathway to regulate neural differentiation of hESCs.
To further confirm the functional role of PARP-1 in neuroectoderm specification of hESCs at the genetic level, we generated PARP-1 knockdown hESC lines by lenti-virus-mediated constitutive expression of PARP-1-specific shRNA. By using GFP as a transgenic marker, we selected one control hESC line and three PARP-1 shRNA-expressing cell lines, two of which highly expressed GFP (#3 and #6), and one of which showed a lower level of GFP expression (#2) (Supporting Information Fig. S3). Consistent with the GFP expression level, PARP-1 protein is largely knocked down in #3 and #6 shRNA lines, while moderately downregulated in #2 line. Expression of control or PARP-1 shRNA did not affect the growth of hESCs, consistent with expression of PARP-1 but without high level of PAR activity at the ESC stage (Fig. 2). The expression of SOX2, OTX2, PAX6 and pluripotent marker, OCT3/4 in the cells expressing PARP-1 shRNA was not changed in ESC stage, compared with control cells (Fig. 4A, left panel). After the transgenic hESCs were differentiated to neuroepithelia for 10 days, the expression of PARP-1 remained highly inhibited in high-GFP expression cell line (#3 and #6), whereas moderately inhibited in the low-GFP expression cell line (#2), when compared with the control cell line (Fig. 4A, right panel). Corresponding to the genetic ablation of PARP-1, the expression of neuroepithelial markers, PAX6, SOX2, and OTX2, was greatly reduced in high-GFP expression cell lines, in comparison to the low-GFP expression cell line and the control cell line (Fig. 4A). OCT3/4 is not expressed in differentiating cells, suggesting that the cells are not proliferating ESCs. The expression of mRNAs for the above neural transcription factors, revealed by real-time PCR, showed a similar pattern (Fig. 4B). Consistent with the above observations, immunostaining, confirmed by FACS analysis (Supporting Information Fig. S4), showed that the expression of PAX6 and SOX2 at day 10 was suppressed in PARP-1 knockdown cells (Fig. 4C and Supporting Information Fig. S5). Together, data from pharmacological intervention and genetic ablation studies strongly support that PARP-1 is required for efficient neuroectoderm specification of hESCs.
PARP-1 Regulates PAX6 Transcription During hESC Neural Differentiation
PAX6 has recently been shown to be a transcriptional determinant of the human neuroectoderm . As PARP-1 regulates hESC neural differentiation, we investigated whether FGF-ERK1/2-PARP-1 pathway directly regulates PAX6 transcription. We first analyzed the protein level of PAX6 in the presence of inhibitors of FGF and ERK1/2. PAX6 protein stability is not changed by treatment with either FGF or ERK1/2 inhibitors. We next tested whether this pathway regulates transcription of PAX6. A direct connection between a signal pathway and the expression of its target gene can be demonstrated by assessing corresponding transcriptional responses in a short term. Neural differentiating cells at day 5 were treated with inhibitors for 2 hours, and transcription of PAX6 was measured by real-time PCR. As shown in Figure 5A, PJ34 alone significantly suppressed the transcription of PAX6, as did SU5402 and U0126, suggesting that FGF-ERK1/2-PARP-1 cascade regulates the transcription of PAX6. PARP-1 regulates its target gene generally by its enzymatic activity [36, 44–48]. It may also bind to DNA as a member of a protein complex, such as in the case that PARP-1 binds to the enhancer region where it regulates the expression of quail PAX6 gene . We therefore examined whether regulation of PAX6 expression is mediated by direct PAX6 promoter occupancy of PARP-1 by ChIP-qPCR with anti-PARP-1 antibody and lysates from self-renewing (day 0) and neural differentiating (day 10) hESCs. Interestingly, PARP-1 was localized specifically at the promoter region of PAX6, but not at SOX2 promoter (Fig. 5B). As the activity of PARP-1 is regulated by FGF-ERK along neural differentiation, the localization of PARP-1 in the PAX6 promoter strongly suggests that FGF-ERK1/2-PARP-1 signal pathway regulates neural induction of hESCs at least in part, by controlling the transcription of neural transcription factor, PAX6.
Embryogenesis, including neuroectoderm specification, is orchestrated by multiple signaling pathways . FGFs and BMPs, the two major and opposing families of extracellular molecules, regulate Xenopus neuroectoderm induction through integration at the SMAD level [8, 12–15, 38]. Recent analyses of neuroectoderm specification using ESC models, especially human ESCs, suggest that FGFs affect neuroectoderm specification independently of BMP-SMAD pathway . Our present study indicates that ERK1/2 is a predominant downstream molecule activated by FGF signaling and responsible for the regulation of neuroectoderm specification. Importantly, we discovered that PARP-1 is directly regulated by the FGF-ERK1/2 pathway during hESC neural differentiation. Furthermore, we show that PARP-1 is localized to the promoter region of PAX6, the human neuroectoderm transcriptional determinant, whereas the PAR activity is regulated by FGF-ERK along neural differentiation. We propose that FGFs regulate neuroectoderm specification of hESCs through the ERK1/2-PARP-1 pathway to control the transcription of PAX6 and hence neuroectoderm specification (Fig. 6).
Under multiple biological systems, FGFs exert their function through activation of MAPK/ERK1/2 [24, 51–55]. Our present study confirms that ERK1/2 is the predominant effector of FGF signaling during hESC neural differentiation. How FGF-ERK regulates neuroectoderm transcription factors, for that matter also for BMP-SMAD, remains an unresolved issue in the field of neural induction. PARP-1 emerges as a potential mediator as it is regulated by ERK1/2 during cell proliferation and differentiation besides its role in DNA repair [26, 27, 29]. PARP-1 activation is required for nerve growth factor-mediated neuronal differentiation of PC12 cells, a process that is mediated by ERK1/2 [42, 43]. Our present study demonstrated that PAR activity of PARP-1 during hESC neural differentiation is regulated by FGF-ERK1/2 activation. Inhibition of FGF receptors by SU5402 or blockage of ERK1/2 activation by U0126 results in corresponding downregulation of PARP-1. The net outcome of FGF-ERK inhibition is similar to that of PARP-1 inhibition by PJ34, suggesting that PARP-1 activity is required for hESC neural differentiation. This conclusion is validated by our genetic knockdown of PARP-1 in hESCs. In this case, knockdown of PARP-1 blocked hESCs from differentiating to neuroepithelia without favoring alternative fates as meso- and endo-derm markers are not increased (Supporting Information Fig. S6). This inhibition of hESC differentiation occurs even in the presence of (endogenous) FGFs. One may question why the growth of hESCs, which are also affected by FGF2 , is not affected. One possibility is that other cellular PARylating activity from members of the PARP family may compensate and rescue the effect of PARP-1 knockdown. Indeed, mice with null mutation of PARP-1 or PARP-2, another member of the PARP family, are viable, but double knockout embryos of PARP-1 and PARP-2 die early in development at the onset of gastrulation . Similarly, PARP-1(−/−) mESCs are viable and expandable [57, 58]. Alternatively, these effects are mediated by different FGFs. As shown in gene expression profiling data (Supporting Information Fig. S1), FGF8 and FGF9 are highly expressed in neural differentiating hESCs, suggesting that these are the potential molecules that activate endogenous FGF signaling pathway in neural differentiating hESCs, at least in our system. Another possibility is that the effect of FGF2 on hESC proliferation is not mediated by PARP-1. This is suggested by the fact that even though PARP-1 is expressed at the ESC stage, PAR activity is not detected at that stage (Fig. 2). Therefore, PARP-1 appears to mediate the FGF effect on neuroectoderm specification rather than on hESC proliferation.
PARP-1 regulates gene transcription in diverse manners. It binds to promoter regions of target genes in collaboration with its coactivators . While its enzymatic activity is dispensable in some cases [35, 60], it is often required, in which other components of the coregulatory complex become the target of PARP-1 enzyme activity [36, 45–48]. During our hESC neural differentiation, PARP-1 expression remains constant, whereas the PAR activity increases along differentiation, corresponding to changes in FGF expression. Inhibition of FGF signaling does not alter the expression of PARP-1 but it does indeed block the PAR activity. Therefore, FGF-ERK regulates hESC neural differentiation by modulating the PAR activity. This is further suggested by the fact that PARP-1 is localized to the promoter of PAX6 along hESC differentiation but PAX6 is not expressed at the ESC stage, possibly because of lack of PAR activity. Along hESC differentiation, FGF increases, so are the PAR activity and the transcription of PAX6. This may be similar to the scenario of the PC12 proliferation/differentiation model that PARP-1-mediated PARylation is involved in NGF-induced differentiation . As PARP-1 appears to directly regulate the transcription of PAX6 (Fig. 5), we propose that FGFs regulate neuroectoderm specification from hESCs by regulating PAR activity. This proposition also explains why the growth of hESCs depends on FGF2 but does not require PAR activity.
We discovered that PARP-1 is localized to the promoter region of PAX6, the human neuroectoderm transcriptional determinant, whereas the PAR activity is turned on by FGF-ERK during neuroectoderm induction. Pharmacological inhibition and genetic ablation of PARP-1 inhibited neural induction from hESCs. We propose that FGFs regulate neuroectoderm specification of hESCs through the ERK1/2–PARP-1 pathway to control the transcription of PAX6 and hence neuroectoderm specification.
This study was supported by the National Institute of Neurological Disorders and Stroke (R01 NS045926) and partly by a core grant to the Waisman Center from the National Institute of Child Health and Human Development (P30 HD03352).