I. Background and Rationale

Top Previous Next References

In multicellular organisms, cellular interactions are controlled by highly coordinated mechanisms. These complex networks are also responsible for responses to infection and wounding, and likewise mediate normal embryonic development. Since the discovery of nerve growth factor (Levi-Montalcini, 1987) and epidermal growth factor (EGF)² (Cohen, 1986), a wide array of polypeptides involved in the modulation of cell proliferation have been identified. Such growth-controlling molecules influence cell growth and differentiation both positively and negatively. Growth factors may have either a paracrine (i.e., released by one cell type and acting on another cell) or autocrine (i.e., acting on the same cell by which it has been secreted) influence on cell proliferation. In addition, the ultimate response of a target cell to a particular growth factor is determined by the cellular context in which the stimulus is received, such that some factors are able to give rise to qualitatively different responses. This occurs during the myeloid differentiation of hemopoietic cells, whose response changes from proliferation to priming in precursor and mature cells, respectively (Cross and Dexter, 1991). However, qualitatively different results may even arise in a concentration-dependent manner from the action of a growth factor on a specific cell type (Cross and Dexter, 1991).

The activation and/or repression of a subset of genes shown to function at critical steps in mitogenic stimulus is the result of the cascade of intracellular biochemical signals ensuing from the interaction of growth factor with specific receptors. Because these growth factors are unable to cross the hydrophobic cell membrane, a fundamental question is how they transduce their signals into the cells. Growth factors exert their effects via binding to cell membrane receptors, and it has been shown in recent years that these receptors are often activated by ligand-induced dimerization or oligomerization (Heldin, 1995, 1996). Receptor dimerization is in some cases the result of the interaction between a symmetrical, dimeric ligand which binds to two receptor molecules (Heldin, 1996) (Fig. 1). On the contrary, other ligands, such as members of the EGF-family, are apparently monomers. It seems, however, that two ligands interact with two EGF-receptors in a symmetrical fashion (Lemmon and Schlessinger, 1994). Another possibility for dimerization is exemplified by basic fibroblast growth factor that binds to a high- and a low-affinity receptor.

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Ligand-induced receptor homodimerization triggering the initial step of cell signaling pathways. The juxtaposition of the cytoplasmic regions of the receptors makes it possible for the kinase domains () to phosphorylate (open square containing P) each other at the tyrosine residues (open oval containing Tyr). The positions of phosphorylated tyrosines, which serve as docking sites for adaptor molecules, are schematically represented.

Many traditional growth factors (e.g., EGF, fibroblast growth factors, and insulin-like growth factors) bind to receptors with protein tyrosine kinase (PTK) activity. All PTK receptors (PTK-R) consist of single transmembrane domains that separate the intracellular kinase domains from the extracellular portions. These latter domains contain one or several copies of Ig-like, EGF-like, or fibronectin type III-like domains (Heldin, 1995). The catalytic (kinase) domains display the highest level of conservation. Structural motifs that are conserved in this region include an ATP-binding site and a tyrosine residue, which corresponds to the major phosphate acceptor site (Yarden and Ullrich, 1988).

The mechanism of ligand-induced dimerization of PTK-R, whereby the receptors are activated, involves juxtaposition of the cytoplasmic parts of the receptors that allows the kinase domains to phosphorylate each other (Fig. 1). Autophosphorylation involves two different classes of tyrosine residues. One phosphorylation occurs on a conserved tyrosine residue within the intracellular kinase domain. It is still not completely clear how autophosphorylation is initiated; one possibility is that the monomeric receptor has a low enough basal kinase activity to phosphorylate and activate the companion receptor resulting from dimerization. This would then be followed by reciprocal phosphorylation. Otherwise, the interaction between the intracellular domains of the receptors in the dimer may induce conformational changes leading to an increased kinase activity. The secondary autophosphorylation sites are normally located outside the kinase domains and serve the fundamental function of creating docking sites for downstream signal transduction molecules containing Src-homology 2 (SH2) domains or phosphotyrosine binding (PTB) tyrosine domains. The SH2 domain is a 100-amino acid motif that folds to form a surface that recognizes phosphotyrosine and three to six C-terminal amino acid residues (Pawson, 1995). The PTB domain has been identified in the amino terminus of Shc; this adaptor protein, also containing SH2 domains, appears to be involved in ras activation. PTB domains are longer than SH2 domains. In contrast to SH2 domains, they recognize phosphotyrosine in the context of an N-terminal residue (van der Geer and Pawson, 1995). Other SH2-containing molecules include phospholipase C (PLC), phosphatidylinositol-3'-kinase (PI3K), p21ras GTPase-activating protein, and growth factor receptor binding 2 (Grb2) protein.

PI3K is a dimer of a regulatory component of 85 kDa and a catalytic component of 110 kDa which phosphorylates phosphoinositides on the D-3 position. The reaction products can act on multiple downstream effectors that include SH2 and Pleckstrin homology domains of serine/threonine and tyrosine kinases and various cytoskeletal proteins (Carpenter and Cantley, 1996). It has been shown that a signaling pathway from PI3K to the serine/threonine protein kinase Akt/PKB may mediate some cellular responses of PI3K (Burgering and Coffer, 1995; Cross et al., 1995; Franke et al., 1995), including protection from apoptosis (Dudek et al., 1997; Franke et al., 1997). Although the exact mechanisms by which Akt/PKB prevents apoptosis are not completely highlighted, it has been proposed that Akt/PKB might mediate cell survival by phosphorylation and inhibition of proteins that are involved in programmed cell death.

Grb2 is an adaptor protein containing another type of domain, Src-homology 3 (SH3), which binds specific proline-rich motifs. These sites interact with a molecule known as sos (son of sevenless), which in turn activates ras. A signaling cascade through a series of kinases is then initiated; ras activates raf-1, which in turn activates mitogen-activated protein (MAP) kinase kinase (MEK); MEK activates MAP kinase, which translocates in the nucleus and activates transcription factors myc, jun, and fos. Interaction between fos and jun leads to the transcription complex activator protein-1, which switches on a number of genes associated with cell growth and differentiation, as well as with the regulation of cell shape and chemotaxis (Yarden and Ullrich, 1988; Heldin, 1995, 1996; Langdon and Smyth, 1995) (Fig. 2).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Schematic representation of the cell signaling pathway mediated by growth factor (GF) receptors with intrinsic tyrosine kinase activity. AP-1, activator protein 1; DAG, diacylglycerol; MAP, mitogen activated protein kinase; MEK, MAP kinase kinase; P, phosphated region; PLC-, phospholipase C-; PI3P, phosphatidylinositol(3)-phosphate; PI4P, phosphatidylinositol(4)-phosphate; PI3,4P2, phosphatidylinositol(3,4,)-bisphosphate; PI4,5P2, phosphatidylinositol(4,5)-bisphosphate; PI1,4,5P3, phosphatidylinositol(1,4,5)-triphosphate; PI3,4,5P3, phosphatidylinositol(3,4,5)-triphosphate; PKC, protein kinase C; PTB, PTB domain; SH2, SH2 domain; SH3, SH3 domain.

The activation of signal transducers and activators of transcription (STATs) family members, which are latent cytoplasmic transcription factors, is an important way whereby PTK-R transduce their signal. Phosphorylation of a tyrosine residue conserved in all STATs family members induces their dimerization, which is followed by translocation into the nucleus, DNA binding, and regulation of numerous genes involved in growth and differentiation, such as p21^WAF1 (Ruff-Jamison et al., 1995; Chin et al., 1996; Xie et al., 1997) and c-fos (Sadowski et al., 1993).

After ligand binding and activation, PTK-R are deactivated by protein tyrosine phosphatases (PTP) that dephosphorylate the autophosphorylated tyrosine residues (Hunter, 1995). Several such PTP have been identified, and some of them have SH2 domains and bind to PTK-R. Dephosphorylation does not represent the only mechanism by which PTK-R are deactivated; receptors are also internalized in endosomes after assembly in coated pits. At this stage, ligands dissociate from the receptors because of the low pH and the receptor is either recycled back to the cell surface or degraded after the endosomes have fused with lisosomes (Heldin, 1996).

Dimerization has also been shown to occur after binding to serine/threonine kinase receptors (Heldin, 1995). Transforming growth factor  represents a prototype for a large family of structurally related factors that interact with such receptors. Ligand binding induces a hetero-oligomeric complex of type I and type II receptors, most likely a heterotetramer made up of two receptors of each type (Yamashita et al., 1994a). Type II receptor, which has a constitutively active kinase, first binds the growth factor. Type I is then recruited and the serine residues in the glycine-serine residues domain are phosphorylated, thereby triggering the signal transduction pathway (Heldin, 1995).

Genetic aberrations in growth factor signaling pathways are strongly connected with developmental abnormalities and a variety of chronic diseases, chief of which is cancer. Tumor evolution is a multistep process requiring, first, that the normally interdependent systems controlling proliferation and differentiation are separated and, second, that proliferation is stimulated in such a way as to result in extensive growth of the transformed cells. Malignant cells arise either from alterations that may take the form of up- or down-regulation of growth factors and/or their receptors, or from a switch from a paracrine to an autocrine mechanism of action. Alternatively, given that many growth factors use a common signal transduction pathway leading to an intracellular biochemical cascade, any mutation in these cascades may affect several growth factor pathways simultaneously. Moreover, some growth factors have the ability to induce the extension of nearby blood vessels and, hence, have the potential to contribute to tumor vascularization. In the absence of vessels, proliferation at the surface of the tumor is balanced by cell death in the center. Once such a tumor begins to release angiogenic factors, the consequent vascularization allows cancer cells to spread through solid tissues (Blood and Zetter, 1990).

In the first part of this review, we describe the growth factor families involved in the most important carcinomas in humans; for this purpose, we analyze in detail the chemistry and molecular organization of the most common growth factors and their specific receptors. The complex cascade of biochemical events ensuing from growth factor/receptor binding is also discussed. We believe that the present knowledge of the role of growth factor biochemical signaling networks in cancer is pivotal to improvements not only in diagnosis and prognosis of this disease, but also in new and more targeted therapeutic intervention. Thus, the second part of this review focuses on the novel pharmacological approaches for cancer therapy that have already been or are being developed with the aim to specifically interfere at various steps of growth factor signaling pathways.

	II. Growth Factor Families

Top Previous Next References

Human epidermal growth factor (EGF) (Table 1) is a 53-amino acid single-chain polypeptide (6 kDa) that arises from the proteolytic cleavage of a large (1207 amino acids) integral membrane protein precursor (Carpenter and Cohen, 1990). This precursor protein is composed of eight extracellular EGF-like domains, only one of which has EGF-like activity. This growth factor, whose gene is positioned on human chromosome 4, stimulates the proliferation of epithelial cells, inhibits gastric acid secretion, and is involved in wound healing. EGF is closely related structurally to transforming growth factor- and to vaccinia growth factor, both of which bind to EGF-receptor (EGF-R).

The EGF-R (1186 amino acids, 170 kDa) (Table 1), also known as c-erbB, whose gene is located on human chromosome 7, was the first PTK-R to be purified and cloned (Ullrich et al., 1984; Ullrich and Schlessinger, 1990). It is the prototype of a subfamily (subclass-I) of four members, namely erbB-2, erbB-3, erbB-4, and EGF-R itself, and is expressed by most cells. ErbB molecules are characterized by the presence of two extracellular cysteine-rich domains and an intracellular portion with a long C-terminal tail carrying most of the autophosphorylation sites. In addition to the already mentioned transforming growth factor- and vaccinia growth factor, heregulin (also termed Neu differentiation factor), amphiregulin, cripto, and schwannoma-derived growth factor bind to the EGF-R family. All of these ligands contain a conserved EGF-like domain and are synthesized as transmembrane precursor proteins (Heldin, 1996).

Binding of EGF to its receptor triggers oligomerization and an increase in receptor affinity. The tyrosine kinase domain then autophosphorylates the tyrosine residues of the receptor (Heldin, 1995). Subsequent events include tyrosine phosphorylation of other proteins, including the neu p185 receptor, breakdown of inositol lipids, and increase in calcium concentration (Petch et al., 1990). More specifically, EGF-R stimulation by its ligand leads to Ras activation that in turn activates cytoplasmic protein Ser/Thr protein kinases from both the MEK and Raf families (Zwick et al., 1999). EGF-R signaling is also mediated by STATs, which, following phosphotyrosine-dependent dimerization, bypass classical kinase cascades and enter the nucleus to directly regulate genes (Sadowski et al., 1993; Ruff-Jamison et al., 1995; Xie et al., 1997).

It has been noted that components of the EGF-R signaling pathway are over-expressed and/or activated in human breast tumors; moreover, transgenic mouse models have indicated that proto-oncogene c-myc and transforming growth factor- (the latter a member of the EGF family, as mentioned above) strongly synergize to induce mammary tumors (Derynck et al., 1987; Travers et al., 1988). In addition, more recently, Nass and Dickson (1998) examined the ability of c-myc expression in cooperation with EGF to abrogate cell cycle regulation in an in vitro mammary epithelial cell model system. These authors report that constitutive, elevated expression of c-myc in breast epithelial cells is not sufficient to force the cells through the cell cycle, but rather leads to altered cell cycle progression in response to EGF, with accelerated passage through G₁. The fundamental role of EGF-R, as well as that of its ligands, in the pathogenesis and progression of breast cancer is well recognized. On average, 45% of breast cancers investigated by 40 different groups were shown to be EGF-R positive (Klijn et al., 1992). This positivity was inversely correlated to the presence of estrogen or progesterone receptors and was associated with poor prognosis and higher risk of relapse, although this finding remains controversial. In vivo, between 20 and 60% of EGF-R expressing tumors also express an EGF-R ligand, TGF-, suggesting that an autocrine loop may be involved in the growth of such tumors (Barrett-Lee et al., 1990; Murray et al., 1993). Finally, EGF-R was recently shown to be associated with an increase in tumorigenesis and to act as a mediator of transformation, evaluated by anchorage-independent growth assay (Ma et al., 1998).

The c-erbB-2/Her-2 (Her-2/neu) proto-oncogene encodes a 185-kDa transmembrane tyrosine kinase with a marked degree of homology to the EGF-R. Since the mid-1980s, many publications have shown that Her-2/neu is a marker of poor prognosis, shorter overall survival, and biological aggressiveness in breast cancer and in several other epithelial-origin cancers. Her-2/neu receptor is commonly overexpressed and involved in autocrine loops in human breast and ovarian cancer (Slamon et al., 1989; Garrett and Workman, 1999). It has also been reported that extracellular ligand-binding domain is shed into the blood of normal individuals and is detected at high levels in women with metastatic breast cancer (Zabrecky et al., 1991). A recent work reported that: 1) c-erbB-2/Her-2 shows a lower sensitivity than CEA and CA 15.3, the most frequently employed tumor markers in breast cancer early diagnosis of relapse and follow-up, in patients with locoregional and metastatic breast tumor; 2) c-erbB-2/Her-2 is an independent prognostic factor in disease-free survival and overall survival; and 3) there is a clear relationship between c-erbB-2/Her-2 overexpression in tissue and in serum (Molina et al., 1998).

Beyond the mechanisms regulating breast cancer cell proliferation, it is necessary to bear in mind that the initial stage of breast cancer development is under hormonal control. In responsive cells, steroid hormones induce a variety of biosynthetic processes leading to increased cell growth. Estrogen-induced stimulation of cellular proliferation involves the binding of the hormone to the estrogen receptor protein that is located predominantly in the cell nucleus. The key to explaining estrogen action lies in defining the mechanisms that come into play between the initial ligand recognition event and the activation of transcription of hormone-responsive genes. On steroid hormone binding, the ligand/receptor complex binds to palindromic DNA binding sites in the vicinity of hormone responsive genes, thereby prompting transcriptional activation. Transcriptional activation by estrogen receptor appears to be mediated by at least two distinct transcription activation factors (TAF), TAF1 and TAF2. Even though TAF1 and TAF2 could potentially act independently, it appears that they interact with the intact receptor in some still undefined way. It has been demonstrated by various biological and/or immunological strategies that conditioned medium from estrogen-treated cells contains numerous proteins, such as protease growth factors and their receptors. These observations prompted interest in understanding the complex "cross talk" occurring between these multiple factors and their regulatory pathways (reviewed in de Cupis and Favoni, 1997).

Several studies have documented transcriptional and/or proliferative synergy between EGF and steroidal hormones (Krusekopf et al., 1991; Modiano et al., 1991) and have shown that progesterone up-regulates EGF-R on the cell membrane (Murphy et al., 1986; Sarup et al., 1988; Modiano et al., 1991). In particular, progestins alone cause a nearly 3-fold increase in the number of EGF-R (Murphy et al., 1986) and a 6-fold increase in EGF mRNA levels (Murphy et al., 1988) in human breast cancer cells. In addition to increasing the number of high affinity EGF-R, progesterone affects the phosphorylation state of the receptor (Modiano et al., 1991). In contrast, progesterone receptors are down-regulated by EGF, thus indicating bi-directional effects of each molecule on the receptor of the other. Moreover, this documented reciprocal regulation suggests the existence of a regulatory loop in which the loss of steroid hormone responsiveness is concurrent with a gain in growth factor-dependent proliferation, as is observed during breast cancer progression (Nicholson et al., 1994; Lupu et al., 1996).

The involvement of EGF-R and its ligand transforming growth factor- is also recognized in head and neck squamous cell carcinoma. Up-regulation of these molecules, at both the mRNA and protein levels, has been identified as an early event in head and neck carcinogenesis (Rubin Grandis and Tweardy, 1993; Rubin Grandis et al., 1996). In particular, reports show it is likely that increased TGF- protein expression precedes elevation of EGF-R (Rubin Grandis et al., 1998), indicating that increases of these proteins are chronologically, and perhaps mechanistically, distinct events in the pathogenesis of head and neck squamous cell carcinoma. In addition, EGF-R expression levels in the premalignant lesion appear to be a sensitive predictive factor of the neoplastic potential of dysplastic tissues, thus implying that the receptor protein may serve as a biological marker to identify high-risk subgroups (Rubin Grandis et al., 1998).

In prostate cancer, experimental studies show EGF-R expression to increase with progression, but clinical trials have given rise to contrasting findings. Nevertheless, up-regulation of EGF, TGF-, and EGF-R in advanced tumor suggests their autocrine expression. In addition, increased EGF levels appear to be associated with the invasive ability of prostate cancer cells (Russell et al., 1998).

Over the past decade, several reports have evinced correlation between the presence of EGF-R and invasive tumors, namely of the prostate, breast, ovary, bladder and lung (Eccles et al., 1995; Ellerbroek et al., 1998), indicating that the receptor plays a role in the malignant phenotype (Damstrup et al., 1998). Moreover, attachment to the basement membrane and cell motility represent two of the first steps involved in tumor cell invasion (Fidler and Nicolson, 1987). Recent findings by Damstrup et al. (1998) indicate that, among a panel of human small cell lung cancer cell lines, only EGF-R positive cell lines, independent of the expression of the enzymes degrading basement membrane (proteinases), had the in vitro invasive phenotype, as demonstrated by the ability to traverse a reconstituted Matrigel membrane.

Acidic fibroblast growth factor (FGF) (aFGF; 115 amino acids, 16 KDa) and basic FGF (bFGF; 155 amino acids; 16-18 kDa) (Table 1) belong to the FGF family, which also includes FGF-3 (or int-2), FGF-4 (or Kaposi FGF), FGF-5, FGF-6, FGF-7 (or keratinocyte growth factor), FGF-8 (or androgen-induced factor), and FGF-9. These growth factors are modulators of cell proliferation, differentiation, motility, and angiogenesis. Both aFGF and bFGF, whose genes are located on human chromosomes 5 and 4, respectively, have a high affinity for heparin and are found to be associated with extracellular matrix components (Basilico and Moscatelli, 1992).

FGFs bind simultaneously to both low- and high-affinity receptors (Klagsburn and Baird, 1991; Johnson and Williams, 1993). Low-affinity receptors are heparan sulfate proteoglycans. Binding of FGFs to heparin and other glycosaminoglycans protects them from degradation and can retain FGFs in the extracellular matrix as a reservoir. These growth factors bind to cell surface heparan sulfate proteoglycans, but it is not completely clear whether these are functional receptors. Four different genes are currently known to encode distinct high-affinity FGF receptors (Table 1) (K_d ranging from 2 × 10⁹ to 5 × 10¹⁰ M): FGF-R1 (flg, 801 amino acids, 160 kDa); FGF-R2 (bek, 800 amino acids, 135 kDa); FGF-R3 (784 amino acids, 135 kDa); and FGF-R4 (778 amino acids, 140 kDa) (Johnson et al., 1991). High-affinity FGF receptors are members of a complex family characterized by the presence of two or three Ig-like domains, a sequence of extracellular acidic residues, and an intracellular PTK domain with a short inserted sequence of 14 amino acids (Fantl et al., 1993). Although their precise ligand-binding specificities remain to be determined, it appears that each receptor in the family binds a subset of members of the FGF ligand family. The genes for FGF-R1 and FGF-R2, located on human chromosome 8 and 10, respectively, may undergo alternative splicing events that give rise to a great variety of products, including soluble receptors (Dionne et al., 1991). Ligand binding to FGF receptors induces dimerization and interchain autophosphorylation. The receptors bind to and activate PLC- and stimulate the phosphoinositol (PI) second messenger pathway (Heldin, 1996).

Numerous studies suggest that alterations in the expression of FGF may contribute to growth deregulation in neoplastic cells (Eguchi et al., 1992; Myoken et al., 1994b). The contribution of aFGF and bFGF to tumor development is supported by the observation that cells transfected with the aFGF or bFGF genes show increased autocrine proliferation in both monolayer cultures and soft agar (Jaye et al., 1988; Sasada et al., 1988). In addition, neutralizing antibodies and antisense oligonucleotides attenuate the growth factor action, indicating that endogenous aFGF and bFGF may contribute to neoplastic cell growth in an autocrine manner (Myoken et al., 1994a). To elucidate the contribution of the autocrine effects of aFGF overexpression to an increased malignant phenotype, Zhang et al. (1998) studied aFGF-transfected MCF-7 breast cancer cells that were retransfected with a vector encoding a truncated FGF-R1. This receptor is truncated immediately after the transmembrane domain (Amaya et al., 1991; Ueno et al., 1992). Although capable of dimerization, it will not cross-phosphorylate a paired wild-type receptor because it lacks the tyrosine kinase domain; this, essentially, nullifies the effect of ligand binding. The "double" transfected MCF-7 cells showed inhibited autocrine growth factor signaling but remained able to produce aFGF, allowing possible paracrine effects to be observed in vivo. Zhang et al. (1998) report that, in ovariectomized mice, truncated receptor expression severely inhibited the ability of aFGF overexpressing breast cancer cells to form tumors in the absence of estrogen. However, rapid formation of large tumors was still observed in estrogen-supplemented mice injected with the same "double"-transfected MCF-7 cells, thus suggesting that paracrine effects of aFGF could act in synergy with mitogenic effects mediated by estrogen. These results imply that aFGF acts as an autocrine modulator of MCF-7 breast cancer cell proliferation under estrogen-depleted conditions. In addition, this growth factor also contributes through paracrine mechanisms of action to the enhancement of tumor growth in estrogen-supplemented animals. Basic FGF, which is a mitogen and a survival factor in fibroblast and endothelial cells, is one of the primary angiogenic factors in breast cancer (Folkman et al., 1989). Paradoxically, as evinced by colony-forming assay (Wang et al., 1998), bFGF also inhibits proliferation of several breast cancer cell lines. Moreover, it has been demonstrated (Wang et al., 1998) that exposure to bFGF promotes drug-induced apoptosis in MCF-7 cells, but has essentially an opposite effect on fibroblasts (NIH3T3), suggesting that bFGF may generate different responses in breast cancer and surrounding cells.

Basic FGF appears to be produced by healthy, normal prostate stromal cells where, acting by an autocrine mechanism of action, it is important for maintaining homeostasis (Story et al., 1989; Sherwood et al., 1992). As prostate cancer occurs and progresses, the production of bFGF becomes androgen-dependent and is regulated by prostate cancer epithelial cells in an autocrine manner (Russell et al., 1998). Because bFGF is angiogenic, as previously mentioned, its increased production in late-stage disease may promote angiogenesis, thus allowing tumor growth and metastases (Folkman, 1990).

In a recent study (Brattstrom et al., 1998), elevated bFGF values were found in sera from patients with nonsmall cell lung cancer, and the authors suggest, in contrast to previous studies carried out on other malignancies, that the bFGF serum levels appear to be a statistically valid prognostic factor.

In ovarian cancer cell lines, bFGF was reported to promote tumor growth and metastases (Di Blasio et al., 1993; Speirs et al., 1993; Crickard et al., 1994), even though no overexpression of bFGF mRNA was revealed in a comparison of malignant and benign tumors (Reynolds et al., 1994). The role of intratumor bFGF as a prognostic marker was recently evaluated in patients with epithelial ovarian cancer (Obermair et al., 1998). It was suggested that cytosolic concentration of bFGF, as well as histological grading and residual tumor mass, affect the overall survival probability. In particular, tumors with high bFGF cytoplasmic levels, which revealed a much greater stromal content, were associated with improved survival. These findings imply that bFGF may induce a fibroblastic response which makes tumors with a high growth factor level less aggressive than those with less stromal tissue (Obermair et al., 1998).

The family of insulin-like growth factor (IGF) ligands includes the single chain polypeptides IGF-I (70 amino acids; 7.6 kDa) and IGF-II (60 amino acids; 7.4 kDa) (Table 1), whose genes are located on human chromosome 12 and 11, respectively (Favoni et al., 1994a). IGFs, which share structural similarity with insulin, play an important role in regulating cell proliferation and differentiation and in suppressing the cell apoptotic pathway (Parrizas and Leroith, 1997). In the fetus, both IGF-I and IGF-II are found at low levels in the serum as well as in most tissues, especially those of mesenchymal origin (Bondy et al., 1990). Synthesized principally by the liver, circulating IGFs are abundant in the newborn human. Although IGF-I primarily mediates the effects of growth hormone after the pituitary axis matures two weeks following birth, the endocrine role of IGF-II is still quite unclear (Voss and Rosenfeld, 1992).

Competitive binding and ligand cross-linking studies were employed to identify the receptors for these growth factors. Although both IGF-I and IGF-II bind weakly to insulin receptor, they have their own specific receptors on the cell membrane (Rechler and Nissley, 1985; Minniti et al., 1992a). The type-I IGF-receptor (IGF-R) (Table 1), a transmembrane PTK-R, binds both IGF-I and IGF-II with high affinity (K_d 1 and 3 nM, respectively) and is also able to bind insulin, albeit less efficiently (100-fold lower affinity). By contrast, the type-II IGF-R (Table 1), which shows high affinity for IGF-II and significantly lower affinity for IGF-I, is identical with the mannose 6-phosphate-receptor that is involved in the transport of lysosomal enzymes. The type-II IGF-R lacks PTK activity and does not appear to be involved in transducing the IGF-II mitogenic signal (Kiess et al., 1987; Minniti et al., 1992a). Instead, extensive evidence demonstrates that IGF-II exerts its mitogenic effects through the activation of the type-I IGF-R and, in certain cases, through activation of IR (Kiess et al., 1987). It is still uncertain whether and how signaling through the type-II IGF-R occurs. One hypothesis is that by binding to IGF-II, this receptor might regulate the growth factor bioavailability and modulate the growth factor interaction with the type-I IGF-R (Ludwig et al., 1995). Unlike the type-II IGF-R, the type I receptor is structurally related to insulin receptor. Both receptors belong to the type II subgroup of PTK-R (Yarden and Ullrich, 1988). Type-I IGF-R, whose gene is located on human chromosome 15, is a disulfide-linked ₂-₂ heterodimeric glycoproteinic complex composed of two entirely extracellular -chains (130 kDa), which provide the growth factor binding site, and two transmembrane-intracellular -subunits (95 kDa), which possess intrinsic PTK activity (Yarden and Ullrich, 1988; Heldin, 1996). Ligand binding induces autophosphorylation of three closely located tyrosine residues (positions 1131, 1135, and 1136) within the kinase domain, which leads to an increase in the catalytic efficiency of the kinase. Thus, several cellular substrates, such as the insulin receptor substrates 1 and 2 (IRS-1 and IRS-2), the SH2/collagen -protein and the growth factor-receptor binding protein-10 (GRB-10), are recruited. These substrates then couple the receptor to downstream signaling pathways by serving as binding sites for effector proteins (Rubin and Baserga, 1995). For instance, IRS-1 activates phosphatidyl inositol-3 kinase (PI3K), Ras/MAP cascade (through Grb2/sos), as well as other still poorly identified pathways involving the adapters Nck and Crk (Myers et al., 1994; Beitner-Johnson et al., 1996).

In plasma and in other biological fluids, IGFs are complexed with specific binding proteins (BP). To date, seven proteins have been identified as belonging to the human IGF-BP family; of these, IGF-BP1 (234 amino acids, 25.2 kDa), IGF-BP2 (289 amino acids, 31.3 kDa), IGF-BP3 (264 amino acids, 28.7 kDa), IGF-BP4 (237 amino acids, 25.9 kDa), and IGF-BP5 (252 amino acids, 28.5 kDa) are the most studied and best characterized (Fig. 3). IGF-BP6 (216 amino acids, 22.8 kDa) and IGF-BP7 (251 amino acids, undetermined molecular weight) (Jones and Clemmons, 1995; Swisshelm et al., 1995; Oh et al., 1996) are the most recently identified. All molecular species of IGF-BPs, except IGF-BP7, bind both IGF-I and IGF-II with high affinity and serve to transport IGFs, prolong their half-lives, and modulate their proliferative and anabolic effects on target cells. At the cellular level, IGF-BPs, depending on cell types and IGF-BP species, can either potentiate or inhibit the mitogenic effects of IGFs. The precise molecular mechanisms involved in the interactions among IGF-BPs, IGF-Rs, and their ligands are still unclear. However, IGF-BPs appear to at least regulate the availability of free IGFs for binding with IGF-Rs (Jones and Clemmons, 1995; Kelley et al., 1995). Finally, the recently identified IGF-BP7 (mac25), binds IGFs with lower affinity than do the other molecular species (Oh et al., 1996).

View larger version (39K): [in this window] [in a new window]   Fig. 3.   Pure human recombinant (hr)-IGF-BPs separated by Western Ligand Blot. Molecular weight standard (MW std) is reported on the right axis, while the MW of IGF-BPs is indicated on the left axis. Lines: A = hr-IGF-BP3; B = hr-IGF-BP2; C = hr-IGF-BP5; D = hr-IGF-BP1; E = hr-IGF-BP4.

Studies seeking to elucidate involvement of IGFs in breast tumorigenesis have focused primarily on their potential autocrine/paracrine interactions. Cultured breast cancer epithelial cells express both transmembrane IGF-Rs and are sensitive to the mitogenic stimuli of IGF-I and IGF-II (Rogler et al., 1994). On the other hand, IGF transcripts have rarely been identified in malignant breast epithelial cells either in vitro or in vivo. Rather, IGF transcripts were found prevalently within fibroblasts surrounding the breast epithelium. Thus, extensive evidence suggests that IGFs exert mainly paracrine effects on breast epithelial cells (Gebauer et al., 1998). The finding that malignant breast epithelial cells can also overexpress the type I IGF-R, regardless of IGF source, suggests that enhanced IGF signaling could have physiological relevance in malignancy (Pekonen et al., 1988). Analysis of breast tumor biopsies showed a different pattern of IGF expression. IGF-I mRNA was localized to stromal fibroblasts surrounding normal breast epithelium, whereas high levels of IGF-II were found in the fibroblasts adjacent to malignant epithelial cells (Yee et al., 1989; Paik, 1992; Rasmussen and Cullen, 1998). Type-I IGF-R is overexpressed in breast tumors as well as in breast cancer cell lines. Resnik et al. (1998) recently reported that type-I IGF-R expression was 14-fold higher in malignant breast tissue than in normal tissue. One possible mechanism underlying this high receptor level is the loss of its repression following mutations of the tumor suppressor gene p53 (Webster et al., 1996; Werner et al., 1996). In addition, estrogens may also regulate the receptor expression (Stewart et al., 1990; de Cupis et al., 1995). Moreover, receptor autophosphorylation and kinase activity were shown to be two or four times higher in neoplastic than in nondiseased tissues (Resnik et al., 1998). Thus, signaling that occurs through the type-I IGF-R may contribute to the pathogenesis of breast cancer by stimulation of cell proliferation and/or inhibition of apoptosis (Resnicoff et al., 1995). Findings also hint at the possibility that IGF signaling may modulate adhesion and invasion of breast cancer cells at local and distant sites (Scholar and Toews, 1994; Doerr and Jones, 1996; Leventhal and Feldman, 1997; Dunn et al., 1998). The estrogen-induced enhancement of the IGF-mediated mitogenic pathway in estrogen-receptor positive breast epithelium is well documented at both the cellular and molecular levels. In vitro, 17-estradiol increases the binding of iodinated-IGF-I to its transmembrane receptor, and also increases the expression of the growth factor-receptor itself (de Cupis et al., 1995; de Cupis and Favoni, 1997). In tumors, while there is a positive correlation between type-I IGF-R and estrogen receptor, no correlation has been demonstrated between type-II IGF-R expression and estrogen receptor status (Pekonen et al., 1988; Mathieu et al., 1990).

Breast cancer cells also secrete various molecular species of IGF-BPs. The predominant secreted IGF-BP appears to correlate with estrogen receptor status of the cell. Estrogen-receptor-negative cells secrete IGF-BP3 and IGF-BP4 as major species, and IGF-BP6 as a minor protein; by contrast estrogen-receptor-positive cells synthesize IGF-BP2 and IGF-BP4 as major species and IGF-BP3 and IGF-BP5 as minor species (Clemmons et al., 1990). These different patterns of IGF-BP expression imply that the IGF system in breast cancer is complex and that the biological significance of cellular response to IGF-BPs may differ according to estrogen-responsiveness. In addition, recent studies have indicated that IGF-BPs, especially IGF-BP3, strongly inhibit breast cancer cell proliferation in an IGF-independent manner. The IGF-independent action of IGF-BP3 requires interaction with cell surface proteins, most likely putative IGF-BP3-specific receptors (Oh, 1998). Recent reports suggest that IGF-BP3 appears to be a major factor in a negative control system involved in regulating breast cancer cell proliferation in vitro (Oh et al., 1995; Gucev et al., 1996). Several studies have been conducted to evaluate whether IGF family members could be prognostic indicators of breast cancers. Although in vitro data suggest that IGF expression should be a reliable prognostic marker in breast tumors, the limited clinical evidence obtained to date does not support these findings (Barni et al., 1994; Bhatavdekar et al., 1994). Similarly to what was ascertained for estrogen receptor, the type-I IGF-R has also been identified as a potentially valuable prognostic marker (Lee et al., 1998). Finally, when analyzed as prognostic factors, high levels of IGF-BPs are generally poor, although the physiological mechanism underlying this observation is not known (Yee et al., 1994; Lee et al., 1998).

In healthy prostate, IGF-I and IGF-II are produced by stromal cells, whereas normal epithelial cells express the type-I IGF-R. Both epithelial and stromal cells secrete mainly IGF-BP4, together with IGF-BP2 and IGF-BP3. On the other hand, controversy surrounds the IGF-mitogenic loop in prostate cancer. Kimura et al. (1996) showed that the DU-145 prostate cancer cell line could proliferate in response to IGF-I, but that it does not produce this protein, suggesting a paracrine mode of action of this growth factor. However, other authors describe an autocrine loop for IGF-I in three different prostate cancer cell lines, which also display constitutively autophosphorylated type-I IGF-R (Pietrzowski et al., 1993). The balance of IGF-BPs produced by cancer cells varies, depending on the cell lines examined. There is also evidence that some IGF-BPs may be under androgen regulation (Russell et al., 1998). Particularly, in vitro data appear to suggest that androgen may indirectly modulate IGF-induced proliferation of prostate cancer cells by regulating IGF-BP3 production (Marcelli et al., 1995). Furthermore, in advanced prostate cancer, IGFs appear to be involved in the development of bone metastasis (Chevalley et al., 1996).

In colorectal cancer, the IGF-II gene was reported to be overexpressed in 30 to 40% of the tumors tested (Lambert et al., 1990, 1991). Recently, it has been suggested that IGF-II modulates, by a paracrine mechanism of action, cellular proliferation of human colorectal cancer cell lines through binding to the type-I IGF-R (Lahm et al., 1994; Lamonerie et al., 1995). Kawamoto et al. (1998) indicated that IGF-II expression, evaluated by immunohistochemical staining of tissue samples from 92 colorectal cancer patients, was correlated with tumor progression, clinic-pathological factors, and patient survival. These findings led the authors to conclude that IGF-II staining might be a useful prognostic factor in colorectal cancer. Finally, a recent in vitro study (Akagi et al., 1998) performed on human colon cancer cell lines demonstrated that IGF-I is able to increase the expression of vascular endothelial growth factor (VEGF, see below), a potent and unique angiogenic protein.

The IGF system is also involved in the modulation of both small cell lung cancer (SCLC) and nonsmall cell lung cancer (N-SCLC) proliferation. Primary lung tumors possess IGF-I binding sites as detected by autoradiography and/or monoclonal antibodies (Maculay, 1992; Kaiser et al., 1993). In addition, iodinated ligand-binding assays revealed two classes of IGF-R (high- and low-affinity) on SCLC cell lines (Rotsch et al., 1992). In a previous study, we demonstrated the presence of one class of high-affinity functional type-I IGF-R on a pool of human N-SCLC cell lines belonging to adenocarcinoma and squamous carcinoma histological subtypes (Favoni et al., 1994a). Data on the type-II IGF-R in human lung cancer are scarce. However, expression of this receptor has been demonstrated in both SCLC and N-SCLC cells. Schardt et al. (1993) characterized type-II IGF-R in SCLC cell lines and demonstrated a 10- to 15-fold higher affinity of the IGF-II ligand for the type-I than for the type-II receptor. Immunoreactive IGF-I is detectable in primary lung tumor tissues to a greater extent than in normal lung. Furthermore, the mitogenic peptide is also detectable both in extracted SCLC and N-SCLC cells and in their conditioned media (Maculay, 1992; Favoni et al., 1994a). It has, however, been demonstrated that IGF-I levels in lung cancer patients' sera are not related to the bulk of disease or response to treatment. Beyond expressing the growth factor and cell surface receptor, lung cancer cells can also synthesize and secrete IGF-BPs, even if the pattern of their expression differs qualitatively and quantitatively depending on the model analyzed (Kiefer et al., 1991; Maculay, 1992; Favoni et al., 1994a). In agreement with the proposal of Sporn and Todaro (1980), the production of the mitogenic growth factor, the expression of the specific receptor and the carrier proteins, together with cellular sensitivity to growth factor action, are conditions that should substantiate the hypothesis of an autocrine mechanism of action in lung cancer cells. Nevertheless, available experimental data do not definitively exclude a paracrine role for the IGF system in lung cancer (Jaques et al., 1988; Maculay, 1992; Rotsch et al., 1992; Favoni et al., 1994a).

Transforming growth factors (TGFs) (Table 1) are polypeptides that, when originally isolated from viral-transformed rodent cells, were found to convert some normal cells to a transformed phenotype. Thus, in the presence of these growth factors, cultured fibroblasts pile up and decrease their anchorage-dependence, but do not become neoplastic (Folkman and Klagsbrun, 1987). Furthermore, it has been demonstrated that TGFs are also angiogenic in vivo. Two structurally distinct TGFs, TGF- and TGF-, have been purified and their structures determined by protein sequencing and cDNA cloning. TGF-, whose gene is located on human chromosome 2, is a small integral membrane protein (50 amino acids; 6 kDa) which shares biological and structural properties with EGF (35% homology) and exerts its action through EGF-R (Yarden and Ullrich, 1988; Massagué et al., 1994; Heldin, 1996). TGF- is a pleiotropic growth factor involved in tissue remodelling, wound repair, development, and hematopoiesis, but its predominant action is to inhibit cell growth. There are three structurally-related TGF- isoforms, TGF-1, TGF-2, and TGF-3, which are encoded by different genes on human chromosomes 19, 1, and 14, respectively. The expressed proteins (25 kDa) are biologically inactive disulfide-linked dimers that are cleaved to active dimers of 112-amino acid disulfide-linked peptides (Miller et al., 1990). These molecules have been shown to exert their actions by binding to heteromeric complexes of serine/threonine kinase receptors (Massagué et al., 1994). Both type I and type II TGF- receptors (Table 1) have small cysteine-rich domains; the type I receptors have a characteristic region rich in glycine-serine residues domain in their cytoplasmic juxtamembrane domain. Both receptors are needed for signaling, and the cytoplasmic parts of the receptors are not interchangeable (Heldin, 1995). TGF- binding induces formation of a hetero-oligomeric complex of the type I and type II receptors, most likely a heterotetramer containing two receptors of each type (Yamashita et al., 1994a). In particular, the type II receptor, which also exists as a dimer in the absence of ligand and has a constitutively activate kinase, is the first to bind the growth factor. The resulting complex then recruits the type I receptor, which cannot bind TGF- without the type II receptor. Thus, phosphorylation of the type I receptor on serine residues occurs in the glycine-serine domain (Chen and Derynck, 1994; Henis et al., 1994). The phosphorylation presumably activates the type I receptor kinase, which acts on downstream components in the signal transduction pathway (Heldin, 1995). In addition to the high-affinity type-I and type II receptors, TGF- also binds to the low-affinity type III receptors, which include -glycan and the endothelial cell-specific type III receptor CD105 (Cheifetz et al., 1992). Beta-glycan has not been shown to transduce signals, but it may function to concentrate TGF- on the cell surface and to present the ligand to its other receptors. It appears that coexpression of type III and type II receptors increases the ability of the high-affinity receptor to bind the ligand (Lopez-Casillas et al., 1994). Little is known about the possible role of the specialized vascular endothelial cell receptor CD105 in the TGF- signaling pathways, although it has been demonstrated to form heteromeric complexes with the signaling receptors on endothelial cells (Yamashita et al., 1994a).

TGF- ligands and receptors are expressed by normal breast epithelial cells. In vivo, TGF- appears to regulate the normal development of ductal and lobulary epithelium in the mammary gland (Jhappan et al., 1993). Moreover, in the adult, this growth factor seems to mediate the cell death and restructuring processes that take place during postlactation involution (Strange et al., 1992). Beyond these physiological functions, there is considerable evidence that TGF- is pivotally involved in breast cancer (Reiss and Barcellos-Hoff, 1997). Several published findings indicate that endogenous TGF-s expressed by tumor cells are autocrine modulators of cell proliferation (Norgaard et al., 1995; Gold, 1999). In addition, estradiol-induced cell proliferation in hormone-dependent breast cancer cell lines is known to be associated with a decrease in TGF-2 and TGF-3 mRNA levels (Arrick et al., 1990). Furthermore, loss of autocrine growth factor-mediated growth regulation due to down-regulation or mutation of TGF- receptors has been associated with tumor progression (Brattain et al., 1996; Koli and Arteaga, 1996). Recently, a particular somatic missense mutation in the TGF- receptor was identified (Chen et al., 1998); it results in a serine to tyrosine substitution at codon 387, within the catalytic core of TGF--R1 serine-threonine kinase. This mutation disrupts the signaling function of the receptor. It has been proposed that the inactivation of the TGF- signaling pathway is probably a relatively late event because the mutation was found predominantly in metastatic lesions (Chen et al., 1998). However, the growth inhibitory effect of TGF- on breast cancer cell line proliferation, documented by in vitro studies, was not confirmed in experimental animal models. Several reports suggest a positive association between TGF-s and breast cancer progression, owing to enhanced angiogenesis and suppressed host immune surveillance (Arteaga et al., 1996). A recent study (Li et al., 1998) carried out on 80 patients with early stage breast cancer proposed that the plasma levels of TGF-3 and type III receptor CD105-TGF-3 complex may be of prognostic value in the early detection of breast cancer metastasis.

In nondiseased prostate, TGF-s exert a role in growth regulation by counterbalancing the mitogenic effects of various growth factors (Russell et al., 1998). Conversely, increasing intracellular expression of TGF-1 mRNA and protein, detected in both epithelial and stromal cells, seems to be important in prostate cancer progression, even though its exact involvement remains uncertain (Truong et al., 1993). In addition, it has been suggested that the progression of prostate cancer is associated with TGF-1 switching from an autocrine/paracrine to a juxtacrine mechanism of action (Russell et al., 1998). Studies on human prostatic cancer cell lines suggest that changes in sensitivity to TGF-1 may be related to progression of the disease. Kim et al. (1996) documented the ability of TGF-1 to inhibit the proliferation of androgen-independent cell lines but not the growth of androgen-sensitive LNCaP cell lines. The insensitivity of LNCaP cells to TGF- has been attributed to a genetic change in their TGF- receptor I gene. This could provide a possible mechanistic explanation for the ability of the prostate cancer cell to escape the growth-inhibitory effects of TGF-. Furthermore, it has been shown that TGF- can modulate extracellular matrix metalloprotease production and can stimulate adhesion of prostate cancer cells to bone cells. These observations imply the potential involvement of TGF- in the promotion of prostate cancer metastases (Sehgal et al., 1996). Finally, in view of the role of TGF- in angiogenesis and as an immunoregulatory molecule, TGF- secretion could have important effects on prostate cancer cell environment.

TGF-1 arrests the cell cycle by transcriptionally activating a series of cyclin-dependent kinase inhibitors including p21^waf1/cip1 (Datto et al., 1995; Halevy et al., 1995). Because both p21^waf1/cip1 and TGF-1 modulate apoptosis and cell cycle progression, protein levels should be correlated with biological outcomes, e.g., survival. It has recently been reported that concordant expression of TGF-1 and p21^waf1/cip1 levels (i.e., high and high, or low and low, protein expression) detected by immunohistochemical analysis (Bennett et al., 1998) predicted 70% disease-free survival at 2000 days follow-up in N-SCLC patients. Although currently available models do not clearly explain these findings, it appears that analysis of both TGF-1 and p21^waf1/cip1 may provide useful information concerning the survival of these patients.

The dimeric molecule VEGF, also known as vascular permeability factor, was first purified from media conditioned by bovine pituitary folliculostellate cells. VEGF owes its discovery to its ability to stimulate angiogenesis by increasing vascular permeability (Senger et al., 1983) and by acting as an endothelial cell mitogen (Ferrara and Henzel, 1989). Four different isoforms of VEGF transcripts (Table 1) encoding polypeptides of 206, 189, 165, and 121 amino acids have been reported to be expressed in human cells; each of these isoforms possesses different biological activity (Houck et al., 1992). VEGF121 and VEGF165 are secreted in soluble forms, whereas the two larger isoforms, VEGF189 and VEGF206, remain associated with cells because of their stronger affinities for cell-surface proteoglycans. The smallest isoform does not bind heparin, whereas the inclusion of more cationic exons in VEGF165 and VEGF189 confers heparin-binding properties (Scott et al., 1998). The largest isoform, VEGF206, has been identified only in a fetal liver library, and little is known about its biological relevance (Houck et al., 1991).

The family of VEGF receptors (Table 1) contains three members, Flt, Flk-1/KDR and Flt-4, each characterized by the presence of seven extracellular Ig-like domains and an intracellular PTK domain. These receptors are expressed predominantly on endothelial cells and have been shown to be of importance in mediating angiogenic response (Heldin, 1996).

VEGF expression has been detected at the mRNA and protein level in a number of malignancies (Brown et al., 1993; Sato et al., 1994; Guidi et al., 1995; Mattern et al., 1995; Olson et al., 1995; Maeda et al., 1996; Mise et al., 1996). In human breast carcinoma, VEGF mRNA expression, as detected by in situ hybridization, was reported to be higher in neoplastic cells than in normal ductal cells (Brown et al., 1995). In addition, the levels of growth factor protein were found to correlate with increased microvessel density and early relapse in the same disease (Toi et al., 1994). Recent studies have shown that VEGF mRNA and protein levels are significantly higher in tumor samples than in matched normal tissues (Yoshiji et al., 1996; Scott et al., 1998). Furthermore, Scott et al. (1998) demonstrated that VEGF mRNA level was elevated in breast tumors expressing the EGF-R, while no differences according to estrogen receptor status or nodal status were observed. Recently, a posttranscriptional mechanism modulating VEGF expression has been identified. The mRNA of growth factors such as VEGF is characterized by a long 5'untranslated region with complex secondary structures that render them inefficiently translated (Kevil et al., 1996). The polypeptide eIF-4E unwinds the 5'untranslated region of the target mRNA, thus facilitating the identification of the translation start site by ribosomes. It has been shown that elevated levels of eIF-4E are associated with increased levels of VEGF protein and increased growth rate (Kevil et al., 1996).

VEGF overexpression was also observed in the sera of N-SCLC patients (Brattstrom et al., 1998; Takigawa et al., 1998), and was related to an increase in microvessel density (Mattern et al., 1995, 1996). However, contrary to the observation reported for breast cancer, the level of VEGF in lung cancer patients appeared to be unrelated to tumor burden, thus casting doubt on the usefulness of this angiogenic growth factor as a clinically reliable tumor marker (Brattstrom et al., 1998; Takigawa et al., 1998). In human lung cancer, increased VEGF level was found to be correlated with nuclear accumulation of p53 (Mattern et al., 1995; Fontanini et al., 1997). The expression of this angiogenic growth factor is induced by hypoxia (Forsythe et al., 1996), a feature common to all solid tumors. Studies have shown that hypoxia increases VEGF expression within 3 to 6 h, whereas normalization of oxygen tension causes cellular VEGF mRNA to return to baseline levels (Ikeda et al., 1995; Shima et al., 1995). Hypoxic induction of VEGF may be modulated by an increase in transcription and/or stabilization of its mRNA (Levy et al., 1995, 1996). Furthermore, a mutationally activated ras oncogene (Larcher et al., 1996; Mazue et al., 1996) or p53 (Kieser et al., 1994) can act synergistically with hypoxia to induce VEGF expression. In contrast, wild-type p53 down-regulates VEGF promoter activity (Mukhopadhyay et al., 1995) and up-regulates the expression of the antiangiogenic factor thrombospondin-1 (Dameron et al., 1994). However, a recent report (Ambs et al., 1998) found no evidence of a direct regulation of VEGF by p53 in N-SCLC tissues.

Data from several laboratories indicate that microvessel counts are strong prognostic factors in human colorectal cancer (Takahashi et al., 1995, 1996, 1998). The finding that VEGF expression correlates with microvessel counts has implicated this growth factor in the regulation of colon cancer angiogenesis (Takahashi et al., 1995, 1996, 1998). Furthermore, on a pool of colon cancer specimens, an association of mutant p53 expression with VEGF and vessel count was observed (Takahashi et al., 1998); this finding appears to suggest that poor prognosis associated with p53 mutation (Kastrinakis et al., 1995) may be due, at least in part, to the ability of p53 protein to promote angiogenesis (Takahashi et al., 1998). Tokunaga et al. (1998) recently reported that, together with the detection of VEGF level, it is also important to examine the growth factor isoform patterns in order to predict the prognosis of colon cancer patients. In addition to hypoxia, factors such as EGF, bFGF, TGF-, TGF-, IL-6, and other cytokines are now being characterized as mediators of VEGF expression in both neoplastic and normal cells. An experimental study by Akagi et al. (1998) showed that IGF-I, but not IGF-BPs, induces VEGF mRNA and protein expression in colon carcinoma cell lines. The increased growth factor level appears to be due to an increase in gene transcription without significant alteration of the mRNA half-life.

Ovarian cancer is characterized by widespread i.p. carcinomatosis and formation of large volumes of ascitic fluid. VEGF may play a major role in the progression of ovarian cancer by modulating tumor proliferation through its promotion of tumor angiogenesis. Moreover, the growth factor may be involved in ascites production by stimulating vascular permeability (Mesiano et al., 1998). Although VEGF has been detected in ovarian cancer (Boocock et al., 1995; Brown et al., 1995; Abu-Jawdeh et al., 1996; Paley et al., 1997), its role as a regulator of angiogenesis is not completely elucidated. However, microvessel density and the level of VEGF expression directly correlate with poor prognosis in ovarian cancer, thus suggesting that angiogenesis, mediated at least in part by VEGF, influences disease progression (Boocock et al., 1995; Abu-Jawdeh et al., 1996; Paley et al., 1997). Finally, Tempfer et al. (1998) recently reported that VEGF appears to be an additional factor for predicting the clinical outcome of epithelial ovarian cancer patients.

Hepatocyte growth factor (HGF) (Table 1), first identified as a mitogen for hepatocytes (Nakamura et al., 1984), has also been described as a growth modulator of kidney cells, melanocytes, keratinocytes, and other cell lines in vitro (Igawa et al., 1991; Kan et al., 1991; Matsumoto et al., 1991). Scatter factor (SF), originally described as a cytokine dispersing cohesive epithelial colonies and stimulating cell motility, was subsequently shown to be identical with HGF (Stoker et al., 1987; Weidner et al., 1990, 1991; Naldini et al., 1991b). Early studies revealed that HGF/SF is produced by cells of mesenchymal origin, such as fibroblasts, and acts on epithelial cells (Sonnenberg et al., 1993). Recently, it was reported that cells of epithelial origin also express HGF/SF (Olivero et al., 1996; Tuck et al., 1996; Jin et al., 1997). In vitro studies showed HGF/SF to be a morphogenetic (Montesano et al., 1991) and an angiogenic factor (Bussolino et al., 1992). Experimental studies indicated that HGF/SF might also induce endothelial secretion of plasminogen activators that are required during the early stages of angiogenesis, in which endothelial cells degrade the extracellular matrix (Grant et al., 1993). The growth factor is produced in an immature form and is then processed by a proteolytic cleavage into a glycosylated heterodimer of a heavy -chain (62 kDa) and a light -chain (34-32 kDa). The -chain contains four kringle domains, the first two of which are necessary for scatter activity; the complete protein, on the other hand, appears to be involved in mitogenic activity. The -chain shows strong homology with the catalytic domain of serine proteases, but does not possess enzymatic activity because the catalytic site is different (Matsumoto and Nakamura, 1992; Bellusci et al., 1994).

The HGF/SF receptor was first identified as the product of the c-met proto-oncogene (Naldini et al., 1991a) (Table 1). The mature protein is a heterodimer composed of 50-kDa - and 145-kDa -subunits. The -chain extends over the membrane and contains the catalytic domain, whereas the -chain remains extracellular. The receptor is autophosphorylated inside the kinase domain, thereby increasing its catalytic activity, as well in the C-terminal tail where two closely related tyrosine residues account for binding of a number of different signal transduction molecules (Bellusci et al., 1994