Journal list menu

Volume 580, Issue 12 p. 2869-2874
Free Access

Fibroblast growth factor signaling in embryonic and cancer stem cells

Petr Dvorak

Corresponding Author

Petr Dvorak

Department of Biology, Faculty of Medicine, Masaryk University, Kamenice 5, 625 00 Brno, Czech Republic

Department of Molecular Embryology, Institute of Experimental Medicine, Academy of Sciences of the Czech Republic, Prague, Czech Republic

Center for Cell Therapy and Tissue Repair, 2nd Faculty of Medicine, Charles University, Prague, Czech Republic

Corresponding author. Fax: +420 549491327.Search for more papers by this author
Dana Dvorakova

Dana Dvorakova

Department of Internal Medicine-Hematooncology, University Hospital Brno, Czech Republic

Search for more papers by this author
Ales Hampl

Ales Hampl

Department of Biology, Faculty of Medicine, Masaryk University, Kamenice 5, 625 00 Brno, Czech Republic

Department of Molecular Embryology, Institute of Experimental Medicine, Academy of Sciences of the Czech Republic, Prague, Czech Republic

Center for Cell Therapy and Tissue Repair, 2nd Faculty of Medicine, Charles University, Prague, Czech Republic

Search for more papers by this author
First published: 28 February 2006
Citations: 75


The fibroblast growth factor 2 (FGF-2) pathway is one of the most significant regulators of human embryonic stem cell (hESC) self-renewal and cancer cell tumorigenesis. Here we summarize recent data on the effects of FGF-2 and its receptors on hESCs and leukemic stem/progenitor cells. Also, we discuss the similarities of these findings with stem cell renewal and differentiation phenotypes.

1 Basics of FGF-2 pathway

Fibroblast growth factors (FGFs) comprise a large family of signaling molecules with various functions in development as well as in adult physiology and pathology. FGF-2 (also called basic FGF; bFGF), a prototype member of the FGF family, is encoded by a single copy gene that is alternatively translated to produce one low (18-kDa) and four high (22-, 22.5-, 24-, and 34-kDa) molecular mass isoforms [1, 2]. The 18-kDa low molecular mass (LMM) FGF-2 lacks a detectable secretion signal sequence and is released via its association with other molecules (e.g. HSP27) by an exocytotic mechanism that is independent of the endoplasmatic reticulum/Golgi pathway [3]. Cytosolic 18-kDa FGF-2 harboring a C-terminal nuclear localization sequence (NLS) is specifically targeted to the nucleus [4, 5] by a mechanism involving the cytoplasmic microtubule-associated protein translokin [6]. The high molecular mass (HMM) isoforms of FGF-2 all contain classical N-terminal NLS, and their nuclear localization can be further enhanced through their association with the 53/55-kDa FGF-2-interacting factor (FIF) [7]. In theory, FIF could interact with LMM FGF-2, which is also translocated into the nucleus, since the FIF-binding motif of FGF-2 is located in the last 155 amino acids of FGF-2 and not in the NH2-extension that is specific for HMM isoforms of FGF-2 [7]. The nuclear accumulation of HMM FGF-2 is associated with its post-translational methylation [8], which may be crucial for protein–protein or protein–RNA interactions in the nuclear environment [9].

The different subcellular localization and trafficking of the FGF-2 isoforms unambiguously determine their mode of action. LMM FGF-2 signals via its interaction with the high-affinity transmembrane FGF receptors (FGFR1-4) in a paracrine or autocrine manner. Importantly, FGFR-ligand binding studies demonstrated that LMM FGF-2 binds predominantly to FGFR1 and 2 [10, 11]. Moreover, high-affinity binding is extracellularly modulated by non-signaling heparin/heparan sulfate (HS) proteoglycans that are subsequently involved in the intracellular processing of FGF-2 [12]. HS proteoglycans are also unconditionally required for ligand-induced dimerization and transphosphorylation of FGFRs, the key events by which tyrosine kinase receptors initiate downstream signaling (reviewed in [13]). The signaling cascade downstream of the FGFRs involves tyrosine phosphorylation of the docking protein FRS2 followed by recruitment of several Grb2 molecules. Next, activated Grb2 associated with FRS2 recruits the nucleotide exchange factor SOS. The formation of FRS2-Grb2-SOS complexes results in activation of the Ras-Raf-MAPK signaling pathway that initiates a cell response. In parallel, FGFRs can also activate the PLCγ–PKC pathway in which Grb2 molecules recruit the docking protein Gab1 leading to the activation of the PI3K-Akt cell survival pathway. Thus, FGF signaling through FGFRs is mediated primarily by the assembly of multi-docking protein complexes and involves multiple layers of regulation [14].

In contrast to the relatively well-understood paracrine/autocrine activities of LMM FGF-2, HMM FGF-2 possess receptor-independent intracrine activities that are not easily analyzed. It has been reported that HMM FGF-2 downregulates PKCε and upregulates PKCδ, an effect which subsequently leads to the activation of ERK1/2 (MAPK) [15]. This finding suggests that HMM FGF-2 preferably utilizes the PLC/PKC signaling pathways and may provide an alternative pathway from LMM FGF-2-mediated signaling [14]. Although the biological role of HMM FGF-2 is largely unknown, these and other data suggest that the signaling and nuclear localization of HMM FGF-2 allows for the regulation of genes that are involved in the pro-survival phenotype of cells [1, 16, 17].

The complexity of FGF-2 signaling via its cognate high-affinity receptors and the known or hypothetical cellular consequences of such signaling with respect to human embryonic stem cells (hESCs) and cancer stem cells are illustrated in Fig. 1 .

figure image
Schematic diagram of FGFR and possible modes of FGF-2 action. (I) Exogenous FGF-2 (red ovals) associates with heparin sulfate proteoglycans (HSPGs) or their side chains (small beige circles). The resulting complex binds to the ligand-binding site (immunoglobulin[Ig]-like domains II and alternatively spliced IIIb/IIIc; DII and DIIIb/IIIc) and the heparin-binding region (Ig-like domain II; DII) of the high-affinity receptor. The subsequent receptor dimerization brings the cytoplasmic tyrosine kinase (TK) domains into close proximity allowing for receptor transphosphorylation (rightarrow, P). This tri-molecular arrangement triggers the “classical” transduction pathway that consists of recruitment, assembly, and phosphorylation (P) of downstream signaling proteins (blue squares). (II) The exogenous LMM isoform of FGF-2 binds to high-affinity receptors, primarily FGFR1, and the resulting complex is internalized and translocated into the nucleus. There, this complex presumably functions to activate genetic programs related to cell growth, differentiation, and (in the case of hESCs) an adaptive response to in vitro conditions. (III) Endogenously synthesized LMM FGF-2, lacking an obvious signal sequence for active secretion via endoplasmatic reticulum/Golgi pathway, is released in complex with various molecules (e.g. HSP27 (light blue oval)) or targeted to the nucleus via association with translokin (pink oval). Additionally, this LMM FGF-2 can be embedded and efficiently exported in membrane-derived microvesicles [our unpublished observation]. Once exported, LMM FGF-2 induces cellular signaling in an autocrine or paracrine manner. (IV) Endogenously synthesized HMM FGF-2 associates with FIF (dark blue oval) and is directly transported into the nucleus where mediates its pro-survival and anti-apoptotic activities. Although the biological impact on each cell type may be entirely different, the complex actions of exogenous and endogenous FGF-2 described above may be coordinated to perform the diverse cellular functions that are shared by human embryonic stem cells and leukemia progenitor/stem cells.

2 Roles of FGF-2 in hESCs

Despite a concentrated effort by many researchers, the signaling pathways involved in the self-renewal and maintenance of the pluripotent status of hESCs are still largely unexplored. It is widely accepted that hESCs require exogenous FGF-2 to sustain self-renewal and the capacity to differentiate into a large number of somatic cell types [18, 19]. However, understanding the molecular regulation of FGF signaling in hESCs remains a major challenge.

The first step in understanding the role of FGF-signaling in hESCs is to examine the expression of individual components of the FGF signaling pathway. Using cDNA microarray analysis, several groups have described elevated levels of key regulators of the FGF pathway in undifferentiated hESCs compared with their differentiated counterparts. These differentiated counterparts were obtained following the formation of embryoid bodies which exhibited long-lasting (3–5 weeks) non-specific differentiation. The upregulated FGF signaling components identified include FGF-2, FGF-11, and FGF-13, and all four FGFRs [20-25]. Importantly, using real time RT-PCR, we determined that the expression of the FGFRs in undifferentiated cells follows a specific pattern with FGFR1 being the most abundant receptor and other receptors showing lower expression in the following order: FGFR3 > FGFR4 > FGFR2. In contrast to the cDNA microarray data, we showed that when hESCs are induced to differentiate for 8–15 days using various stimuli, the expression of all four FGFRs increases [26]. The inconsistency of the results obtained when using long-lasting or shorter differentiation protocols suggests a tight regulation of FGFRs regulation during cell specification. Also, it is interesting that the most abundant receptor, FGFR1, is the dominant target of exogenous FGF-2. In neural cells, FGFR1 has been demonstrated to enter into the nucleus with its ligand (FGF-2) where it acts as a general transcriptional regulator of the genes involved in cell growth and differentiation [27]. The expression of FGF-2 and its receptors in undifferentiated hESCs and hESC-derived embryoid bodies was also examined using massively parallel signature sequencing [28]. These results also clearly demonstrate that hESCs express the docking protein FRS2, which is a major target of activated FGFRs. In conjunction with transcriptome analysis, we determined that undifferentiated hESCs as well as hESCs at the early stages of differentiation express several molecular mass isoforms of FGF-2, including those that are exportable as well as those that are nuclear localized [26]. Recently, we demonstrated that, when FGF-2-enriched hESCs are subjected to an extended differentiation protocol (2–3 weeks of differentiation as embryoid bodies), the expression of endogenous FGF-2 remarkably diminishes [our unpublished data]. Together, several studies have clearly demonstrated that undifferentiated as well as differentiated hESCs are ready to accept and transmit FGF signals. In addition, the differential expression of multiple exportable and nuclear forms of FGF-2 clearly suggests that this signaling pathway may function in an autocrine or intracrine manner in undifferentiated cells, while the exogenous FGF-2 may represent the key activity driving these cells to form specialized cell types during their differentiation.

The second step in understanding the complexities of FGF signaling in hESCs is to examine the biological effects of exogenous FGF-2 on the growth of hESCs. FGF-2 supports growth of undifferentiated cells and increases cloning efficiency [18, 19, 29, 30]. Although it is widely used, the molecular mechanism(s) responsible for this effect is not understood. Interestingly, in our experiments, the outgrowth of hESC colonies in medium containing increased concentrations of exogenous FGF-2 (>5 ng/ml) was reduced while their proliferation rate was not affected [26]. This observation suggests that exogenous FGF-2 can modulate cell attachment and spreading, likely via the induction of cell adhesion molecules. Such indirect effect might also be involved, at least partially, in the phenomenon of culturing hESCs in feeder-free conditions which absolutely requires extremely high concentration of FGF-2 (∼40 ng/ml), either alone [31] or in combination with noggin [32, 33]. The simplest explanation for this result is that exogenous FGF-2 synergizes with noggin to repress trophoblast-inducing BMP signaling and thus sustains the undifferentiated growth of hESCs [32, 33]. The importance of the FGF signaling pathway in maintaining the pluripotency of hESCs was recently demonstrated by two experiments in which inhibition of the FGFR tyrosine kinase activity caused rapid differentiation. Our study [26] showed that LMM FGF-2 is released by hESCs into the medium where it can signal via an autocrine pathway. We further demonstrated that the inhibition of such autocrine FGF signaling using the synthetic inhibitor of FGFRs, SU5402, suppressed the activation of signaling molecules downstream of FGFRs, downregulated Oct-4, upregulated p27Kip1, and led to rapid differentiation. Recently, we performed cDNA microarray analysis to examine mRNA expression patterns in hESCs cultured either in medium without FGF-2 supplemented with SU5402 (to inhibit autocrine signaling) or in standard FGF-2-containing medium (4 ng/ml). Cells cultured in medium without FGF-2 served as controls. We observed characteristic gene expression changes under both conditions; however, number of differentially expressed genes as well as the magnitude of the fold-changes of these genes were consistently much higher in cells maintained with SU5402 [our unpublished data]. Therefore, it is possible that the endogenously-produced FGF-2 including the exportable 18-kDa isoform may significantly regulate the phenotype of hESCs. Other studies [34] also demonstrated that treatment of hESCs with SU5402 causes hESC differentiation and suggested that FGF-2 acts as a competence factor in the activin/nodal/TGFβ pathway. In addition, they described a chemically defined medium supplemented with a combination of activin or nodal and FGF-2 in which hESCs maintain long-term pluripotency in the absence of feeder cells, conditioned medium, or serum replacer.

Together, it is now clear that FGF signaling is unconditionally required for the sustained self-renewal and pluripotency of hESCs. However, understanding the basic molecular mechanisms regulating this effect and its consequences on other signaling pathways is still very limited.

3 Roles of FGF-2 and FGFRs in human hematologic malignancies

A number of studies describing the positive effects of FGF-2 on normal hematopoietic development have been reported [35-37]. These effects typically utilize the common activities of several mesoderm-inducing factors and are involved in the early stages of hematopoietic commitment. During adult hematopoiesis, FGF-2 has been shown to synergize with other hematopoietic growth factors to stimulate the growth of progenitor cells. For example, LMM FGF-2 was shown to cooperate with v-Myb to control the sustained proliferation of erythroid progenitors, which would otherwise effectively differentiate [38]. It was also shown that the clonogenic growth of hematopoietic progenitor cells obtained from granulocyte macrophage colony-stimulating factor-mobilized bone marrow cells is augmented in the presence of high concentrations of stem cell factor (SCF) and FGF-2 (100 ng/ml) [39]. Exogenous FGF-2 also enhances the growth of bone marrow mesenchymal stem cells [40]. This effect is likely mediated by the FGF-2-induced stimulation of telomere elongation that was shown to prolongs the life span and modulates the differentiation potential of bone marrow stromal cells [41]. In addition to the relatively well-understood effects of exogenous FGF-2 on hematopoietic stem/progenitor cells, normal adult hematopoiesis might also be regulated by FGF-2 intracrine or autocrine signals. Human CD34+ progenitor cells have been shown to synthesize and secrete numerous growth factors, including FGF-2. It has been suggested that FGF-2 produced by these cells stimulates the proliferation of other cell types, such as endothelial cells [42], although, its intracrine activities remain completely unknown.

Importantly, the biological effects of either endogenous or exogenous FGF-2 may have different significance in transformed cells, an effect that may potentially relate to their malignant phenotype. There are several, rather generally applicable, examples of non-hematopoietic cancers that support this hypothesis. It has been shown that elevated levels of FGF-2 in the micro-environment of metastatic prostate tumors provides a mechanism by which cancer cells are able to counteract the anti-proliferative and cell-killing effects of chemotherapy. Since these protective effects could be reversed with the addition of an FGF-2 antibody, it was suggested to involve specific interactions between FGF-2 with its cognate receptors [43]. Another example of how FGF-2 contributes to the malignant phenotype of cancer cells involves the targeting of rat bladder carcinoma cells with nuclear 24-kDa FGF-2. These cells become highly tumorigenic and metastatic [44], likely due to the dramatic activation of their cell survival program [45].

Although FGF-2 has been shown to promote the neovascularization of solid tumors, its effects in hematological malignancies are largely unknown. Current models of this effect suggest that aberrant expression of specific growth factors in a variety of cancer cell types, including leukemia cells, can confer a growth advantage and, thus, contribute to a tumorigenic phenotype. Therefore, our laboratory has focused on the level of plasma FGF-2 and the spectrum of the FGF-2 molecular mass isoforms in peripheral white blood cells (WBCs) from a large sample of patients with various neoplastic disorders of the hematopoietic system, such as B cell chronic lymphocytic leukemia (B-CLL), chronic myeloid leukemia (CML), myelodysplatic syndrome, acute myeloid leukemia, acute lymphoblastic leukemia, and hairy cell leukemia. We demonstrated that the level of plasma FGF-2 is elevated in about 50% of patients suffering from B-CLL and CML. Importantly, the WBCs from B-CLL patients express both exportable and nuclear-localized variants of FGF-2, whereas the WBCs from CML patients have a weak expression of nuclear FGF-2 isoforms. Furthermore, while FGFRs expressed in WBCs from B-CLL patients bind only weakly 18-kDa FGF-2, the FGFRs expressed in WBCs from CML patients bind FGF-2 more strongly. As a consequence of this stronger binding, exogenous FGF-2 activates the signal cascade downstream of FGFRs in and stimulates the growth of WBCs from CML patients [46]. These results were confirmed in model leukemia cell lines [47]. We also demonstrated that nuclear HMM FGF-2 associates with FIF [47], a finding similar to that observed in hESCs. Therefore, we concluded that WBCs in CML patients are able to utilize 18-kDa FGF-2 provided by other FGF-2-hyperproducing cells to activate downstream targets and presumably modulate their cell growth. While searching for the source of this FGF-2 in the blood of CML patients, we identified a very minor subpopulation of CML CD34+ BCR-ABL+ progenitor cells which contain low levels of secretable 18-kDa FGF-2 and very high levels of nuclear 22-, 22.5-, 24-, and 34-kDa FGF-2. These cells most likely represent the source of the HMM FGF-2 detected in the extract from total WBCs [48]. Considering the activities of exportable LMM and nuclear HMM FGF-2 isoforms in various cell types, there are several abnormalities of CD34+ BCR-ABL+ cells that have been previously ascribed to FGF-2-unrelated mechanisms which may involve endogenous FGF-2. For example, CML CD34+ cells are resistant to anti-leukemic drug-induced apoptosis and have sustained activation of the MAPK pathway, aberrant cycling and differentiation, and activated transcription and production of several hematopoietic cytokines that serve as autocrine mediators of proliferation. It is likely that the abnormal accumulation of FGF-2 in these cells produces a pro-leukemic effect(s) in addition to the key molecular event, the BCR-ABL translocation.

Neoplastic transformation is often related to abnormal activation of growth factor receptors. For example, in multiple myeloma (MM), the presence of the translocation t(4;14) results in the dysregulated expression of FGFR3 [49]. As a consequence, the constitutively activated kinase domain of FGFR3 signals to promote MM cell proliferation and survival. Additionally, fusions between the catalytic domain of FGFR1 and several partner genes (the most common is the zinc finger transcription factor ZNF198) result in the constitutive activation of the FGFR1 tyrosine kinase that is highly transforming and associated with stem cell myeloproliferative disorder (MPD) [50]. An alternative fusion that occurs very rarely in CML cells is the fusion between FGFR1 and BCR [51]. This fusion also shows high transforming activity and potentiates the growth factor independence of malignant progenitor cells. Therefore, in both malignancies, dysregulated FGFR1 and 3 provide important molecular targets of kinase-specific small molecule inhibitors which could mediate the decreased viability and growth arrest of cancer cells [51, 52].

Work by our laboratory has also demonstrated that the expression of the myeloma-associated oncogene FGFR3 is increased in cancer cells from patients with CML and that this expression is decreased following the successful transplantation of stem cells [53]. Furthermore, we determined that highly enriched CML CD34+ cells express significantly increased levels of FGFR3 compared to either their normal counterparts or differentiated CML CD34- progeny. Interestingly, this increased level of FGFR3 was efficiently inhibited by the BCR-ABL tyrosine kinase inhibitor, imatinib [54]. This finding again suggests the involvement of FGFR3 in malignant hematopoiesis.

Together, these findings related to cancer stem cells have clear implications for the involvement of the FGF signaling pathway in the genesis of hematological malignancies. However, the practical beneficial effects of potential therapies can only result from a greater understanding of the molecular biology of the individual components of the FGF pathway. Such an understanding can even be achieved by investigating normal stem cells.

4 Common elements and functional relationships of the FGF-2 pathway in hESCs and leukemic stem/progenitor cells

As described in this review, hESCs maintained in long-term cultures and hematopoietic stem/progenitor cells obtained directly from patients suffering from leukemia show similarly increased expression of endogenous FGF-2 that is either exported out of cells or targeted into the nucleus. Both types of stem cell are well-equipped to accept and transmit exogenous FGF signals and to execute an appropriate cellular response. Thus, the functional involvement of the FGF signaling pathway in the regulation of stem cell fate is likely very important.

Our studies suggest that FGF-2 signaling participates in the self-renewal and maintenance of an undifferentiated state of hESCs, whereas dysregulated expression of FGF-2 and its receptors in cancer cells increases their transforming activity. It is likely that the expression of endogenous FGF-2 in undifferentiated hESCs may reflect adaptation to long-term culture in vitro followed by selection and forced activation of those signaling pathways that are advantageous for survival and the maintenance of the undifferentiated phenotype. Utilizing the cancer stem cell model, the dysregulation of the FGF-2 signaling pathway in malignant cells may resemble such a “culture adaptation” of abnormal cells in vivo, which would provide a growth advantage and self-renewal stimulation to cancer stem cells. As a consequence, this “adaptation”, which is similarly manifested in both stem cell types, provides an important advantage for maintaining long-term culture of undifferentiated hESCs and an important disadvantage for the selective elimination of cancer stem cells by various apoptosis-inducing drugs. Therefore, it is tempting to speculate that the inhibition of the FGF signaling pathway may induce similar differentiation effects in both stem cell types.

5 Future directions

Accumulation of endogenous FGF-2 and its autocrine/intracrine activities important for the maintenance of the undifferentiated phenotype are common events in both embryonic stem cells and leukemic stem/progenitor cells. Therefore, it will be important to specifically and effectively inhibit nuclear-targeted FGF-2 expression (e.g. by short hairpin RNAs) to assess accurately the importance of this signaling pathway on cellular response and cell fates. Additionally, it will be important to examine the molecular mechanisms by which the expression of endogenous FGF-2 becomes downregulated upon differentiation. The manipulation of one or several components of the FGF signaling pathway may have critical implications in the controlled differentiation of hESCs which may be important for the establishment of certain cell therapies and, in the case of leukemic stem/progenitor cells, the development of a targeted approach for the treatment of leukemia.


Supported by Grant Agency of the Czech Republic (301/03/1122), the Academy of Sciences of the Czech Republic (AV0Z50390512), and the Ministry of Education, Youth, and Sports (1M0021620803).