As discussed above, HS-GAG chains are a central part of the FGF signalling complex (Figure 2). Moreover, the sulphate group at the 6-O position of glucosamine has been shown to be essential for the association of ligand and receptor, making contact with FGFR (Pellegrini, 2001). The contact of the HS-GAGs with FGFR and FGF ligand draws the signalling complex together and stabilizes it (Schlessinger et al., 2000). In addition, the presence of the 6-O sulphate group also has been shown to be critical for the mitogenic activity of FGF (Lundin et al., 2000). It is therefore not surprising that enzymes controlling the presence of the 6-O sulphate group can influence FGF signalling.

FIGURE 2. The FGF signalling complex.


The FGF signalling complex. FGF ligands (shown as light blue ovals) interact with extracellular immunoglobulin domains of the FGF receptors. The GAG chains of HSPGs make contact with both the FGF ligands and the FGF receptors driving the formation of (more...)

During the biosynthesis of HSPGs, 6OST acts in the Golgi to add 6-O-sulphate groups to glucosamine in some regions of the HS chain; however, not all disaccharides are modified, leading to structural diversity along the HS chain (Turnbull et al., 2001). Further modification can come subsequent to HSPG synthesis: the enzymes Sulf1 and Sulf2 are HS-6-O-endosulfatases that remove the 6-O-sulphate group from HS chains (Dhoot et al., 2001; Morimoto-Tomita et al., 2002). There are some excellent examples of the complementary expression of these enzymes with opposite activities (Figure 3), suggesting that the 6-O modification of HSPGs is very important for developmental patterning. In vivo studies have found that Drosophila lacking 6OST have defective trachea (Kamimura et al., 2001; Kamimura et al., 2006) which is known to require FGF during its morphogenesis. 6OST genes have distinct expression patterns during zebrafish development (Cadwallader and Yost, 2006), and zebrafish 6OST mutants have abnormalities in the development of skeletal muscle and blood (Bink et al., 2003; Chen et al., 2005).

FIGURE 3. Sulf2 and 6OST are expressed in complementary regions of the developing Xenopus brain.


Sulf2 and 6OST are expressed in complementary regions of the developing Xenopus brain. Vibratome sections subsequent to in situ hybridisation (ISH) shows that the expression of Sulf2 is restricted to the ventral part of the midbrain and hindbrain, while (more...)

It is well established that Sulf1 and Sulf2 are negative regulators of FGF signalling (Freeman et al., 2008; Lai et al., 2003; Wang et al., 2004). Overexpression of Sulf1 inhibits the activation of dpERK in Xenopus explants treated with FGF and in ovarian cancer cells. Knockdown of Sulf1 up-regulates dpERK in vivo during Xenopus development, indicating Sulf1 is an endogenous regulator of FGF signalling. Gene targeting of Sulf1 and Sulf2 increases the response of mouse embryonic fibroblasts (MEFs) to FGF (Lamanna et al., 2008). Interestingly, the expression of the 6OST genes is up-regulated in the Sulf1/2 knockout MEFs, suggesting a feedback mechanism where Sulf activity represses 6OST gene expression (Lamanna et al., 2008). MEFs where 6OST has been knocked out have a reduced response to FGF (Sugaya et al., 2008), but any effect on Sulf gene expression has not been determined.


Sprouty (Spry) was first identified as a Drosophila mutant (Hacohen et al., 1998) that had high levels of FGF activity, indicating that the normal function of the Sprouty gene is to restrict FGF signalling, and RTK signalling in general (Casci et al., 1999). In mammals, there are four Spry homologues and all possess a cysteine-rich Spry domain in the C-terminal half of the protein. At the N-terminus, there is a conserved tyrosine residue that is essential for the inhibitory function of the Spry proteins (see reviews of Cabrita and Christofori, 2008; Mason et al., 2006). Mutating this tyrosine results in a dominant negative Spry that enhances MAPK signalling downstream of FGF (Hanafusa et al., 2002), and it has been suggested that this tyrosine is needed for Spry to complex with other proteins to efficiently inhibit MAPK signalling (Mason et al., 2006). Sprouty proteins can act at multiple levels to restrict the MAPK signalling pathway. Spry has been shown to act upstream of Ras and compete with FRS2 for binding to Grb2/SOS complex (Hanafusa et al., 2002), while other studies have shown that Spry proteins inhibit Raf activation (Sasaki et al., 2003). Importantly, the expression of Spry genes is activated by FGF signalling (Branney et al., 2009; Hacohen et al., 1998; Sivak et al., 2005), and this creates a negative feedback loop whereby FGF restricts its own activity by activating the expression of Spry. Spry proteins are also regulated posttranscriptionally by FGF where phosphorylation of the conserved N-terminal tyrosine is required for its ability to inhibit the MAPK pathway (Mason et al., 2004). In addition, Spry proteins are phosphorylated at two serine residues that are dephosphorylated in response to FGF, activating Spry. These same serine residues are recognised by the ubiquitin ligase CBL, which competes with the phosphatase activated by FGF signalling, such that Spry is either dephosphorylated and activated by FGF or targeted to the proteasome by CBL (Lao et al., 2007).


Activation of the MAPK pathway by FGF results in a series of phosphorylation events that are reversible. The reversible nature of ERK phosphorylation means that its activity relies on a balance between the kinase that phophorylates it (MEK) and the phophatase that dephosphorylates it. MAPK phosphatases (MKPs) are dual specificity phosphatases that target and dephosphorylate activated MAPKs. There are several MAPKs including ERK, JNK, and p38 that are inactivated by specific MKPs/dual-specificity phosphatases (DUSPs). dpERK is inactivated when the phosphorylation of the tyrosine or the threonine within its activation loops is removed by a DUSP. There are 11 genes coding for DUSPs in mammalian genomes (Alonso et al., 2004) and some are highly specific to their MAPK substrate; for instance, MKP3/DUSP6 (also known as Pyst1) specifically inactivates dpERK (Keyse, 2000). The binding of MKP3 to dpERK stimulates the phosphatase activity of MKP; the activation and nuclear translocation of ERK is blocked. In this way, MKPs/DUSPs function as intracellular brakes of FGF signal transduction (Eblaghie et al., 2003).

Interesting, the expression of some MKPs/DUSPs were found in regions of developing embryos known to be active in FGF signalling (Dickinson et al., 2002; Eblaghie et al., 2003; Gomez et al., 2005; Lewis et al., 1995; Li et al., 2007). In addition, MKP1, MKP3, and DUSP5 have all been found to require FGF signalling for their expression. Consistent with these findings, the transcriptional activation of the MKP/DUSP genes is an early and robust response to FGF signalling (Branney et al., 2009; Eblaghie et al., 2003). Gene targeting of MKP3/DUSP6 results in increased dpERK levels and dwarfing phenotypes similar to those seen in embryos where strong activating FGFRs are expressed (Li et al., 2007). Overexpression of MKP3 or DUSP5 blocks mesoderm induction by FGF in Xenopus tissues (Branney et al., 2009; Umbhauer et al., 1995).

These observations provide another example where FGF signalling activates the expression of a negative regulator of FGF. It is possible that this negative feedback provides a mechanism to control the duration of signal transduction or even create an oscillatory response downstream of FGF.


Sef (Similar Expression to FGF) is a transmembrane protein originally identified in zebrafish and negatively regulates FGF signalling in development and cell culture (Furthauer et al., 2002; Tsang et al., 2002). The Sef protein associates with FGFRs, and it has been reported that the extracellular, transmembrane, and intracellular domains of Sef are involved in FGF antagonism (Ren et al., 2007). However, the exact mode of action of Sef has yet to be elucidated.


The three FLRT (Fibronection-Leucine-Rich-Transmembrane) genes were originally identified in human adult skeletal muscle (Lacy et al., 1999). It was subsequently shown Xenopus FLRT3 associates with FGFRs and, unlike the previously describe modulators, can act as a positive regulator of FGF signalling and is required for normal responses to FGF signalling in embryo tissues (Bottcher et al., 2004). In contrast, a recent report indicates that transcriptional responses to FGF signalling are normal in FLRT3 knockout mice (Maretto et al., 2008).

The sprouties, MKPs, Sef, and FLRTs are members of the FGF synexpression group that share similar expression patterns with the FGFs and are transcriptionally regulated by FGF signalling (Bottcher et al., 2004). Thus, presence of multiple negative and positive feedback loops operating downstream of FGF signalling illustrates the importance of fine-tuning the levels and extent of FGF signalling in the vertebrate embryo.