Cancer Therapy Vol 2, 317-328, 2004

Division of Molecular Oncology, University of Torino Medical School, Candiolo, Italy

__________________________________________________________________________________

Key words: Receptor tyrosine kinases, Cancer therapy, Signal transduction

Abbreviations: dermatofibrosarcoma protuberans, (DFSP); Epidermal growth factor receptors, (EGFRs); Fibroblast growth factor receptors, (FGFRs); gastrointestinal stromal tumors, (GIST); Hepatocyte growth factor receptor, (HGFR); human epidermal receptors, (HERs); Insulin-like growth factor I receptor, (IGF-IR); monoclonal antibodies, (mAbs); myeloproliferative disorders, (MPDs); non-small-cell lung cancer, (NSCLC); oligonucleotides, (ODNs); phosphotyrosine binding, (PTB); Platelet-derived growth factor receptors, (PDGFRs); Receptor tyrosine kinases, (RTKs); small interfering RNAs, (siRNAs); Src homology 2, (SH2); Vascular endothelial growth factor receptors, (VEGFRs)

Tumorigenesis is an extremely complex multistep process, which leads to the pathological expansion of a tissue resulting in morbidity . Tumor cells gain an unlimited replicative potential as a result of imbalance among the finely tuned proliferative, growth-inhibitory and apoptotic signals. In parallel, dynamic changes in the host microenvironment are required to convey signals that favor the expansion of the tumor mass. The orchestrated execution of the different strategies operated during tumorigenesis can eventually lead to cell invasion and settling in secondary body sites, i.e. to the acquisition of a malignant phenotype. Cells need to breach several defense barriers in order to acquire the cancer phenotype. Disruption of the tightly regulated network that controls tissue homeostasis occurs as a result of perturbed signal transduction, whereby certain transduction pathways, namely those affecting proliferation, survival, cell migration, invasiveness and angiogenesis, are preferentially targeted by oncogenic changes. Ligand-mediated activation of receptor tyrosine kinases (RTKs) plays a pivotal role in these processes and is frequently deregulated in cancer cells. When mutated, overexpressed or structurally altered, RTKs can become potent oncoproteins, and actually they are aberrantly activated in most tumor types. Hence, RTKs establish a promising class of molecular targets for the design of tailored therapies aimed at the selective eradication of tumor cells.

RTKs are a class of approximately 60 molecules divided into 20 subfamilies, whose enzymatic activity is the transfer of g -phosphate groups from ATP to the hydroxil group of tyrosine residues on signal transduction molecules. They are transmembrane receptors that share a similar structure (Figure 1): an extracellular domain, usually glycosilated, is connected to the cytoplasmic region by a single transmembrane helix. The extracellular portion contains the binding site for polypeptide growth factors, whereas in the cytoplasmic domain are located the conserved protein tyrosine kinase core and several regulatory sequences . RTKs transduce

Figure 1. Schematic structure of the main human RTK families. Abbreviations are: TK, tyrosine kinase domain; CRD, cystein-rich domain; LD, leucine domain; FNIII, fibronectin type III-like domain; AB, acidic box; CadhD; cadherin-like domain; LRD, leucine-rich domain; IgD, immunoglubulin-like domain.

signals from the extracellular environment to the cytoplasm, ultimately modulating gene transcription within the nucleus. When strictly controlled, their activity regulates several fundamental cell processes, as metabolism and growth, survival, proliferation, differentiation and migration . RTK enzyme activity is stimulated by the selective binding of cognate growth factors. Ligand binding elicits or stabilizes receptor oligomerization, causing the reciprocal transphosphorylation of receptor protomers on tyrosine residues in the activation loop of the catalytic region and in regulatory domains. Thus, receptor autophosphorylation enables full kinase activation, but also provides docking sites for molecular recruitment. In fact, phosphorylated tyrosines are recognized by SH2 (Src homology 2)- and PTB (phosphotyrosine binding)-domain containing proteins, which in turn harbor other interaction motifs and promote the assembly of molecular platforms based upon modular interactions between definite peptide domains . Signaling molecules in these complexes include adaptor proteins, scaffold and docking molecules and enzymes. Hence, in response to RTK stimulation the formation of molecular piers facilitates the activation of specific effectors, induces the amplification of single transduction pathways and promotes the bridging among different cascades, thus eliciting a network of receptor-dependent effects. Among the best characterized intracellular signaling pathways downstream to RTK activation there are: the MAP kinase cascades, composed by three protein kinases organized in a modular amplification sequence; the nuclear translocation of the latent cytoplasmic transcription factors STATs; and the stimulation of phosphoinositol metabolism, which generates multiple second messengers . The final output of these signaling events is the phosphorylation and activation of specific transcription factors, whose targets are involved in the control of a plethora of metabolic processes, including cell cycle, cell migration, proliferation, differentiation, survival and cell shape. It remains an open question how a given biological response is obtained by the stimulation of relatively common repertoires of signal transducers . Nonetheless, pathway specificity can be determined through different strategies. These include: the combinatorial recruitment of specific molecular subsets at each step of the pathway; the insulation through scaffolding and subcellular compartimentalization of molecular assemblies; differences in signal threshold caused by variations in input duration and amplitude; differences in the molecular complement of specific signal transducers expressed in a given cell in a certain moment; and signal redundancy due to the cross-talks with other metabolic pathways .

Deregulated activation of such carefully orchestrated signaling networks can lead to the transformation of RTKs into potent oncoproteins, and more than half of the known RTKs are associated with human malignancies . Multiple brakes operate as safeguard autoinhibitory mechanisms to ensure a correct RTK activation. The acquisition of oncogenic properties results from relief or perturbation of these control devices, leading to RTKs with constitutive or enhanced signaling capacity. The resulting stimulation of signaling pathways enables cells to survive independently of their environment and can make them resistant to genotoxic therapies . This aberrant activity of RTKs can be originated by receptor overexpression or mutation, or by formation of ligand-receptor autocrine loops (Robertson et al, 2000; Trusolino and Comoglio, 2002; Zwick et al, 2002). RTK overexpression can occur following gene amplification or reduced degradation , raising the response of cancer cells to growth factor levels and favoring receptor oligomerization and activation even in the absence of the specific ligand . Somatic and germline mutations include chromosomal translocations, which result in fusion protein formed by the RTK catalytic domain and peptides providing constitutive dimerization capabilities , and gain-of-function point mutations or small deletions either within the extracellular domain or in the catalytic region . In addition, mutations in the transmembrane motif can lead to ligand-independent kinase activation . The establishment of autocrine-paracrine circuitries requires that a RTK is aberrantly expressed in the presence of its cognate ligand, or vice versa, thus relieving cells from dependence on a paracrine supply of growth factors .

Since RTKs are major players in the evolution of a variety of cancer types, efforts are underway to develop treatment strategies that target these molecules or their downstream effectors. This approach has great potential as it relies on blocking specific transduction pathways by taking advantage of the molecular differences between normal and tumor cells . The final goal is to obtain a higher therapeutic index and safety with respect to traditional chemotherapeutics, which induce a generic cytotoxic effect that hardly distinguishes malignant from non-malignant cells .

The recent development of a breed of agents selectively directed against RTKs, and their ability to bridle the proliferation of tumor cells expressing the target RTK in vivo, show that inhibition of a deregulated, dominant oncogenic receptor is often enough to slow tumor progression. Strategies towards the prevention or interception of aberrant RTK enzymatic activity include the development of modified protein ligands, as inhibitory monoclonal antibodies or new generation immunoconjugates (Sanz et al, 2004; Wels et al, 2004), and of small inhibitor molecules, as kinase inhibitors or antisense oligonucleotides (Figure 2) .

The success of these approaches implies a dominant role of the oncogenic RTK in subverting cellular signaling networks (oncogene addiction), so that targeted therapies could be specifically detrimental to tumor cells .

Mechanistically, monoclonal antibodies (mAbs) directed against the extracellular domain of RTKs might specifically interfere with ligand binding, hence neutralizing the growth factor-driven signaling, and/or induce receptor downregulation or internalization. It is now evident that mAbs can also alter the intracellular signaling pathway in target cells, leading to growth inhibition or apoptosis (Cragg et al, 1999; Zwick et al, 2001). In addition, mAbs can bind epitopes on cancer cells, eliciting immune responses such as complement-mediated cell lysis or antibody-dependent cytotoxicity by macrophages or natural killer cells . Furthermore, the use of mAbs in combination therapy sensitizes neoplastic cells to a range of conventional therapeutic regimens such as DNA-damaging chemotherapy or irradiation. Improved response to the traditional chemotherapeutics is generally obtained without additional toxicity. However, conventional therapeutic approaches are in principle highly mutagenic, and should therefore be used with the utmost attention in order not to generate unwanted resistance to targeted agents .

The applicability of mAbs in cancer therapy is reduced by immunogenicity, suboptimal biodistribution and severe side effects. With the aim to circumvent these problems, in the last few years several recombinant technologies were developed. Recombinant molecules include: (I) chimeric or humanized antibodies, composed by fused human and animal portions; (II) bivalent or multivalent antibody derivatives, which simultaneously bind antigens on target tumor cells and on immune effector cells; (III) intracellular antibodies (intrabodies); (IV) single-chain Fv, i.e. antibody sequences that lack constant regions and Fc domains; (V) antibodies linked to toxins (immunotoxins) by chemical coupling or recombinant technology; (VI) immune cells genetically modified for exposing antibody-based receptors on the surface or for secreting therapeutic antibodies. Several of these compounds are undergoing clinical trials (Sanz et al, 2004, Wels et al, 2004).

A promising approach to inhibit aberrant RTK signaling is the development of small-molecule drugs that selectively interfere with their intrinsic tyrosine kinase activity and block receptor autophosphorylation and activation. The three-dimensional structure of the kinase domain and its modes of regulation have been rather well elucidated by NMR, crystallographic and biochemical approaches for several RTKs (Mohammadi et al, 1996;

Figure 2. Strategies to selectively prevent RTK signaling. Monoclonal antibodies (mAbs) can be used alone, associated to a toxin (immunotoxin) or to another antibody recognized by effector cells (Bispecific Abs). Engineered receptors (decoy soluble receptors or mini-receptors) or decoy ligands (e.g. NK4) act as dominant negative molecules and abrogate signal transduction pathways. Small tyrosine kinase inhibitors (TKI) compete with ATP in binding the activation domain. Antisense oligonucleotides DNA (ODN)/RNA, small RNA interfering (siRNAs) and ribozymes inhibit RTK transcription and/or translation.

Till et al, 2001; Ekman et al, 2002; Bohmer et al, 2003; Chiara et al, 2003, 2004; Jorissen et al, 2003; Mol et al, 2003; Schiering et al, 2003; Manley et al, 2004). Hence, the rational design of kinase inhibitors can now be pursued by the integration of high-throughput screens, as the use of peptide chips of kinase substrates, with combinatorial chemistry (Schlessinger, 2002, Heldin, 2001, Noble et al, 2004). These small-molecule drugs can be substrate competitive, ATP competitive, i.e. they target the Mg-ATP complex pocket in the enzyme catalytic domain, or bisubstrate competitive (competitive against both the substrate and ATP). Despite the high sequence homology among the ATP binding sites of RTKs, these inhibitors are frequently endowed with an elevated degree of specificity towards their targets . The first classes of these molecules were natural products, such as quercetin, staurosporine, herbimycin A and genistein. However, their selectivity and efficacy was low, and they have mainly served as starting points for the development of many classes of tyrosine kinase inhibitors. These include: (I) compounds that compete at the ATP binding site of RTKs, as quinazolines, pyridopyrymidines or phenylamino-pyrimidines; (II) molecules that act in a non-competitive fashion against ATP or peptide substrates, as indoles or oxindoles; and (III) tyrphostins, which are competitive inhibitors of RTKs at either or both ATP and substrate binding sites . Many of these compounds are moving to clinical trials and they are displaying an enormous clinical potential, either as single agent therapeutics or, more frequently, when supplied in association with more traditional therapeutic approaches.

The use of antisense oligonucleotides (ODNs), antisense RNA, triple helix or small interfering RNAs (siRNAs) is aimed at the inhibition of RTK transcription and/or translation . ODNs are short fragments of synthetic DNA or RNA that are designed to interact with the mRNA to block the transcription and the expression of specific target proteins by the production of heteroduplex, recognized and cleaved by RNase H . Even though ODNs can be degraded upon internalization by cellular nucleases, preclinical and clinical studies suggest that antisense therapy could be useful for the treatment of solid tumors . Antisense RNA vectors containing fragments of RTK cDNA cloned in 3¢ -5¢ orientation have given promising results in animal models, inducing massive apoptosis of tumor cells. The triple helix approach utilizes an oligoribonucleotide capable of forming a triple helix in target DNA, inhibiting the passage of RNA polymerase and thus the transcription of specific genes . siRNAs take advantage of the cell RNA interference machinery. Specific enzymes recognize double-strand RNA and cleave it in 21-28 nucleotide fragments (siRNAs) that are incorporated in a protein complex, where they direct the degradation of complementary mRNA sequences . Recent works report a negative regulation of RTK expression when targeted by siRNA approaches .

In the following sections we focus on therapeutic strategies targeting RTK families involved in the main tumor phenotypes.

The EGFRs or human epidermal receptors (HERs) form a family of four related RTKs broadly expressed in epithelial and mesenchimal tissues: EGFR (HER1, erbB1), HER2 (neu, erbB2), HER3 (erbB3) and HER4 (erbB4) . They bind six known ligands, including EGF, TGFa and neuregulins . Ligand binding induces homo- or hetero-dimerization of HERs, thus eliciting the transduction of cell growth, survival and proliferation signals. HER2/neu lacks any specific ligand, but it is the preferred dimerization partner for the other HERs . HERs are overexpressed and activated in two thirds of all solid tumors , where they play a pivotal role in angiogenesis, invasion and metastasis . HER aberrant activation correlates with poor prognosis and can result from mutation, overexpression or stimulation through autocrine loops . Hence, HERs are promising targets for the development of cancer therapeutics , and a variety of anti-EGFR agents and strategies are currently used or are undergoing pre-clinical or clinical trials (Ciardiello and Tortora, 2004; Thomas and Grandis, 2004).

The anti-HER2/neu humanized antibody trastuzumab (herceptin) was the first therapeutic agent applied on the basis of the genetic characteristics of a tumor . Trastuzumab is now the standard therapy in women with HER2-overexpressing metastatic breast cancer, and in clinical studies it induces long-term stable remission . Recently, the combination of trastuzumab with chemotherapeutics such as taxanes, platinum salts and gemcitabine has proved to be active also against HER2-overexpressing early-stage breast cancers (Burstein et al, 2003; Spigel and Burstein, 2003; Montemurro et al, 2004). Trastuzumab is currently undergoing clinical trials in non-small cell lung cancer patients . Another anti-HER2/neu antibody, pertuzumab, could have additional potentialities with respect to trastuzumab, as it blocks HER2/neu association with its partner receptors, thus inhibiting the downstream signaling cascades. Pertuzumab is currently in phase II clinical trials (Franklin et al, 2004; Badache and Hynes, 2004). These antibodies could act by inducing receptor down-regulation, antibody-dependent cell cytotoxicity or alteration of vessel development. A better comprehension of their mode of action is highly warranted in order to tailor adequate therapies for the consistent number of non-responsive patients .

Cetuximab (C225, Erbitux) is a human-murine chimeric anti-EGFR antibody that mediates complement fixation and inhibits cell cycle progression and cellular proliferation by receptor down-modulation . Cetuximab is used either as a monotherapy or in combination with chemo- and radiotherapy, and is currently in phase II/III clinical trials in non-small cell lung and head-and-neck cancer patients (Janmaat and Giaccone, 2003; Caponigro, 2004; Ciardiello et al, 2004; Laskin and Sandler, 2004). ABX-EGF is a fully humanized monoclonal antibody that binds EGFR with high affinity. It eradicates xenograft tumors in animal models, probably by inactivating EGFR-mediated growth pathways. ABX-EGF displays a good anti-tumoral activity with minimal side effects in phase I trials, and in combination therapy it enhances the activity of chemotherapeutics . MDX-447 (Medarex) is a bispecific antibody that simultaneously binds to EGFR on target cells and to CD64 on monocytes, macrophages and neutrophils, thus increasing antibody-dependent cell-mediated cytotoxicity .

Among small-molecule EGFR kinase inhibitors, the quinazoline derivatives Gefitinib (Iressa, ZD1839) and Erlotinib (Tarceva, OSI-774) act by selectively preventing EGFR phosphorylation. These compounds inhibit proliferation, cell cycle progression and angiogenesis, and increase apoptosis. Gefitinib has a significant anti-tumor effect in cells that co-express EGFR and TGFa and overexpress HER2/neu . Gefitinib enhances tumor growth inhibition in combination therapies with paclitaxel or cisplatin , and it is undergoing phase II clinical trials for the treatment of several solid tumors . Erlotinib phase III studies are now ongoing in non-small-cell lung cancer (NSCLC) and pancreatic cancer patients (Herbst, 2003; Thomas and Grandis, 2004). Bispecific inhibitors targeting both EGFR and HER2 (GW572016 and CI-1033) are presently in early clinical trials .

Antisense oligonucleotides decrease EGFR expression, inhibiting proliferation and inducing apoptosis of EGFR-expressing cells. Although these molecules have anticancer activity in xenograft models, they have not entered clinical testing. The only exception is TP-38 , an ODN conjugated to the TGFa /mutated Pseudomonas exotoxin presently in phase I clinical trial for the treatment of head and neck squamous cell .

The formation of new blood vessels from pre-existing ones, termed angiogenesis, requires the integration of highly complex and coordinated processes and is crucial for the expansion and progression to malignancy of solid tumors . The signaling pathways triggered by binding of vascular endothelial growth factor isoforms (VEGFs) to their cognate receptors (VEGFRs) are mandatory for tumor neovascularization . VEGFs bind two high-affinity RTKs, VEGFR-1 and VEGFR-2, on endothelial cells. VEGFR-1 expression is upregulated by hypoxia and controls the release of tissue-specific growth factors. VEGFR-2 is the major mediator of the mitogenic, angiogenic and permeability-enhancing effects of VEGF .

Since formation of solid tumors is angiogenesis dependent, it is expected that blocking angiogenesis will be an efficient therapeutic approach against many tumor types . VEGFs and VEGFRs are expressed at high levels in many types of human solid tumors, including glioma, lung, breast, renal, ovarian and gastrointestinal tract carcinomas . Strategies for targeting the VEGF signaling utilize a vast array of molecules, and the combination of anti-angiogenesis therapies with chemo- or radiotherapy results in greater antitumor effects.

Bevacizumab (rhuMab VEGF; Avastin) is a recombinant humanized monoclonal antibody raised against VEGF that potently suppresses angiogenesis and growth of human tumor xenografts in nude mice as a single agent or synergistically with chemotherapy. Phase I clinical trials showed that bevacizumab was relatively non-toxic, and phase II trials displayed encouraging efficacy in metastatic prostate cancer, NSCLC and in colorectal cancer (Sridhar and Shepherd, 2003; Manley et al, 2004). Phase III trials are under way on NSCLC and renal-cell cancer. In metastatic colorectal cancer, Bevacizumab in combination with a 5-fluorouracil/leucovorin/irinotecan-based chemotherapy increases response rates, time to disease progression and patient survival and has been recently approved as a first-line treatment .

The VEGF-trap is a decoy-soluble receptor created by fusing two distinct extracellular domains of the receptor to an IgG constant region, with the aim of increasing the binding affinity to VEGFR. In vivo the VEGF-trap suppresses tumor growth and vascularization, and at present it is in phase I clinical trials for several different cancer types (Holash et al, 2002; Ferrara et al, 2004).

Small-molecule inhibitors target the ATP-binding pocket of VEGFRs. The first generation of these compounds includes indoline derivatives, the phthalazine derivative PTK787 (Vatalanib), and ZD6474, that is based upon a quinazoline template. Despite in vitro inhibition of endothelial cell proliferation and in vivo anti-tumor activity in mice xenografts, indoline derivatives show an unacceptable toxicity profile. PTK787 and ZD6474 are instead well tolerated and have anti-tumor activity in animal models. PTK787 has been used as a single agent or in combination studies and phase III trials in metastatic colorectal cancer have now commenced. ZD6474 potently inhibits several RTKs and is undergoing phase III clinical trials for a variety of solid tumors (Manley et al, 2004; Ferrara et al, 2004; Sridhar and Shepherd, 2003).

The development of second-generation kinase inhibitors is largely based upon molecular modeling aimed at targeting the active ATP-binding pocket. AAL993 is a molecule designed upon the PTK787 structure. It displays good biopharmaceutical properties and an excellent oral bioavailability in animal models. Clinical trials have not yet initiated . CP-547,632 acts specifically against VEGFR-2 tyrosine kinase activity. In vitro it inhibits endothelial cell proliferation and in xenografts it retards tumor growth. Phase I data suggest encouraging safety profile and pharmacokinetic parameters . Bay 43-9006 is a broad spectrum RTK inhibitor, now in phase III trials for metastatic renal-cell carcinoma .

A different approach is constituted by angiozyme, a stabilized ribozyme directed against the pre-mRNA of the VEGFR-1 gene. Angiozyme is now in phase II studies on patients with metastatic colorectal cancer with promising results. The high cost of manufacturing and the necessity for parental application are limiting features of this drug .

PDGFRs are composed by one a and one b chain that can variously homo- or hetero-dimerize upon binding of five different PDGF isoforms. Their activation induces cell proliferation, survival and migration, and is required for a plethora of biological processes. These include wound healing, regulation of the interstitial fluid pressure and villus formation in the gut. Moreover, PDGFRs are crucial for the development of several cell types, as mesangial cells in the kidney, smooth muscle cells in vessels and alveoli, hair follicle cells and oligodendrocytes . Tumorigenic activation of PDGFRs can occur by gene amplification or mutation in high-grade gliomas (Fleming et al, 1992, Clarke and Dirks, 2003) and in a subset of patients with gastrointestinal stromal tumors (GIST) , or by chromosomal translocation in myeloproliferative disorders (MPDs) and in dermatofibrosarcoma protuberans (DFSP) . PDGFR signaling plays important roles in the autocrine stimulation of cancer cells, in the paracrine stimulation of stromal fibroblasts and of perivascular cells, and in angiogenesis .

The most important inhibitor of PDGFRs is STI571 (Gleevec, Glivec, Imatinib), which is also used to inhibit the oncogenic activity of Kit and Abl in GIST and chronic myelogenous leukemia, respectively . In seminal studies on MPD patients with PDGFR-activating translocations, STI571 causes a persistent disappearance of translocation-positive cells and a rapid normalization of blood counts . Case reports of DFSP patients display marked tumor reduction, even though not all patients are responsive . Furthermore, in neuroblastoma and pancreatic adenocarcinoma xenografts, STI571-mediated tumor regression is associated with suppression of PDGFR activity (Hwang et al, 2003; Beppu et al, 2004). An antiangiogenic activity of STI571 has been reported in prostate cancer bone metastases, possibly due to the inhibition of PDGFR-mediated pericyte recruitment , and in human ovarian cancers grafted in nude mice, due to apoptosis of tumor-associated endothelial cells . These results pave the way for the combination of PDGFR inhibitors with anti-endothelial compounds .

As PDGFRs are expressed in the stroma of the majority of tumors, STI571 has been used in animal models that express PDGFRs in stromal and perivascular cells. When supplied as a single agent, STI571 is ineffective on tumor growth. Instead, a striking response is obtained by using STI571 in association with chemotherapeutics. Remarkably, the increased efficacy of chemotherapy is not due to enhanced tumor cell sensitivity or to reduced angiogenesis, but to an augmented drug uptake by tumor cells, possibly related to a PDGFR antagonist-mediated reduction in tumor interstitial fluid pressure (Pietras et al, 2001, 2002, 2003).

Recently, other small-molecule PDGFR inhibitors have been developed. PKC412 is effective for treatment of STI571-resistant MPD in animal models . SU11248 and SU6668 are broad spectrum RTK inhibitors that potently enhance the regression of different PDGFR-expressing tumor types in preclinical models, particularly when supplied in combination with chemo- or radiotherapy (Abrams et al, 2003; Garofalo et al, 2003; Mendel et al, 2003; Schueneman et al, 2003; Marzola et al, 2004).

Synthetic agents that bind to PDGFs can antagonize the activation of PDGFRs. One such molecule, GFB-111, binds PDGF with a high degree of selectivity over other growth factors. GFB-111 inhibits PDGFR autophosphorylation and activation of downstream effectors, and abolishes tumor growth in nude mice grafted with different tumor cell lines, possibly interfering with angiogenesis .

IGF-IR is an ubiquitously expressed RTK composed of two extracellular a subunits, which bind the IGF-I and IGF-II ligands, and of two cytoplasmic b subunits that contain the kinase domain. The signaling cascades triggered by the binding of IGFs to the receptor are mainly involved in cell survival and proliferation, but they also affect migration, invasion, and cell-substrate adhesion. Tumorigenesis by IGF-IR requires the activation of anti-apoptotic pathways following constitutive overexpression and/or hyperactivation of the receptor, which may also result from autocrine or paracrine stimulation. IGF-IR is involved in the development of several malignancies, as carcinomas of breast, prostate, liver, colon, pancreas and lung, and many compounds fully accomplish their tumorigenic potential only in the presence of IGF-IR .

Several strategies to inactivate IGF-IR are being developed. Blocking antibodies or single-chain humanized scFv-Fc immunoconjugates raised against the a subunit of the receptor downregulate IGF-IR, inhibit the survival of cancer cells and cause regression of pancreatic tumor xenografts in mice (Hailey et al, 2002; Maloney et al, 2003; Sachdev et al, 2003). The low-molecular weight kinase inhibitor NVP-ADW742 abrogates the growth of small cell lung cancer cells , whereas NVP-AEW541 reduces the growth of IGF-IR-driven fibrosarcomas in mouse xenografts . Mutant IGF-IRs delivered to target cells act as dominant-negative receptors. In particular, truncated IGF-IR mutants lacking the b subunit cause massive apoptosis of cancer cells in animals, reducing tumor growth and metastatization and also displaying bystander effects on neighboring cells (Hongo et al, 2003; Min et al, 2003; Sachdev et al, 2004). Furthermore, since the C terminal tail inhibits the anti-apoptotic activity of the receptor, the expression of this portion of IGF-IR linked to a membrane-anchoring signal (mini-receptor) abrogates tumorigenesis in nude mice by inducing apoptosis of tumor cells .

Alternative strategies to target IGF-IR or IGFs are still in early discovery phases, and include: (I) increase of receptor internalization by molecule that act on chaperones; (II) use of ligand mimetic peptides, whose efficacy in vivo remains to be determined; (III) antisense oligodeoxynucleotides, antisense RNAs, ribozymes or siRNAs, which are currently under intensive investigation in animal models (Bohula et al, 2003a, 2003b; Surmacz, 2003; Grzmil et al, 2004; Nielsen et al, 2004).

FGFRs make up a family of four receptor types that bind to 23 different ligands, the fibroblasts growth factors (FGFs), whose prototypic members are acidic FGF (aFGF or FGF-1) and basic FGF (bFGF or FGF-2). FGFs can be secreted from cells or localized on the cell surface, and they also interact with heparin or heparan sulphate proteoglycans . FGFR activation triggers a number of transduction events in epithelial and mesenchimal cells . Biological responses include proliferation, differentiation, migration and survival of cells, and also organogenesis, inflammation, hematopoiesis and wound healing (Ornitz and Itoh, 2001; Cronauer et al, 2003). In addition, the activation of FGFRs plays a major role in angiogenesis and in the degradation of extracellular matrix, thus facilitating metastatization (Christofori, 2003; Manetti and Botta, 2003). FGFs and/or FGFRs are aberrantly expressed in several malignancies, and a high FGF serum level correlates with poor prognosis and resistance to chemotherapeutics . An aberrant tyrosine kinase activity of FGFR1 is generated by chromosomal translocations that generate fusion proteins responsible for myeloproliferative disorders .

The solving of the crystallographic structures of the FGFR TK domain has provided the basis for the rational design of small-molecule kinase inhibitors, some of which are undergoing preclinical trials . The development of pyrido-pyrimidine derivatives led to the development of an array of agents (PD166285, PD173074, PD161570, PD166866) that selectively inhibit FGFR autophosphorylation and signaling in various cell models . The association of a broad-spectrum (PD166285) and of a selective (PD173074) kinase inhibitor of this family in combination with photodynamic therapy causes tumor regression in animal models . The quinazoline derivatives ZD4190 and ZD6474 abrogate the proliferation of endothelial cells, and are currently under preclinical evaluation . The oxindole derivative compounds SU6668, SU5416 and SU5402 inhibit vascularization and growth of tumor xenografts of diverse origins. They also induce regression of established tumors and abrogation of metastasis formation, with a massive apoptosis of tumor and endothelial cells. Currently SU6668, which is also used in combination with chemotherapeutics, is undergoing phase II clinical trials (Laird et al, 2000; Shaheen et al, 2001; Garofalo et al, 2003; Manetti and Botta, 2003; Udayakamur et al, 2003).

The HGFR, encoded by the proto-oncogene MET, is functionally deregulated in a variety of cancer types. Met overexpression strictly correlates with higher metastatic potential and poor prognosis in a plethora of aggressive tumors, including thyroid and colorectal carcinomas, whereas its tyrosine kinase activity is increased by germ-line and somatic mutations in papillary renal, gastric and hepatocellular carcinomas, and in lymph-node metastases of head and neck squamous-cell carcinomas. HGF-dependent autocrine loops are found associated with osteosarcomas, rhabdomyosarcomas and breast carcinomas (Danilkovitch-Miagkova and Zbar, 2002; Trusolino and Comoglio, 2002). To date, Met is the only known proto-oncogene whose activation can elicit and orchestrate all the biological programs required for invasive growth, a complex process in which apparently independent events such as proliferation, migration, survival, polarization and matrix degradation, are integrated to confer cells the capability to invade secondary districts. Hence, selective Met inhibitors would be endowed with the unique feature of hampering cancer progression towards malignancy and metastasis spreading. Nevertheless, Met inhibitors are far from being tested in clinical trials, even though in recent years several prototype molecules have been developed in pre-clinical studies.

The furanosylated indolocarbazole K252a and the pyrrole-indolinone PHA-665752 competitively inhibit the binding of ATP to the Met catalytic domain. Both compounds abrogate the activation of the Met pathway and prevent the in vitro oncogenic properties driven by c-Met. Unfortunately, K252a is inactive in vivo , whereas the K252a analogues CEP-751 and CEP-701 have to be tested on Met-driven tumors. Instead, PHA-665752 causes growth inhibition or regression of tumors at well-tolerated doses in mouse models . The indoline derivative SU11274 is a highly selective Met inhibitor that induces G1 cell cycle arrest and apoptosis by targeting key regulators of the PI3K pathway .

Tumorigenesis induced by mutant forms of Met is strictly dependent on ligand stimulation and can therefore be inhibited by HGF antagonists . An alternatively spliced form of HGF, NK2, antagonizes the growth of Met-expressing melanoma xenografts in NK2-HGF bitransgenic mice, but proved not to be useful for therapeutic development as it facilitates metastasis . Another HGF fragment, NK4, binds but does not activate the receptor, acting as an antagonist of the biological activity of HGF, and also as an angiogenesis inhibitor . In animal models, NK4 suppresses tumor growth and inhibits metastasis even in cases of advanced diseases (Maemondo et al, 2002; Matsumoto and Nakamura, 2003). A similar result has been recently obtained with a soluble decoy Met receptor . Alternatively, peptides directed against the cytosolic tail of Met can impair the catalytic and biological properties of the receptor, thus providing another selective targeting strategy .

Furthermore, the comprehension of the machinery that recycles the receptor on plasma membrane will allow its selective down-modulation. Geldanamycins are antibiotics that interfere with the chaperone function of Hsp90, thus reducing the expression on plasma membrane of several RTKs by not completely understood mechanisms . These drugs down-regulate Met protein expression and inhibit HGF/SF-mediated cell motility and invasion in NIH3T3 cells transformed by mutant forms of Met . The use of receptor as a possible apoptosis enhancer constitutes an attractive therapeutic possibility. We have recently observed that, in ovarian cancer cells overexpressing Met, the apoptotic effect of chemotherapeutics is increased by stimulation with HGF . HGF constructs devoid of pro-invasive properties could improve the feasibility of this approach .

Targeting either HGF or Met with recombinant ribozymes carrying antisense sequences (anti-HGF or anti-Met U1snRNA/ribozyme) reduces HGF/Met expression and abolishes Met-driven biological responses. When delivered in mouse glioma xenografts, these ribozymes inhibit cancer growth by inducing tumor cell apoptosis and blocking tumor angiogenesis .

The development of targeted RTK inhibitors represents a major breakthrough both in the understanding of the molecular mechanisms of cancer and in the rational design of therapeutic tools. Novel-generation compounds will be designed following the integration of combinatorial chemistry, high-throughput screening, genomic and/or proteomic profiling, comprehension of the three-dimension molecular structure, together with a better understanding of signal transduction mechanisms.

Several important issues need to be addressed in order to maximize the enormous potential of receptor targeting strategies. The efficacy of this approach depends upon the differential magnitude of receptor expression by the tumor vs normal cells, as well as by accessibility of tumor cells to drug administration. Receptor levels might not simply correlate with response, and the selection of potentially responsive patients must be accurately achieved by molecular profiling and by improved methods to find activated receptors. New inhibitors must be carefully designed in order to mitigate toxicity and to increase selectivity, but non-selective compounds could be required in some cases for overcoming redundancies in deregulated signal cascades. The abrogation of signaling cascades at the cell membrane is generally required for a complete inhibition of downstream transduction networks, but in certain situations retaining a subset of signals could be advantageous. In this frame, the development of animal models that better reflect the pathogenesis of human malignancies is highly warranted. Moreover, potential applications for RTK inhibitors in chemoprevention and in the chronic or long-term treatment settings are being investigated.

Finally, it remains to be fully understood why does targeted therapy works, as it is not specific for receptor mutations. This possibly depends on the particular sensitivity of tumor cells to the disruption of dominant oncogenic pathways (oncogene addiction). Therefore, tumor resistance could appear following selection of cells no longer addicted to the targeted oncogene. The targeting of cancer cell microenvironment, an accurate molecular characterization of the tumor and the improvement of crossing approaches with more conventional anticancer treatments will help to tailor more adequate and efficient therapies.

We are indebted with Prof. Paolo M. Comoglio for continuous support.

Andrea Rasola