The role of the EGFR signaling in tumor microenvironment^†

It has been long established that the epidermal growth factor receptor (EGFR) and its ligands, belonging to the epidermal growth factor (EGF)-like family of growth factors, play an important role in tumorigenesis. These proteins are frequently expressed in the majority of human carcinomas (Salomon et al., 1995; Normanno et al., 2003). More importantly, the EGFR and its cognate growth factors support autocrine and paracrine circuits that regulate several mechanisms underlying tumor pathogenesis and progression, including proliferation, survival, tissue invasion, and ability to form metastases (Salomon et al., 1995; Normanno et al., 2003, 2005a).

Tumor progression is a complex phenomenon that involves the interaction of tumor cells with surrounding normal tissues. The EGFR is expressed in almost all types of non-transformed cells, with the only exception of mature cells of the lymphohemopoietic system (Salomon et al., 1995; Normanno et al., 2006). Therefore, activation of the EGFR signaling in non-malignant cell populations of the tumor microenvironment might play an important role in cancer progression. In particular, evidence suggests that the EGFR system is involved in two important phenomena that promote tumor progression such as angiogenesis and bone destruction that is necessary for the development of metastatic bone disease.

Angiogenesis, the formation of new blood vessels from the existing vasculature, is essential for the growth of the primary tumor and for the formation of metastasis (Folkman, 1971). A number of different growth factors and cytokines can affect the proliferation and migration of endothelial cells and have been implicated in tumor-associated angiogenesis (Jain, 2003; Ferrara and Kerbel, 2005). Actually, angiogenesis is mediated by positive and negative regulators whose balance determines the rate of vessel formation (Ferrara and Kerbel, 2005). Vascular endothelial growth factor (VEGF) is the main inducer of tumor-associated neo-angiogenesis, although additional growth factors such as interleukin-8 (IL-8) and basic fibroblast growth factor (bFGF) might in turn play a role in this phenomenon (Ferrara and Kerbel, 2005). As we will discuss, the EGF receptor/ligand system regulates tumor-associated angiogenesis by different mechanisms, which include both direct and indirect effects on endothelial cells.

Tumor-induced stimulation of osteoclast-mediated bone resorption is the main mechanism responsible for bone destruction in cancer patients (Roodman, 2001). In this respect, bone marrow stromal cells (BMSCs) represent critical mediators for the pathogenesis of bone metastases. In fact, these cells are able to synthesize different growth factors that can induce differentiation and activation of osteoclast (Boyle et al., 2003). Interestingly, recent findings suggest that BMSCs express functional EGFR, whose signaling regulates the ability of these cells to sustain osteoclast activation (Normanno et al., 2005b).

In this short review, we describe the role of the EGFR and its ligands in the cross-talk between tumor cells and the above-mentioned non-malignant cell types of the tumor microenvironment. In particular, we will illustrate how these interactions might affect tumor growth and progression and how these mechanisms may offer important potential targets for novel therapeutic approaches.

The ErbB Receptor/Ligand Network

The ErbB family of tyrosine kinase receptors includes four members, namely EGFR, ErbB-2, ErbB-3, and ErbB-4 (Table 1; Normanno et al., 2005a, 2006). These proteins have an extracellular ligand-binding domain, a single hydrophobic transmembrane domain and a cytoplasmic tyrosine kinase-containing domain (Olayioye et al., 2000). The intracellular tyrosine kinase domain of ErbB receptors is highly conserved. However, the kinase domain of ErbB-3 contains substitutions of critical amino acids and therefore lacks kinase activity (Guy et al., 1994). The extracellular domains of the ErbB receptors are less conserved, suggesting that they are able to bind different ligands (Olayioye et al., 2000; Yarden, 2001; Yarden and Sliwkowski, 2001). In fact, the receptors of the ErbB family are activated following binding to peptide growth factors of the EGF-family (Table 1). EGF-like growth factors can be produced by the same cells that express ErbB receptors (autocrine secretion) or by surrounding cells (paracrine secretion), including stromal cells (Olayioye et al., 2000; Yarden and Sliwkowski, 2001). The EGF-like peptides are characterized by the presence of an EGF-like domain composed of three disulfide-bonded intramolecular groups, which confers binding specificity, and additional structural motif such as immunoglobulin-like domains, heparin-binding sites and glycosylation sites. Proteins of this family can be divided into three groups (Table 1; Yarden and Sliwkowski, 2001; Normanno et al., 2003). The first group includes EGF, transforming growth factor α (TGF-α), amphiregulin (AR), and the recently identified ligand epigen that binds specifically the EGFR (Strachan et al., 2001). The second group includes betacellulin (BTC), heparin-binding growth factor (HB-EGF), and epiregulin (EPR), which are capable of binding both EGFR and ErbB-4. The third group is composed of the neuregulins (NRGs) and tomoregulin, and can be divided in two subgroups based upon their capacity to bind ErbB-3 and ErbB-4 (NRG-1 and NRG-2) or only ErbB-4 (NRG-3, NRG-4, and tomoregulin) (Carraway et al., 1997; Chang et al., 1997; Zhang et al., 1997; Harari et al., 1999). None of the EGF family of peptides binds ErbB-2.

The ErbB receptors form either homo- or hetero-dimers following ligand binding. Formation of dimers is essential for receptor activation and subsequent transmission of intracellular signals (Olayioye et al., 2000). In fact, following dimerization, auto- and trans-phosphorylation in tyrosine residues of the ErbB receptors occur. The tyrosine-phosphorylated receptors become able to interact with adaptor proteins that couple the receptors to intracellular signaling pathways (Olayioye et al., 2000). The ErbB receptors are able to activate different intracellular signaling cascades, including the phosphatidylinositol 3-kinase (PI3K)/Akt, the ras/raf/MEK/mitogen-activated protein kinase (MAPK), and the signal transducer and activator of transcription (STAT) pathways. Activation of these signaling proteins regulates cellular functions involved in cancer pathogenesis and progression (Normanno et al., 2006).

The Role of EGFR in Angiogenesis

It has long been demonstrated that EGFR ligands such EGF and TGFα are potent pro-angiogenic factors (Schreiber et al., 1986). However, the effects of these peptides on tumor-associated angiogenesis depend on two distinct phenomena: an indirect effect, due to the ability of the EGFR system to regulate the production of angiogenic factors in tumor cells, and a direct effect on endothelial cells.

The Role of EGFR in the Pathogenesis of Bone Metastases

The pathogenesis and progression of bone metastases represent complex phenomena that involve the interaction of tumor cells with different cell types of the bone microenvironment (Roodman, 2001). Although it has been shown that tumor cells can directly resorb bone, evidence suggests that the main mechanism responsible for bone destruction in cancer patients is tumor-induced stimulation of osteoclast-mediated bone resorption (Roodman, 2001). Osteoclasts are the specialized progeny of hemopoietic precursors committed to the monocyte/macrophage lineage that, upon appropriate stimuli, fuse by giving rise to mature bone resorbing cells (Boyle et al., 2003). Development and functional activation of mature osteoclasts are regulated through the interplay of several cytokines and growth factors. However, two factors are both necessary and sufficient for osteoclast formation and activation: macrophage colony stimulating factor (M-CSF), which induces proliferation and differentiation of pre-osteoclast cells, and receptor activator of NF-kB ligand (RANKL) that is involved in fusion and activation of these cells (Boyle et al., 2003). Activation of osteoclasts is also regulated by osteoprotegerin (OPG), a soluble decoy receptor for RANKL that functions by sequestering the secreted form of RANKL and, therefore, by preventing the binding to its cognate receptor (Boyle et al., 2003).

Cancer cells are able to synthesize many growth factors and cytokines that can lead to the activation of osteoclasts (Roodman, 2001). In detail, parathyroid hormone related protein (PTHrP) is believed to represent the main mediator of breast cancer-induced bone resorption (Burtis et al., 1990). PTHrP has been purified from breast cancer cells, and it is expressed in up to 60% of human primary breast carcinomas (Burtis et al., 1990; Southby et al., 1990). However, most osteotropic factors do not directly stimulate osteoclast, but rather act indirectly by binding to accessory cells of the bone marrow microenvironment (Roodman, 2001). This heterogeneous cell compartment comprehends specialized endothelial cells, as well as mesenchymal stem cells (MSC), which maintain a level of self renewal and give rise to different specialized connective tissue cells such as “reticular cells,” osteoblasts, chondrocytes, adipocytes, and SMCs (Clark and Keating, 1995; Deans and Moseley, 2000). MSC and their progeny are collectively referred to as the BMSC compartment that is known to support hemopoiesis.

MSC, marrow stromal cells, and osteoblasts support osteoclast differentiation within the bone (Takahashi et al., 1988; Udagawa et al., 1989; Mbalaviele et al., 1999). In this regard, tumor cells have been shown to regulate the synthesis of osteoclastogenic factors in BMSC. In fact, PTHrP induces in BMSC the expression of RANKL (Horwood et al., 1998). Interestingly, no expression of RANKL was demonstrated in breast cancer cell lines (Thomas et al., 1999). In agreement with these data, these cells did not act as surrogate osteoblasts to support osteoclast formation in co-culture experiments (Thomas et al., 1999). However, when breast cancer cells overexpressing PTHrP were added to co-cultures of murine osteoblasts and hematopoietic cells, osteoclast formation took place without the addition of any osteotropic agents (Thomas et al., 1999). RANKL has also been shown to increase in BMSC the production of IL-6 that is involved in activation of osteoclasts; IL-6 in turn can upregulate RANKL and downregulate OPG in osteoblasts through PGE2 (Giuliani et al., 2004; Liu et al., 2005). Finally, it has been demonstrated that human MSC positively regulate osteoclastogenesis as undifferentiated progenitor cells (Mbalaviele et al., 1999). Co-culture with CD34+ hematopoietic stem cells induces MSC to secrete cytokines that support osteoclast formation, such as IL-6, IL-11, leukemia inhibitory factor (LIF), and M-CSF (Mbalaviele et al., 1999).

Several findings suggest a potential role of the EGFR pathway in regulating the ability of BMSC to promote osteoclast formation and activation. Addition of recombinant TGFα to long-term human bone marrow cultures, which contain MSC, stromal cells, osteoblasts, and osteoclast-precursors, has been shown to stimulate the formation of bone resorbing osteoclasts (Takahashi et al., 1986; Tanaka et al., 1998). In addition, the ability of TGFα and EGF to stimulate bone resorbing activity in organ culture systems, such as the fetal rat long bone and neonatal mouse calvarial systems, has long been described (Tashjian et al., 1985; Ibbotson et al., 1986). Several studies have also demonstrated expression of EGFR in BMSC and in osteoblast-like cells, as well as biological effects of EGF and/or TGFα in osteoblasts (Ibbotson et al., 1986; Drake et al., 1994; Sodek et al., 1995; Satomura et al., 1998). For example, EGF and TGFα caused a concentration-dependent inhibition of alkaline phosphatase activity in osteoblast-like cells (Ibbotson et al., 1986). EGF has also been found to upregulate the expression of mRNA for osteopontin in osteoblasts; osteopontin is a bone matrix protein that is synthesized by osteoblastic cells and mediates osteoclast adherence to the bone matrix (Sodek et al., 1995). Interestingly, PTH induced an increase in the levels of expression of EGFR in osteoblast-like cells, suggesting that PTH-induced growth of osteoblasts might be mediated by EGFR (Drake et al., 1994). Since expression of EGF has been demonstrated to occur in osteoclasts (Symons, 2003), it is conceivable that paracrine circuits involving EGFR and its ligands are operating between osteoclasts and osteoblasts. More recently, the functional role of EGFR signaling in MSC has been investigated. Krampera et al. (2005) reported that activation of EGFR by HB-EGF increased cell proliferation and prevented adipogenic, osteogenic, and chondrogenic differentiation in human bone marrow-derived MSC. In contrast, a different study described that the differentiation of human MSC into bone forming cells was stimulated by EGF but not PDGF (Kratchmarova et al., 2005). The use of different EGFR-ligands might at least in part justify such contrasting results. Furthermore, slight differences among preparations of primary MSC might also give rise to cells with divergent ability to differentiate in specialized bone marrow populations.

The involvement of the EGFR in the pathogenesis of bone metastases is suggested by an unexpected phenomenon that was observed in clinical trials of gefitinib in breast cancer patients. Albain et al., 2002 enrolled in their study of gefitinib in breast cancer 12 patients with bone metastases and bone pain. Surprisingly, 5 out of 12 patients had a significant relief of bone pain, leading to the complete withdrawal of all scheduled narcotics in several cases. Due to the impressive effects on bone pain palliation, two patients were maintained on gefitinib despite objective progression of the disease. A significant improvement in bone pain was also reported in a patient enrolled in a different trial of gefitinib in metastatic breast cancer (von Minckwitz et al., 2005).

Recent findings from our group have shed light on the role of EGFR signaling in the pathogenesis of bone metastases. In fact, we found that treatment of MSC-like cells with gefitinib resulted in significant reduction of the basal levels of activation of the EGFR and AKT, without affecting MSC proliferation (Normanno et al., 2005b). However, gefitinib treatment produced a significant decrease in the levels of secreted M-CSF and cell-associated RANKL in MSC (Normanno et al., 2005b). Furthermore, the ability of conditioned medium from gefitinib-treated MSC-like cells to sustain the differentiation of pre-osteoclasts was significantly reduced as compared with untreated cells (Normanno et al., 2005b). These results demonstrated for the first time that the EGFR regulates the ability of MSC to induce osteoclast differentiation. Our findings were confirmed by Angelucci et al. (2006) who demonstrated that treatment with gefitinib significantly reduced the ability of conditioned medium from prostate cancer cells to induce expression of RANKL in osteoblasts. In this regard, it is well established that EGFR and several of its ligands are expressed by prostate cancer cell lines and human primary prostatic carcinomas (Leverton and Gullick, 2000). Interestingly, expression of EGF-like peptides in BMSC has been also demonstrated (Mahtouk et al., 2004; Normanno et al., 2005b).

Although the above-mentioned study suggest a direct role of the EGFR in MSC-mediated osteoclast differentiation, different mechanisms are likely to be involved in the effects of anti-EGFR agents on the pathogenesis and progression of bone metastases (Normanno and Gullick, 2006). In fact, anti-EGFR agents might have a direct effect on cancer cells. In this regard, it has been shown in an experimental model of prostate cancer cells implanted in the bone of nude mice that tumor cells growing adjacent to bone tissues express high levels of activated EGFR whereas tumor cells invading the muscle do not (Kim et al., 2003b). This observation might suggest that activation of EGFR is essential for the ability of prostate cancer cells to grow within the bone. In addition, EGFR signaling modulates in tumor cells the expression of molecules that play an important role in the metastatic cascade. For example, the urokinase-type plasminogen-activator (uPA)/uPA receptor (uPAR) system and matrix metalloproteinases (MMPs) are involved in the invasive ability of tumor cells and in the formation of bone metastases (Guise and Mundy, 1998; Nemeth et al., 2002). In this regard, treatment with gefitinib significantly reduced the expression and activation of uPA and MMP-9 in prostate cancer cells, and this phenomenon was associated with a significant decrease in the ability of these cells to form bone metastases (Angelucci et al., 2006). Furthermore, treatment of human squamous cell carcinoma of head and neck (SCCHN) cells with the anti-EGFR blocking antibody C225 (cetuximab) significantly reduced their ability to invade surrounding tissues, including bone, and this inhibition was associated with downregulation of MMP-9 expression (Huang et al., 2002). Neo-angiogenesis is also essential for the formation of bone metastases. Expression of the EGFR in the endothelial cells of experimental models of bone metastases has been described (Kim et al., 2003b; Weber et al., 2003). The ability of anti-EGFR agents to inhibit the formation of bone metastases might be related at least in part to the induction of apoptosis in EGFR-expressing endothelial cells. Interestingly, it has been also demonstrated that osteoclasts express the VEGF receptor-1 (VEGFR-1), and that VEGF can induce osteoclast differentiation and activation (Niida et al., 1999). Therefore, the decrease in VEGF synthesis that is induced by anti-EGFR agents might lead to a reduced activation of osteoclasts. Finally, expression of EGFR mRNA in osteoclasts has been described (Wang et al., 2004). The EGFR tyrosine kinase inhibitor AG1478 reduced the formation of osteoclasts from bone marrow cells. However, expression of functional EGFR protein in osteoclasts has not yet been formally documented.

Conclusions and Perspectives

The above-summarized data demonstrate that the ErbB receptors and their ligands are involved in the cross-talk between cancer cells and different cell types of the tumor microenvironment (Fig. 1). In fact, several different studies have shown that the signaling of ErbB receptors regulates the production of pro-angiogenic factors in tumor cells. In addition, the majority of human carcinomas are able to synthesize and to secrete EGF-like growth factors that can bind to ErbB receptors expressed in accessory cells of the tumor microenvironment (Normanno et al., 2001). This interaction has been shown to regulate important mechanisms of tumor progression, such as proliferation and motility of endothelial cells and production of pro-angiogenic and pro-osteoclastogenic cytokines in tumor and stromal cells. Interestingly, expression of ErbB ligands has also been demonstrated to occur in endothelial cells (Arkonac et al., 1998; Iivanainen et al., 2003, 2007). Endothelial-derived EGF-like growth factors can have multiple functions: (a) regulation of VEGF synthesis in tumor cells; (b) activation of autocrine circuits in endothelial cells; and (c) direct activity on SMCs that are implicated in vessel maturation. In fact, SMCs express EGFR and ErbB-2, and their migration is induced by activation of the EGFR (Iivanainen et al., 2003). Furthermore, the EGFR signaling regulates the synthesis of VEGF in SMCs (Sini et al., 2005). Finally, BMSCs secrete EGF-like peptides and angiogenic growth factors that can contribute to boost the above-mentioned paracrine circuits (Gupta et al., 2001; Mahtouk et al., 2004; Normanno et al., 2005b).

The receptors and ligands of the EGFR family mediate complex interactions between tumor cells and the neoplastic environment that may ultimately result in enhanced tumor growth and progression. This observation makes of the EGFR an interesting target for therapeutic intervention even in tumors with EGFR-independent growth. Interestingly, clinical activity of anti-EGFR agents in patients carrying EGFR-negative tumors has been already demonstrated (Normanno et al., 2003). In this respect, the finding that EGFR-targeting drugs can affect mechanisms that support cancer progression in the tumor microenvironment might at least in part explain this phenomenon.

Two different hypotheses on the mechanism of action of anti-EGFR agents with respect to inhibition of angiogenesis have been formulated. Several studies have shown that anti-EGFR agents produce a significant reduction in the synthesis of pro-angiogenic cytokines in experimental tumors. It has also been demonstrated that downregulation of angiogenic factors precede the reduction of microvessel density (Perrotte et al., 1999). Therefore, these data suggest that the anti-angiogenic effect of EGFR targeting drugs is mainly indirect. On the other hand, it has been suggested that tumor-associated endothelial cells express the EGFR and that blockade of this receptor can lead to apoptosis of endothelial cells (Baker et al., 2006). It is conceivable that both these mechanisms might be involved in the anti-tumor effect of anti-EGFR agents. Finally, the observation that the EGFR regulates the production of pro-osteoclastogenic cytokines in BMSC suggests that anti-EGFR drugs might be employed to control the progression of bone metastases. However, clinical trials of anti-EGFR agents in cancer patients with metastatic bone disease are definitely required to confirm this hypothesis.

An interesting therapeutic approach in cancer patients is the combined blockade of EGFR and VEGF receptor signaling, which both contribute to tumor progression. In this regard, different tyrosine kinase inhibitors that are able to block the activation of both receptors have been synthesized and are in clinical development. The anti-tumor efficacy of these agents has been demonstrated in different experimental models of human tumors including bone metastases (Ciardiello et al., 2003; Yokoi et al., 2005; Younes et al., 2005; Morelli et al., 2006; Wu et al., 2007). Preliminary results of clinical trials with these drugs suggest a good tolerability and a significant anti-tumor activity (Morabito et al., 2006). However, larger clinical trials are definitely required to confirm the efficacy of these compounds.

In conclusion, the evidence summarized in this article suggests that the EGFR system is an important mediator of autocrine and paracrine circuits within the tumor microenvironment. Since the EGFR is expressed in all the cell types with the exception of mature hemopoietic cells, these findings also highlight the importance of studying the effects of anti-EGFR agents in the different non-cancer cell types of the neoplastic microenvironment that might be involved in tumor growth and progression. In this respect, targeting EGFR-mediated circuits of tumor microenvironment might represent a novel important therapeutic approach.