Under physiologic conditions, epithelial cells proliferate in response to mitogens of the epidermal growth factor (EGF) family of ligands that are secreted in autocrine or paracrine fashion, or presented via cell-cell contact and activate the EGF receptor (EGFR) (Fig. 108.2A). For instance, cells are stimulated to proliferate via EGF during wound healing. The signal is initiated by 1 of 11 members of the EGF family. All EGF family ligands exist in pro-form as transmembrane precursor proteins. They are converted to an active diffusible form by extracellular cleavage of the membrane protein. Six members of the EGF family are expressed in most SCC cell lines and activate the EGFR: EGF, transforming growth factor alpha (TGFa), HB-EGF, amphiregulin, betacellulin, and heregulin (45). HB-EGF is particularly interesting. The proform may either be cleaved to generate a diffusible factor by matrix metalloproteases (MMPs) that degrade extracellular matrix during cell migration (46), or it may activate EGFR as a transmembrane protein via cell-cell interaction, acting in a juxtacrine fashion. The effects on the target cell are different depending on whether HB-EGF is soluble or membrane bound. (Singh AB, Harris R. Autocrine, paracrine and juxtacrine signaling by EGFR ligands. Cell Signal 2005;10:1183-1193.)

Growth factors present in the interstitial space can also act on other receptors of the EGFR family (also called the ErbB family) that include ErbB1 (EGFR), ErbB2 (HER2/cneu), ErbB3 (Her3), and ErbB4 (Her4) (45). In absence of ligand, most ErbB-family receptors are present at the cell surface of epithelial cells in the form of monomeric inactive transmembrane tyrosine kinase receptors. Binding of bivalent extracellular ligand triggers receptor dimerization, which brings catalytic cytoplasmic tails into close proximity, allowing trans-autophosphorylation of the cytoplasmic tails on tyrosine, serine, and threonine residues (47, 48). ErbB receptors present at the cell surface are not a static pool of proteins. The receptors constantly cycle between an endosomal reservoir and the cell surface, where they become exposed to the ligands (49). After being activated, the receptor is endocytosed and either degraded or recycled (47, 49).

Upon activation and transphosphorylation, active EGFR triggers three major signaling pathways. First, it activates the Ras-Map kinase pathway. Autophosphorylation sites on the cytoplasmic tail become docking sites for the adaptor protein SHC, which in turn recruits Grb2, which recruits Sos. Sos is a guanosine exchange factor for Ras, a small GTPase bound to the inner leaflet of the plasma membrane where it cycles between GDP and GTP bound forms (50, 51). Ras acts as a molecular switch which is turned “on” when bound to GTP (52). The Sos activates Ras by mediating the exchange of GDP for GTP. Active Ras recruits the scaffolding protein Raf to the membrane (53). This leads to activation of the map kinase cascade, ultimately leading to ERK
and Jun kinase (JNK) phosphorylation and activation (54). JNK is a serine/threonine kinase, which phosphorylates the c-Jun transcription factor. Phosphorylated c-Jun forms a heterodimer with phosphorylated c-fos, which together constitutes the AP-1 early response transcription factor that drives cell proliferation (55). ERK can also phosphorylate and activate transcription factors responsible for cell proliferation and cell migration.

The second signaling pathway activated by EGFR is the Jak/STAT pathway. Inactive STAT (signal transducer and activator of transcription) proteins are located in the cytoplasm. Activated EGFR recruits and activates the transcription factor STAT3 via two mechanisms: via Janus kinase (Jak) or via direct interaction with STAT3. Jak is a tyrosine kinase that binds to the juxtamembrane intracellular portion of activated receptors. Active Jak then phosphorylates
and activates STAT1 or STAT3 proteins (56). An alternate mode of activation of STAT3 is via direct interaction with EGFR (57). STAT3 is recruited to active EGFR, where it binds to specific phosphotyrosine residues of the receptor via its Src homology domain (SH2). STAT3 is then directly phosphorylated by the receptor on critical tyrosine and serine residues. Phosphorylation of STAT3, either via Jak or via EGFR, leads to its dimerization, either as STAT3 homodimers or as STAT3:STAT1 heterodimers (58). Activated STAT transcription factor dimers translocate either to mitochondria where they regulate metabolism characteristic of cancer cells (59, 60) or to the nucleus where they regulate growth factor/cytokine-directed gene expression (56) and increase cell motility (61, 62).

Finally, EGFR also activates the PI3 kinase/Akt pathway that promotes cell survival by suppressing apoptosis. Activated EGFR recruits and activates phospholipase-Cγ, which directs the hydrolysis of the phosphatidylinositol biphosphate (PIP2) at the plasma membrane. This hydrolysis generates an insoluble diacyl glycerol, which activates PKC, and a soluble phosphatidyl (1, 4, 5) triphosphate in the cytosol (32). There, IP3 (1, 4, 5) releases Ca2+ stores from the endoplasmic reticulum. Activated EGFR can also recruit and activate PI3 kinase, via the SH2 domain of the kinase. PI3 kinase then phosphorylates PIP2 to generate phosphatidylinositol-(3, 4, 5) triphosphate (PIP3) that remains membrane bound (63). Proteins with a pleckstrin homology (PH) domain such as Akt and PDK migrate to and associate with PIP3 located at the plasma membrane (63). Close proximity of Akt and PDK leads to Akt phosphorylation and activation. Akt renders the cell more resistant to apoptosis and increases proliferation. In addition, Akt also induces a motile and invasive cell phenotype. PTEN, an important tumor suppressor that is frequently mutated in cancers, counteracts Akt effects (64, 65). PTEN’s sequence resembles serine, threonine, and tyrosine phosphatases, but it acts principally as a lipid phosphatase converting PIP3 into PIP2 (64). Thus, PTEN tempers the effect of PI3K activation by limiting PIP3-mediated signaling. PI3 kinase lipid signaling pathway is a major pathway linking EGFR activation to increased cell proliferation and motility and decreased cell death, and is often involved in carcinogenesis.

EGFR is frequently upregulated in HNSCC (66, 67, 68), and in general EGFR upregulation correlates with a worse prognosis (68, 69, 70), including more rapid disease progression and the presence of lymphatic metastasis, through its effect on proliferation, motility, and angiogenesis (71). The mechanisms leading to EGFR overexpression in HNSCC are still unclear. Genome sequencing suggests that EGFR is not mutated in HNSCC, in contrast to other tumor types (3). It is rarely amplified (less than 10% of HNSCC). However, a peculiarity of HNSCC is the common overexpression of a truncated mutant of the receptor (EGFRvIII) in 40% of HNSCC (72). EGFRvIII, which is constitutively active and has degradation prolonged half-life, is found exclusively in tumors, making it a prime target for anticancer therapy with specific monoclonal antibodies. Because of the high frequency of EGFR upregulation HNSCC, EGFR-targeting strategies have been extensively investigated. Both small molecule EGFR tyrosine kinase inhibitors and monoclonal antibodies have been tested for activity against HNSCC. Cetuximab, an EGFR-blocking antibody, demonstrated activity against HNSCC, with improved locoregional control and survival compared to radiotherapy alone (73) and with improved response rate and survival when used in combination chemotherapy (74). Cetuximab (Erbitux) is now used routinely to potentiate radiotherapy. To date, while tyrosine kinase inhibitors targeting EGFR (such as gefitinib and erlotinib) have activity in HNSCC, they have not demonstrated prolonged survival in Phase III clinical studies (75). This discrepancy suggests that EGFR-blocking antibody has effects other than inhibition of the signaling pathway; the antibody could have an effect on immunity (76, 77), or on EGF-independent effects such as DNA repair (78, 79).

In addition to targeting EGFR itself, considerable effort has been invested in inhibiting downstream targets. STAT3 homodimers, which have oncogenic properties, are overexpressed in HNSCC and are the target of anticancer therapies (80). Specifically, STAT3 inhibitory strategies using dimerization-disrupting agents or competitive inhibitors of the DNA-binding domain of STAT3 have been proposed to combat HNSCC (80, 81, 82); the latter was used in a clinical trial with promising results. HNSCC genome sequencing found that 5% of tumors display H-Ras mutations, leading to constitutive activation of the small GTPase (3). Whether Ras is mutated or simply subject to sustained activation by EGFR, it is a seductive therapeutic target. Ras may be inhibited via blocking farnesylation of the protein (63), an essential posttranslational modification required for its targeting to the plasma membrane where it mediates its effects. In addition, the PIP3 lipid pathway is also relatively frequently mutated in HNSCC. PTEN is mutated in 7% of HNSCC with some cancers bearing both H-Ras and PTEN mutations (3, 83). The inactivation of PTEN promotes PIP3 accumulation and activation of downstream targets such as Akt. PI3KCA, a catalytic subunit of PI3 kinase, is also mutated in HNSCC (3, 84). These mutations underscore how frequently the inositol lipid pathway is the target of carcinogenic mutations. While the contribution of these mutations to protein function or prognosis is yet unknown, they will likely be the subject of exciting HNSCC tumor biology discoveries in the near future. To date, therapeutic strategies targeting the PI3K/Akt pathway in HNSCC remain mostly limited to the preclinical setting (63). Pharmacologic inhibitors being developed include a water-soluble form of wortmannin, an inhibitor of PI3K.
Importantly, mTOR, a downstream target of Akt, can be inhibited with rapamycin, the RAD001(Everolimus), and CCI-779 (Temsirolimus) compounds and is the focus of innovative anti-HNSCC therapeutic strategies that are already being tested in Phase I and II clinical trials (85).

The transforming growth factor beta (TGF-β) superfamily of ligands regulates both embryologic processes and adult tissue homeostasis. These ligands act as tumor suppressors by inhibiting proliferation of epithelial cells, and by sensitizing cells to apoptosis; in that respect, TGF-β effects are opposite to those of EGFR. It is therefore not surprising that TGF-β signaling is also the target of carcinogenic mutations.

The TGF-β superfamily includes TGF-βs (TGF-β 1, 2, and 3), bone morphogenic protein, growth and differentiation factors, nodal, activin (involved in embryogenesis and bone formation), and anti-mullerian hormone (86, 87). TGF-β is secreted as inactive precursors that are proteolytically activated by extracellular proteinases (86, 88). The biologically active cytokine is present in the interstitial compartment; the signal is triggered when TGF-β binds a dimeric Type II TGFβ receptor. TGF-β receptors are single-span transmembrane receptors with serine/threonine kinase activity (Fig. 108.2B). Upon activation, dimeric type II receptors recruit and associate with dimeric Type I receptors, another transmembrane serine/threonine kinase receptor, to form a tetrameric receptor complex (86, 87, 88). Both receptors are then in close proximity, leading to the phosphorylation and activation of the Type I receptor by the Type II receptor. In turn, activated Type I receptor recruits an adaptor protein named SARA (Smad anchor for receptor activation) that mediates transient binding between the Type I receptor and SMADs, a family of transcription factors present in the cytoplasm. The SMAD family of proteins has eight members, SMAD 1-8, which come in three types: receptor-activated SMADS (R-SMADS), co-SMADS, and inhibitory SMADS (I-SMADS) (89). R-SMADS involved in TGF-β signaling are SMAD2 and SMAD3. SARA brings SMAD 2 and 3 in close proximity to the catalytic site of the Type I receptor, which then phosphorylates the R-SMAD C-terminus (87, 88). This posttranslational modification induces a conformational change that allows interaction with a co-SMAD (SMAD4). The newly assembled SMAD:co-SMAD complex translocates to the nucleus where it interacts with other transcription regulators, to control expression of genes involved in apoptosis, and cell cycle repressors including p15 and p21(87). However, TGF-β signaling does not necessarily involve SMADs, and SMADs are not necessarily activated by Type I receptors. Recently, important cross talk between the MAP kinase pathway and SMADs has been revealed: ERK has been shown to phosphorylate SMADS (90). Reciprocally, TGF receptors can regulate important targets in a SMAD-independent manner, including the Map kinase pathway (90) and the Rho family of small GTPases (91), which are “molecular switches” similar to Ras, but are involved in the regulation of actin dynamics, cell migration, and metastasis (51). This represents one way in which TGF-β regulates cell morphology and migration.

The TGF-β pathway is negatively regulated at three levels. First, at the level of the extracellular matrix, the small proteoglycan decorin inhibits TGF-β receptor signaling (92). Decorin has antitumoral activity in vitro. Second, intracellular I-SMADs prevent binding of R-SMADs to the activated Type I receptor, and thus negatively impact SMAD signaling. Third, both the receptor and SMADs can be targeted for proteasome-mediated degradation, regulated by E3 ubiquitin ligases SMURF1 and SMURF2 (93, 94). Interestingly, SMURF1 not only downregulates the TGF-β signaling pathway but also enhances cancer cell migration and invasion by targeting the GTPase Rho A for degradation, thereby reducing stress fibers and enhancing cell protrusions and motility (91). In summary, TGF-β signaling is at the crossroads of cell proliferation, cell death, and migration, and it is involved in many aspects of carcinogenesis.

The role of TGF-β signaling in tumorigenesis is multifaceted. While TGF-β has tumor suppressive effects, cancer cells can circumvent the suppressive component of the pathway and use TGF-β stimulation to their advantage to promote invasion and secretion of mitogens and pro-metastatic cytokines. This explains why TGF-β signaling is inhibited in certain cancers and upregulated in others. HNSCC is no exception. For instance, TGF-β receptor and SMADS are often downregulated in HNSCC (95). The downregulation may occur via loss of heterozygosity of chromosome q18, which contains TGFβRI and SMAD genes, or via DNA methylation-mediated gene repression (88). In addition, inactivating mutations in TGFβRI or II and SMADs have been reported in HNSCC, and may be present in up to 20% of neck metastasis (TGFβRI) (96, 97, 98). However, exosome sequencing of 80 HNSCC primary tumors found no mutations in TGF-β pathway, indicating that inactivating mutations are probably not a frequent event (3). An elegant demonstration that the downregulation of TGF-β signaling participates in HNSCC formation comes from a genetic mouse model in which the loss of TGFβRII, in the presence of an activating Ras mutation, induces head and neck tumors that closely resemble human HNSCC.

As above, TGF-β1 is actually frequently overexpressed in HNSCC, by about five times the normal levels in 40% of HNSCC (99). TGF-β1 is not only upregulated in the tumor but also in the surrounding normal-appearing tissue, suggesting that the tumor microenvironment contributes to tumorigenesis. This hypothesis was confirmed in a
transgenic animal model where the induction of TGF-β1 overexpression in oral mucosa led to mucosal hyperproliferation and angiogenesis (99). This suggests that TGF-βR ligand stimulation may be upregulated in HNSCC and have a profound effect on tumorigenesis and tumor biology. In addition, modulation of TGF-β signaling mediators downstream of the receptor also contributes to HNSCC tumorigenesis. For instance, SMAD2 and 4, but not SMAD3, are frequently downregulated in HNSCC, correlating with a worse prognosis (100, 101). Here again, a genetic animal model in which SMAD4 is deleted demonstrates that loss of SMAD4 interferes with DNA repair pathways and results in the spontaneous development of invasive head and neck tumors (101). In a similar vein, TGF-β inhibitors interfering with SMAD signaling also contribute to tumorigenesis. The overexpression of SMURF2, which enhances cell migration and invasion in vitro, is a marker of bad prognosis in esophageal SCC (102); whether SMURFs have prognostic value in HSNCC has not yet been tested.