Although the detection of several components of the fibroblast growth factor (FGF) signaling pathway in human embryonic stem cells (hESCs) has been reported, the functionality of that pathway and effects on cell fate decisions are yet to be established. In this study we characterized expression of FGF-2, the prototypic member of the FGF family, and its receptors (FGFRs) in undifferentiated and differentiating hESCs; subsequently, we analyzed the effects of FGF-2 on hESCs, acting as both exogenous and endogenous factors. We have determined that undifferentiated hESCs are abundant in several molecular-mass isoforms of FGF-2 and that expression pattern of these isoforms remains unchanged under conditions that induce hESC differentiation. Significantly, FGF-2 is released by hESCs into the medium, suggesting an autocrine activity. Expression of FGFRs in undifferentiated hESCs follows a specific pattern, with FGFR1 being the most abundant species and other receptors showing lower expression in the following order: FGFR1 → FGFR3 → FGFR4 → FGFR2. Initiation of differentiation is accompanied by profound changes in FGFR expression, particularly the upregulation of FGFR1. When hESCs are exposed to exogenous FGF-2, extracellular signal-regulated kinases are phosphorylated and thereby activated. However, the presence or absence of exogenous FGF-2 does not significantly affect the proliferation of hESCs. Instead, increased concentration of exogenous FGF-2 leads to reduced outgrowth of hESC colonies with time in culture. Finally, the inhibitor of FGFRs, SU5402, was used to ascertain whether FGF-2 that is released by hESCs exerts its activities via autocrine pathways. Strikingly, the resultant inhibition of FGFR suppresses activation of downstream protein kinases and causes rapid cell differentiation, suggesting an involvement of autocrine FGF signals in the maintenance of proliferating hESCs in the undifferentiated state. In conclusion from our data, we propose that this endogenous FGF signaling pathway can be implicated in self-renewal or differentiation of hESCs.
Human embryonic stem cells (hESCs) are pluripotent stem cells derived from the human blastocyst-stage embryo. In common with their mouse counterparts (mESCs), hESCs have the capacity to self-renew and, under appropriate conditions, to differentiate into a diverse range of specialized cell types. These two major properties of hESCs, combined with their untransformed character, render them an ideal resource for the study of human development and for transplantation therapy in numerous pathologies. Regarding hESC-based therapies, the current most important challenge in this field is the development of culture systems that are chemically defined and free of animal products and which moreover allow sustained proliferation of undifferentiated cells and/or directed differentiation of hESCs into specific cell types, efficiently and to homogeneity. Here, as with mESCs, it is considered that growth factors represent the key components of such defined culture systems; however, despite great efforts, those growth factors and signaling pathways maintaining pluripotency and regulating self-renewal and/or differentiation of hESCs are largely unknown. Although activation of the JAK/STAT3 pathway by leukemia inhibitory factor (LIF) leads to maintained self-renewal of mESCs, with input also from bone morphogenetic proteins [1–3], this particular pathway is dispensable to the self-renewal of hESCs . It has been suggested that Wnt signaling positively regulates expression of transcription factors Oct4, Rex-1, and Nanog, which represent key molecules implicated in the state of hESC stemness and therefore may contribute to undifferentiated growth of hESCs . Nonreceptor tyrosine kinase, cYes, also has been identified as contributing to the maintenance of undifferentiated mESCs and to hESC self-renewal . Although cYes can be activated by LIF or other components present in serum, its inhibition does not interfere with LIF-induced JAK/STAT3 or extracellular signal-regulated kinases (ERK1/2) phosphorylation, suggesting the existence of a new, LIF-independent pathway . It is generally accepted that hESCs in culture depend on the presence of feeder cells to sustain self-renewal and the capacity to differentiate. By convention, hESCs are cultured on feeder layers of mitotically inactivated mouse embryonic fibroblasts (MEFs), using medium supplemented with serum replacement and fibroblast growth factor (FGF)-2. There are several reports in which undifferentiated hESCs are, in addition to FGF-2, grown with LIF and/or transforming growth factor β1 [7–9]. However, it is not yet clearly established that such modified culture conditions offer significant improvement for culture of undifferentiated hESCs. Undoubtedly, hESCs maintained under standard conditions may use numerous and as-yet undefined factors produced by the feeder cells. Moreover, hESCs may use several intracrine or autocrine signaling pathways to retain their capacity to proliferate indefinitely in an undifferentiated state or to modify their developmental program. Such autonomous signaling mechanisms might not have relevance in vivo because hESCs represent a greatly expanded cell population that is similar, although not identical, to inner cell mass–derived primitive ectoderm in the developing embryo. The signaling pathways in hESCs may reflect their adaptation to in vitro conditions and therefore may be somewhat simplified compared with signaling processes occurring during normal development. Nevertheless, an understanding of intracrine and autocrine hESC-specific signaling pathways is crucial not only for development of new culture conditions for propagating cells in the undifferentiated state but also for developing efficient differentiation protocols. In this context, using cDNA microarray analysis, several groups have detected high/elevated levels of components of the FGF/fibro-blast growth factor receptor (FGFR) signaling pathway in undifferentiated hESCs compared with their differentiated progeny, human tissue, and mESCs. These components included FGFR-1, -2, -3, and -4, as well as FGF-2, -11, and -13 [10–12]. However, there is as yet no report addressing the action of endogenous FGFs in undifferentiated hESCs.
In contrast to the current limited understanding of the actions of other FGFs and FGFRs in undifferentiated hESCs, FGF-2 has been shown to induce development of ectodermal and mesodermal cells from predifferentiated hESCs  and to support hESC differentiation into neural lineages [14–18]. However, the effects of exogenous FGF signaling during in vitro differentiation protocols may be peripheral, as studies in mESCs have indicated that for neural specification, autocrine FGF signaling is preferentially involved . Taken together, these data are somewhat contradictory, especially as individual components of the FGF signaling are elevated in undifferentiated hESCs and downregulated during differentiation, in which, on the other hand, the importance of FGF regulatory pathways has been repeatedly proven.
Therefore, as a first step to understanding the role of the FGF/FGFR signaling pathways in hESCs, we determined the expression pattern of molecular isoforms of endogenous FGF-2 in undifferentiated and differentiating hESCs. Then, using a quantitative method, we analyzed and compared the expression of all four FGFRs under the same conditions. The response of hESCs to exogenous FGF-2 was analyzed at both biochemical and cellular levels. Finally, we used an inhibitor of FGFR tyrosine kinases to investigate the autocrine FGF signaling pathway. Our results clearly demonstrate the existence of an operational, autocrine FGF signaling pathway in undifferentiated hESCs, inhibition of which leads to rapid differentiation. We therefore suggest that autocrine FGF signaling through FGFRs, and specifically FGFR1, is unconditionally required for proliferation of hESCs in the undifferentiated state.
Materials and Methods
Experiments were performed using hESC lines derived from the Department of Molecular Embryology, Institute of Experimental Medicine, Academy of Sciences of the Czech Republic. Complete information on derivation, characterization, and maintenance of hESCs is available at http://www.isscr.org/science/sclines.htm.
For this study, undifferentiated hESCs were grown on feeder cells in medium supplemented with human recombinant FGF-2 (4 ng/ml) and were passaged using either mechanical scraping with a glass pipette (to provide low-density cultures of undifferentiated cells) or enzymatic dissociation with collagenase (to provide high-density cultures of undifferentiated cells).
To induce less advanced states of differentiation in hESCs, cells were cultured without FGF-2 supplementation and in the absence of feeder layers for 8 days (designated D8 of spontaneous differentiation) or were cultured as aggregates in suspension (without FGF-2 supplementation in the medium) for 8 days to form embryoid bodies (D8 embryoid bodies). To induce more advanced states of differentiation in hESCs, cells were cultured as aggregates in suspension (without FGF-2 supplementation in the medium) for 5 days, followed by 10 days of adherent culture with 10 ng/ml FGF-2 (D5 + 10 of two-step differentiation). The conditions for maintenance and differentiation of hESCs used in this study are summarized in Figure 1.
Expression analyses for FGF-2 and FGFRs were performed in hESC lines CCTL (Center for Cell Therapy Line) 9, 10, 12, and 14 at passage numbers 20 through 30. All experiments to assess the effects of exogenous FGF-2 and to investigate autocrine FGF signaling were performed using hESC lines CCTL12 and 14 at passage numbers 30 through 50. Note that all hESC lines used in this study have normal 46XX (CCTL9,14) and 46XY (CCTL10) karyotypes. However, for CCTL12, trisomy of chromosome 12 was later detected at passage 42 (47XX +12); therefore, the last experiment on the chemical inhibition of FGF receptor kinases was performed in hESC line CCTL14 and confirmed in CCTL9.
Western Blot Analysis
For Western blot analysis of endogenous FGF-2 and FGF-2–interacting factor (FIF), hESCs were harvested by gently scraping colonies with a glass pipette washed with phosphate-buffered saline (PBS) pH 7.4 and proteins were extracted in 50 mM Tris-HCl (pH 7.4), 150 mM sodium chloride, 1% NP-40, 1 mM EDTA, 50 mM sodium fluoride, 8 mM β-glycerophosphate, 0.1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, and 10 μg/ml tosylphenylalanine chloromethane for 30 minutes on ice. Protein concentrations in extracts were determined using the DC protein assay (BioRad, Hercules, CA, http://www.bio-rad.com). After equalizing protein concentrations, the samples were mixed with × 2 Laemmli sample buffer and boiled for 5 minutes. For Western blot analysis of Oct-4, lamin B, cytokeratin 8 (the TROMA-1 antigen), and nestin, hESCs were harvested and washed as above and then directly lysed in Laemmli sample buffer and boiled for 5 minutes. All samples were subjected to SDS-PAGE, electrotransferred onto Hybond P membrane (Amersham Pharmacia Biotech, Buckinghamshire, U.K., http://www1.amershambiosciences.com), and probed with the following primary antibodies: mouse monoclonal antibody to FGF-2 (clone FB-8, F6162) was purchased from Sigma (St. Louis, http://www.sigmaaldrich.com), rabbit polyclonal antibody to Oct-4 (sc-9081) and goat polyclonal antibody to lamin B (sc-6217) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, http://www.scbt.com), rabbit polyclonal antibody to FIF was kindly provided by Dr. H. Prats (Institut Louis Bugnard, Toulouse, France, http://ifr31.toulouse.inserm.fr/2005.shtml), human-specific mouse monoclonal antibody to nestin (MAB5326) was purchased from Chemicon (Temecula, CA, http://www.chemicon.com), and mouse monoclonal antibody to TROMA-1 was developed from hybridoma cells obtained from the Developmental Studies Hybridoma Bank (University of Iowa, Iowa City, IA, http://www.uiowa.edu/∼dshbwww). After incubation with appropriate peroxidase-conjugated secondary antibodies, protein bands were visualized using the chemiluminescence detection reagent ECL+Plus (Amersham).
Determination of FGF-2 in Conditioned Medium
To isolate FGF-2 from conditioned medium, aliquots of 50 ml of culture medium conditioned for 24 hours by hESCs growing on feeder cells and by feeder cells alone were gently rocked with heparin-coated agarose beads (100 μl, 4% beads; Sigma) overnight at 4°C. The beads were then extensively washed with PBS pH 7.4, boiled with Laemmli sample buffer for 5 minutes, and assayed for FGF-2 by Western blotting. To measure the concentration of FGF-2 in conditioned medium collected from low- and high-density cultures, we used commercially available sandwich enzyme-linked immunosorbent assay (ELISA) kit (Quantikine; R&D Systems, Wiesbaden, Germany, http://www.rndsystems.com).
Cells were cultured in Permanox chambers (Nunc, Naperville, IL, http://www.nuncbrand.com), fixed with ice-cold ethanol/acetic acid (vol/vol, 95% EtOH, 1% acetic acid), rehydrated, blocked with 5% normal goat serum in PBS pH 7.4, and incubated with primary antibodies diluted in blocking solution. Unbound antibody was removed by extensive washing, and cells were incubated with secondary antibody conjugated with fluorescein isothiocyanate. Cell nuclei were counterstained with propidium iodide, and cells were mounted in Mowiol containing 1,4-diazobicyclo-[2.2.2]-octane to prevent fading. Microscopical analysis was performed using an Olympus BX60 microscope equipped with Fluoview confocal laser scanning unit (Olympus C&S Ltd., Prague, Czech Republic, http://www.olympus.com).
Total RNA isolated from cell extract using RNeasy Protect Mini Kit (Qiagen, Valencia, CA, http://www1.qiagen.com) was used to synthesize random hexamer-primed cDNA by MuLV reverse transcription (Applied Biosystems, Beaconsfield, U.K.. http://www.appliedbiosystems.com). The human-specific FGFR primers and probes were designed by Primer Express 2.0 (Applied Bio-systems). Table 1 lists the primers and probes for real-time reverse transcription–polymerase chain reaction (RT-PCR). To identify suitable control gene, we initially assayed Abelson (ABL) and β-glucuronidase (GUS) genes. Both housekeeping genes that are absent of pseudogenes had a stable expression in undifferentiated and differentiating hESCs. However, compared with GUS, ABL showed higher expression level (900 vs. 350 transcripts per 1 ng RNA) and therefore was selected for the quantification of FGFRs. The primers and probe for the ABL gene were designed using a standardized protocol from a Europe Against Cancer Program . Real-time RT-PCR reactions contained forward and reverse primers each at 0.2-μM concentration and 0.1 μM TaqMan probe in a total volume of 25 μl, and amplifications were performed in an ABI PRISM 5700 SDS (Applied Biosystems) for 2 minutes at 50°C, 10 minutes at 95°C, followed by 50 cycles of 15 seconds at 95°C and 1 minute at 60°C. To construct standard curves for the quantification of FGFR transcripts, we cloned RT-PCR products obtained by amplification of FGFR cDNA from K562 cells into the pCR 2.1 vector using the TA Cloning Kit (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). The ABL control plasmid was purchased from Ipsogen (Marseille, France, http://www.ipsogen.com). The plasmids were serially diluted (106 − 10 copies for FGFR genes and 105 − 103 for the ABL gene) and run in duplicates in each 96-well reaction plate. Resulting standard curves generated a mean slope −3.4 and intercept 40.86 ± 1 Ct. Real-time RT-PCR values are shown as a ratio between calculated numbers of transcripts for both FGFR and ABL, expressed as a percentage.
Table Table 1.. Primers and probes used for real-time reverse transcription–polymerase chain reaction
The primer pairs were located in two separate exons and were confirmed not to produce amplification artifacts and allowed for high efficiency of amplification with a slope close to −3.32 and sensitivity (at least 100 plasmid copies). TaqMan probes were labeled by a 5′ FAM reporter and 3′ BHQ1 quencher.
Total Tyrosine Phosphorylation and ERK1/2 Activation Assays
To assess tyrosine phosphorylation and activation of ERK1/2, hESCs cultured without recombinant FGF-2 for 5 days were treated with 5 ng/ml FGF-2 for 7 minutes, washed with PBS, pH 7.4, and harvested into Laemmli sample buffer heated to 95°C. Samples were then boiled for an additional 3 minutes, subjected to SDS-PAGE, and electrotransferred onto Hybond P membranes for Western blotting; the phosphorylation state of tyrosine residues was determined by probing with peroxidase-conjugated antiphosphotyrosine antibody (RC20; Transduction Laboratories, Lexington, KY, http://www.bdbiosciences.com), and activation of ERK1/2 was assessed by immunostaining with anti–phospho-ERK1/2 antibody (sc-7383; Santa Cruz Biotechnology) followed by peroxidase-conjugated secondary antibody. To compare the levels of unphosphorylated proteins in FGF-2–treated and control cells, membranes were either stained with amido black (tyrosine phosphorylation) or stripped and reprobed with anti–ERK1/2 antibody (sc-93; Santa Cruz Biotechnology).
Cells were cultured on feeder layers in 96-well plates in medium supplemented with FGF-2 (at 5, 10, and 20 ng/ml) or without FGF-2 for 2, 3, or 5 days. At each time point, the effect of FGF-2 on cell proliferation was assayed by (a) measuring the metabolic activity of FGF-2–stimulated hESCs using a WST-1–based colorimetric assay (Cell Proliferation Reagent WST-1; Roche Diagnostics, Mannheim, Germany, http://www.roche.com/home.html) and (b) determining incorporation of 5-bromo-2′-deoxyuridine (BrdU) into DNA using an ELISA-based BrdU Labeling and Detection Kit III, according to the manufacturer's instructions (Roche Diagnostics).
Cells were cultured on feeder layers in 24-well dishes in medium supplemented either with FGF-2(at 5,10, and 20 ng/ml) or without FGF-2 for 4, 5, or 6 days. At each time point, hESCs were photographed at low magnification and the size of colonies in millimeters was determined using one axis per colony, which was selected at day 4. A minimum of 80 colonies was measured for each culture condition. The colonies and the positions of their axes, as selected at day 4 of culture, remained constant throughout the experiment. Average colony size (colony outgrowth) was calculated for each culture condition and time point, and the observed increase in size at days 5 and 6 was expressed as a percentage of the initial measurement at day 4.
Inhibition of FGFR Tyrosine Kinases
Cells grown on feeder layers in medium supplemented with FGF-2 (at 4 ng/ml) or with out FGF-2 were exposed to increasing concentrations of inhibitor SU5402 (Calbiochem-Novabiochem, Bad Soden, Germany, http://www.merckbiosciences.co.uk/home.asp). Cells cultured in standard hESC medium, and in medium supplemented with the diluent for SU5402, dimethylsulfoxide, served as controls (control 1 and control 2, respectively). To determine the activity of mitogen-activated protein kinase kinase (MEK1/2) and ERK1/2 in SU5402-treated hESCs, cells were exposed to inhibitor for 4 days and either treated (7 minutes) or not treated with exogenous FGF-2. Cells were then harvested and subjected to phosphorylation assay as described above. The activation of MEK1/2 was assessed by immunostaining with anti–phospho MEK1/2 antibody followed by reprobing with anti-MEK1/2 antibody (9121S and 9122, respectively; New England Biolabs, Beverly, MA, http://www.neb.com). To assess differentiation-associated changes in the expression of p27, SU5402-treated and control hESCs were probed with anti-p27 antibody (K25020; Transduction Laboratories).
Expression and Release of FGF-2 in hESCs
Human FGF-2 is synthesized in five isoforms that originate from alternative translation start codons within a single mRNA species. Different molecular isoforms of FGF-2 act through distinct pathways: four high-molecular-mass isoforms (22-, 22.5-, 24-, and 34-kDa) are likely to be involved in intracrine regulation of cell proliferation and apoptosis, and the low-molecular-mass isoform (18 kDa), which is released from cells, signals via transmembrane receptors and regulates both proliferation and differentiation in either an autocrine or paracrine manner, depending mainly on the cell type concerned.
An antibody that recognizes all five in vivo FGF-2 isoforms was used here to analyze the FGF-2 expression pattern in hESCs. As this antibody recognizes both human and mouse FGF-2, it was also used to investigate the expression and release of FGF-2 in CF-1 feeder cells. It was revealed that undifferentiated hESCs (maintained in recombinant FGF-2–free medium for 24 hours), unlike mESCs , express 18-, 22-, 22.5-, and 24-kDa isoforms of FGF-2. The 34-kDa FGF-2 molecule was not detected (Fig. 2A). Indirect immunofluorescence with the same antibody demonstrated strong nuclear and nucleolar staining, with less-intensive cytoplasmic staining, which is consistent with the preponderance in hESCs of the high-molecular-mass isoforms of FGF-2, which are nuclear-localized (Fig. 2B). We also demonstrated that hESCs contain nuclear protein FIF (Fig. 2A, lower panel), which interacts with high-molecular-mass FGF-2 and exhibits antiapoptotic properties .
We next analyzed the expression of endogenous FGF-2 in differentiating/differentiated hESCs (Fig. 2A). The pattern of expression of all four (18-, 22-, 22.5-, and 24-kDa) molecular isoforms of FGF-2 in D8 embryoid bodies, and in adherent cells resulting from the two-step differentiation protocol, is indistinguishable from that in undifferentiated cells. Cells produced by our standard, two-step protocol (D5 + 10) still contain detectable levels of Oct-4, a marker of undifferentiated hESCs; therefore, we prolonged the differentiation period to a total of 24 days, comprising 6 days in suspension and 18 days in adherent culture (D6 + 18), after which time immunoreactivity for Oct-4 is undetectable, and probed these cells for the presence of FGF-2. Even in such highly differentiated cells, there is no observable difference in the gross pattern of FGF-2 expression compared with undifferentiated hESCs.
The capacity of hESCs in culture to release the specific, 18-kDa isoform of FGF-2, mediating autocrine signaling, was examined. Culture medium that had been conditioned for 24 hours by hESCs, maintained on a feeder layer in recombinant FGF-2–free medium, was subjected to enrichment of heparin-binding growth factors. As revealed by Western blot analysis, exportable 18-kDa FGF-2 isoform was recovered from medium conditioned by hESCs on feeder cells but not from medium conditioned by feeder cells alone (Fig. 2C). Crucially, the concentration of 18-kDa FGF-2 in hESC-conditioned medium increased with increasing number of hESCs and reached 80 to 100 pg/ml in high-density cultures. Therefore, it is deduced that hESCs have the capacity, in terms of exported low-molecular-mass isoform of FGF-2, to activate FGFRs in an autocrine manner, and such activity may reach high levels, especially in high-density hESC cultures. It is notable also that feeder layers prepared from CF-1 strain MEFs express no potentially FGFR-stimulatory FGF-2 isoforms in this culture system.
Expression of FGFRs in hESCs
The biological effects of FGFs are mediated by their binding to four FGFR kinases, FGFR1 through FGFR4, each possessing different ligand-binding specificities and cell type–specific and tissue-specific expression. The ligand-binding specificities of FGFR1, FGFR2, and FGFR3 are primarily achieved by alternative splicing of the exons encoding the C-terminal region of extracellular, immunoglobulin-like domain III. Such alternative splicing event results in b and c isoforms of FGFR1, FGFR2, and FGFR3, whereas FGFR4 exists in one sole spliced form.
Quantitative real-time RT-PCR was used to analyze the expression of all four FGFRs in undifferentiated hESCs and in hESCs at various stages of differentiation. The primers were designed to detect human-specific FGF-2–binding, IIIc alternatives of FGFR1, FGFR2, and FGFR3 and the ligand-binding domain of FGFR4. Real-time RT-PCR data were normalized by comparison with expression of the ABL gene that was found to be eligible for quantification of FGFR gene transcripts in the different hESC samples and are summarized in Table 2. To avoid contamination with feeder cells, colonies of hESCs were carefully separated by mechanical scraping with a glass pipette. First, by analyzing four independent hESC lines, it was revealed that undifferentiated cells that are mechanically passaged and cultured at low densities express all four FGFRs in a very stable, relative pattern: in all the lines, FGFR1 was dominant with the other receptors showing lower expression; FGFR1 → FGFR3 → FGFR4 → FGFR2. Notably, this relative expression pattern changed to FGFR1 → FGFR4 → FGFR3 → FGFR2 in two selected hESC lines, CCTL12 and 14, after adaptation to enzymatic passaging, for the purpose of achieving higher cell-culture densities, and this pattern reflected upregulation of FGFR1 and downregulation of FGFR3 expression compared with low-density cultures. A similar trend in expression of FGFRs was observed also in cells that had initiated differentiation, whether by simple 8-day adherent culture without FGF-2 and feeder cells (D8 spontaneous differentiation) or by 8-day culture in aggregates (D8 embryoid bodies). Also for hESCs that had undergone more advanced differentiation by the two-step protocol (D5 + 10), the relative order of expression was FGFR1 → FGFR4 → FGFR3 → FGFR2. Here, compared with undifferentiated hESCs, expression of all four FGFRs was dramatically elevated, with FGFR1 being increased about sixfold, FGFR2 about twofold, and FGFR3 and FGFR4 about sevenfold.
Table Table 2.. Expression of FGFRs in undifferentiated and differentiated human embryonic stem cells
Real-time RT-PCR values are expressed as FGFR/Abelson ratios. To determine the specificity of real-time RT-PCR samples (either from mitotically inactivated or normal), CF-1 mouse embryonic fibroblasts were processed using the same primers and probes. No amplification of FGFR1, 2, 3, and 4 was detected. The results are representative of at least three independent repetitions of the experiment.
Abbreviations: CCTL, Center for Cell Therapy Line; FGFR, fibroblast growth factor receptor; RT-PCR, reverse transcription–polymerase chain reaction.
Undifferentiated (low density)
Undifferentiated (high density)
D8 spontaneous differentiation
D8 embryoid bodies
D5 + 10 two-step differentiation
In summary, our data clearly indicate that hESCs are well equipped to accept and transmit FGF-2 signals via all four FGFRs, with FGFR1 potentially representing a dominant target, and that the relative levels of expression of FGFRs are tightly coupled to conditions that direct hESCs to differentiate.
Biological Effects of Exogenous and Endogenous FGF-2 on hESCs
Recombinant FGF-2 added exogenously to culture media may stimulate hESCs only via their cognate receptors, FGFRs. In contrast, FGF-2 produced endogenously by hESCs may function in two ways, depending on the presence or absence of the nuclear localization sequence: high-molecular-mass isoforms may be targeted to the nucleus and operate independently of cell-surface receptors, in an intracrine manner, whereas the low-molecular-mass isoforms may be exported from the cells and act via FGFRs as an autocrine or paracrine factor.
Stimulation of cells by FGF-2 leads to overall tyrosine phosphorylation of various proteins and to activation of extracellular signal-regulated kinases, ERK1/2 in particular, and so we used these two phenomena as criteria for the capacity of hESCs to respond to FGF-2. As evident from Figure 3, a short exposure of hESCs to 5 ng/ml recombinant FGF-2 induces a rapid and significant increase in levels of tyrosine phosphorylation of proteins (Fig. 3A), including ERK1/2 (Fig. 3B). This concentration of recombinant FGF-2 is routinely used for hESC culture and, as shown here, can be further increased by endogenously produced FGF-2. It is also of note that undifferentiated hESCs maintained in FGF-2–free medium possess an unexpectedly high basal level of phosphorylation of ERK1/2, which contrasts with the almost undetectable ERK1/2 phosphorylation typically demonstrated by mESCs D3 cultured under feeder-free conditions with serum- and LIF-supplemented medium (Fig. 3C).
To determine whether the observed increased tyrosine phosphorylation and concomitant activation of growth-related kinases result in accelerated proliferation of hESCs, we used two complementary assays: measurement of the number of viable/metabolically active cells by the cleavage of tetrazolium salt WST-1 and analysis of the proportion of cells in S-phase by incorporation of BrdU into newly synthesized DNA. Although overall metabolic activity as determined by the cleavage of WST-1 was increased slightly in FGF-2–treated hESC line CCTL12 compared with untreated controls, such effect was not observed in CCTL14. We also determined that the number of cells in S-phase remained unchanged at any tested concentration of FGF-2 (Fig. 4A). From this data, metabolic activity of hESCs rather than progression of their cell cycle is influenced by exogenously added FGF-2. The absence or presence of recombinant FGF-2 also had no effect on the expression of Oct-4 (Fig. 4B). We therefore conclude that exogenously added FGF-2 at concentrations up to 20 ng/ml is dispensable for growth and self-renewal of hESCs maintained under standard conditions with feeder cells.
Although no effect of FGF-2 on the proliferation of hESCs was found, there are still observable differences in hESC cultures depending on the concentration of exogenous FGF-2, which was manifest by the ability of cells to spread on the culture substratum. To quantify this effect, we used a simple strategy based on determining the change with time in size of hESC colonies in cultures that were exposed to various concentration of exogenous FGF-2 for 4, 5, and 6 days. Interestingly, the outgrowth of hESC colonies grown in medium containing 10 and 20 ng/ml FGF-2 increased at day 5 of culture and then was much less than the outgrowth in medium without or with only 5 ng/ml FGF-2 at day 6 of culture. This reduction in colony size at day 6 of culture cannot be attributed to reduced proliferation (see above) but instead reflected a morphological change to more compacted colonies, with well-defined borders and without the habitual, flattened cells on their periphery (Figs. 5A, 5B).
To test whether autocrine FGF signaling is essential for proliferation of undifferentiated hESCs, cultures of hESCs in FGF-2–free medium were treated with the pharmacological inhibitor of FGFR tyrosine kinases, SU5402, which specifically interacts with intra-cellular catalytic domain of FGFRs . After 2 days of continuous exposure to SU5402 at a concentration of 10–30 μM, cells in the centers of colonies acquired flattened morphology that resembled those spontaneously differentiating hESCs that occasionally occur in standard cultures, or the crater cells obtained during neuronal differentiation using HepG2-conditioned medium . Upon further culture totaling 5 days in the presence of SU5402, such morphology (Fig. 6A, upper panel), accompanied by loss of alkaline phosphatase activity (Fig. 6A, lower panel), finally developed in all hESC colonies . Furthermore, decreased phosphorylation of MEK1/2 and its substrate ERK1/2 were both observed in cells maintained in medium with inhibitor (Fig. 6B, – FGF-2). A similar effect was observed also in hESCs that were treated with 10 ng/ml recombinant FGF-2 (Fig. 6B, +FGF-2). Crucially, no signs of decreased cell viability were observed in SU5402-treated cultures (determined by cleavage of lamin B; Fig. 6C, upper left panel); instead, the presence of flattened cells was accompanied by downregulation of Oct-4 and upregulation of cyclin-dependent kinase inhibitor p27 (Fig. 6C, upper right panel), a common event that characterizes differentiation of cells of early embryonic origin  and slower proliferation (Fig. 6C, lower panel). Moreover, differentiation process was manifested by upregulation of TROMA-1, a marker for primitive endoderm, and of nestin, a marker for developing neuroepithelium and for epithelial precursors in the embryonic pancreas (Fig. 6D, Western blots and immunofluorescence). Recombinant FGF-2 at concentrations up to 20 ng/ml did not reverse the phenotype produced by the noncompetitive inhibitor, SU5402 (results not shown), thus supporting our hypothesis of a critical role for autocrine FGF signaling in maintaining the pluripotent state of hESCs.
In the present study, we show that the components of the FGF signaling family are highly expressed in undifferentiated hESCs and, in the case of receptor tyrosine kinases, are uniformly regulated when cells are stimulated to differentiate under various conditions or maintained at high density using enzymatic passaging. Moreover, we demonstrate that endogenously produced FGF-2, as well as exogenously added, recombinant FGF-2, may function to sustain the undifferentiated state in hESCs. Consequently, we propose that endogenously produced FGF-2 and its intracrine and autocrine activities represent a category of regulatory mechanisms that are not shared by the mouse counterparts, mESCs, and that furthermore may be exploited either to sustain undifferentiated proliferation or to initiate directed differentiation of hESCs.
Here we found that the exportable, low-molecular-mass isoform of FGF-2, and those nuclear-localized, high-molecular-mass isoforms of FGF-2 excluding the 34-kDa isoform, are abundantly expressed in both undifferentiated and differentiated hESCs. It is notable that the 34-kDa isoform of FGF-2, which in other systems has been shown to permit cell survival in low-serum conditions , was undetectable here, in hESCs. Low-molecular-mass FGF-2 normally binds to its membrane receptors, principally FGFR1, and the resulting complex is translocated to the nucleus . Once deposited in the nucleus, FGF-2 may function by activating specific genetic programs related either to cell growth/differentiation or to adaptive responses to culture conditions. We have detected FGF-2 in nuclei of hESCs and also have determined hESCs to be rich in FGFR1, supporting a functional role for FGF-2 and the so-called integrative nuclear FGFR1 signaling in this cell type. Relevant to this proposed nuclear activity of FGF-2 is our detection of nuclear FGF-associating protein FIF, which mediates the antiapoptotic properties of nuclear and/or nucleolar isoforms of FGF-2 . In addition to a specific, nuclear activity for high-molecular-mass FGF-2, a distinct, nucleolus-related function for the 18-kDa form of FGF-2 also has been suggested [27, 28], and this role is supported by the significant amounts of FGF-2 that we observed in nucleoli of hESCs.
The expression of high-molecular-mass isoforms of FGF-2, which subsequently are transported directly to the nucleus owing to the presence of a nuclear localization signal , is associated with activation of ERKs . One of the most striking findings in this study was the unusually high basal activity of ERK1/2 in undifferentiated hESCs; for mESCs, in contrast, the activity of ERK1/2 becomes elevated only when they are stimulated to enter the differentiation pathway . Although for both undifferentiated hESCs and differentiating mESCs, a high activity of ERK1/2 correlates with large amounts of endogenous, high-molecular-mass FGF-2 (this study and ), supporting a common mode of interaction, there are other major differences between hESCs and mESCs regarding their response to FGF-2. Stimulation of hESCs with exogenous FGF-2 results in heightened ERK1/2 phosphorylation that is not, however, manifested by either changes in cell proliferation or differentiation. But in undifferentiated mESCs, activation of ERK1/2 antagonizes the LIF/gp130/STAT3 signaling pathway and effectively promotes differentiation . It is relevant that we have been unable to detect the phosphorylated form of STAT3 in hESCs whether treated or untreated with exogenous FGF-2 (results not shown); and although it has been reported that the stimulation of hESCs with the interleukin-6 family of cytokines, including LIF, results in a weak increase of STAT 3 phosphorylation, such treatment causes robust phosphorylation of ERK1/2 [4, 5]. Hence, there seems to be no mechanistic basis for the LIF/gp130/STAT3 to be actively involved in sustaining the self-renewal in hESCs. Although here we did not address these mechanisms in more detail, our data underline fundamental differences between hESCs and mESCs in their molecular pathways for controlling self-renewal.
It is widely regarded that the routine, serum-free culture of hESCs on mouse feeder cells requires soluble FGF-2 (mostly at 4 ng/ml) for undifferentiated proliferation. In agreement with our conclusion from the colony-spreading assay, the effects of exogenous FGF-2 are to support growth of smaller cells in tighter colonies and to increase cloning efficiency . Furthermore, two other studies have reported a positive effect of increased concentrations of FGF-2 (8 ng/ml) in medium conditioned by feeder cells in feeder-free hESC cultures [32, 33]. Studies on various cell types have indicated that soluble FGF-2 has the potential to modulate cell attachment and spreading in feeder-free or feeder-dependent cultures, either by the induction of cell adhesion molecules [34–36] or by formation of tri-molecular complexes with FGFR1 expressed on growing cells and with feeder cell–associated heparan sulfates . As shown here, both undifferentiated and differentiated hESCs express FGFR1 in large amounts, and so there is potential for formation of growth-modulating, FGF-2/FGFR1/heparan sulfate complexes also in this system. Moreover, nearly terminated colony outgrowth without negative effects on cell number in cultures with 10 and 20 ng/ml FGF-2 at day 6 suggests a specific kinetics of proposed mode of action. However, to establish the functional significance of such a mechanism in hESCs, the effect of soluble FGF-2 on the expression and physical interactions of specific adhesion molecules requires investigation.
Several investigators have included FGF-2 at a concentration of 10 ng/ml also in protocols promoting the differentiation of hESCs toward a neural lineage [14–18]. However, in all these experiments, FGF-2 treatment was applied to hESCs that had undergone preliminary differentiation in embryoid bodies. Thus, it emerges that hESCs use exogenous FGF-2 in different ways, depending on their state of differentiation. Such pleiotropy of FGF-2 action may reflect interactions with specific genetic programs driven by other stimuli. For example, the observed differentiation-dependent changes in levels of expression of FGFRs, which most likely affect the signaling capacities of these receptors, may represent one of the molecular events accompanying such developmental reprogramming.
Most important, we showed that inhibition of autocrine FGF signaling by a synthetic inhibitor of FGFR kinase activity suppresses activation of signaling molecules downstream of FGFRs, downregulates Oct-4, upregulates p27, and leads to rapid changes in morphology of hESCs. The effect on cell morphology was observed as early as day 2 of treatment, suggesting that the inhibition occurs already in undifferentiated cells. Moreover, we observed that this differentiation process starts from the centers of colonies, where autocrine signaling should be most pronounced, and that it is accompanied by upregulation of TROMA-1 and nestin, markers of endoderm and developing epithelium, respectively. These results substantiate the previously established model in which growth factor activities in hESCs inhibit differentiation into specific cell lineages rather than inducing differentiation into specific cell types . It is noteworthy that such a scenario has been well documented in Xenopus, in which inhibition of FGF signaling in animal cap explants promotes endoderm formation . In this regard, it is somewhat surprising that SU5402-treated hESCs strongly upregulate intermediate filament nestin, a typical marker of developing neuroepithelium and neural stem cells. However, nestin was recently shown to be expressed by epithelial progenitor cells of the embryonic pancreas , which is derived from definitive endoderm. Many studies, including our own, have previously shown that FGF signaling is necessary for neural and mesodermal specification [3, 40–42], and we now postulate that endodermal differentiation in hESCs also can be manipulated and enhanced by elimination of FGF signaling by using inhibitors such as SU5402.
This research was supported in part by the Grant Agency of the Czech Republic (301/03/1122, 305/05/0434), Ministry of Education, Youth, and Sports(1M0021620803), Ministry of Health (MZ 00065269705), and Academy of Sciences of the Czech Republic (AV0Z50390512). We are very grateful to Dr. Elena Notarianni for valuable comments on the manuscript and for having revised the English of this manuscript.