|Home | About | Journals | Submit | Contact Us | Français|
Fibroblast growth factor (FGF), transforming growth factor (TGF)/Nodal, and Insulin/insulin-like growth factor (IGF) signaling pathways are sufficient to maintain human embryonic stem cells (ESCs) and induced pluripotent stem cells in a proliferative, undifferentiated state. Here, we show that only a few FGF family members (FGF2, FGF4, FGF6, and FGF9) are able to sustain strong extracellular-signal-regulated kinase (ERK) phosphorylation and NANOG expression levels in human ESCs. Surprisingly, FGF1, which is reported to target the same set of receptors as FGF2, fails to sustain ERK phosphorylation and NANOG expression under standard culture conditions. We find that the failure of FGF1 to sustain ES is due to thermal instability of the wild-type protein, not receptor specificity, and that a mutated thermal-stable FGF1 sustains human ESCs and supports both differentiation and reprogramming protocols.
Human embryonic stem cells (ESCs) and human induced pluripotent stem cells (iPSCs) can proliferate without limit and have the potential to generate cell types in all three germ layers [1–6]. These properties make them ideal model systems for studies of human embryogenesis, for drug discovery, and for clinic applications.
Human ESCs and iPSCs closely resemble mouse epiblast stem cells (EpiSCs) in embryogenesis [7, 8] and are maintained in cell culture by the six growth factors that activate fibroblast growth factor (FGF), TGF/Nodal, and Insulin/IGF pathways [9–13]. The FGF pathway, in particular, has been implicated in all phases of mammalian ESC culture, such as cell survival, proliferation, pluripotency, and lineage determination during differentiation [10, 11, 14–16]. Among the 18 FGF proteins in mammals [16, 17], FGF2 is most commonly used to maintain self-renewal and pluripotency of human ESCs and iPSCs  and mouse EpiSCs .
The FGF pathway is activated through the binding of FGF ligands to FGF receptors, which in turn may trigger the activation of various downstream signaling pathways such as the rat sarcoma (RAS)-mitrogen-activated protein (MAP) kinase pathway, the P-I-3 kinase-AKT pathway, and the phospholipase C (PLC)γ pathway [16, 19]. Multiple factors may contribute to the differential regulation of various FGF proteins in vivo. First, each FGF protein has a different affinity to each of the four FGF receptors (FGFRs) that activate specific pathways [20–22]. Second, the combination of differential expression of FGFs and FGFRs leads to specific physiological roles in specific tissues . However, even the existence of these multiple factors is not sufficient to explain how specific FGFs regulate certain processes in human ESC culture. To better modulate human pluripotent stem cells through FGF pathways, it is important for us to understand all the critical mechanisms potentially involved in determining FGF function.
FGF2’s ability to support pluripotency is just one of the unknowns of FGF regulation in stem cell culture. The high concentrations of FGF2 used to support pluripotency in defined long-term culture—up to 100 ng/ml—is also of interest because these concentrations are higher than those used on other cell types, which usually range from 1 to 10 ng/ml [18, 23]. It has also been suggested that FGF signaling is dosage-dependent, and that a high level of FGF2 is probably required to satisfy a specific signaling threshold  or to prevent inhibitions such as protein degradation [15, 24]. Furthermore, heparin and heparan sulfate were reported to promote pluripotency [25, 26]. It remains unclear how heparin and heparan sulfate directly function through the FGF pathway to regulate human ESCs and iPSCs.
Given the importance of the FGF pathway to pluripotency and differentiation in human ESCs, we decided to perform a systematic study of FGF proteins. Our goal was to identify novel regulatory mechanisms that contribute to the function of FGF proteins in human ESCs and iPSCs. In addition to identifying a specific set of FGFs that activate FGFR in human ESCs, we found that thermal stability is another deciding factor in determining a specific FGF’s capacity to support self-renewal. We demonstrated that modulating FGF stability with heparin or point mutation could significantly impact FGF’s ability to control various aspects of stem cell culture, ranging from pluripotency and differentiation to reprogramming. Our studies thus provide a new strategy: manipulating human ESCs by altering FGF thermal stability.
Human ESCs were usually maintained in E8 media on Matrigel-coated tissue culture plates . Cells were repassaged routinely with EDTA as described previously . Briefly, cells were washed twice with phosphate buffered saline (PBS)/EDTA medium (0.5 mM EDTA in PBS, osmolarity 340 mOsm), then incubated with PBS/EDTA for 5 minutes at 37°C. PBS/EDTA was removed, and cells were washed off swiftly with a small volume of corresponding media.
The antibodies used include: anti-OCT4 antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), anti-phospho-ERK1/2 (Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com), and anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Millipore, Billerica, MA, http://www.millipore.com).
FGF family proteins were purchased from R&D Systems (Minneapolis, MN, http://www.rndsystems.com). For simplicity of presentation, we used FGF1 to represent full-length acidic FGF (aFGF, 154 aa, from 2 to 155) and FGF2 to represent basic FGF. We also used truncated FGF1 (aFGF, 140 aa, from 16 to 155) (Supporting Information Fig. S2B). All homemade FGF1 proteins are truncated forms, with specific names emphasizing point mutations (Fig. 3B).
Cell growth measurement followed the procedure previously described unless otherwise specified . All experiments were done on 12-well plates, usually in triplicate for each treatment. Prior to the addition of cells, 500 μl of medium was loaded into each well. Cells were dissociated with TrypLE (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for 5 minutes or until fully detached from the plate, neutralized with equal volumes of media, counted, washed, and diluted to 100,000–300,000 cells per milliliter, and 100 μl of cells were then added into each well. At specific time points, cells were again dissociated with 0.4 ml TrypLE, neutralized with equal volumes of 10% fetal bovine serum (FBS) in Dulbecco/Vogt modified Eagle’s minimal essential medium, harvested with pipettes, and counted by flow cytometry. As an internal control, 5,000 Count-bright beads (Invitrogen, Grand Island, NY, http://www.invitrogen.com) were added to each sample, and usually approximately 200 beads were counted for each sample. For proliferation experiments, media were changed daily till the day of analysis, and cells were counted as described above.
Protein expression was done in Rosetta2 (DE3) pLysS cells (Novagen) using Magic Media (Invitrogen) at 37°C for 24 hours. FGF proteins were purified with heparin column and cation column. FGF1 derivatives were a gift from Dr. Malgorzata Zakrzewska and Dr. Jacek Otlewski. FGF2-K128N mutant was made in this study.
For differentiation, H1 ESCs were dissociated by TrypLE (Invitrogen) and seeded at the density of 4 × 104 cells per square centimeter in E8 medium containing 5 ng/ml human BMP4 (R&D Systems) with different FGFs (10 ng/ml). The cells were harvested 48 hours after treatment.
FGF and TGFβ/NODAL pathways sufficiently support pluripotency in human ESCs and iPSCs [13, 10]. To study systematically the FGF family, we first established reliable biological assays to analyze FGF functions on biochemical and gene expression levels. An ERK1/2 phosphorylation assay was designed to show the effect of FGFR activation by FGF in short-term treatments ranging from 15 minutes to a couple of days. Here, we found that FGF2, not TGFβ, stimulated MAP kinase ERK1/2 phosphorylation after short-term incubation (15 minutes) (Fig. 1A). Pluripotency marker genes such as NANOG were used to measure the long-term impact of FGF on pluripotency. We found that NANOG expression was suppressed by an ERK inhibitor (Fig. 1B).
ERK phosphorylation and NANOG expression were then used to evaluate the functional difference of FGF family members in human pluripotent stem cells. FGF family proteins were screened for their ability to support pluripotency. Only a few FGF proteins (i.e., FGF2, FGF4, FGF6, and FGF9) were able to sustain strong ERK phosphorylation after 24-hour incubation in cell culture (Fig. 1C). This pattern was consistent with NANOG expression levels in ESCs under the treatment of different FGFs (Fig. 1D).
We then screened for FGFs that could activate FGFRs after short-term incubation. Strong ERK phosphorylation was observed in ESCs treated with FGF1, FGF2, FGF4, FGF6, and FGF9 for 15 minutes (Fig. 1E, Supporting Information Fig. S1A for FGF18). FGF1 could stimulate ERK phosphorylation after 15 minutes but failed to do so after 24 hours. We next explored FGF1’s activity during incubation at 37°C. After 6 hours of 37°C preincubation, FGF1 lost all its ERK phosphorylation activity (Fig. 1F); at the same time, a few others maintained their activities (i.e., FGF2, FGF4, FGF6, and FGF9), which was consistent with their ability to maintain 24-hour ERK phosphorylation and NANOG expression (Fig. 1C). This indicates that FGF1 is not as stable at 37°C when compared with other FGF proteins, such as FGF2 and FGF4, a factor that might also contribute to its inability to maintain ERK phosphorylation and pluripotency in human ESCs. Based on the observation of dramatic loss of FGF1 activity in 6 hours, we hypothesized that other FGFs’ activity could also be affected by longer 37°C incubation in a regular 24-hour medium change cycle.
Even for FGF2, a high-level of FGF protein is often required to maintain pluripotency of human ESCs in culture. Thus we hypothesized that this phenomenon might also be associated with thermal instability of FGF. We first examined the most commonly used FGF2s in human ESCs, human FGF2, and zebrafish FGF2 [18, 23, 27]. Both human and zebrafish FGF2s have a similar ability to induce ERK phosphorylation (Supporting Information Fig. S1B), but they both lose most of their activity after 24-hour incubation at 37°C (Fig. 2A, Supporting Information Fig. S1C).
Consistent with heparin’s role in maintaining pluripotency, coincubation with heparin helped maintain the activity of both human and zebrafish FGF2 proteins (Fig. 2B, Supporting Information Fig. S1C). Several other sulfate-rich polymers also preserved FGF2 activity during incubation (Supporting Information Fig. S1D, S1E), while coincubation with bovine serum albumin (BSA) or reducing reagents failed to preserve the activity (Supporting Information Fig. S1C). We also found that heparin has to be present at the time of incubation to protect FGF activity, and heparin cannot recover the activity after the protein is heat inactivated (Supporting Information Fig. S1D).
We found that 37°C incubation led not to degradation but to aggregation of FGF2, and that such aggregation was prevented by heparin (Fig. 2C). Heat-induced aggregation was also observed in zebrafish FGF2 proteins (Supporting Information Fig. S1E). Heparin’s role in maintaining FGF2 was consistent with its ability to improve the expression of pluripotency markers, such as NANOG (Fig. 2D).
Based on our observation of differential thermal stability in FGF family members, we hypothesized that thermal stability of protein could be a critical factor for specific FGF function in stem cells, so we chose FGF1 as a model molecule to test this hypothesis. FGF1 is very unstable at 37°C, and it could not support self-renewal of human ESCs in cultures fed daily (Fig. 1). However, more frequent feeding of fresh FGF1 significantly improved short-term cell culture (Supporting Information Fig. S3A). This suggests that FGF1 stability, not receptor specificity, might play a determinant role in its function. Unlike FGF2, FGF1 lost most of its activity after 6 hours at 37°C even with heparin (Fig. 3A). This indicates that FGF1 thermal stability is also controlled by factors other than heparin binding. We found that widely available, truncated FGF1 protein was also unstable at 37°C with dynamics similar to full-length FGF1 (Supporting Information Fig. S2B), which suggests that the N-terminal sequence of FGF1 does not play an important role in thermal stability.
To further improve FGF1 thermal stability, we used well-characterized truncated FGF1 derivatives whose thermal stability was significantly improved by point mutations [29, 30] (Fig. 3B, Supporting Information Fig. S2C). Amino acids at Q40, S47, and H93 contribute to thermal instability, while a specific lysine in FGF1 (K112) is important to heparin binding, and mutation to asparagine (FGF1 K112N) leads to weaker heparin affinity. By mutating the first three sites (Q40P, S47I, and H93G, FGF1-3X), the FGF1 protein is significantly stabilized, while FGF activity is maintained . This increased stability can even compensate for reduced affinity to heparin when additional heparin binding site K112 is mutated to asparagine (FGF1-4X) .
FGF1 derivative proteins were purified in Escherichia coli (Supporting Information Fig. S2C). Each mutated FGF1 and the wild-type FGF1 induced ERK phosphorylation in human ESCs after a 15-minute treatment (Supporting Information Fig. S2D). However, although each mutated FGF1 sustained ERK phosphorylation in human ESCs during a 24-hour incubation, the wild-type FGF1 failed to do so (Fig. 3C).
We further tested the loss of activity after preincubating the FGFs at 37°C in the absence of cells. After a prior 24-hour 37°C treatment, FGF1-wild type (WT) and FGF1-K112N lost their activity to trigger ERK phosphorylation in human ESCs (Fig. 3D). After the same 37°C treatment, FGF1-3X activity was reduced in the absence of heparin but was preserved in the presence of heparin. The heparin-dependent stability in FGF1-3X is thus similar to human FGF2 (Fig. 3D). To our surprise, FGF1-4X was thermally stable independent of heparin (Fig. 3D).
When these FGF proteins were applied to human ESCs, mutated FGF1 proteins, especially FGF1-3X and FGF1-4X, significantly improved cell growth (Fig. 3E). Stabilized FGF1 proteins also helped maintain pluripotency of human ESCs (Fig. 3F, Supporting Information Fig. S2E). Human ESCs were cultured in media with FGF1-3X and FGF1-4X in the place of FGF2 for more than 2 months, and cells maintained normal karyotypes (Supporting Information Fig. S2E).
Since FGF can play different roles in various stages of stem cell culture, we further explored how FGF thermal stability could contribute to differentiation and reprogramming. We first looked at how FGF function is affected in spontaneous differentiation, which, in human ESCs, is usually initiated after a few passages in the absence of TGFβ. We found that thermal stability of FGF significantly affected the upregulation of a few differentiation genes. With increased thermal stability, mRNA expression was suppressed among differentiation-related genes, such as GATA2 and HAND1 (Fig. 4A, 4B). The suppression by FGF1-3X or FGF2 was further improved with additional heparin, which is consistent with heparin’s role in stabilizing FGFs (Fig. 4A). These results indicate that thermal stability could play other roles beyond maintaining pluripotency.
To further demonstrate the impact of thermal stability on differentiation, we tested FGF function in mesoderm-specific differentiation, where FGF activation is essential for the emergence of mesoderm lineage . Different FGF1 derivatives were used to initiate differentiation in the presence of BMP4, and we found that thermal stability of FGF directly affected the cell differentiation to mesodermal-specific cells (Fig. 4B, 4C). NANOG was elevated with increased FGF stability; at the same time, mesoderm marker Brachurary (T) was upregulated accordingly. This finding is consistent with the notion that the FGF pathway maintains NANOG activity to promote mesoderm-specific differentiation.
We also tested how specific FGFs could reprogram to generate iPSCs. In limited tests, we found that a specific FGF’s ability to promote reprogramming was consistent with its ability to sustain ERK phosphorylation (Supporting Information Fig. S4, Fig. 1). We also found that thermal-stabilized FGF1 mutants significantly improved fibroblast cell growth of human fibroblasts (Fig. 4E), and that reprogramming efficiency was also enhanced significantly (Fig. 4F). This indicates that thermal stability of FGF is also a factor essential to the reprogramming process.
In summary, we identified thermal stability as a critical factor controlling the function of FGF proteins, a finding that has the potential to impact all aspects of ESCs and iPSC research, from pluripotency to differentiation and reprogramming.
During the course of these studies, we systematically analyzed the regulation of FGF family members using human ESCs. We found that human ESCs specifically responded to a few FGFs in cell culture, and we also identified thermal stability of protein as a major mechanism determining specific FGF’s ability to support pluripotency in cell culture, thus providing a new strategy to modulate cell culture conditions for human pluripotent stem cells.
We first established the fact that the capacity of specific FGFs to maintain pluripotency is directly associated with its ability to stimulate and maintain MAP kinase phosphorylation in human ESCs (Fig. 1). It is surprising to find that only a few FGFs are able to stimulate MAP kinase phosphorylation, even though all FGFs have their corresponding receptors expressed in human ESCs [24, 32]. This observation can also help explain why mouse Fgfs, such as Fgf4, Fgf5, and Fgf8 have different functions during mouse embryogenesis, even though all of them are expressed at around that same time. Further studies are needed to find the mechanisms leading to such phenomenon. We also noticed that the FGF ligand–receptor activation pattern in human pluripotent stem cells is different from previously observed in Ba/F3 cell lines [20, 22]. This indicates that when these FGF receptors are expressed in different cell types, different FGF receptors might have different downstream responses upon exposure to ligands.
Our study also provided a possible explanation for why high concentrations of FGF or FGF-related factors such as heparin are beneficial to human ESC/iPSC culture. The addition of heparin (Fig. 2B–2D) or a high dose of FGF2 likely helps offset the great loss of FGF activity during 24-hour incubation at 37°C. This conclusion is different from the hypothesis that a high dosage of FGF2 leads to high FGFR activation for specific pathways . Our further characterization suggested otherwise, and we found that FGFR activation was usually maintained at steady but relatively low levels in regular ESC culture. FGFR downstream ERK phosphorylation decreased gradually after initial induction even when active FGF2 was still maintained in culture (Supporting Information Fig. S3A, S3B). In continuous cell culture, ERK phosphorylation was usually maintained at a relatively low level while pluripotency was sustained (Supporting Information Fig. S3C). These results indicate that higher FGF concentrations are probably not necessary for stem cells, and that a negative feedback curbs excessive activation. This is consistent with our hypothesis that a higher FGF level is required to counter the excessive loss caused by heat inactivation at 37°C (Fig. 2A); in this way, the FGF-MAPK pathway is continuously maintained for self-renewal (Supporting Information Fig. S3A–S3C).
The relationship between thermal stability and FGF specificity in ESCs is further confirmed by our study on FGF1 derivatives, which could also be useful tools in modulating stem cell culture. Our results demonstrated that the thermal instability of FGF1 is the main reason why it failed to support human ESC culture. At the same time, we identified two FGF1 mutants, FGF1-3X and FGF1-4X, as stable FGFs capable of maintaining ESC pluripotency (Fig. 3) [29, 30]. Because of their different thermal stability, each derivative could be a potentially useful reagent to stimulate FGF receptors in either long-term or transient exposure.
To our surprise, we found that FGF1 was stabilized by the mutations to heparin binding site K112. FGF1-K112N was more potent than wild-type FGF1 (Fig. 3C), while FGF1-4X (FGF1-3X with additional K112N mutation) was more stable than FGF1-3X (Fig. 3D). This indicates that K112 might contribute to the instability countered by heparin binding. This phenomenon was confirmed by our study on FGF2 where we mutated the conserved heparin-binding site K128 to asparagine (FGF2-K128N). The mutation helped FGF2 maintain its activity during 37°C incubation (Supporting Information Fig. S2F, S2G). This discovery suggests a potential link between FGF and heparin during molecular evolution. It is thus possible that thermal instability provides a regulation mechanism that allows FGF to have a long-lasting function only at sites with a rich poly-sulfate environment, such as heparin and surface heparan sulfate (Supporting Information Fig. S1C, S1G).
The significant stability difference between FGF1 and FGF2 also led us to rethink how we can use specific FGFs in different cell culture procedures. Even though stable FGF2 is beneficial to pluripotency and mesodermal differentiation, the unstable FGF1 might perform better than FGF2 in other processes. These and other possibilities will require more in-depth studies in the future.
This is the first study to analyze systematically how FGF family proteins impact human pluripotent stem cells. Protein stability is a major mechanism regulating cellular processes [34–36], and our study elucidates how thermal stability contributes to FGF specificity for stem cell pluripotency in culture. It not only provides researchers with a few practical tools, such as FGF1 and FGF2 derivatives with different stability but also provides a simple mechanistic explanation for choosing high FGF levels or heparin in human ESC culture. This study demonstrates that modulating growth factor stability could be an effective way to regulate pluripotency or differentiation and has significant potential to benefit both basic and clinical research.
This work was supported by the Charlotte Geyer Foundation, the Morgridge Institute for Research, NIH Grant UO1ES017166 (to J.A.T.), and NIH contract RR-05-19 (to J.A.T.). We thank Krista Eastman for editorial assistance.
Disclosure of Potential Conflicts of Interest
J.A.T. is a founder, stockowner, consultant, and board member of Cellular Dynamics International (CDI). He also serves as scientific advisor to and has financial interests in Tactics II Stem Cell Ventures.
Author contributions: G.C.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; D.R.G.: conception and design, collection and/or assembly of data, and data analysis and interpretation; P.Y. and Z.H.: collection and/or assembly of data and data analysis and interpretation; J.A.T.: conception and design, data analysis and interpretation, financial support, and final approval of manuscript. G.C. and D.R.C. contributed equally to this article.