Fibroblast Growth Factor Signaling during Early Vertebrate Development
http://www.100md.com
《内分泌研究杂志》
Abstract
Fibroblast growth factors (FGFs) have been implicated in diverse cellular processes including apoptosis, cell survival, chemotaxis, cell adhesion, migration, differentiation, and proliferation. This review presents our current understanding on the roles of FGF signaling, the pathways employed, and its regulation. We focus on FGF signaling during early embryonic processes in vertebrates, such as induction and patterning of the three germ layers as well as its function in the control of morphogenetic movements.
I. Introduction
II. FGFs and FGF Receptors (FGFRs)
III. FGFR Signal Transduction
A. Ras/MAPK pathway
B. PLC/Ca2+ pathway
C. PI3 kinase/Akt pathway
IV. The Biological Activities of FGFs
A. FGF signaling pathway in mesoderm formation and gastrulation movements
B. Chemotactic action of FGFs in morphogenetic movements
C. FGF signaling in neural induction and AP patterning
D. FGF signaling in endoderm formation
V. Regulation of FGF Signaling
A. Regulation by HSPGs
B. Transmembrane regulators
C. Intracellular regulators
VI. Concluding Remarks
I. Introduction
Top
Abstract
I. Introduction
II. FGFs and FGF...
III. FGFR Signal Transduction
IV. The Biological Activities...
V. Regulation of FGF...
VI. Concluding Remarks
References
FIBROBLAST GROWTH FACTORS (FGFs) constitute a large family of polypeptide growth factors found in a variety of multicellular organisms, including invertebrates. The first member of this family, FGF2, was identified by its ability to stimulate proliferation of mouse 3T3 fibroblasts (1); however, the function of FGFs is not restricted to cell growth. Instead, FGFs are involved in diverse cellular processes including chemotaxis, cell migration, differentiation, cell survival, and apoptosis. The common feature of FGFs is that they are structurally related and generally signal through receptor tyrosine kinases. FGFs play an important role in embryonic development in invertebrates and vertebrates. This review focuses on the role of FGF signaling during early vertebrate development. Excellent reviews dealing with the role of FGFs in Drosophila (2, 3, 4), Caenorhabditis elegans (5, 6), vertebrate limb development (7, 8, 9), central nervous system development (10, 11), skeleton formation (12), and cancer (13, 14, 15) are available.
II. FGFs and FGF Receptors (FGFRs)
Top
Abstract
I. Introduction
II. FGFs and FGF...
III. FGFR Signal Transduction
IV. The Biological Activities...
V. Regulation of FGF...
VI. Concluding Remarks
References
The human FGF protein family consists of 22 members that share a high affinity for heparin as well as a high-sequence homology within a central core domain of 120 amino acids (16), which interacts with the FGFR (17, 18, 19) (Fig. 1A). FGFs can be classified into subgroups according to structure, biochemical properties, and expression (for reviews see Refs.16 and 20). For example, members of the FGF8 subfamily (FGF8, FGF17, FGF18) share 70–80% amino acid sequence identity and similar receptor-binding properties and have overlapping expression patterns. FGFs appear to be diffusible (21), and they are able to act in a dose-dependent fashion, e.g. to induce different marker genes at different concentrations (22, 23). This raises the question of whether FGFs act physiologically as morphogens. Indeed, in neural progenitors, graded FGF signaling appears to be translated into a distinct motor neuron Hox pattern (24).
FGFs induce their biological responses by binding to and activating FGFRs, a subfamily of cell surface receptor tyrosine kinases (RTKs). The vertebrate Fgfr gene family consists of four highly related genes, Fgfr1–4 (25). These genes code for single spanning transmembrane proteins with an extracellular ligand-binding region and an intracellular domain harboring tyrosine kinase activity. The extracellular region is composed of Ig-like domains that are required for FGF binding. The Ig-like domains also regulate binding affinity and ligand specificity (Fig. 1B). Located between Ig-like domains I and II is a stretch of acidic amino acids (acidic box domain) followed by a heparin-binding region and a cell adhesion homology domain. These domains are required for the interaction of the receptor with components of the extracellular matrix, in particular heparan sulfate (HS) proteoglycans (HSPGs), and cell adhesion molecules (CAMs). The intracellular part of the receptor includes the juxtamembrane domain, the split tyrosine kinase domain, and a short carboxy-terminal tail. In addition to the enzymatic activity, the intracellular domain harbors protein binding and phosphorylation sites (including protein kinase C and FRS2 sites) as well as several autophosphorylation sites that interact with intracellular substrates.
Alternative splicing of Fgfr transcripts generates a diversity of FGFR isoforms (25, 26). In vitro assays show that different isoforms have distinct FGF-binding specificities (27, 28). The tissue-specific alternative splicing encompassing the carboxy-terminal half of Ig domain III strongly affects ligand-receptor binding specificity (29, 30, 31). In vivo, FGFR-FGF complex assembly and signaling is further regulated by the spatial and temporal expression of endogenous HSPGs (32, 33). HSPGs are required to activate effectively the FGFR and determine the interaction between specific FGF-FGFR pairs (see Section V.A.).
III. FGFR Signal Transduction
FGFRs, like other receptor tyrosine kinases, transmit extracellular signals after ligand binding to various cytoplasmic signal transduction pathways through tyrosine phosphorylation. Binding of FGFs causes receptor dimerization and triggers tyrosine kinase activation leading to autophosphorylation of the intracellular domain (34). Tyrosine autophosphorylation controls the protein tyrosine kinase activity of the receptor but also serves as a mechanism for assembly and recruitment of signaling complexes (35). These phosphorylated tyrosines function as binding sites for Src homology 2 and phosphotyrosine binding domains of signaling proteins, resulting in their phosphorylation and activation (36, 37). A subset of Src homology 2-containing proteins such as Src-kinase and phospholipase C (PLC) possesses intrinsic catalytic activities, whereas others are adapter proteins. FGF signal transduction, as analyzed during early embryonic development, can proceed via three main pathways described below (Fig. 2).
A. Ras/MAPK pathway
The most common pathway employed by FGFs is the MAPK pathway. This involves the lipid-anchored docking protein FRS2 (also called SNT1) that constitutively binds FGFR1 even without receptor activation (38, 39, 40). Several groups have demonstrated the importance of FRS2 in FGFR1-mediated signal transduction during embryonic development (41, 42, 43). After activation of the FGFR, tyrosine-phosphorylated FRS2 functions as a site for coordinated assembly of a multiprotein complex activating and controlling the Ras-MAPK signaling cascade and the phosphatidylinositol 3 (PI3)-kinase/Akt pathway. The FRS2 tyrosine phosphorylation sites are recognized and bound by the adapter protein Grb2 and the protein tyrosine phosphatase (PTP) Shp2 (39, 44). Grb2 forms a complex with the guanine nucleotide exchange factor Son of sevenless (SOS) via its SH3 domain. Translocation of this complex to the plasma membrane by binding to phosphorylated FRS2 allows SOS to activate Ras by GTP exchange due to its close proximity to membrane-bound Ras. Once in the active GTP-bound state, Ras interacts with several effector proteins, including Raf leading to the activation of the MAPK signaling cascade. This cascade leads to phosphorylation of target transcription factors, such as c-myc, AP1, and members of the Ets family of transcription factors (reviewed in Ref.45).
B. PLC/Ca2+ pathway
The PLC/Ca2+ pathway involves binding of PLC to phosphorylated tyrosine 766 of FGFR1 (46, 47). Upon activation, PLC hydrolyzes phosphatidylinositol-4,5-diphosphate to form two second messengers, inositol-1,4,5-triphosphate and diacylglycerol. Diacylglycerol is an activator of protein kinase C, whereas inositol-1,4,5-triphosphate stimulates the release of intracellular Ca2+. This cascade has been implicated in the FGF2-stimulated neurite outgrowth (48, 49) and in the caudalization of neural tissue by FGFR4 in Xenopus (50).
C. PI3 kinase/Akt pathway
The PI3 kinase/Akt pathway can be activated by three mechanisms after FGFR activation (Fig. 2), and the phospholipids thereby generated regulate directly or indirectly the activity of target proteins such as Akt/PKB.
Among other processes, the PI3 kinase signaling branch is involved during Xenopus mesoderm induction acting in parallel to the Ras/MAPK pathway. Overexpression of a dominant negative form of the PI3 kinase-regulatory subunit p85 interferes with Xenopus mesoderm formation. Conversely, coexpression of activated forms of MAPK and PI3 kinase leads to synergistic mesoderm induction (51).
IV. The Biological Activities of FGFs
Ealy developmental processes during gastrulation stages of Xenopus, zebrafish, chicken, and mouse include mesoderm formation and gastrulation movements, neural induction and AP patterning, and endoderm formation (Fig. 3). FGF signaling plays important roles in early vertebrate development during these processes as discussed below and summarized in Table 1. Genes that function in FGF signaling and for which loss-of-function analysis revealed an early embryonic phenotype are summarized in Table 2.
A. FGF signaling pathway in mesoderm formation and gastrulation movements
Induction and patterning of mesoderm is one of the earliest events during vertebrate body axis formation. Although TGF-? signaling plays a pivotal role in mesoderm induction and patterning and has received most attention in this context recently, FGF signaling is also essential. Indeed, the first identified mesoderm inducer was basic FGF (52), and this represents an evolutionarily conserved role, because FGF signaling is required for mesoderm induction in the primitive chordate Ciona (53). In Xenopus, mouse, and zebrafish, disruption of FGF signaling by Fgfr knock out or by overexpression of a dominant negative FGFR1 strongly affects body axis formation (54, 55, 56, 57). In these embryos, phenotypic changes are observed mostly in posterior regions, such as defects of trunk and tail structures. Despite some organism-specific differences, two functions for FGFs are conserved. First, FGFs control the specification and maintenance of mesoderm by regulation of T box transcription factors in Xenopus (58, 59), mouse (60, 61), and zebrafish (57, 62, 63). Second, they have a direct and/or indirect role in morphogenetic movements during gastrulation.
The effect of FGFs on mesoderm formation was first demonstrated and most extensively studied in Xenopus (64). Since the first studies that discovered the mesoderm-inducing ability of FGFs (52, 65, 66), many components of the FGF signaling pathway were shown in Xenopus to be required for mesoderm development. These components include FGFRs (54, 67), HSPGs (68), adaptor proteins FRS2/SNT-1, Nck, and Grb2 (42, 43, 69), Src-kinase Laloo (70), PTPs (71, 72), Ras, Raf, MAPK kinase, ERK (73, 74, 75, 76), and the transcription factors AP1 and Ets2 (77, 78). Their inhibition blocks mesoderm formation and induces posterior and gastrulation defects. Conversely, in gain-of-function experiments they induce mesodermal markers and phenocopy FGF overexpression. The role of FGF signaling during mesoderm formation is that of a competence factor. It is required for cells to respond to TGF-?-like mesoderm inducers, because activin-mediated mesoderm induction in Xenopus animal caps, for example, is blocked by a dominant negative FGFR1 (79, 80, 81). Similarly, the trunk-inducing activity of dorsal mesoderm, which depends on Nodal signaling, is inhibited by a dominant negative FGFR1 (82). The ability of FGFs to regulate Nodal signaling may be due to phosphorylation of Smad2 by MAPKs (83).
FGF signaling needs to continue during gastrulation for mesoderm maintenance as shown by transgenic Xenopus embryos overexpressing a dominant negative FGFR1 (84). One of the FGF mesodermal targets is the T box transcription factor brachyury (Xbra), required for posterior mesoderm and axis formations in mouse, zebrafish, and Xenopus (58, 85, 86, 87). Xbra and eFGF form an autoregulatory loop in which Xbra induces eFGF and vice versa (88, 89). eFGF is also required for XmyoD expression in the myogenic cell lineage of Xenopus. eFGF is a candidate factor mediating the community effect, whereby cells differentiate into muscle once a critical cell mass is reached, due to the accumulation of diffusible factors (90, 91).
In Xenopus embryos, FGF signaling has both direct and indirect roles on gastrulation movements. Nutt et al. (92) provided evidence for a direct role of FGF signaling in cell movements during convergence extension. They identified and characterized XSprouty2 as an intracellular antagonist of FGF-dependent calcium signaling. Overexpression of XSprouty2 inhibits convergent extension movements without affecting MAPK activity, mesoderm induction, and patterning. FGF signaling also affects gastrulation movements indirectly via regulation of Xbra induction and maintenance. Xbra acts as a switch promoting convergent extension and actively repressing cell migration by inhibiting adhesion of migrating cells to fibronectin (93, 94). Also, Xbra directly activates the expression of Xwnt11 and prickle (95, 96), which regulate convergent extension movements via the noncanonical Wnt pathway (95, 96, 97).
In zebrafish, as in Xenopus, FGF signaling acts as competence factor to control mesoderm induction by TGF-?s, such as activin and Nodal (63, 98). Through the transcriptional regulation of three mesodermal T box transcription factors ntl (86), tbx6 (99), and spadetail (spt), a regulator of mesoderm formation in the trunk (100), FGF signaling regulates tail and trunk development in zebrafish (57, 62, 101, 102). As in Xenopus and mouse, the T box genes also serve as an important link between cell fate and morphogenesis. One important role of spt is the regulation of convergence movements of paraxial mesoderm during gastrulation (103) mediated by paraxial protocadherin (papc) (104). Additionally, FGFs may affect gastrulation movements via snail. Consistent with results in mouse (61), overexpression of eFGF causes abnormal widespread expression of snail1 throughout the hypoblast of the late zebrafish gastrula (57).
In addition to its effects on anterior-posterior (AP) patterning of the embryo and gastrulation movements, FGF signaling affects dorsoventral patterning of the early zebrafish embryo. Activation of the FGF8/FGFR1/Ras signaling pathway leads to the expansion of dorsolateral derivatives at the expense of ventral and posterior domains by an early inhibition of ventral bone morphogenetic protein (BMP) expression (105, 106, 107). It was therefore suggested that FGFs act in concert with secreted BMP antagonists (Spemann organizer factors Follistatin, Noggin, and Chordin) in dorsoventral patterning (105).
FGFR1 knockouts gave the first evidence that FGF signaling is also essential for proper gastrulation in the mouse (55, 56). Fgfr1–/– mutant embryos die late in gastrulation, exhibiting defects in cell migration, cell fate specification, and patterning. Chimera analysis indicate that Fgfr1-deficient cells are impaired in epithelial to mesenchymal transition (EMT) and, as a consequence, fail to cross the primitive streak (108). The mutant cells accumulate in the primitive streak and overexpress E-cadherin, and this expression is accompanied by the down-regulation of mSnail. mSnail is a key mediator of EMT in development and cancer by repressing E-cadherin (reviewed in Ref.109). Thus, Fgfr1–/– cells may fail to migrate properly because they maintain high levels of E-cadherin in the streak as a result of mSnail down-regulation. The regulation of E-cadherin by FGFR1 via mSnail also provides a molecular link between FGF and Wnt signaling pathways because ectopic E-cadherin expression in Fgfr1–/– embryos attenuates Wnt signaling by sequestering cytoplasmic ?-catenin to the plasma membrane (61).
Similar to Fgfr mutants, Fgf8–/– embryos display severe gastrulation defects, lacking mesodermal and endodermal tissues. However, epiblast cells move into the streak and undergo EMT in the Fgf8–/– embryos, but fail to move away from the streak (60). This indicates that FGF family members are involved in different stages of mesoderm formation from EMT to subsequent migration and that these two processes are independent.
B. Chemotactic action of FGFs in morphogenetic movements
The activity of FGF signaling during gastrulation movements in Xenopus, zebrafish, and mouse has been mostly linked to the regulation of target gene expression, which in turn regulates the migratory behavior of gastrulating cells. However, there is increasing evidence for a direct chemotactic response of migrating cells in response to FGFs both in vertebrates and invertebrates (3, 5). FGF2 and FGF8 function as potent chemoattractants during migration of mesencephalic neural crest cells (110), and FGF2 and FGF4 attract mesenchymal cells during limb bud development in the mouse (111, 112). FGF10, a key regulator of lung-branching morphogenesis, exerts a powerful chemoattractive effect to direct lung epithelial buds to their destinations during development (113). Recently, an elegant study by Yang et al. (114) showed that FGFs act as chemotactic signals to directly coordinate cell movements during gastrulation in the chick embryo. They analyzed the trajectories of cells emerging from different AP positions from the fully extended primitive streak and found that cells show different movement patterns depending on their position along the streak. The cells are guided by signals from the surrounding tissue, and beads soaked in FGFs alter their trajectories. FGF4 acts as a chemoattractant and FGF8b acts as a repellent for migrating streak cells, whereas FGF2 does not elicit a strong chemotactic response (114). In line with the expression of Fgf8 in the primitive streak and Fgf4 in Hensen’s node and forming head process, the movement away from the streak may be the result of a chemorepellent effect of FGF8, whereas the inward movement of cells results from attraction by FGF4 secreted by cells in the head process (114).
C. FGF signaling in neural induction and AP patterning
Neural induction is the process whereby naive ectodermal cells are instructed to adopt a neural fate. Several findings, mainly from experiments in Xenopus, led to the default model of neural induction (reviewed in Ref.115): BMP signals in the early embryo instruct ectoderm cells to form epidermis whereas ectodermal cells will adopt a neural fate (their default state) in the absence of cell-cell signaling. In this model, neural induction is mediated during gastrulation by diffusible BMP antagonists released by Spemann’s organizer. However, this model is certainly oversimplified, and increasing evidence suggests a pivotal role of FGF in neural induction and places the onset of neural induction before gastrulation. FGFs act as neural inducers in lower chordates such as ascidians (116) and planarians (117), indicating that this role is evolutionarily conserved.
The best case for FGFs as neural inducer is the chick embryo. Here, induction of neural tissue precedes the formation of the organizer (Hensen’s node) (118, 119). FGFs are sufficient to induce preneural markers (118, 120, 121, 122), and inhibition of FGF signaling using various inhibitors blocks neural induction by Hensen’s node (118). FGF signaling functions by repressing BMP expression because Fgf3 overexpression leads to a down-regulation of BMP mRNA and the subsequent induction of neural tissue. Conversely, when FGF signaling is inhibited, BMP mRNA expression is maintained and neural fate is blocked (119). A recent study by Sheng et al. (123) suggests a mechanism by which FGFs may block BMP signaling during neural induction in chick. FGFs induce the transcriptional activator Churchhill (ChCh), which is required for the expression of Sip1 (Smad-interacting protein 1) in the neural plate. Sip1 binds to and represses Smad1/5 transcriptional activation of target genes, thereby blocking BMP signaling. ChCh/Sip1 sensitizes cells to neural-inducing signals and represses mesodermal genes (123). However, FGFs promote neural fates also independent of BMP inhibition because beads soaked with the BMP antagonists Noggin or Chordin are insufficient for neural induction under most experimental conditions (119, 124).
In Xenopus, the role of FGF signaling in neural induction and patterning has been studied extensively. However, conflicting results were obtained concerning the role of FGFs in neural induction. Some reports showed that FGFs could directly induce neural tissue in animal cap explants. However, in most cases a sensitized background was used in which BMP signaling was partially attenuated (23, 125, 126). In addition, inhibition of FGF signaling by dominant negative FGFR1 blocks neural induction by Noggin or Spemann organizer (127), Chordin (128), or notochord tissue (129). On the other hand, several studies using dominant negative FGFR1 obtained contradictory results showing that neural tissue forms apparently even when FGF signaling is inhibited (84, 126, 129, 130, 131). However, these experiments were carried out using the dominant negative form of the FGFR1, but FGF signaling required for anterior neural induction may be mediated through FGFRs other than FGFR1. Indeed, Hongo et al. (132) showed that anterior neural development is more sensitive to inhibition of FGFR4a than to FGFR1. Likewise, FGF8 stimulates neuronal differentiation preferentially through FGFR4a (133), and FGF signaling through FGFR1, but not FGFR4a, is required for neural crest formation (134). The two receptors preferentially activate different downstream pathways, suggesting that they may have distinct functions during development. FGFR4a can signal through PLC in neural induction (50) whereas FGFR1 preferentially activates the Ras-MAPK pathway, involved in posteriorization of already neuralized ectoderm (50, 126).
In addition to their early effects in neural induction, FGFs are potent posteriorizing agents during AP patterning of the neural plate (135). They are able to convert anterior neural tissue to more posterior neural cell types (50, 125, 126, 131, 136). For example, when FGF-soaked beads are placed in the anterior neural plate, they inhibit anterior neural marker expression (137, 138), and this may involve a cross-talk with the Wnt pathway (130, 139, 140, 141). Similarly, overexpression of FGF3 in zebrafish expands posterior neural tissue (142). Conversely, dominant negative FGFR1 or dominant negative Ras block induction of posterior neural markers (126, 131).
Patterning of neuroectoderm and mesoderm is linked to the patterned expression of Hox genes along the AP axis of the embryo (143). There is strong evidence from studies in mouse, chick, and Xenopus that FGF signaling regulates early AP patterning via the regulation of Hox genes. Ectopic application or overexpression of FGFs results in the up-regulation of posterior members of the Hox gene family (137, 144, 145, 146, 147). Conversely, overexpression of dominant negative FGFR1 represses posterior Hox gene expression (137, 148). Caudals are immediate-early response genes to FGF signaling and mediate Hox regulation by FGFs (137, 147, 149, 150). A recent study demonstrates that graded FGF signaling is translated into a distinct motor neuron Hox pattern (24).
An alternative mode of action for FGFs as posteriorizing factor was described for chick neural tube in which they mediate the maintenance of neural stem cells at Hensen’s node. Therefore, FGFs might act to prolong the time window during which cells are exposed to other posteriorizing factors, such as retinoic acid or Wnts, thereby promoting the continuous development of the posterior nervous system (151).
D. FGF signaling in endoderm formation
Although endodermal organ development is widely studied, little is known about how endoderm is initially formed and regionally specified in vertebrates. Accumulating data, especially in the mouse, point toward a function of FGF signaling in early endoderm maintenance and patterning. In chimeric embryos, Fgfr1–/– cells are unable to populate endodermal derivatives (61, 108). Similarly, embryoid bodies derived from Fgfr1–/– embryonic stem cells lack visceral endoderm. This is associated with a lowered expression of the endodermal marker -fetoprotein. However, some endoderm is eventually formed, as indicated by the unchanged expression of GATA4 (152). Endoderm deficiency was also obtained in embryoid bodies overexpressing a dominant negative FGFR2, where inhibition of three endodermal markers (-fetoprotein, HNF4, and Evx1) was observed (153). Fgf8–/– embryos develop no embryonic mesoderm- and endoderm-derived tissue, and this is accompanied by inhibition of Fgf4 expression (60). Fgf4–/– embryos die shortly after implantation and fail to make detectable mesoderm and endoderm (154), a phenotype similar to the targeted disruption of Fgfr2 (155, 156). A study by Wells and Melton (157) further supports the role of FGF4 in the specification of primitive endoderm in the mouse because recombinant FGF4 affects the patterning in isolated endoderm in a concentration-dependent manner. Other analyzed FGFs (aFGF, FGF2, 5, 8b) or other growth factors (TGF, TGF?, epidermal growth factor, IGF) had no activity in this assay (157). In conclusion, FGF signaling appears not to be required for induction but rather for the maintenance and patterning of endoderm, because some endodermal markers continue to be expressed even when FGF signaling is inhibited.
An important caveat in analyzing endoderm development is the potent influence of mesoderm on its differentiation and patterning (158, 159). Thus, factors may regulate endoderm via an effect on mesoderm unless this is rigorously controlled. In Xenopus this can be achieved by explanting vegetal caps, which are devoid of mesoderm and only develop into endoderm. FGF signaling-dependent MAPK activity is detectable in the prospective endodermal cells of the Xenopus blastula embryo (160). However, studies with such vegetal caps gave conflicting results regarding the role of FGF signaling in endoderm. Expression of the dorsal endodermal marker Xlhbox8 and the pan-endodermal marker Mix1 can be blocked by injection of dominant negative FGFR1 (160, 161). In contrast, other studies failed to observe the requirement of FGF signaling for Xlhbox8 expression but rather report an inhibition of Xlhbox8 after FGF treatment (162, 163). Modulation of FGF signaling does not affect the expression of pan-endodermal markers in the majority of vegetal explants experiments, and FGF does not induce a number of endodermal genes in animal caps (128, 164). It is therefore suggested that FGF signaling is not essential for endoderm formation. However, defined levels of FGF activity may be required for endodermal patterning.
In summary, FGF signaling plays simultaneous roles during gastrulation to induce different cell fates and morphogenetic movements, raising the question of how the specific response to FGFs in different germ layers is mediated. On the one hand, different FGFs may mediate neural or mesodermal induction (e.g. Ref.133). On the other hand, two recent papers show that in the ascidian Ciona and in chick the competence of cells to respond to FGF signals with neural induction is regulated by the neural-specific transcription factors GATA and ChCh, respectively (123, 165). Thus, in addition to qualitative differences among different FGF family members, the spatial and temporal restricted expression of cofactors provides another mechanism to separate the different functions of FGF during gastrulation.
V. Regulation of FGF Signaling
Developmental signaling that involves intercellular communication and regulates cell fate must be precisely regulated regarding timing, duration, and spread of such a signal. Precision and robustness of these features can be achieved by both positive and negative feedback loops (reviewed in Ref.166). Negative feedback stabilizes a system when confronted with environmental fluctuations (167). Furthermore, it can limit the absolute level or duration of a signal and can lead to signal oscillations (168). FGF signaling is regulated at many levels in the extracellular space and within the cell. Many of these regulatory mechanisms are not specific for FGF signaling but rather activate or attenuate RTK signaling in general (169). Yet, a growing number of proteins were identified in recent years that specifically regulate the activity of FGFs or signaling pathways activated through FGFRs. Interestingly, many of these regulatory components are themselves regulated by FGF signaling and are tightly coexpressed with Fgfs, forming a synexpression group (107, 168, 170). Synexpression groups are genetic modules in which expression of different genes are closely correlated and whose members function in a common pathway (171). Most of the FGF synexpression group members inhibit FGF signaling and thereby establish negative feedback loops.
A. Regulation by HSPGs
An important feature of FGF biology involves the extracellular interaction between FGF and heparin or HSPG (32). HSPGs are sulfated glycosaminoglycans covalently bound to a core protein. Enormous structural heterogeneity can be generated through specific heparin sulfate chain modifications during their biosynthesis, as well as from the diverse nature of their core proteins (reviewed in Ref.172). HSPGs are required for FGFs to effectively activate the FGFR and determine interaction between individual FGF family members with certain FGFRs (173, 174). The crystal structure of ternary FGF-heparin-FGFR complexes have been determined; heparin bridges two FGFRs to form FGFR dimers, which bind two FGFs (175). In an elegant study, a ligand and carbohydrate engagement assay was employed to examine the spatio-temporal changes in HS to promote ternary complex formation with FGFs and FGFRs. These changes depend on glycosyltransferases and sugar sulfotransferases. The analysis reveals that synthesis of HS domains is tightly regulated throughout development, and different FGF-FGFR pairs have unique HS-binding requirements (33). Other than promoting FGF binding to its receptors, HSPGs protect the FGFs from thermal denaturation and proteolysis (176). In addition, these interactions limit FGF diffusion and release into interstitial spaces, thereby regulating FGF activity within a tissue (177, 178). The importance of HSPGs and enzymes involved in HS biosynthesis for FGF signaling during development has been demonstrated in vivo in a number of studies (68, 179, 180, 181). For example, mice mutated in UDP-glucose dehydrogenase (Ugdh), one of the key enzymes for HS biosynthesis, fail to gastrulate properly and show defects of mesoderm and endoderm migration similar to FGF pathway mutants (181). In the Ugdh mutant embryos, FGF signaling is specifically blocked whereas signaling by Wnt3 and Nodal appears unaffected. Ugdh mutant Drosophila embryos also have phenotypes similar to FGFR mutants (180), indicating an evolutionary conserved requirement for HSPGs in FGF signaling.
The nature of the proteoglycan core protein essential for FGF signal transduction has not been identified because single knock-out mice mutant for diverse core proteins (including Syndecan1, -3, and -4; Glypican2 and -3; and Perlecan) survive past gastrulation.
B. Transmembrane regulators
An emerging theme is the regulation of FGF signaling by transmembrane modulators. The transmembrane protein Sef is a member of the FGF synexpression group in zebrafish and mouse, expressed in the characteristic pattern of FGF8 and regulated by FGF signaling (107, 170, 182). Overexpression studies in zebrafish embryos show that Sef inhibits FGF signaling via interaction with FGFR1 and -4a (170) and that it functions at the level or downstream of MAPK kinase (107). Interfering with Sef function by antisense morpholino oligonucleotide (MO) injection induces phenotypic changes reminiscent of embryos dorsalized by ectopic expression of Fgf8 (107, 170). The molecular mechanism of Sef action is controversial. Sef may inhibit FGF signaling at the level of FGFR (183, 213). Overexpression of mouse Sef inhibits tyrosine-phosphorylation of FGFR1 and FRS2 and inhibits activation of the MAP kinase and Akt pathway, possibly due to reduced recruitment of Grb2 to FRS2 (183). In contrast, other reports argue Sef acts downstream of or at the level of MEK (107, 214, 215, 216). Torii et al. (216) showed that hSef acts to restrict ERK activity to the cytoplasm. hSef binds to activated forms of MEK and inhibits the dissociation of the MEK-ERK complexes. Thereby, nuclear translocation of ERK is blocked without inhibiting its activity in the cytoplasm (216).
In Xenopus, XFLRT3, a member of a family of leucine-rich repeat transmembrane proteins, is another component of the FGF signaling pathway (184). Its expression is very similar to that of the FGF8 synexpression group, and it is regulated by FGFs. When overexpressed, FLRT3 phenocopies FGF signaling and activates the MAPK pathway. Inhibition of XFLRTs by MO injection interferes with FGF signaling. XFLRT3 binds to FGFRs and modulates FGFR signaling by an unknown mechanism. In cell culture, FLRT3 promotes neurite outgrowth of primary cerebella granule neurons (217). Since FGF promotes axon outgrowth (218, 219, 220), it is possible that potentiation of FGF signaling by FLRT3 underlies its effects on neurite growth.
Finally, FGFRs interact with CAMs such as N-CAM, N-cadherin, and L1 via their extracellular domains. In neurons, N-CAM binding to FGFR1 activates the PLC/Ca2+ pathway (49, 185, 186). In tumor ?-cells and L cells, Cavallaro et al. (187) show a direct association of FGFR4 and N-CAM. This interaction induces the formation of a multiprotein signaling complex that mediates neurite outgrowth and cell-matrix adhesion (187).
C. Intracellular regulators
1. PTPs.
It is well established that RTK signaling is controlled by PTPs, which act as positive or negative regulators on RTK signaling (reviewed in Refs.188 and 189). The dual-specificity PTP MAPK phosphatase 3 (MKP3, also called Pyst1) antagonizes the MAPK pathway via ERK1/2 inactivation. In vertebrates, its expression is very similar to that of the FGF8 synexpression group, and it is regulated by FGF8 (190, 191). In the developing chick limb MKP3 mediates the antiapoptotic signaling of FGF8 (191), and manipulation of MKP3 activity by short interfering RNA inhibition or overexpression disrupts limb morphology and outgrowth (190, 191). In contrast, the PTP SHP-2 and Xenopus low-molecular weight PTP1 (XLPTP1) act as positive regulators of FGF signaling upstream of Ras during Xenopus embryogenesis (71, 72). Inhibition of XLPTP1 by antisense MOs blocks FGF-induced MAPK activation, leading to a shortened AP axis (72).
2. Sprouty protein.
The feedback inhibitor Sprouty was identified in Drosophila as a negative regulator of FGF signaling during tracheal development (192). Four mammalian Sprouty proteins (Spry1–4) and three Spreds (Sprouty-related EVH1 domain proteins) that share a highly conserved cysteine-rich domain at the carboxy terminus have been identified. This domain is critical to target them to the plasma membrane and to inhibit the MAPK pathway (193). Sprouty2 and -4 are members of the FGF8 synexpression group and are regulated by FGF signaling (106, 194, 195, 196). Sprouty proteins interfere with FGF signaling through several mechanisms, depending on the cellular context and/or stimulation. Different Sprouty proteins exhibit different activities and have different interaction partners, including but not limited to Grb2 (197) and Raf1 (198) (reviewed in Ref.199).
Gain- and loss-of-function experiments confirm the function of Sprouty proteins as intracellular antagonists of FGF signaling. Reduction of mSprouty2 expression during mouse lung development using antisense oligonucleotides established Sprouty as an inhibitor of branching morphogenesis, as in Drosophila (192, 200). Overexpressing Sprouty during chick limb development causes a reduction in limb bud outgrowth consistent with reduced FGF signaling (195). Furthermore, Fürthauer et al. (106) showed that inhibiting Sprouty4 in zebrafish by MO injection elicits phenotypes similar to that obtained after up-regulation of Ffg8. Sprouty4 is required for dorsoventral mesoderm patterning and for head development (106). In Xenopus embryos, XSprouty2 regulates gastrulation movements by a Ca2+-dependent pathway (92).
VI. Concluding Remarks
Substantial progress has been made in recent years in the understanding of the role of FGF signaling during early vertebrate development. FGFs play an important role in regulating cell fate and specification of all three germ layers. The role of FGF signaling during mesoderm formation and neural induction is now well documented and appears evolutionarily conserved in chordates. A surprising new role is the ability of FGFs to orchestrate gastrulation movements by acting as chemoattractants and repellents. This probably involves direct regulation of the cytoskeleton.
FGF downstream pathways and the components employed are characterized in some detail. A growing number of these components are themselves tightly regulated by FGFs and form an FGF8 synexpression group. The FGF8 synexpression group is an autoregulatory genetic module, conserved from frogs to mice, which is employed, for example, in mesoderm formation and neural patterning during early development. Part of this group are the novel transmembrane regulators Sef and FLRT3, which bind to and modulate FGFRs, but their precise function needs further investigation. It has also become clear that different ligand receptor pairs may preferentially couple to different pathways but how this is regulated is unclear. In the future it will be important to assign the pathways employed by FGFs in different cells and define the determinants for signaling specificity. One emerging class of determinants are HSPGs and the enzymes involved in their synthesis. Two classes of downstream target genes, which mediate FGF signaling are T-box and Hox genes, and this is evolutionary conserved in all vertebrates. Furthermore, there is substantial cross-talk between FGFs with TGF-?- and Wnt signaling pathways. We are only beginning to understand the mechanisms of these cross-regulations, which are at the heart of the ability of FGFs to act as competence factor, e.g. during mesoderm induction by Nodals. Finally, do FGFs act as morphogens? They have many characteristics of morphogens, including their ability to induce marker genes in a dose-dependent manner. However, the proof that FGFs act directly in a concentration-dependent fashion rather than by relay and the demonstration of a physiological requirement for this is missing.
Acknowledgments
We thank Dr. Sonia Pinho for advice and critical reading of the manuscript.
Footnotes
First Published Online October 26, 2004
Abbreviations: AP, Anterior-posterior; BMP, bone morphogenetic protein; CAM, cell adhesion molecule; EMT, epithelial to mesenchymal transition; FGF, fibroblast growth factor; FGFR, FGF receptor; HS, heparan sulfate; HSPG, heparan sulfate proteoglycan; MO, morpholino oligonucleotide; PI3, phosphatidylinositol-3; PLC, phospholipase C; PTP, protein tyrosine phosphatase; RTK, receptor tyrosine kinase; SOS, Son of sevenless.
References
Yamaguchi TP, Harpal K, Henkemeyer M, Rossant J 1994 fgfr-1 is required for embryonic growth and mesodermal patterning during mouse gastrulation. Genes Dev 8:3032–304
Griffin K, Patient R, Holder N 1995 Analysis of FGF function in normal and no tail zebrafish embryos reveals separate mechanisms for formation of the trunk and the tail. Development 121:2983–2994
Smith JC, Price BM, Green JB, Weigel D, Herrmann BG 1991 Expression of a Xenopus homolog of Brachyury (T) is an immediate-early response to mesoderm induction. Cell 67:79–87(Ralph T. B?ttcher and Chr)
Fibroblast growth factors (FGFs) have been implicated in diverse cellular processes including apoptosis, cell survival, chemotaxis, cell adhesion, migration, differentiation, and proliferation. This review presents our current understanding on the roles of FGF signaling, the pathways employed, and its regulation. We focus on FGF signaling during early embryonic processes in vertebrates, such as induction and patterning of the three germ layers as well as its function in the control of morphogenetic movements.
I. Introduction
II. FGFs and FGF Receptors (FGFRs)
III. FGFR Signal Transduction
A. Ras/MAPK pathway
B. PLC/Ca2+ pathway
C. PI3 kinase/Akt pathway
IV. The Biological Activities of FGFs
A. FGF signaling pathway in mesoderm formation and gastrulation movements
B. Chemotactic action of FGFs in morphogenetic movements
C. FGF signaling in neural induction and AP patterning
D. FGF signaling in endoderm formation
V. Regulation of FGF Signaling
A. Regulation by HSPGs
B. Transmembrane regulators
C. Intracellular regulators
VI. Concluding Remarks
I. Introduction
Top
Abstract
I. Introduction
II. FGFs and FGF...
III. FGFR Signal Transduction
IV. The Biological Activities...
V. Regulation of FGF...
VI. Concluding Remarks
References
FIBROBLAST GROWTH FACTORS (FGFs) constitute a large family of polypeptide growth factors found in a variety of multicellular organisms, including invertebrates. The first member of this family, FGF2, was identified by its ability to stimulate proliferation of mouse 3T3 fibroblasts (1); however, the function of FGFs is not restricted to cell growth. Instead, FGFs are involved in diverse cellular processes including chemotaxis, cell migration, differentiation, cell survival, and apoptosis. The common feature of FGFs is that they are structurally related and generally signal through receptor tyrosine kinases. FGFs play an important role in embryonic development in invertebrates and vertebrates. This review focuses on the role of FGF signaling during early vertebrate development. Excellent reviews dealing with the role of FGFs in Drosophila (2, 3, 4), Caenorhabditis elegans (5, 6), vertebrate limb development (7, 8, 9), central nervous system development (10, 11), skeleton formation (12), and cancer (13, 14, 15) are available.
II. FGFs and FGF Receptors (FGFRs)
Top
Abstract
I. Introduction
II. FGFs and FGF...
III. FGFR Signal Transduction
IV. The Biological Activities...
V. Regulation of FGF...
VI. Concluding Remarks
References
The human FGF protein family consists of 22 members that share a high affinity for heparin as well as a high-sequence homology within a central core domain of 120 amino acids (16), which interacts with the FGFR (17, 18, 19) (Fig. 1A). FGFs can be classified into subgroups according to structure, biochemical properties, and expression (for reviews see Refs.16 and 20). For example, members of the FGF8 subfamily (FGF8, FGF17, FGF18) share 70–80% amino acid sequence identity and similar receptor-binding properties and have overlapping expression patterns. FGFs appear to be diffusible (21), and they are able to act in a dose-dependent fashion, e.g. to induce different marker genes at different concentrations (22, 23). This raises the question of whether FGFs act physiologically as morphogens. Indeed, in neural progenitors, graded FGF signaling appears to be translated into a distinct motor neuron Hox pattern (24).
FGFs induce their biological responses by binding to and activating FGFRs, a subfamily of cell surface receptor tyrosine kinases (RTKs). The vertebrate Fgfr gene family consists of four highly related genes, Fgfr1–4 (25). These genes code for single spanning transmembrane proteins with an extracellular ligand-binding region and an intracellular domain harboring tyrosine kinase activity. The extracellular region is composed of Ig-like domains that are required for FGF binding. The Ig-like domains also regulate binding affinity and ligand specificity (Fig. 1B). Located between Ig-like domains I and II is a stretch of acidic amino acids (acidic box domain) followed by a heparin-binding region and a cell adhesion homology domain. These domains are required for the interaction of the receptor with components of the extracellular matrix, in particular heparan sulfate (HS) proteoglycans (HSPGs), and cell adhesion molecules (CAMs). The intracellular part of the receptor includes the juxtamembrane domain, the split tyrosine kinase domain, and a short carboxy-terminal tail. In addition to the enzymatic activity, the intracellular domain harbors protein binding and phosphorylation sites (including protein kinase C and FRS2 sites) as well as several autophosphorylation sites that interact with intracellular substrates.
Alternative splicing of Fgfr transcripts generates a diversity of FGFR isoforms (25, 26). In vitro assays show that different isoforms have distinct FGF-binding specificities (27, 28). The tissue-specific alternative splicing encompassing the carboxy-terminal half of Ig domain III strongly affects ligand-receptor binding specificity (29, 30, 31). In vivo, FGFR-FGF complex assembly and signaling is further regulated by the spatial and temporal expression of endogenous HSPGs (32, 33). HSPGs are required to activate effectively the FGFR and determine the interaction between specific FGF-FGFR pairs (see Section V.A.).
III. FGFR Signal Transduction
FGFRs, like other receptor tyrosine kinases, transmit extracellular signals after ligand binding to various cytoplasmic signal transduction pathways through tyrosine phosphorylation. Binding of FGFs causes receptor dimerization and triggers tyrosine kinase activation leading to autophosphorylation of the intracellular domain (34). Tyrosine autophosphorylation controls the protein tyrosine kinase activity of the receptor but also serves as a mechanism for assembly and recruitment of signaling complexes (35). These phosphorylated tyrosines function as binding sites for Src homology 2 and phosphotyrosine binding domains of signaling proteins, resulting in their phosphorylation and activation (36, 37). A subset of Src homology 2-containing proteins such as Src-kinase and phospholipase C (PLC) possesses intrinsic catalytic activities, whereas others are adapter proteins. FGF signal transduction, as analyzed during early embryonic development, can proceed via three main pathways described below (Fig. 2).
A. Ras/MAPK pathway
The most common pathway employed by FGFs is the MAPK pathway. This involves the lipid-anchored docking protein FRS2 (also called SNT1) that constitutively binds FGFR1 even without receptor activation (38, 39, 40). Several groups have demonstrated the importance of FRS2 in FGFR1-mediated signal transduction during embryonic development (41, 42, 43). After activation of the FGFR, tyrosine-phosphorylated FRS2 functions as a site for coordinated assembly of a multiprotein complex activating and controlling the Ras-MAPK signaling cascade and the phosphatidylinositol 3 (PI3)-kinase/Akt pathway. The FRS2 tyrosine phosphorylation sites are recognized and bound by the adapter protein Grb2 and the protein tyrosine phosphatase (PTP) Shp2 (39, 44). Grb2 forms a complex with the guanine nucleotide exchange factor Son of sevenless (SOS) via its SH3 domain. Translocation of this complex to the plasma membrane by binding to phosphorylated FRS2 allows SOS to activate Ras by GTP exchange due to its close proximity to membrane-bound Ras. Once in the active GTP-bound state, Ras interacts with several effector proteins, including Raf leading to the activation of the MAPK signaling cascade. This cascade leads to phosphorylation of target transcription factors, such as c-myc, AP1, and members of the Ets family of transcription factors (reviewed in Ref.45).
B. PLC/Ca2+ pathway
The PLC/Ca2+ pathway involves binding of PLC to phosphorylated tyrosine 766 of FGFR1 (46, 47). Upon activation, PLC hydrolyzes phosphatidylinositol-4,5-diphosphate to form two second messengers, inositol-1,4,5-triphosphate and diacylglycerol. Diacylglycerol is an activator of protein kinase C, whereas inositol-1,4,5-triphosphate stimulates the release of intracellular Ca2+. This cascade has been implicated in the FGF2-stimulated neurite outgrowth (48, 49) and in the caudalization of neural tissue by FGFR4 in Xenopus (50).
C. PI3 kinase/Akt pathway
The PI3 kinase/Akt pathway can be activated by three mechanisms after FGFR activation (Fig. 2), and the phospholipids thereby generated regulate directly or indirectly the activity of target proteins such as Akt/PKB.
Among other processes, the PI3 kinase signaling branch is involved during Xenopus mesoderm induction acting in parallel to the Ras/MAPK pathway. Overexpression of a dominant negative form of the PI3 kinase-regulatory subunit p85 interferes with Xenopus mesoderm formation. Conversely, coexpression of activated forms of MAPK and PI3 kinase leads to synergistic mesoderm induction (51).
IV. The Biological Activities of FGFs
Ealy developmental processes during gastrulation stages of Xenopus, zebrafish, chicken, and mouse include mesoderm formation and gastrulation movements, neural induction and AP patterning, and endoderm formation (Fig. 3). FGF signaling plays important roles in early vertebrate development during these processes as discussed below and summarized in Table 1. Genes that function in FGF signaling and for which loss-of-function analysis revealed an early embryonic phenotype are summarized in Table 2.
A. FGF signaling pathway in mesoderm formation and gastrulation movements
Induction and patterning of mesoderm is one of the earliest events during vertebrate body axis formation. Although TGF-? signaling plays a pivotal role in mesoderm induction and patterning and has received most attention in this context recently, FGF signaling is also essential. Indeed, the first identified mesoderm inducer was basic FGF (52), and this represents an evolutionarily conserved role, because FGF signaling is required for mesoderm induction in the primitive chordate Ciona (53). In Xenopus, mouse, and zebrafish, disruption of FGF signaling by Fgfr knock out or by overexpression of a dominant negative FGFR1 strongly affects body axis formation (54, 55, 56, 57). In these embryos, phenotypic changes are observed mostly in posterior regions, such as defects of trunk and tail structures. Despite some organism-specific differences, two functions for FGFs are conserved. First, FGFs control the specification and maintenance of mesoderm by regulation of T box transcription factors in Xenopus (58, 59), mouse (60, 61), and zebrafish (57, 62, 63). Second, they have a direct and/or indirect role in morphogenetic movements during gastrulation.
The effect of FGFs on mesoderm formation was first demonstrated and most extensively studied in Xenopus (64). Since the first studies that discovered the mesoderm-inducing ability of FGFs (52, 65, 66), many components of the FGF signaling pathway were shown in Xenopus to be required for mesoderm development. These components include FGFRs (54, 67), HSPGs (68), adaptor proteins FRS2/SNT-1, Nck, and Grb2 (42, 43, 69), Src-kinase Laloo (70), PTPs (71, 72), Ras, Raf, MAPK kinase, ERK (73, 74, 75, 76), and the transcription factors AP1 and Ets2 (77, 78). Their inhibition blocks mesoderm formation and induces posterior and gastrulation defects. Conversely, in gain-of-function experiments they induce mesodermal markers and phenocopy FGF overexpression. The role of FGF signaling during mesoderm formation is that of a competence factor. It is required for cells to respond to TGF-?-like mesoderm inducers, because activin-mediated mesoderm induction in Xenopus animal caps, for example, is blocked by a dominant negative FGFR1 (79, 80, 81). Similarly, the trunk-inducing activity of dorsal mesoderm, which depends on Nodal signaling, is inhibited by a dominant negative FGFR1 (82). The ability of FGFs to regulate Nodal signaling may be due to phosphorylation of Smad2 by MAPKs (83).
FGF signaling needs to continue during gastrulation for mesoderm maintenance as shown by transgenic Xenopus embryos overexpressing a dominant negative FGFR1 (84). One of the FGF mesodermal targets is the T box transcription factor brachyury (Xbra), required for posterior mesoderm and axis formations in mouse, zebrafish, and Xenopus (58, 85, 86, 87). Xbra and eFGF form an autoregulatory loop in which Xbra induces eFGF and vice versa (88, 89). eFGF is also required for XmyoD expression in the myogenic cell lineage of Xenopus. eFGF is a candidate factor mediating the community effect, whereby cells differentiate into muscle once a critical cell mass is reached, due to the accumulation of diffusible factors (90, 91).
In Xenopus embryos, FGF signaling has both direct and indirect roles on gastrulation movements. Nutt et al. (92) provided evidence for a direct role of FGF signaling in cell movements during convergence extension. They identified and characterized XSprouty2 as an intracellular antagonist of FGF-dependent calcium signaling. Overexpression of XSprouty2 inhibits convergent extension movements without affecting MAPK activity, mesoderm induction, and patterning. FGF signaling also affects gastrulation movements indirectly via regulation of Xbra induction and maintenance. Xbra acts as a switch promoting convergent extension and actively repressing cell migration by inhibiting adhesion of migrating cells to fibronectin (93, 94). Also, Xbra directly activates the expression of Xwnt11 and prickle (95, 96), which regulate convergent extension movements via the noncanonical Wnt pathway (95, 96, 97).
In zebrafish, as in Xenopus, FGF signaling acts as competence factor to control mesoderm induction by TGF-?s, such as activin and Nodal (63, 98). Through the transcriptional regulation of three mesodermal T box transcription factors ntl (86), tbx6 (99), and spadetail (spt), a regulator of mesoderm formation in the trunk (100), FGF signaling regulates tail and trunk development in zebrafish (57, 62, 101, 102). As in Xenopus and mouse, the T box genes also serve as an important link between cell fate and morphogenesis. One important role of spt is the regulation of convergence movements of paraxial mesoderm during gastrulation (103) mediated by paraxial protocadherin (papc) (104). Additionally, FGFs may affect gastrulation movements via snail. Consistent with results in mouse (61), overexpression of eFGF causes abnormal widespread expression of snail1 throughout the hypoblast of the late zebrafish gastrula (57).
In addition to its effects on anterior-posterior (AP) patterning of the embryo and gastrulation movements, FGF signaling affects dorsoventral patterning of the early zebrafish embryo. Activation of the FGF8/FGFR1/Ras signaling pathway leads to the expansion of dorsolateral derivatives at the expense of ventral and posterior domains by an early inhibition of ventral bone morphogenetic protein (BMP) expression (105, 106, 107). It was therefore suggested that FGFs act in concert with secreted BMP antagonists (Spemann organizer factors Follistatin, Noggin, and Chordin) in dorsoventral patterning (105).
FGFR1 knockouts gave the first evidence that FGF signaling is also essential for proper gastrulation in the mouse (55, 56). Fgfr1–/– mutant embryos die late in gastrulation, exhibiting defects in cell migration, cell fate specification, and patterning. Chimera analysis indicate that Fgfr1-deficient cells are impaired in epithelial to mesenchymal transition (EMT) and, as a consequence, fail to cross the primitive streak (108). The mutant cells accumulate in the primitive streak and overexpress E-cadherin, and this expression is accompanied by the down-regulation of mSnail. mSnail is a key mediator of EMT in development and cancer by repressing E-cadherin (reviewed in Ref.109). Thus, Fgfr1–/– cells may fail to migrate properly because they maintain high levels of E-cadherin in the streak as a result of mSnail down-regulation. The regulation of E-cadherin by FGFR1 via mSnail also provides a molecular link between FGF and Wnt signaling pathways because ectopic E-cadherin expression in Fgfr1–/– embryos attenuates Wnt signaling by sequestering cytoplasmic ?-catenin to the plasma membrane (61).
Similar to Fgfr mutants, Fgf8–/– embryos display severe gastrulation defects, lacking mesodermal and endodermal tissues. However, epiblast cells move into the streak and undergo EMT in the Fgf8–/– embryos, but fail to move away from the streak (60). This indicates that FGF family members are involved in different stages of mesoderm formation from EMT to subsequent migration and that these two processes are independent.
B. Chemotactic action of FGFs in morphogenetic movements
The activity of FGF signaling during gastrulation movements in Xenopus, zebrafish, and mouse has been mostly linked to the regulation of target gene expression, which in turn regulates the migratory behavior of gastrulating cells. However, there is increasing evidence for a direct chemotactic response of migrating cells in response to FGFs both in vertebrates and invertebrates (3, 5). FGF2 and FGF8 function as potent chemoattractants during migration of mesencephalic neural crest cells (110), and FGF2 and FGF4 attract mesenchymal cells during limb bud development in the mouse (111, 112). FGF10, a key regulator of lung-branching morphogenesis, exerts a powerful chemoattractive effect to direct lung epithelial buds to their destinations during development (113). Recently, an elegant study by Yang et al. (114) showed that FGFs act as chemotactic signals to directly coordinate cell movements during gastrulation in the chick embryo. They analyzed the trajectories of cells emerging from different AP positions from the fully extended primitive streak and found that cells show different movement patterns depending on their position along the streak. The cells are guided by signals from the surrounding tissue, and beads soaked in FGFs alter their trajectories. FGF4 acts as a chemoattractant and FGF8b acts as a repellent for migrating streak cells, whereas FGF2 does not elicit a strong chemotactic response (114). In line with the expression of Fgf8 in the primitive streak and Fgf4 in Hensen’s node and forming head process, the movement away from the streak may be the result of a chemorepellent effect of FGF8, whereas the inward movement of cells results from attraction by FGF4 secreted by cells in the head process (114).
C. FGF signaling in neural induction and AP patterning
Neural induction is the process whereby naive ectodermal cells are instructed to adopt a neural fate. Several findings, mainly from experiments in Xenopus, led to the default model of neural induction (reviewed in Ref.115): BMP signals in the early embryo instruct ectoderm cells to form epidermis whereas ectodermal cells will adopt a neural fate (their default state) in the absence of cell-cell signaling. In this model, neural induction is mediated during gastrulation by diffusible BMP antagonists released by Spemann’s organizer. However, this model is certainly oversimplified, and increasing evidence suggests a pivotal role of FGF in neural induction and places the onset of neural induction before gastrulation. FGFs act as neural inducers in lower chordates such as ascidians (116) and planarians (117), indicating that this role is evolutionarily conserved.
The best case for FGFs as neural inducer is the chick embryo. Here, induction of neural tissue precedes the formation of the organizer (Hensen’s node) (118, 119). FGFs are sufficient to induce preneural markers (118, 120, 121, 122), and inhibition of FGF signaling using various inhibitors blocks neural induction by Hensen’s node (118). FGF signaling functions by repressing BMP expression because Fgf3 overexpression leads to a down-regulation of BMP mRNA and the subsequent induction of neural tissue. Conversely, when FGF signaling is inhibited, BMP mRNA expression is maintained and neural fate is blocked (119). A recent study by Sheng et al. (123) suggests a mechanism by which FGFs may block BMP signaling during neural induction in chick. FGFs induce the transcriptional activator Churchhill (ChCh), which is required for the expression of Sip1 (Smad-interacting protein 1) in the neural plate. Sip1 binds to and represses Smad1/5 transcriptional activation of target genes, thereby blocking BMP signaling. ChCh/Sip1 sensitizes cells to neural-inducing signals and represses mesodermal genes (123). However, FGFs promote neural fates also independent of BMP inhibition because beads soaked with the BMP antagonists Noggin or Chordin are insufficient for neural induction under most experimental conditions (119, 124).
In Xenopus, the role of FGF signaling in neural induction and patterning has been studied extensively. However, conflicting results were obtained concerning the role of FGFs in neural induction. Some reports showed that FGFs could directly induce neural tissue in animal cap explants. However, in most cases a sensitized background was used in which BMP signaling was partially attenuated (23, 125, 126). In addition, inhibition of FGF signaling by dominant negative FGFR1 blocks neural induction by Noggin or Spemann organizer (127), Chordin (128), or notochord tissue (129). On the other hand, several studies using dominant negative FGFR1 obtained contradictory results showing that neural tissue forms apparently even when FGF signaling is inhibited (84, 126, 129, 130, 131). However, these experiments were carried out using the dominant negative form of the FGFR1, but FGF signaling required for anterior neural induction may be mediated through FGFRs other than FGFR1. Indeed, Hongo et al. (132) showed that anterior neural development is more sensitive to inhibition of FGFR4a than to FGFR1. Likewise, FGF8 stimulates neuronal differentiation preferentially through FGFR4a (133), and FGF signaling through FGFR1, but not FGFR4a, is required for neural crest formation (134). The two receptors preferentially activate different downstream pathways, suggesting that they may have distinct functions during development. FGFR4a can signal through PLC in neural induction (50) whereas FGFR1 preferentially activates the Ras-MAPK pathway, involved in posteriorization of already neuralized ectoderm (50, 126).
In addition to their early effects in neural induction, FGFs are potent posteriorizing agents during AP patterning of the neural plate (135). They are able to convert anterior neural tissue to more posterior neural cell types (50, 125, 126, 131, 136). For example, when FGF-soaked beads are placed in the anterior neural plate, they inhibit anterior neural marker expression (137, 138), and this may involve a cross-talk with the Wnt pathway (130, 139, 140, 141). Similarly, overexpression of FGF3 in zebrafish expands posterior neural tissue (142). Conversely, dominant negative FGFR1 or dominant negative Ras block induction of posterior neural markers (126, 131).
Patterning of neuroectoderm and mesoderm is linked to the patterned expression of Hox genes along the AP axis of the embryo (143). There is strong evidence from studies in mouse, chick, and Xenopus that FGF signaling regulates early AP patterning via the regulation of Hox genes. Ectopic application or overexpression of FGFs results in the up-regulation of posterior members of the Hox gene family (137, 144, 145, 146, 147). Conversely, overexpression of dominant negative FGFR1 represses posterior Hox gene expression (137, 148). Caudals are immediate-early response genes to FGF signaling and mediate Hox regulation by FGFs (137, 147, 149, 150). A recent study demonstrates that graded FGF signaling is translated into a distinct motor neuron Hox pattern (24).
An alternative mode of action for FGFs as posteriorizing factor was described for chick neural tube in which they mediate the maintenance of neural stem cells at Hensen’s node. Therefore, FGFs might act to prolong the time window during which cells are exposed to other posteriorizing factors, such as retinoic acid or Wnts, thereby promoting the continuous development of the posterior nervous system (151).
D. FGF signaling in endoderm formation
Although endodermal organ development is widely studied, little is known about how endoderm is initially formed and regionally specified in vertebrates. Accumulating data, especially in the mouse, point toward a function of FGF signaling in early endoderm maintenance and patterning. In chimeric embryos, Fgfr1–/– cells are unable to populate endodermal derivatives (61, 108). Similarly, embryoid bodies derived from Fgfr1–/– embryonic stem cells lack visceral endoderm. This is associated with a lowered expression of the endodermal marker -fetoprotein. However, some endoderm is eventually formed, as indicated by the unchanged expression of GATA4 (152). Endoderm deficiency was also obtained in embryoid bodies overexpressing a dominant negative FGFR2, where inhibition of three endodermal markers (-fetoprotein, HNF4, and Evx1) was observed (153). Fgf8–/– embryos develop no embryonic mesoderm- and endoderm-derived tissue, and this is accompanied by inhibition of Fgf4 expression (60). Fgf4–/– embryos die shortly after implantation and fail to make detectable mesoderm and endoderm (154), a phenotype similar to the targeted disruption of Fgfr2 (155, 156). A study by Wells and Melton (157) further supports the role of FGF4 in the specification of primitive endoderm in the mouse because recombinant FGF4 affects the patterning in isolated endoderm in a concentration-dependent manner. Other analyzed FGFs (aFGF, FGF2, 5, 8b) or other growth factors (TGF, TGF?, epidermal growth factor, IGF) had no activity in this assay (157). In conclusion, FGF signaling appears not to be required for induction but rather for the maintenance and patterning of endoderm, because some endodermal markers continue to be expressed even when FGF signaling is inhibited.
An important caveat in analyzing endoderm development is the potent influence of mesoderm on its differentiation and patterning (158, 159). Thus, factors may regulate endoderm via an effect on mesoderm unless this is rigorously controlled. In Xenopus this can be achieved by explanting vegetal caps, which are devoid of mesoderm and only develop into endoderm. FGF signaling-dependent MAPK activity is detectable in the prospective endodermal cells of the Xenopus blastula embryo (160). However, studies with such vegetal caps gave conflicting results regarding the role of FGF signaling in endoderm. Expression of the dorsal endodermal marker Xlhbox8 and the pan-endodermal marker Mix1 can be blocked by injection of dominant negative FGFR1 (160, 161). In contrast, other studies failed to observe the requirement of FGF signaling for Xlhbox8 expression but rather report an inhibition of Xlhbox8 after FGF treatment (162, 163). Modulation of FGF signaling does not affect the expression of pan-endodermal markers in the majority of vegetal explants experiments, and FGF does not induce a number of endodermal genes in animal caps (128, 164). It is therefore suggested that FGF signaling is not essential for endoderm formation. However, defined levels of FGF activity may be required for endodermal patterning.
In summary, FGF signaling plays simultaneous roles during gastrulation to induce different cell fates and morphogenetic movements, raising the question of how the specific response to FGFs in different germ layers is mediated. On the one hand, different FGFs may mediate neural or mesodermal induction (e.g. Ref.133). On the other hand, two recent papers show that in the ascidian Ciona and in chick the competence of cells to respond to FGF signals with neural induction is regulated by the neural-specific transcription factors GATA and ChCh, respectively (123, 165). Thus, in addition to qualitative differences among different FGF family members, the spatial and temporal restricted expression of cofactors provides another mechanism to separate the different functions of FGF during gastrulation.
V. Regulation of FGF Signaling
Developmental signaling that involves intercellular communication and regulates cell fate must be precisely regulated regarding timing, duration, and spread of such a signal. Precision and robustness of these features can be achieved by both positive and negative feedback loops (reviewed in Ref.166). Negative feedback stabilizes a system when confronted with environmental fluctuations (167). Furthermore, it can limit the absolute level or duration of a signal and can lead to signal oscillations (168). FGF signaling is regulated at many levels in the extracellular space and within the cell. Many of these regulatory mechanisms are not specific for FGF signaling but rather activate or attenuate RTK signaling in general (169). Yet, a growing number of proteins were identified in recent years that specifically regulate the activity of FGFs or signaling pathways activated through FGFRs. Interestingly, many of these regulatory components are themselves regulated by FGF signaling and are tightly coexpressed with Fgfs, forming a synexpression group (107, 168, 170). Synexpression groups are genetic modules in which expression of different genes are closely correlated and whose members function in a common pathway (171). Most of the FGF synexpression group members inhibit FGF signaling and thereby establish negative feedback loops.
A. Regulation by HSPGs
An important feature of FGF biology involves the extracellular interaction between FGF and heparin or HSPG (32). HSPGs are sulfated glycosaminoglycans covalently bound to a core protein. Enormous structural heterogeneity can be generated through specific heparin sulfate chain modifications during their biosynthesis, as well as from the diverse nature of their core proteins (reviewed in Ref.172). HSPGs are required for FGFs to effectively activate the FGFR and determine interaction between individual FGF family members with certain FGFRs (173, 174). The crystal structure of ternary FGF-heparin-FGFR complexes have been determined; heparin bridges two FGFRs to form FGFR dimers, which bind two FGFs (175). In an elegant study, a ligand and carbohydrate engagement assay was employed to examine the spatio-temporal changes in HS to promote ternary complex formation with FGFs and FGFRs. These changes depend on glycosyltransferases and sugar sulfotransferases. The analysis reveals that synthesis of HS domains is tightly regulated throughout development, and different FGF-FGFR pairs have unique HS-binding requirements (33). Other than promoting FGF binding to its receptors, HSPGs protect the FGFs from thermal denaturation and proteolysis (176). In addition, these interactions limit FGF diffusion and release into interstitial spaces, thereby regulating FGF activity within a tissue (177, 178). The importance of HSPGs and enzymes involved in HS biosynthesis for FGF signaling during development has been demonstrated in vivo in a number of studies (68, 179, 180, 181). For example, mice mutated in UDP-glucose dehydrogenase (Ugdh), one of the key enzymes for HS biosynthesis, fail to gastrulate properly and show defects of mesoderm and endoderm migration similar to FGF pathway mutants (181). In the Ugdh mutant embryos, FGF signaling is specifically blocked whereas signaling by Wnt3 and Nodal appears unaffected. Ugdh mutant Drosophila embryos also have phenotypes similar to FGFR mutants (180), indicating an evolutionary conserved requirement for HSPGs in FGF signaling.
The nature of the proteoglycan core protein essential for FGF signal transduction has not been identified because single knock-out mice mutant for diverse core proteins (including Syndecan1, -3, and -4; Glypican2 and -3; and Perlecan) survive past gastrulation.
B. Transmembrane regulators
An emerging theme is the regulation of FGF signaling by transmembrane modulators. The transmembrane protein Sef is a member of the FGF synexpression group in zebrafish and mouse, expressed in the characteristic pattern of FGF8 and regulated by FGF signaling (107, 170, 182). Overexpression studies in zebrafish embryos show that Sef inhibits FGF signaling via interaction with FGFR1 and -4a (170) and that it functions at the level or downstream of MAPK kinase (107). Interfering with Sef function by antisense morpholino oligonucleotide (MO) injection induces phenotypic changes reminiscent of embryos dorsalized by ectopic expression of Fgf8 (107, 170). The molecular mechanism of Sef action is controversial. Sef may inhibit FGF signaling at the level of FGFR (183, 213). Overexpression of mouse Sef inhibits tyrosine-phosphorylation of FGFR1 and FRS2 and inhibits activation of the MAP kinase and Akt pathway, possibly due to reduced recruitment of Grb2 to FRS2 (183). In contrast, other reports argue Sef acts downstream of or at the level of MEK (107, 214, 215, 216). Torii et al. (216) showed that hSef acts to restrict ERK activity to the cytoplasm. hSef binds to activated forms of MEK and inhibits the dissociation of the MEK-ERK complexes. Thereby, nuclear translocation of ERK is blocked without inhibiting its activity in the cytoplasm (216).
In Xenopus, XFLRT3, a member of a family of leucine-rich repeat transmembrane proteins, is another component of the FGF signaling pathway (184). Its expression is very similar to that of the FGF8 synexpression group, and it is regulated by FGFs. When overexpressed, FLRT3 phenocopies FGF signaling and activates the MAPK pathway. Inhibition of XFLRTs by MO injection interferes with FGF signaling. XFLRT3 binds to FGFRs and modulates FGFR signaling by an unknown mechanism. In cell culture, FLRT3 promotes neurite outgrowth of primary cerebella granule neurons (217). Since FGF promotes axon outgrowth (218, 219, 220), it is possible that potentiation of FGF signaling by FLRT3 underlies its effects on neurite growth.
Finally, FGFRs interact with CAMs such as N-CAM, N-cadherin, and L1 via their extracellular domains. In neurons, N-CAM binding to FGFR1 activates the PLC/Ca2+ pathway (49, 185, 186). In tumor ?-cells and L cells, Cavallaro et al. (187) show a direct association of FGFR4 and N-CAM. This interaction induces the formation of a multiprotein signaling complex that mediates neurite outgrowth and cell-matrix adhesion (187).
C. Intracellular regulators
1. PTPs.
It is well established that RTK signaling is controlled by PTPs, which act as positive or negative regulators on RTK signaling (reviewed in Refs.188 and 189). The dual-specificity PTP MAPK phosphatase 3 (MKP3, also called Pyst1) antagonizes the MAPK pathway via ERK1/2 inactivation. In vertebrates, its expression is very similar to that of the FGF8 synexpression group, and it is regulated by FGF8 (190, 191). In the developing chick limb MKP3 mediates the antiapoptotic signaling of FGF8 (191), and manipulation of MKP3 activity by short interfering RNA inhibition or overexpression disrupts limb morphology and outgrowth (190, 191). In contrast, the PTP SHP-2 and Xenopus low-molecular weight PTP1 (XLPTP1) act as positive regulators of FGF signaling upstream of Ras during Xenopus embryogenesis (71, 72). Inhibition of XLPTP1 by antisense MOs blocks FGF-induced MAPK activation, leading to a shortened AP axis (72).
2. Sprouty protein.
The feedback inhibitor Sprouty was identified in Drosophila as a negative regulator of FGF signaling during tracheal development (192). Four mammalian Sprouty proteins (Spry1–4) and three Spreds (Sprouty-related EVH1 domain proteins) that share a highly conserved cysteine-rich domain at the carboxy terminus have been identified. This domain is critical to target them to the plasma membrane and to inhibit the MAPK pathway (193). Sprouty2 and -4 are members of the FGF8 synexpression group and are regulated by FGF signaling (106, 194, 195, 196). Sprouty proteins interfere with FGF signaling through several mechanisms, depending on the cellular context and/or stimulation. Different Sprouty proteins exhibit different activities and have different interaction partners, including but not limited to Grb2 (197) and Raf1 (198) (reviewed in Ref.199).
Gain- and loss-of-function experiments confirm the function of Sprouty proteins as intracellular antagonists of FGF signaling. Reduction of mSprouty2 expression during mouse lung development using antisense oligonucleotides established Sprouty as an inhibitor of branching morphogenesis, as in Drosophila (192, 200). Overexpressing Sprouty during chick limb development causes a reduction in limb bud outgrowth consistent with reduced FGF signaling (195). Furthermore, Fürthauer et al. (106) showed that inhibiting Sprouty4 in zebrafish by MO injection elicits phenotypes similar to that obtained after up-regulation of Ffg8. Sprouty4 is required for dorsoventral mesoderm patterning and for head development (106). In Xenopus embryos, XSprouty2 regulates gastrulation movements by a Ca2+-dependent pathway (92).
VI. Concluding Remarks
Substantial progress has been made in recent years in the understanding of the role of FGF signaling during early vertebrate development. FGFs play an important role in regulating cell fate and specification of all three germ layers. The role of FGF signaling during mesoderm formation and neural induction is now well documented and appears evolutionarily conserved in chordates. A surprising new role is the ability of FGFs to orchestrate gastrulation movements by acting as chemoattractants and repellents. This probably involves direct regulation of the cytoskeleton.
FGF downstream pathways and the components employed are characterized in some detail. A growing number of these components are themselves tightly regulated by FGFs and form an FGF8 synexpression group. The FGF8 synexpression group is an autoregulatory genetic module, conserved from frogs to mice, which is employed, for example, in mesoderm formation and neural patterning during early development. Part of this group are the novel transmembrane regulators Sef and FLRT3, which bind to and modulate FGFRs, but their precise function needs further investigation. It has also become clear that different ligand receptor pairs may preferentially couple to different pathways but how this is regulated is unclear. In the future it will be important to assign the pathways employed by FGFs in different cells and define the determinants for signaling specificity. One emerging class of determinants are HSPGs and the enzymes involved in their synthesis. Two classes of downstream target genes, which mediate FGF signaling are T-box and Hox genes, and this is evolutionary conserved in all vertebrates. Furthermore, there is substantial cross-talk between FGFs with TGF-?- and Wnt signaling pathways. We are only beginning to understand the mechanisms of these cross-regulations, which are at the heart of the ability of FGFs to act as competence factor, e.g. during mesoderm induction by Nodals. Finally, do FGFs act as morphogens? They have many characteristics of morphogens, including their ability to induce marker genes in a dose-dependent manner. However, the proof that FGFs act directly in a concentration-dependent fashion rather than by relay and the demonstration of a physiological requirement for this is missing.
Acknowledgments
We thank Dr. Sonia Pinho for advice and critical reading of the manuscript.
Footnotes
First Published Online October 26, 2004
Abbreviations: AP, Anterior-posterior; BMP, bone morphogenetic protein; CAM, cell adhesion molecule; EMT, epithelial to mesenchymal transition; FGF, fibroblast growth factor; FGFR, FGF receptor; HS, heparan sulfate; HSPG, heparan sulfate proteoglycan; MO, morpholino oligonucleotide; PI3, phosphatidylinositol-3; PLC, phospholipase C; PTP, protein tyrosine phosphatase; RTK, receptor tyrosine kinase; SOS, Son of sevenless.
References
Yamaguchi TP, Harpal K, Henkemeyer M, Rossant J 1994 fgfr-1 is required for embryonic growth and mesodermal patterning during mouse gastrulation. Genes Dev 8:3032–304
Griffin K, Patient R, Holder N 1995 Analysis of FGF function in normal and no tail zebrafish embryos reveals separate mechanisms for formation of the trunk and the tail. Development 121:2983–2994
Smith JC, Price BM, Green JB, Weigel D, Herrmann BG 1991 Expression of a Xenopus homolog of Brachyury (T) is an immediate-early response to mesoderm induction. Cell 67:79–87(Ralph T. B?ttcher and Chr)