NT$ 67,500.00


Fibroblast  Growth  Factor FGF


Knowledge about how growth factors and hormones function is pivotal in order to rationally try to stimulate or inhibit their effects for therapeutical purposes. The larger this knowledge is, the more possible intervention sites we may expect to see. Today, the distance from basic science to applied therapy is not always long.

A wide array of chemical substances can influence cell growth and differentiation. This can either occur by cell surface receptor interactions, as in the case with peptide growth factors, or by interactions with intracellular molecules, as in the case with steroid hormones. Sometimes though, molecules that do not conform to this paradigm are discovered. Proteins that act both at the cell surface and intracellularly will be discussed in this thesis, with particular emphasis on the fibroblast growth factors.

Epidermal growth factor (EGF) and fibroblast growth factor 2 (FGF-2) induce the proliferation of neural precursor cells isolated from specific regions of the embryonic and adult brain. However, the lineage relationship between the EGF- and FGF-2-responsive cells is unknown. In this study we used phosphorylation of the transcription factor cAMP response element-binding protein as a functional readout to identify cells responding to EGF and FGF-2. In primary cultures of mouse embryonic day 14 (E14) striatum, maintained in vitro for 24 hr, 12% of the cells responded to FGF-2, whereas no response to EGF could be detected. Seventy-five percent of these FGF-2-responsive cells were beta tubulin III (TuJ1)-positive neurons, and 25% expressed nestin, a marker for neuroepithelial precursors. After growth factor treatment for 6 d, a population of nestin-positive cells responding to both EGF and FGF-2 were identified. The 6-d-old cultures also contained a small number of TuJ1-positive cells that responded to FGF-2 only. Priming of striatal cells for 24 hr with FGF-2 but not with EGF was sufficient to induce the appearance of EGF- and FGF-2 responsive cells after only 2 d in vitro. Thus, neural precursor cells from the mouse E14 striatum initially responding to FGF-2 only acquire EGF responsiveness later during in vitro development. At this stage EGF and FGF-2 act on the same cells. The acquisition of EGF responsiveness is promoted by FGF-2.

Fibroblast growth factors, or FGFs, are a family of growth factors involved in wound healing and embryonic development. The FGFs are heparin-binding proteins and interactions with cell-surface associated heparan sulfate proteoglycans has been shown to be essential for FGF signal transduction.

In humans, 22 members of the FGF family have been identified all of which are structurally related signaling molecules (named FGF1, FGF2, etc; Finklestein and Plomaritoglou, 2001). However, FGFs 11-14, also known as FGF homologous factors 1-4 (FHF1-FHF4), have been shown to have distinct functional differences compared to the FGFs. Although these factors possess remarkably similar sequence homology, they do not bind FGFRs and are involved in intracellular processes unrelated to the FGFs (Olsen et al., 2003). FGF molecules bind to a family of receptor molecules consisting of 4 members (FGFR1, FGFR2, FGFR3, and FGFR4). Alternate mRNA splicing gives rise to two distinct forms (b and c) of FGFRs1-3 which differ significantly in their ligand-binding profiles. The signaling complex at the cell surface is believed to be a ternary complex formed between two identical FGF ligands, two identical FGFR subunits and either one or two heparan sulfate chains.


Fibroblast growth factor was found in a cow brain extract by Gospodarowicz and colleagues and tested in a bioassay which caused fibroblasts to proliferate (first published report in 1974). They then further fractionated the extract using acidic and basic pH and isolated two slightly different forms that were named "acidic fibroblast growth factor" (FGF1) and "basic fibroblast growth factor" (FGF2). These proteins had a high degree of amino acid identity but were determined to be distinct mitogens.

Not long after FGF1 and FGF2 were isolated, another group isolated a pair of heparin-binding growth factors which they named HBGF-1 and HBGF-2, whilst a third group isolated a pair of growth factors that caused proliferation of cells in a bioassay containing blood vessel endothelium cells which they called ECGF-1 and ECGF-2. These proteins were found to be identical to the acidic and basic FGFs described by Gospodarowicz and coworkers.


One of the most important functions of bFGF is the promotion of endothelial cellproliferation and the physical organization of endothelial cells into tube-like structures. It thus promotes angiogenesis, the growth of new blood vessels from the pre-existing vasculature. bFGF is a more potent angiogenic factor than VEGF (vascular endothelial growth factor) or PDGF (platelet-derived growth factor). As well as stimulating blood vessel growth, bFGF is an important player in wound healing. It stimulates the proliferation of fibroblasts that give rise to granulation tissue, which fills up a wound space/cavity early in the wound healing process.

Peptide growth factors and their receptors

A large number of peptides and proteins are today known to modulate cell functions, such as growth, differentiation, motility and protein expression and secretion. These hormones and growth factors can be categorized in several ways, one is by the type of receptors that they activate. The receptors can be grouped into three main groups: 1) Receptors with intracellular kinase activity, 2) Cytokine receptors without kinase activity, 3) Receptors with seven transmembrane domains. These categories do not cover all receptors and receptor-like molecules. For instance, receptors with guanylyl cyclase activity and receptor-like molecules with phosphatase activity have also been identified.

Since the thesis is focussed on the fibroblast growth factors which group into category 1, I will only briefly describe receptors of the two other categories, and then focus on receptors with intracellular kinase activity.

Fibroblast growth factors

The fibroblast growth factors (FGF) comprise a growing family 14 members (or of 10 members (230) and 4 homologues factors (301)), showing amino acid sequence homology and heparin affinity. The first of these polypeptides to be identified were acidic FGF (aFGF, or FGF1) and basic FGF (bFGF, or FGF2) (reviewed in (28)). Due to many different sources used for their purification as well as several target cell types, many different names have been used for these growth factors. Also their affinity for heparin has made heparin-binding growth factors (HBGF) a commonly used name.

aFGF and bFGF, often referred to as the FGF prototypes, are polypeptides containing 154 amino acids after the initiator methionines are cleaved off (28,140). The open reading frame of aFGF is flanked by termination codons, and the initiator methionine is in a Kozak consensus sequence for initiation of translation (167). During purification from tissue, full-length aFGF(1-154), also referred to as HBGF-1b, can be cleaved after glycine-20. aFGF(21-154) (or HBGF-1a) is thereby formed (28). With few exceptions (202), these two forms of aFGF have similar properties. Experimentally, aFGF(21-154) is often used, as is the case in papers II-VIII, and this truncated aFGF will in the present thesis be denoted simply aFGF.

The initiation methionine of bFGF is also in a Kozak sequence, but it is not flanked by termination codons. Instead, there are 3 alternative upstream CUG-initiation codons in the cDNA, giving rise to N-terminally extended isoforms that start with the amino acid leucine. These leucine isoforms, often referred to as high molecular weight forms, tend to localize to the nucleus instead of being secreted like the methionine isoform (see below, under AbFGF in ADiscussion).

A common feature of FGFs is conservation of 2 cysteine residues, which are in position 30 and 97 in aFGF. aFGF has one additional non-conserved cysteine, and bFGF has two. The cysteines in aFGF do not form intramolecular disulfide bonds.

Cellular responses to FGF


When FGF is added to cells in vitro, several different responses can be seen, partly dependent on which cells are used. The FGFs are potent inducers of DNA synthesis in a multitude of cells. Cells of all three germ-layers are responsive. Often concentrations as low as 100 pg/ml - 1ng/ml of aFGF give optimal stimulation of DNA-synthesis. This cellular response will be further discussed in the general discussion as a relationship between transport of aFGF to the nucleus of cells and mitogenesis is a central issue of the thesis.


FGFs also stimulate cells to migrate chemotactically (48,174,276). This is of importance both in angiogenesis and in wound healing. FGF induced chemotaxis and mitogenesis require mutually exclusive cytoskeletal arrangements. In BALB/c 3T3 cells high concentrations of aFGF stimulated chemotaxis while lower concentrations stimulated cell proliferation

Protease secretion

Furthermore, the FGFs stimulate cells to secrete proteases such as plasminogen activator , collagenase and gelatinase. Plasminogen activator cleaves a peptide bond in the serum protein plasminogen, converting it to the protease plasmin. Such proteases can digest basal membranes, which may obstruct cell migration.

Together, these FGF stimulated cellular functions, cell proliferation, migration and protease secretion, provide the basis for angiogenesis which is an important physiological function of FGF.


FGF also stimulates certain cells to differentiate. Differentiation has been studied extensively in PC12 pheochromocytoma cells. FGF induces neurite outgrowth in these cells. Similar effects have also been studied in other cell lines and in primary neurons. FGF can stimulate endothelial cells to form tubular structures when grown in special collagen gels. Lens epithelial cells differentiate into lens fiber cells under the influence of FGF in vitro and in vivo . FGFs induce differentiation of preadipocyte fibroblasts into adipocytes . Experiments with mice expressing a dominant negative FGFR specifically in keratinocytes, indicate that FGF/FGFR is of importance for differentiation also of these cells.

There are also examples of differentiation events which are inhibited by FGF, such as the myogenic differentiation of MM14 skeletal muscle myoblasts. FGF can also protect cells from undergoing apoptosis (programmed cell death).

Secretion of FGFs

aFGF, bFGF , FGF9 and FGF11-14 , as opposed to other FGFs, lack a signal sequence required for secretion through the classical endoplasmic reticulum (ER)-Golgi apparatus pathway. Therefore, much effort has been put into determining the mechanisms of secretion of these FGFs.


In response to heat shock (incubation at 42 oC for 2 h) aFGF is secreted in a biologically inactive form unable to bind heparin. Secretion is dependent on cysteine residues in aFGF and the secreted form exists as a homodimer which requires treatment with reducing agents to be biologically active. It is known that after copper ion catalysed oxidation in vitro, aFGF exists as a biologically inactive, disulfide bonded heterodimer.

Recently, it was discovered that the secreted aFGF homodimer exists in a high molecular weight complex together with the extravesicular domain of synaptotagmin-1 (p40 Syn-1) and the S100A13 protein (37,318). Synaptotagmin-1 (Syn-1) is a transmembrane protein of synaptic vesicles involved in regulation of vesicle transport while S100A13 is a small calcium binding protein. Release of aFGF in response to heat shock depends on p40 Syn-1 since both the use of an antisense Syn-1 gene and a Syn-1 deletion mutant lacking 95 amino acids of the extravesicular domain repress aFGF export. Furthermore, amlexanox, an anti-inflammatory drug that binds to S100A13, inhibits the release of both aFGF and p40 Syn-1. Although the actual mechanism of aFGF export is not clarified by these findings, they point out Syn-1 and S100A13 as important players. The data also suggest that interaction with phosphatidylserine may be relevant since all three proteins involved bind phosphatidylserine, while the Syn-1 deletion mutant that repress aFGF export, does not.

aFGF in high molecular weight complexes requiring reducing agents for biological activity, is reported to be secreted to the medium from HIV-TAT-transformed primary murine embryonic fibroblasts independent of heat shock. Primary murine embryonic fibroblasts retrovirally transduced with aFGF cDNA also released aFGF in similar high molecular weight complexes in response to serum starvation. Treatment of cells with oxidized LDL is reported to trigger aFGF release, although in this case secreted aFGF was found as a monomer.


bFGF is secreted through an energy dependent non-ER-Golgi pathway. Recently, the a-subunit of Na+,K+-ATPase, which is the target of inotropic drugs like digoxin and digitoxin, was shown to play a central role in this pathway. Cardenolides (cardioglycosides like digitoxin and their aglycone derivatives) which bind to and inhibit the Na+,K+-ATPase a-subunit, inhibited bFGF export. Also cotransfection of bFGF with the a-subunit had this effect. Furthermore, the a-subunit and bFGF were coimmunoprecipitated with each other. These data do not reveal the mechanisms by which bFGF is exported, but indicate that the a-subunit of Na+,K+-ATPase is central in this process.

Fibroblast growth factor receptors

Three classes of cell surface binding proteins for FGFs have been recognized. Best studied are the high affinity receptors, normally denoted fibroblast growth factor receptors (FGFR). Fibroblast growth factors also bind to heparan sulfate proteoglycans (HSPG) with lower affinity, but with high capacity . A third class of binding protein has also been caracterized, the so-called cysteine-rich receptor.

High affinity FGFR

The high affinity FGFRs encompass a family of four different genes, FGFR1-4. The overall amino acid identity between the human genes is in the range of 50-75 percent. FGFR 1 and FGFR 2 are most identical, while FGFR 1 and FGFR 4 are least identical.. FGFR 1-3 all have different splicing variants, making the FGFR family rather complex.

The FGFRs (Fig. 4) all consist of an extracellular domain, a single transmembrane (TM) domain and an intracellular domain. The extracellular domain contains 2 or 3 immunoglobulin like (Ig) domains, dependent on splicing. Between Ig domain I and II is a stretch of acidic residues often referred to as the acid box. It also contains an essential heparin binding domain in the second Ig-domain. In addition, a signal peptide at the amino-terminus is cleaved off after translocation of the newly synthesized receptor into the endoplasmic reticulum. The intracellular part of the receptor contains a juxtamembrane (JM) stretch and a C-terminal tail (C-tail) in addition to a split tyrosine kinase domain. Seven tyrosine residues in the intracellular part of FGFR 1 have been shown to be phosphorylated upon receptor activation.

Fig. 4. Schematic view of the FGFR.(143)

Numbers in parenthesis indicate the last amino acid of each domain in the FGFR1 protein.

As mentioned above, FGFR1-3 have several different splicing variants.The receptors can have two or three Ig-domains. Two Ig-domain variants may or may not have the acid box, and the receptor may even lack the transmembrane and intracellular parts. What appears to be physiologically most important, however, is the identity of the membrane proximal half of the third Ig-domain, which is termed IIIb or IIIc. IIIb and IIIc isoforms have different ligand binding specificities and are expressed in a different spatiotemporal pattern during development. This makes up a family of seven different FGFRs (see table 1).

Table 1. Ligand specificity of the FGF receptors (54). The specificity was determined by dose response titrations of FGF-mediated proliferation. Mitogenic activity was normalized to that of aFGF (100%): +++,>70%; ++,46-70%; +,20-45%; -,<20%. The chromosome location for each receptor gene is given in brackets.

Heparan sulfate proteoglycans

Proteoglycans are large molecules found predominantly on the cell surface and in the extracellular matrix. They consist of a protein part and a carbohydrate part. The protein part may be transmembrane or it may be linked to the membrane by a GPI (glycosyl-phosphatidylinositol) anchor. The carbohydrate parts, called glycosaminoglycans, are polymers of disaccharide repeats, which are mostly highly sulfated and negatively charged. The main glycosaminoglycans in proteoglycans are chondroitin sulfate, dermatan sulfate, heparan sulfate, heparin and keratan sulfate. Heparan sulfate contains regions of low sulfation separated by regions of high sulfation, while heparin is more uniformly highly sulfated.

It appears that heparin or heparan sulfate proteoglycans are required for FGF-induced biological responses. However, exactly what is the basis for this requirement is somewhat unclear. Heparin can potentiate the biological activity of aFGF, but not of bFGF in cells expressing heparan sulfate proteoglycans. A requirement for heparin or heparan sulfate proteoglycans for binding of bFGF to the high affinity FGFR was reported. In contrast, also binding between bFGF and FGFR in the absence of heparin or heparan sulfate proteoglycans was observed. aFGF appears to be able to bind to FGFR in the absence of heparin or heparan sulfate proteoglycans. Furthermore, a heparin-binding sequence in Ig domain II of FGFR1 was also discovered and suggested to be pivotal for FGF-FGFR binding, and heparin was reported to be able to activate FGFR4, but not FGFR1, in the absence of FGF. Finally, heparin or heparan sulfate proteoglycans protect aFGF and bFGF from inactivation in the extracellular milieu.

Despite of some discrepancies in the detailed explanation for the requirement for heparin or heparan sulfate proteoglycans, the leading hypothesis on FGFR activation by FGF in concert with heparin or heparan sulfate proteoglycans is as follows: Several FGF ligand molecules bind at different sites of a HSPG molecule (or of a soluble heparin molecule often included under experimental conditions). Each of these HSPG-bound FGF molecules also bind one FGFR molecule, thereby bringing several FGFR molecules in close contact with each other (dimerizing the receptor, see under receptor tyrosine kinases), leading to activation of the intracellular kinase, which in turn activates several different signal

transducing pathways.

Cysteine-rich FGF receptor

The third class of membrane-bound FGF-binding proteins, the so-called cysteine-rich FGF receptor, consists of a large extracellular domain comprising 16 cysteine-rich repeated units, a transmembrane part and a small intracellular domain of only 13 amino acids. The cysteine-rich FGF receptor molecule appears to contain one binding site where FGF binds directly, and one where FGF binds via haparan sulfate. The cysteine-rich FGF receptor was reported to bind FGF with an affinity of -10-9 M, while FGFRs and Heparan sulfate proteoglycans exhibit approximately one order of magnitude higher and lower affinities, respectively.

A protein with high sequence identity, possibly identical, to the cysteine-rich FGF receptor, containing a 70-amino acid N-terminal extension, was shown to be a ligand for the cell adhesion molecule E-selectin. The cysteine-rich FGF receptor was suggested to regulate intracellular FGF levels. No further clear functions have been attributed to this receptor.

FGF receptor signal transduction

Activation of the kinase

The tyrosine kinase of FGFR is inactive when the molecule exists as a monomer. Upon ligand binding and oligomerization of the receptor, the kinase is activated. How is this accomplished? The answer to this question is not clear in every detail, but determination of the crystal structure of the kinase domain of FGFR1 has given some hints.

A kinase contains two functionally important binding sites, the ATP binding site and the substrate binding site. The access to one or both of these sites are blocked in other kinases as an autoinhibitory mechanism. In FGFR1 kinase the ATP binding site is accessible, while the substrate binding site is blocked by two amino acids (Arg-661 and Pro-663) in the so-called activation loop. This loop is rather mobile, and it is hypothesized that it may take several conformations, some of which are compatible with substrate binding. In other words, most of the time the receptor is inactive, but every now and then it takes the active conformation, and if a substrate (another FGFR molecule) then is close, it may be phosphorylated. The activation loop contains two tyrosine residues (Tyr-653 and Tyr-654) which, upon phosphorylation, upregulates the kinase activity.

After upregulation of the FGFR kinase by phosphorylation of Tyr-653 and Tyr-654, also other tyrosine residues become phosphorylated. In FGFR1, Tyr-463, Tyr-583, Tyr-585, Tyr-730 and Tyr-766 (122,210,212) are additional residues shown to be phosphorylated.

Downstream signalling cascades

Phospholipase Cg

Through its SH2 domain phospholipase Cg (PLCg) has been shown to bind directly to FGFR1 at phosphorylated Tyr-766. PLCg is itself tyrosine phosphorylated upon binding, and phosphorylation enhances its catalytic activity and is required for activation of PLCg in intact cells. PLCg hydrolyses phosphatidylinositol bisphosphate to diacylglycerol and inositol trisphosphate.

Inositol trisphosphate mediates release of Ca2+ from the endoplasmic reticulum by binding to specific inositol trisphosphate receptors. The intracellular Ca2+ concentration modulates the activity of numerous proteins and is involved in regulation of a vast array of biological processes.

Diacylglycerol activates protein kinase C (PKC) isoforms of the conventional and novel PKC subfamilies. The conventional PKC isoforms also depend on Ca2+ for activity. PKC also has a plethora of functions, such as regulation of glycogen metabolism, regulation of transcription and cell growth as well as a role in learning and memory. PKC is also involved in desensitization of growth factor receptors. Phosphorylation by PKC of the EGF receptor decreases the affinity of the EGF receptor for its ligand. Phosphorylation by PKC of threonine-424 in Xenopus FGFR1 has been shown to mediate downregulation of receptor signalling. Both aFGF and bFGF can be phosphorylated by PKC in vitro and in vivo.

The biological function of PLCg activation by FGFR, however, is not clear. There are several reports that mutation of tyrosine-766 into phenylalanine, which renders FGFR1 unable to activate PLCg and stimulate phospholipid turnover, has little or no effect on functions such as mitogenesis and neuronal differentiation. Yet, other reports find that PLCg activation is pivotal. This is further discussed below (under ARole of FGFR in transmitting signals from cell adhesion molecules).

The Ras/MAP kinase pathway

Two proteins seem to be able to link activation of FGFR to activation of the Ras/MAP kinase cascade, Shc and FRS2. Shc (SH2 domain and collagen-like)is tyrosine phosphorylated upon FGFR1 activation, and so is FRS2 (FGF receptor substrate 2). Shc also becomes phosphorylated upon activation of other receptors, such as PDGF receptor (45) and insulin receptor. FRS2, which is a myristoylated, PTB domain containing protein, is phosphorylated in response to nerve growth factor and aFGF, but not in response to PDGF, EGF or insulin. Possibly, a novel drosophila protein, Dof (Downstream of FGFR), has a similar role as FRS2 and Shc.

Fig. 5. The Ras/MAP kinase pathway. RTK; receptor tyrosine kinase.

Both Shc and FRS2 can then recruit the adaptor protein Grb2 (growth factor receptor bound 2). Grb2 can bind directly to activated PDGF and EGF receptors, but not to FGFR, and recruitment of Grb2 in FGFR signalling therefore appears to require adaptors such as Shc and FRS2. Grb2, in turn binds through its SH3 domain to the guanine nucleotide-releasing factor SOS (mammalian homolog of the Drosophila son of sevenless protein). SOS stimulates GDP/GTP exchange on Ras, which is thereby activated. Growth factor independent activation of Ras by SOS can be accomplished by targeting of SOS to the membrane.

GTP-bound Ras activates the so-called MAP (mitogen-activated protein) kinase cascade by binding to the N-terminal domain of the serine/threonine kinase Raf . As for SOS, membrane targeting of Raf has been reported to be sufficient for its activation. Raf then phosphorylates and activates the dual specificity kinase MEK (or MAP kinase kinase) which in turn phosphorylates and activates ERK1 and ERK2 (extracellular signal-regulated kinase, or MAP kinase). The ERKs can phosphorylate cytosolic substrates such as phospholipase A2, but a fraction of activated ERKs translocates to the nucleus

In the nucleus the ERKs phosphorylate transcription factors, such as Elk-1 and SAP 1 which bind to the serum response element, a major regulatory element of many promotors, for instance the c-fos promotor. Phosphorylation regulates transcription factor activity, thereby regulating expression of proteins required for a given response.

It appears that also alternative pathways to MAP kinase downstream of Raf may be used for FGFR signalling.

Src family kinases

Src family kinases are non-receptor tyrosine kinases that have been implicated in cell transformation and mitogenic signalling. c-Src (hereafter denoted Src) itself was identified as an oncogene,v-Src, from Rous sarcoma virus (3). The Src family contains 9 members, with a common modular structure (see Fig. 6). N-terminally there is a unique region (U), which varies between family members. Then follows a SH3 domain and that binds to proline-rich sequences forming a left-handed helix and a SH2 (Src homology 2) domain which binds to phosphorylated tyrosine residues. C-terminal to the SH2 domain is the kinase domain (SH1 domain), and then follows a regulatory domain, containing a conserved tyrosine (Tyr-527 in Src) which, upon phosphorylation by Csk (C-terminal or cellular Src kinase) turns off the enzyme.

Fig. 6. Schematic structure of Src . See text for explanation.

Kinase activity of Src is turned off due to (at least partially) an intramolecular binding between the SH2 domain and the phosphorylated Tyr-527, and it is believed that phosphorylated tyrosine residues of activated growth factor receptors replace Tyr-527 in binding to the SH2 domain, thereby transiently activating the enzyme. Src family members bind directly to the PDGF receptor and to the receptor for colony stimulating factor 1. Src is also regulated in other ways, for instance it is activated by phosphorylation of Tyr-416 in the activation loop of the kinase.

Src is attached to the plasma membrane through N-terminal myristoylation (M in Fig. 6), a posttranslational modification required for the transforming activity of v-Src. N-terminal myristoylation is not considered to be sufficient for stable membrane association, and in Src a cluster of positively charged residues contributes to membrane localization by interaction with the negatively charged phospholipid headgroups of the inner surface of the membrane. Several other Src family kinases are both myristoylated and palmitoylated. In its inactive state, Src is mainly found associated with detergent soluble, vesicular membranes, while upon activation it relocates to the plasma membrane, particularly to focal adhesion sites. At focal adhesion sites, the cell is bound tightly to the substratum. Intracellularly, the actin cytoskeleton is anchored at these sites and they contain many signalling molecules.

Upon long term aFGF exposure of BALB/c 3T3 cells, tyrosine-phosphorylated proteins of about 60 and 80-85 kDa were observed, among others. The latter was later identified as cortactin, an f-actin binding protein originally isolated as a substrate for v-Src, and the former was identified as Src. Upon activation, FGFR1 was shown to interact with Src in a biphasic manner and Src was shown to interact with cortactin with similar kinetics. Also FGF-induced tyrosine phosphorylation of cortactin exhibited similar biphasic kinetics. These data strongly suggest that upon activation of FGFR, it binds Src, which thereby is activated and phosphorylates cortactin, possibly among other substrates. A number of proteins are phosphorylated by Src, among them several cytoskeletal proteins, and this may be of importance for reduced substratum attachment observed upon PDGF stimulation or in v-Src transformed cells. Furthermore, impaired phosphorylation of Src and cortactin upon FGF stimulation was shown to correlate with inability to proliferate in response to FGF in senescent HUVEC (human umbilical vein endothelial) cells.

Also another group reported alterations in tyrosine phosphorylation of Src upon FGF stimulation, although in this case variations between different cell lines were observed. In some cell lines Src phosphorylation was increased in response to FGF, while the opposite was the case in other cell lines. In this study, coprecipitation of Src with the FGFR was not observed. Evidence for a role of FGFR-stimulated PKC in Src downregulation was also obtained.

Growth factor induced entry into S-phase after serum starvation can be blocked by overexpression of kinase-inactive Src. Constitutive expression of the transcription factor Myc has been reported to overcome this block, suggesting Src in a signal transduction pathway leading to Myc transcription. However, the precise role of Src in growth factor signalling is not clear.

Phosphoinositide 3-kinase

Phosphoinositide 3-kinases (PI 3-kinases) are a family of enzymes reported to be involved in regulation of diverse processes such as mitogenic signalling and cell transformation , vesicle transport, cell motility and invasiveness, insulin regulated glucose uptake  and apoptosis, to mention some. PI 3-kinases phosphorylate phosphoinositides at position 3 of the inositol ring, thereby, depending on the substrate, forming either phosphatidylinositol -trisphosphate [PtdInsP3], PtdInsP2 or PtdInsP (see Fig. 7).

Fig. 7. Phosphoinositide 3-kinase substrates and products

The PI 3-kinases have been divided into three classes due to substrate specificity. Class I will phosphorylate all three substrates in vitro, but it is believed that in vivo the preferred substrate is PtdInsP2. Class II phosphorylates PtdIns and PtdIns(4)P, but not PtdInsP2, while class III phosphorylates only PtdIns. Class I can be further subdivided into class IA and class IB by which type of adaptor proteins are used to link the enzymes to upstream regulatory signals. Class IA molecules bind to, and are regulated by SH2 domain containing adaptors of which the 85kDa regulatory subunits are best characterized. Class IB molecules are activated by bg-subunits of heterotrimeric G-proteins. Also other adaptors associate with these molecules .

It is the class IA PI 3-kinases that have been involved in tyrosine kinase receptor signalling. The prototype of this subclass is the p85/p110 heterodimer. The regulatory p85 subunit contains one SH3 domain, two SH2 domains, one bcr homology region which may bind to small G-proteins, two proline-rich regions known to bind to SH3 domains and the inter-SH2 region that mediates binding to the catalytic subunit. The catalytic p110 subunit contains a p85 binding site, a Ras binding site, a PI kinase region (PIK) of unknown function and the C-terminal kinase domain (see Fig. 8).

Fig. 8. Schematic drawing of the p85/p110 PI 3-kinase associated with the intracellular domain of a tyrosine kinase receptor.

The main mechanism of p85/p110 PI 3-kinase activation is believed to be interaction of the p85 SH2 domains with phosphorylated tyrosine residues of growth factor receptors, such as the PDGF receptor, or their substrates, such as insulin receptor substrate 1. The phosphorylated tyrosine residues that bind the p85 SH2 domains are in the consensus sequence pTyr-X-X-Met. This sequence is conserved in several growth factor receptors and their substrates, and are often present in more than one copy.

By binding to the catalytic p110 subunit, active Ras can stimulate enzymatic activity of PI 3-kinase. Also Cdc42 and Rac, two other small G-proteins, have been reported to activate the enzyme by binding to the regulatory subunit. Tyrosine phosphorylation of the regulatory subunit of PI 3-kinase also has a stimulatory effect, while serine autophosphorylation downregulates the enzyme.

In FGFR1, the pTyr-X-X-Met sequence is present in one copy, at Tyr-730, and it is conserved in all four FGFR genes. As a pentapeptide at high concentrations, this sequence was shown to inhibit binding in vitro of PI 3-kinase to activated PDGF receptor. However, a direct interaction of FGFR and PI 3-kinase has been difficult to demonstrate. There is one recent report that in extracts from Xenopus blastulae, p85 of PI 3-kinase was found to co-immunoprecipitate with FGFR1.

There have also been some conflicting reports on whether PI 3-kinase is stimulated by activated FGFR or not . However, a relatively weak and transient PI 3-kinase stimulation upon FGFR activation seems to take place. It appears that both the cell types used and the applied methods for measurement of PI 3-kinase activity are important in this context. Thus, in the neuroectoderm-derived cell line SKF5, bFGF stimulated PI 3-kinase activity in whole cells, but not in anti-phosphotyrosine immunoprecipitates. In contrast, in anti-phosphotyrosine immunoprecipitates from PC12 cells, PI 3-kinase activity was indeed stimulated by pretreating the cells with bFGF.

Also signal transducing molecules downstream of PI 3-kinase, such as Akt (also called protein kinase B and RAC)(and paper VII) and p70 S6-kinase are activated in response to FGF stimulation. We have obtained evidence that PI 3-kinase activity is required for transport of externally added aFGF to the nucleus and for transport from the cytosol to the nucleus of a fusion protein of aFGF and diphtheria toxin A-fragment, as well as for stimulation of DNA synthesis by aFGF (paper VII).

In addition to the signal transduction pathways described above, stimulation of cells with FGF has been reported to stimulate several enzymes (124). For example, phospholipase A2 was activated through the MAP kinase pathway (277) and this effect also seemed to depend on a heterotrimeric G-protein (276). Also, a 85 kD serine kinase of unknown function was shown to co-immunoprecipitate with FGFR4 (325). In addition, so-called cross-talk between the different signalling systems (modulation of one signalling cascade by elements grouped in another signalling cascade) takes place (58,170,327).

Fibroblast growth factors in development

In addition to the important roles the FGFRs play in development and growth of the skeleton (see below), the FGFs also are crucial in several other processes during embryonal and fetal development.

FGF family members are well known inducers of mesoderm. Addition of FGFs induces mesoderm and blocking FGFR by expression of a dominant negative FGFR leads to deficient mesoderm formation. An aFGF mutant (Lys132Glu, see paperV) with strongly reduced mitogenic activity, but still able to activate the FGFR was as effective as the wild-type aFGF in mesoderm induction in Xenopus animal caps. Targeted disruption of the FGFR1-gene in mice did not block mesoderm induction, but the mesodermal patterning was aberrant. These knock-out mouse embryos displayed severe growth retardation, defects in axial structures and died prior to or during gastrulation.

The FGFs have also been involved in the patterning of the central nervous system (CNS). FGF is able to induce formation of the neuroectoderm. Furthermore, addition of FGF at a site along the formed neuroectoderm in animal caps induced a more posterior CNS identity. Therefore, FGF is thought to be a posteriorizing factor in the CNS. When animal caps were treated with FGF together with a anteriorizing signal, such as noggin, the neuroectoderm was organized into a well-formed anteroposterior axis . FGFs have also been implicated in regulation of CNS development at later stages.

Two lines of evidence support a role for FGF in the molecular cascade leading to limb induction. First, when an FGF-soaked bead (aFGF, bFGF or FGF 3) was applied to the flank of the chick embryo, an additional limb was formed. The entire flank could produce additional limbs, but wings were formed anteriorly and legs posteriorly. Second, limb buds failed to emerge in mouse embryos with a targeted disruption of the FGFR 2 gene. Both FGF 4 and FGF 8 appear to play roles in the complex process of limb induction and development, but possible molecular mechanisms will not be further discussed here.

Mice expressing a dominant negative FGFR specifically in keratinocytes, suffered from epidermal hyper-thickening. Keratinocyte organization was disrupted, indicating that FGF/FGFR is of importance also for development of the skin. FGF 5 and FGF 7 (keratinocyte growth factor) have been shown to be regulators of hair growth. Furthermore, skin manifestations are found in several of the FGFR-associated craniosynostosis syndromes described below. In Crouzon syndrome and Skeletal-Skin-Bone syndrome, some of the patients suffer from acanthosis nigricans, which is thickening and hyperpigmentation of the skin. In Beare-Stevenson cutis gyrata syndrome, acanthosis nigricans is combined with furrowed skin. In Apert syndrome, atypical generalized acne can be seen. Interestingly, the same mutation in FGFR2 that causes Apert syndrome was recently detected in localized acne, demonstrating a somatic mutation causing epidermal mosaicism in that case.

Skeletal disorders associated with FGFR mutations

Over the last years it has become evident that mutations in the FGFRs cause several inherited skeletal disorders. These are syndromes where either growth of the long bones is affected (chondrodysplasias) or fusion of the cranial sutures is premature (craniosynostosis).

Achondroplasia, which is the most common genetic form of dwarfism, is clinically characterized by a pronounced shortening of the proximal long bones of the limbs, relative macrocephaly, exaggerated lumbar lordosis and other skeletal abnormalities. Although achondroplasia is inherited as an autosomal dominant trait with 100% penetrance, most of the cases are sporadic. Achondroplasia homozygotes have a much more severe phenotype and rarely survive beyond early infancy (295). In 99% of achondroplasia cases analysed, glycine-380 in the transmembrane domain of FGFR3 is substituted with an arginine (54,295). In a few individuals, other missense mutations have been identified that also reside within or close to the transmembrane domain of FGFR3, such as a change of glycine-375 into cysteine.

A more severe form of neonatal dwarfism, thanatophoric dysplasia (TD), also results from mutations in FGFR3. In these patients the ribs, as well as the arms and legs, are shortened. Based on differences in manifestations such as curved short femurs and cloverleaf shaped skull, TD patients are grouped into type I and type II. TD type I either have missense mutations that introduce a cysteine residue in or around the third Ig-domain, or a mutation in the stop codon resulting in a carboxy-terminal extension of the protein. TD type II is caused by a single point mutation, lysine-650 in the kinase domain of FGFR3 into glutamic acid.

A less severe form of dwarfism is hypochondroplasia, which has no associated craniofacial abnormalities. Also here, only one missense mutation in the kinase domain of FGFR3 is found, asparagine-540 into lysine.

Whereas mutations only in FGFR3 have been found in the condrodysplasias, mutations causing cranial synostosis syndromes appear in FGFR 1, 2 and 3. There are several such FGFR-associated cranial synostosis syndromes, often named after the clinicians who first described them. In most of these syndromes several different mutations are found, and also mutations in different FGFRs are found to cause the same syndrome.

Premature fusion of the cranial sutures results in an abnormal skull shape usually recognized as a tall forehead, widely spaced and prominent eyes, and mid-face hypoplasia. Different syndromes are distinguished by the severity of the craniofacial anomalies and associated limb and skin manifestations. They are inherited as autosomal dominant traits.

Mutations seen in cranial synostosis are often missense mutations that cause a single amino acid change, but also deletions, insertions and splice site mutations are found. Most of these mutations are localized in the third Ig-domain or membrane proximal to it. Also mutations in the transmembrane domain as well as a few mutations in other parts of the FGFR-molecule are found.

What appears to be common for most, if not all, of the FGFR mutations found in both chondrodysplasia and cranial synostosis syndromes, are that they cause ligand-independent activation of the receptor kinase. This is accomplished in different ways. In several cases the mutation creates a free cysteine-residue (not involved in an intra-molecular disulfide bond) in the extracellular part of the receptor molecule. It is believed that such free cysteine residues can form intermolecular disulfide bonds, thereby dimerizing and activating the receptor tyrosine kinase independently of ligand. Mutations in the transmembrane receptor domain, such as the one seen in achondroplasia, are thought to facilitate intermolecular hydrogen bonds which also may allow ligand independent receptor dimerization. The mutations found in the kinase domain of the receptor are thought to directly activate the kinase.

This hypothesis is further strengthened by the finding that when the gene for FGFR3 was disrupted in mice (AFGFR 3 knock out mice), the animals suffered from severe and progressive bone dysplasia with enhanced and prolonged endochondral growth (the process by which the long bones grow).

It therefore appears that FGFR 3 is an important negative regulator of long bone growth, and that FGFRs 1, 2 and 3 are involved in skeletal maturation and development.

FGF in cancer, angiogenesis and atherosclerosis

FGFs have been implicated in a variety of human neoplasms. Several of the FGFs were discovered because of their oncogenic potential. FGFs can directly promote tumour cell growth due to their mitogenic activity and they can stimulate formation of new blood vessels, required for a tumour to grow thicker than 1-2 mm. Below this size, diffusion of nutrients and waste products can keep the tumour cells alive. The mitogenic and angiogenic modes of action may function additively.

Normal melanocytes require the synergistic action of several growth factors to proliferatein vitro. Among these are members of the FGF family. Characteristic to melanomas, on the other hand, is the ectopic expression of growth factors, in particular bFGF , which stimulate melanoma cells to grow through an autocrine loop. Melanoma cell proliferation can be inhibited by interfering with bFGF action in various ways, such as by using peptides that block interaction with receptor, antisense oligonucleotides or neutralizing antibodies. Aberrant expression of bFGF appears to be an early step in the development of malignant melanoma, since it can be detected already in atypical naevi in situ . Expression is due to gene activation rather than amplification or rearrangement, and the bFGF promoter has been reported to be negatively regulated by p53, a central tumour suppressor.

In human pancreatic cancer, a disease with poor prognosis, several growth factors are frequently overexpressed, such as aFGF, bFGF, FGF5 and FGF7. The former two are mainly overexpressed in the cancer cells, while the latter two are predominantly found in surrounding cells. Also all four FGFRs are expressed in cancer of the pancreas. It appears that the two Ig-domain containing forms are often expressed in cancerous tissue, while the three Ig-domain forms are predominantly expressed in normal pancreas.

Also in breast cancer FGFs are overexpressed and FGFRs are expressed and frequently FGFR genes are amplified. aFGF is expressed in most breast cancer tumours, and aFGF overexpressing cells demonstrate enhanced in vitro growth under estrogen-deprived conditions.

Glioblastoma multiforme is the most common and most malignant human brain cancer, and both FGFR1 and FGFR2 as well as bFGF are expressed in these cancers. Nuclear accumulation of FGFR and bFGF has been correlated with proliferation in astrocytes in vivo and in vitro. Inhibition of bFGF action in different ways has been shown to inhibit glioma growth in vitro and in nude mice.

In AIDS (acquired immune deficiency syndrome) related Kaposi sarcoma, bFGF appears to synergise with tat protein (see discussion under Aviral proteins and transcription factors@) in stimulating tumour cell growth and clinical trials with drugs that inhibit bFGF function are in progress, although these early results are not too promising.

As mentioned above, in addition to FGFs directly stimulating tumour cell growth, these growth factors are also potent angiogenic factors. By using different angiogenesis assays, bFGF and aFGF, were the first angiogenic factors to be discovered (111). Together with the vascular endothelial growth factors, aFGF and bFGF are considered the most common tumour angiogenic factors.

Tumours regularly start recruiting new blood vessels, the so-called angiogenic switch, at the stages of late dysplasia or carcinoma in situ . These neoplasms, as is the case for avascular micrometastases, may exist for extended periods of time without net expansion before the angiogenic switch, while afterwards they often grow rapidly. Before the angiogenic switch, even though the tumours are dormant, the fraction of cells dividing may be rather high. However, also the fraction of cells undergoing apoptosis is high so that cell division is balanced. In fibrosarcoma development, the angiogenic switch has been correlated to a switch to the export of bFGF.

In addition to the requirement of blood vessels for transport of nutrients and waste products, angiogenesis is also needed for the process of metastasis. Some tumours are mainly spread through the lymph, others mainly through the blood stream. A positive correlation between the vascular density of the primary tumour and the number of metastases formed has been reported in several cancers. Inhibition of FGFs may prove to be useful as cancer therapy.

FGFs are also involved in angiogenesis in other contexts than cancer. FGFs are believed to be important in wound healing , in which formation of new blood vessels is an important part of the process.

Many patients with ischemic heart disease are not candidates for coronary surgery, for instance because the occluded vessels are too small (too peripheral) to be bypassed or because they suffer from concomitant diseases making surgery impossible. These patients might benefit from angiogenesis-based therapy. In preclinical trials bFGF has proven capable of inducing angiogenesis in ischemic heart disease and the first clinical trials are also on their way. Also in peripheral vascular disease clinical trials with bFGF are starting up.

Paradoxically, in view of this, aFGF and bFGF have also been implicated as pathogenic factors in atherosclerosis. aFGF, bFGF and FGFRs have been found to be highly expressed in atherosclerotic plaques, although some conflicting data concerning bFGF expression have been reported. Oxidized low-density lipoprotein and foam cell derived lipids as well as shear stress have been shown to induce release of FGFs and it is believed that these growth factors may influence atherogenesis by stimulating endothelial and smooth muscle cell proliferation. Whether this will be a complicating factor in FGF-based angiogenesis therapy is still to early to know.

Role of FGFR in transmitting signals from cell adhesion molecules

Normally, FGFRs transmit signals into the cell upon binding of FGF-ligand. However, a growing body of evidence indicates that FGFRs can also be activated by cell adhesion molecules (CAMs). Three CAMs, NCAM, N-cadherin and L1, have been shown to activate FGFR signalling. Several studies have assayed neurite outgrowth in PC12 cells or primary neurons grown on support cells transfected with one of these CAMs. In this experimental system, each of the three CAMs stimulate neurite outgrowth .

A CAM homology domain was recognized in the FGFR and a putative CAM homology domain-binding motif in each of the CAMs was also found. Antibodies against the CAM homology domain as well as CAM homology domain-derived peptides inhibited neurite outgrowth. It was demonstrated that FGFR was phosphorylated upon CAM activation and expression of a dominant negative FGFR inhibited CAM induced neurite outgrowth, both in transfected PC12 cells and in neurons isolated from transgenic mice. The CAM homology domain overlaps with the heparin binding domain in FGFR1 mentioned above (under Aheparan sulfate proteoglycans).

It appears that interaction between CAMs expressed in the neurite and CAMs expressed in the support cells, is required for CAM-stimulated neurite outgrowth. The neuronal CAM then interacts with FGFR, which is activated and initiates several signal transduction cascades.

By using a pharmacological strategy, it was suggested that the important signalling pathway downstream of FGFR in this context, is activation of phospholipase Cg (PLCg) to generate diacylglycerol which again gives rise to arachidonic acid by the action of diacylglycerol lipase. Arachidonic acid, or some of its metabolites, are thought subsequently to activate calcium channels causing an increase in intracellular calcium which in turn activates calcium-calmodulin dependent kinase (CaM kinase). CaM kinase then is thought to phosphorylate target molecules, such as cytoskeletal proteins, involved in growth cone motility. These, mainly pharmacological, findings were supported by the demonstration that FGF induced neurite outgrowth in cerebellar neurons was inhibited by a cell permeable peptide which specifically inhibited PLCg.

However, controversy exists on FGFR-mediated neurite outgrowth. A point mutation in FGFR1 where tyrosine 766 was changed into phenylalanine, eliminated PLCg activation, but not FGF induced neurite outgrowth in PC12 cells. Tyrosine 766 was previously found to be the docking-site for the SH2 domain of PLCg on FGFR1.

It was formerly demonstrated that FGFR1 was much more potent than FGFR3 in mediating neurite outgrowth in PC12 cells, and that this difference was mediated by the transmembrane or intracellular parts of the receptor molecule (184). Based on this, several FGFR1-FGFR3 hybrid receptors were generated to pin down which part of FGFR1 is important for neurite outgrowth. In that study, tyrosine 766 and PLCg activation was found to be of moderate importance, while a membrane proximal part of FGFR1 was highly important. The presence of this membrane proximal part of FGFR1 in the hybrid receptors correlated with their ability to activate the MAP kinase pathway. Activation of MAP kinase is by many thought to be central in PC12 cell differentiation.

In yet another study, it was suggested that expression of endogenous aFGF in response to exogenously added FGF was responsible for the observed differentiation response. Neurite outgrowth did not correlate with MAP kinase activation in that study, while the expression level of endogenous or overexpressed aFGF did.

The reasons for these apparent discrepancies are not clear. The different experimental conditions used (primary neurons or PC12 cells grown on support cells versus PC12 cells grown alone in cell culture dishes) is one possibility. Also transfection of mutated receptors or aFGF may introduce artefacts due to overexpression. A third possibility is that pharmacological and peptide approaches can be less specific than one thinks.

Basic FGF and FGFR

In an the early report of FGFs entering the nucleus, it was demonstrated by indirect immunofluorescence microscopy that addition of bFGF to ABAE (adult bovine aortic endothelial) cells resulted in bFGF immunoreactivity in nucleoli, parts of the nucleus responsible for transcription of ribosomal RNA (rRNA). The investigators also found a correlation between nucleolar localisation of bFGF and increased rRNA transcription and they could demonstrate that addition of bFGF to isolated nuclei stimulated rRNA transcription. Using the same cells, these authors later reported that transport of bFGF to the nucleus was cell cycle specific. Thus, transport of bFGF to the nucleus was mainly observed in G1 phase and was not observed in density arrested cells. These data were mainly obtained by incubation of cells with 125I-labelled bFGF, followed by fractionation of cells in the absence of detergent. In vitro data demonstrated that bFGF directly binds to and appears to stimulate casein kinase II. Casein kinase II is required at rather distinct periods of time during the cell cycle, and one of its substrates is nucleolin, which is involved in modulation of rRNA synthesis. bFGF was also reported to differentially modulate gene expression in isolated nuclei, although rather high concentrations were required.

As mentioned in the introduction, upstream of the classical AUG initiation codon in the mRNA of bFGF, there are three alternative CUG initiation codons, at which initiation of translation gives rise to the so-called high molecular weight forms of bFGF. bFGF translated from the AUG initiation codon is often referred to as the Met-isoform or the 18-kD isoform. The N-terminal extension was demonstrated to be a nuclear targeting signal, since, in cells transfected with various constructs, it could mediate the nuclear translocation of bFGF as such and of chimeric proteins with b-galactosidase and chloramphenicol acetyltransferase. It contains several glycine-arginine repeats, where the arginine was shown to be methylated. One such sequence was suggested to be the minimal motif required for directing translocation to the nucleus of bFGF. Nuclear localisation of bFGF in vivo has been demonstrated by immunohistochemistry.

Reports on cellular responses to overexpression of either the 18-kD isoform or the high molecular weight forms of bFGF have been somewhat confusing. Bikfalvi et al. found that only expression of the high molecular weight forms enabled the cells to proliferate at low serum concentrations. Dono et al. reported that cells expressing the 18-kD isoform were able to proliferate at low serum concentration, while cells expressing the high molecular weight forms stayed growth arrested. However, these authors seem to agree that the level of expression is also an important determinant for the cellular response. This is in accordance with the findings of Quarto et al. who reported that cells producing either the 18 kD isoform or the high molecular weight forms of bFGF were transformed at high expression levels, but not at low ones. It was also reported that the high molecular weight forms of bFGF were preferentially expressed in transformed and stressed cells.

Upon overexpression, the high molecular weight forms of bFGF appear to enter the nucleus and do not become secreted, while the 18 kD isoform is predominantly cytosolic and is in part secreted.

Stachowiak and coworkers, who have worked with adrenal medullary cells and astrocytes, studied the role of bFGF within cells that express the growth factor, so-called intracrine effect. They found that translocation from the cytosol to the nucleus of bFGF was correlated with proliferation. In non-transformed astrocytes, contact inhibition between cells was associated with redistribution of bFGF to the cytosol, while bFGF was constitutively localised to the nucleus in glioblastoma cells. By several methods, FGFR1 was also shown to exist in the nucleus. Nuclear localisation of both bFGF and FGFR1 was augmented by forskolin treatment, a drug that stimulates protein kinase A. The reported data that PKC stimulation has the opposite effect, are somewhat difficult to interpret, since long-term treatment with PMA (or TPA, a phorbol ester) was used to stimulate PKC. Long-term treatment with PMA, on the contrary, is known to downregulate PKC.

In Swiss 3T3 cells, FGFR1 was shown to translocate from the cell surface to the nucleus in response to externally added bFGF in a time and dose dependent manner. Also a splice variant of FGFR3, in which the transmembrane part was deleted, was reported to be localised in the nucleus.

Prudovsky et al. (253) could by confocal immunofluorescence microscopy demonstrate that FGFR1 was transported to a perinuclear location in response to aFGF. Upon cell fractionation the receptor was mainly found in the nuclear fractions, demonstrating that not everything which is in a nuclear fraction necessarily is within the nucleus. The ability of FGFR1 to traffic to this perinuclear location was suggested to vary between different splicing variants.

Signal Transduction Pathway

Cardiac hypertrophy describes an abnormal condition where the heart becomes enlarged. Under stresses such as high blood pressure, or reduced blood flow through the coronary arteries, the heart must work harder. Instead of dividing and increasing in number, individual cells grow larger and genes normally expressed in the embryonic ventricle are reexpressed. Initially this compensation is effective, but excessive hypertrophy can kill more cells, which increases the stress on the heart, causing surviving cells to grow even larger, which in turn leads to an ever accelerating cycle that can eventually result in heart failure. Cardiac hypertrophy can also cause diseases such as myocardial infarction and arrhythmia, and therefore it is important to try and better understand the molecular mechanisms underlying the development of this condition.

Fibroblast growth factors (FGFs) represent one of five classes of peptide growth factors. Several in vitro studies have shown that FGF-2 (one of the many members of the FGF family) induces cardiac hypertrophy, and in vivo studies have shown that FGF-2 is naturally produced by cardiac myocytes and non-myocytes in the heart, where it plays an autocrine or a paracrine role.

The binding of FGF-2 to it's receptor FGFR - a transmembrane receptor tyrosine kinase (RTK) - induces receptor dimerisation and autophosphorylation. These phosphorylated tyrosines act as binding sites for cytosolic proteins with Src homology (SH2) domains, such as growth factor receptor bound protein 2 (GRB2). GRB2, with Son of sevenless protein (SOS) bound to it, then binds to the RTK, which activates SOS. SOS is a guanine nucleotide exchange factor (GEF) which activates the low-molecular-weight GTPase Ras, by inducing it to release GDP and exchange it for GTP. GTPase activating proteins (GAPs) accelerate the intrinsic GTP hydrolytic activity of Ras, thereby promoting the formation of the inactive, GDP-bound form of Ras (see below). Active Ras triggers a cascade of protein phosphorylation involving mitogen-activated protein kinase kinase kinase (Raf), mitogen activated ERK activating kinase (MEK), and extracellular signal regulated kinase (ERK). Upon activation, the ERKs phosphorylate cytoplasmic targets and they also translocate to the nucleus where they stimulate gene expression through the activation of transcription factors.

Through the common component Ras, activated FGFR is able to initiate several distinct cascades of protein phosphorylation (MAP kinase cascades). Other effectors such as protein kinase C (PKC) and phosphatidylinositol 3-kinase (PI3K) can also interact with the autophosphorylated tyrosine sites on the receptor, and they can also act further downstream in the signalling pathway, as activated Ras protein can induce other distinct responses, as shown in  below. It is likely that for a given stimulus, only a limited subset of all the possible interactions shown here actually occur. It has been suggested that the specific effect of a stimulus will depend on the comple