Abstract
Recent efforts on developing more direct and effective targets for cancer therapy have revolved around a family of transcription factors known as STATs (signal transducers and activators of transcription). STAT proteins are latent cytoplasmic transcription factors that become activated in response to extracellular signaling proteins. STAT proteins have been convincingly reported to possess oncogenic properties in a plethora of human cancers, including oral and oropharyngeal cancer. Signal transduction pathways mediated by these oncogenic transcription factors and their regulation in oral cancer are the focus of this review.
Introduction
Squamous cell carcinoma (SCC) of the head and neck, including the oral cavity and oropharynx, is the sixth most common cancer worldwide (Vokes et al., 1993). Despite recent advances in surgery, radiation, and chemotherapy, the prognosis for head and neck cancer remains dismal, especially if malignant tumors are not diagnosed during the early stages of the disease. Several cellular and molecular events that underlie the occurrence and progression of head and neck SCC, however, are gradually unfolding, including oncogenic alterations and de-regulations in cell death (Todd et al., 1997; Williams, 2000; Nagler et al., 2002; Loro et al., 2003). In this regard, p53 mutations (Maestro et al., 1992; Ahomadegbe et al., 1995) as well as overexpression of cyclin D1 (Jares et al., 1994; Callender et al., 1994; Gimenez-Conti et al., 1996; Quon et al., 2001) and epidermal growth factor receptor (EGFR) (Ozanne et al., 1986; Maxwell et al., 1989) are common alterations. Moreover, overproduction of EGFR and its ligand, transforming growth factor-α (TGF-α), has been described as an early event in head and neck carcinogenesis (Grandis and Tweardy, 1993; Grandis et al., 1996). Elucidation of such mechanisms responsible for critical tumorigenic abnormalities that influence tumor behavior in head and neck cancer are valuable in the diagnosis and development of more effective treatments.
The STAT (signal transducer and activator of transcription) family of proteins includes 7 members (Stat-1, 2, 3, 4, 5A, 5B, and 6), encoded by distinct genes in mammalian cells (Darnell, 1997). The STAT family members are latent cytoplasmic transcription factors that become activated in response to extracellular signaling proteins, including growth factors, cytokines, hormones, and peptides (Darnell et al., 1994; Darnell, 1997; Bromberg and Darnell, 2000). The transmission of signals involves activation of various tyrosine kinases that phosphorylate STAT proteins. Activated STAT proteins transmit signals by translocating into the nucleus and driving transcription of specific genes through interactions with DNA elements in the promoters of target genes (Darnell, 1997; Bromberg and Darnell, 2000).
The STAT family of transcription factors plays a critical role in regulating physiological responses to stimulation by cytokines and growth factors: STAT knockout mice phenotypically present defects similar to those observed in knockouts of cytokines and their receptors (Darnell et al., 1994; Zhong et al., 1994; Schindler and Darnell, 1995; Darnell, 1997; Takeda and Akira, 2000; Levy and Darnell, 2002). In contrast, persistent STAT activation, in particular Stat3 and Stat5, has been convincingly implicated in oncogenesis (Bowman et al., 2000; Bromberg and Darnell, 2000; Bromberg, 2001, 2002; Levy and Darnell, 2002). This review provides a synopsis of recent advances that elucidate the role of STAT proteins in oral cancer, highlighting the important implications of these molecules in oral carcinogenesis, growth, and survival, and possible applications for cancer therapeutics.
STAT Signaling: Activation and Outcome
Activation of STAT molecules involves their tyrosine phosphorylation by several tyrosine kinases. These kinases are activated as a result of interactions between cytokines or growth factors and their respective receptors, inducing receptor dimerization or oligomerization. Cytokine receptors, devoid of intrinsic kinase activity, require the mediation of Janus kinases (JAKs) to induce STAT receptor docking and activation. Upon binding of a cytokine to its corresponding receptor, associated JAKs undergo auto-phosphorylation and subsequently phosphorylate tyrosine residues within the cytoplasmic tail of receptors. These receptors recruit and serve as docking sites for STAT monomers through their phosphotyrosine residues and SH2 domain, respectively (Darnell, 1997; Bromberg and Darnell, 2000; Levy and Darnell, 2002). In contrast, receptor tyrosine kinases such as EGFR possess intrinsic tyrosine kinase activity and facilitate recruitment and subsequent direct activation of STAT proteins (Darnell, 1997; Bromberg and Darnell, 2000; Levy and Darnell, 2002). In addition to cytokine and growth factor receptors, non-receptor tyrosine kinases such as Src and Abl can, either directly or through association with JAKs, induce STAT phosphorylation (Chaturvedi et al., 1997; Olayioye et al., 1999; Wang et al., 2000; Garcia et al., 2001; Schreiner et al., 2002).
Regardless of the mode of activation, phosphorylated STAT proteins form immediate dimer complexes through reciprocal SH2 phosphotyrosine interactions. STAT dimers undergo an immediate disengagement from the receptor or kinase and subsequently translocate to the nucleus. Once in the nucleus, STAT dimers bind to DNA and induce transcription of specific target genes (Darnell, 1997; Levy and Darnell, 2002). Interestingly, different STAT complexes interact with different co-activators and induce distinct genes based on their binding affinities for promoter response elements (Seidel et al., 1995; Ehret et al., 2001).
Structural Elements of STAT Molecules
The STAT molecule is structurally characterized by the presence of distinct N-terminal, C-terminal, and DNA binding regions, which are essential for activation and function (Fig. 1).
(a ) N-terminal region
The N-terminal region consists of an oligomerization domain, which stabilizes the association of two adjacent DNA-bound STAT dimers (Vinkemeier et al., 1996). The N-terminal region is also involved in nuclear translocation and STAT de-activation, thus exerting transcriptional control (Shuai et al., 1996b; Strehlow and Schindler, 1998).
(b ) DNA-binding region
The DNA-binding domain of STATs is an essential element for function as well as regulation. The centrally positioned DNA-binding domain spans the region separating the C- and N-terminal domains and enables STAT to interact with specific response elements within the promoter sequence of target genes (Darnell, 1997). Moreover, a linker domain of unknown function, spanning conserved residues between the DNA-binding and SH2 domains, may be implicated in DNA-binding stability and transcriptosome assembly (Yang et al., 2002).
(c ) C-terminal region
Within the C-terminal region resides a well-conserved Src homology 2 (SH2) domain. This structural motif is critical for STAT interaction with specific phosphotyrosine residues of other proteins (Gupta et al., 1996). This interaction facilitates tyrosine phosphorylation of a STAT molecule at a tyrosine residue (~ 701) by upstream kinases, followed by dimerization through reciprocal SH2 phosphotyrosine binding with another STAT molecule. Preceded by the SH2 domain, the transactivation domain (TAD) lies within the extreme C-terminal region of the STAT molecule and contributes to functional specificity and transcriptional activation of target genes (Kim and Baumann, 1997; Moriggl et al., 1997; Hoey and Schindler, 1998; Leonard and O’Shea, 1998). The latter function is mediated through interaction of the C-terminal TAD of STAT proteins with other transcriptional co-activators and mediators such as p300/CBP, c-Jun, and histone acetyltransferases (HATs) (Levy and Darnell, 2002). In some STATs, particularly Stat1 and Stat3, a critical serine residue 727 (Ser727), within TAD, enhances the DNA-binding affinity and transcriptional activity (Wen et al., 1995; Bromberg et al., 1998; Turkson et al., 1999).
Negative Regulation of STAT Signaling
Several classes of negative regulators of STAT proteins have been discovered.
(a ) Cytoplasmic inhibitors of STATs
Suppressors of cytokine signaling (SOCS) proteins are transcriptional targets of STATs and function in a classic negative feedback loop to inhibit further STAT activation by interacting with either JAK catalytic or receptor sites (Starr and Hilton, 1998, 1999; Kile and Alexander, 2001; Krebs and Hilton, 2001; Levy and Darnell, 2002). Cytoplamic tyrosine phosphatases, including SH2-containing protein tyrosine phosphatase-1 (SHP-1) and SHP-2, also inhibit STATs by dephosphorylating upstream receptors and associated JAK kinases (David et al., 1995; Starr and Hilton, 1999; You et al., 1999; Aoki and Matsuda, 2000; Kile et al., 2001; Qu, 2002).
(b ) Nuclear inhibitors of STATs
The protein inhibitors of activated STATs (PIAS) are nuclear inhibitors of activated STAT proteins that function to repress DNA binding and dimer-dependent transcription (Chung et al., 1997a; Shuai, 2000). At least two members, PIAS1 and PIAS3, have been shown specifically to inhibit Stat1 and Stat3, respectively (Chung et al., 1997a; Shuai, 2000). However, these proteins are also involved in STAT independent processes, and their specificity to STATs remains controversial. In addition to PIAS, naturally occurring truncated STATs, lacking the transactivation domain TAD, can act as non-functional STATs and obstruct DNA binding of functional STAT proteins (Fig. 1) (Wen et al., 1995; Caldenhoven et al., 1996). Finally, nuclear tyrosine phosphatases that function to dephosphorylate active nuclear STATs have also been recently identified (Shuai, 2000; ten Hoeve et al., 2002).
Regulation and Crosstalk in STAT Signaling
(a ) Interactions with MAPKs
Other signal transduction pathways within the cell have been shown to regulate STAT proteins, through indirect kinase-mediated cross-interactions. Among these kinases, mitogen-activated protein kinases (MAPKs) have been implicated in the regulation of STAT proteins through crosstalk signaling (Decker and Kovarik, 2000). ERK-dependent repression of Src- or Jak2-mediated Stat3 signaling involves inhibition of tyrosine phosphorylation, DNA binding, and transcriptional activity (Jain et al., 1998). In this regard, ERK2 has been shown to interact physically with Stat3, involving Ser727 of the Stat3 molecule (Jain et al., 1998). Similar to Stat3, it has been shown that Stat5a contains a putative ERK phosphorylation site and also physically interacts with ERK1/2 (Pircher et al., 1999). Moreover, the inhibitory effect of ERK signaling on IL-6-mediated Stat3 activation has been shown to involve repression of the upstream Jak1 and Jak2 (Sengupta et al., 1998). In addition to ERKs, various inflammatory and stress agents are able to inhibit STAT signaling via a p38 MAPK-mediated mechanism, possibly involving repression of Jak1 (Ahmed and Ivashkiv, 2000) or induction of SOCS3 expression (Bode et al., 1999, 2001). Moreover, activation of the JNK family of MAPK by various stresses or by its upstream kinase MMK7 has also been shown to repress EGF-mediated Stat3 activation (Lim and Cao, 1999).
(b ) Crosstalk with other cellular proteins
In addition to crosstalk with MAPKs, STAT proteins are also subject to regulation by other cellular proteins, including protein kinase Cδ (PKCδ), glucocorticoids, and protein kinase R (PKR) (Decker and Kovarik, 2000). Moreover, the inhibition of Stat3 by a novel inhibitor GRIM-19 (gene associated with retinoid-IFN-induced mortality 19) (Lufei et al., 2003; Zhang et al., 2003) has been documented. The interaction between Grim-19 and Stat3 has also been shown to involve the Stat3 Ser727 residue (Zhang et al., 2003).
(c ) Serine phosphorylation and its potential significance
Serine phosphorylation of STAT proteins has been suggested to play an important role in STAT function (Decker and Kovarik, 2000). In general, STAT serine (in addition to tyrosine) phosphorylation occurs in response to growth factor and cytokine stimulation (Wen et al., 1995; Wen and Darnell, 1997). This activation is mediated by either MAPK or PKCδ signaling in a pathway-dependent manner (Decker and Kovarik, 2000). However, the function of STAT serine phosphorylation in the regulation of STATs remains elusive, in that serine phosphorylation enhances maximum transcriptional activity of STAT proteins (especially Stat1 and Stat3) (Wen et al., 1995; Darnell, 1997; Wen and Darnell, 1997), while causing decreases in STAT tyrosine phosphorylation (Chung et al., 1997b). With respect to the former, overexpression of a phosphoserine mutant of Stat3 was shown to abolish cell transformation by v-Src (Bromberg et al., 1998). The precise role of serine phosphorylation of STAT proteins is further complicated by the findings that MAPKs, which can exert negative regulation on STAT signaling, also induce serine phosphorylation of STAT proteins (Decker and Kovarik, 2000). In light of the fact that repression of Stat3 tyrosine phosphorylation by MEK/ERK signaling can occur independent of Ser727 phosphorylation (Sengupta et al., 1998), the exact role and functional significance of STAT serine phosphorylation necessitates further elucidation.
STATs and Tumorigenesis
(a ) Oncogenic role of STATs
Several lines of evidence support a significant role of STAT molecules, especially Stat3 and Stat5, in carcinogenesis. The ability of STAT molecules to act as oncogenes was established through experiments with a constitutively active mutant of Stat3 (Stat3C), which was capable of inducing malignant transformation in fibroblasts (Bromberg et al., 1999). In addition, dominant-negative STAT mutants induce apoptosis and growth arrest in cancer cells (Grandis et al., 1998; Garcia et al., 2001), suggesting a fundamental role for constitutively active STAT molecules in malignancy. However, it should be noted that not all STATs promote survival. In fact, there is compelling evidence that Stat1 induces cell-cycle arrest and promotes apoptosis through activation of the Cdk inhibitor, p21 (Chin et al., 1996; Sahni et al., 1999) and caspase 1 (Chin et al., 1997). Some investigators have proposed that Stat1 may function as a tumor suppressor, in that Stat1-/- mice potentiate tumor development (Kaplan et al., 1998; Bromberg, 2001). In contrast, over-activation of Stat3 and Stat5 signaling induces specific target genes which prevent apoptosis and stimulate cell proliferation (Bowman et al., 2000; Bromberg, 2001, 2002).
(b ) Aberrations in STAT regulators
With the exception of Stat5 chromosomal translocation in a subset of acute promyelocytic-like leukemias (Ahmed and Ivashkiv, 2000), mutations in STAT proteins that result in aberrant activation have not been reported in cancer. Thus, STAT oncogenic signaling has been linked to aberrations in STAT cellular regulators, including genetic alterations in activators and repressors of STAT signaling. Indeed, several viral as well as cellular oncogenes, including various oncogenic tyrosine kinases, induce transformation selectively by activating STAT proteins (Migone et al., 1995; Yu et al., 1995; Garcia et al., 1997; Lund et al., 1997; Bromberg et al., 1998; Zong et al., 1998; Wen et al., 1999; Bowman et al., 2000; Bromberg, 2001, 2002). The JAK family of kinases has been shown to drive the neoplastic transformation by oncogenic STAT proteins (Schwaller et al., 1998; Bowman et al., 2000). In addition, oncogenic transformation by Src has been consistently linked to STATs (Yu et al., 1995; Cao et al., 1996; Chaturvedi et al., 1997; Garcia et al., 1997; Bromberg et al., 1998; Bowman et al., 2000), and transformation by the Src oncoprotein has been shown to require Stat3 activation (Bromberg et al., 1998; Turkson et al., 1998; Zhang et al., 2000). Interestingly, Jak1 was shown to be required for maximal Stat3 activation by Src kinases in mammalian cells (Zhang et al., 2000). Moreover, v-Abl and BCR-Abl oncoproteins lead to STAT constitutive activation through JAK-dependent and -independent mechanisms (Shuai et al., 1996a; Danial et al., 1998; Danial and Rothman, 2000).
Genetic alterations in negative regulators of STAT proteins have also been reported in several cancer models. Interestingly, SOCS1 gene hypermethylation has been reported in several types of cancer, including hepatocellular carcinomas (Yoshikawa et al., 2001), multiple myeloma (Galm et al., 2003), hepatoblastoma (Nagai et al., 2003), leukemia (Chen et al., 2003), and pancreatic ductal neoplasms (Fukushima et al., 2003), suggesting an important role for SOCS1 in tumor progression. The methylation-induced silencing of SOCS1, leading to constitutive Stat3 activation in hepatocellular carcinoma cell lines, provides a possible mechanism for aberrant activation of STATs in these cells (Yoshikawa et al., 2001). Finally, PIAS3 expression is also repressed in anaplastic large cell lymphoma, possibly contributing to constitutive Stat3 levels (Zhang et al., 2002).
STAT Molecules in Tumor Growth and Survival
(a ) Cell death regulation
Regulation of apoptosis by modulating cell death regulatory proteins is an integral function of oncogenic STAT signaling (Bowman et al., 2000; Bromberg, 2001, 2002). Stat3 and Stat5 especially possess potent anti-apoptotic properties. Constitutive Stat3 signaling has been shown to be essential for the survival of myeloma cells (Catlett-Falcone et al., 1999). Moreover, v-Src-mediated constitutive activation of Stat3 up-regulates bcl-X gene expression in transfected NIH3T3 cells (Catlett-Falcone et al., 1999). Stat3 regulates transcription of the Bcl-X gene, most likely through multiple Stat3 binding elements found within the Bcl-X promoter (Seidel et al., 1995). In head and neck SCC, Stat3 antisense gene therapy leads to decreased Bcl-XL protein expression in xenograft models, revealing Stat3-mediated Bcl-XL anti-apoptotic functions in vivo (Grandis et al., 2000a; Song and Grandis, 2000). Stat5 activation also maintains cell survival by inducing Bcl-XL expression (Lord et al., 2000), since Stat5a and Stat5b double-knockouts result in abrogation of Bcl-XL expression (Dumon et al., 1999; Silva et al., 1999; Socolovsky et al., 1999). In addition to Bcl-XL induction, up-regulation of the apoptosis inhibitor Mcl-1 expression has also been shown to mediate the anti-apoptotic actions of Stat3 and Stat5 (Epling-Burnette et al., 2001; Huang et al., 2002).
(b ) Cell -cycle regulation
In addition to an anti-apoptotic response, oncogenic STAT properties also involve regulation of cell-cycle control genes (Bowman et al., 2000; Bromberg, 2001, 2002). Several cell-cycle regulatory genes—including c-Myc, cyclin D1/D2, p21, and p27—have been linked to the proliferative actions of either Stat3 or Stat5 (Bowman et al., 2000; Bromberg, 2001, 2002). In transformed fibroblasts, abrogation of Stat3 signaling by the dominant-negative Stat3β protein inhibits v-Src-induced and platelet-derived growth-factor-mediated malignant transformation through repression of c-Myc expression (Bowman et al., 2001). Similarly, v-Src transformed fibroblasts exhibit up-regulation of p21, cyclin D1, and cyclin E in a Stat3-dependent manner (Sinibaldi et al., 2000). Importantly, overexpression of cyclin D1 in head and neck SCC has been linked to aberrant activation of Stat3 (Masuda et al., 2002). Expression of cyclin D2 (Martino et al., 2001) and c-Myc (Lord et al., 2000) has also been linked to Stat5-mediated cell-cycle control (Huang et al., 2002).
STAT Proteins in Specific Cancer Types
STAT constitutive activation has been detected in a plethora of primary tumors and tumor cell lines (Bowman et al., 2000; Bromberg, 2001, 2002), including breast cancer (Watson and Miller, 1995; Garcia et al., 1997; Sartor et al., 1997; Bromberg, 2000), head and neck cancer (Grandis et al., 2000a; Song and Grandis, 2000), non-small-cell lung cancer (Fernandes et al., 1999), prostate cancer (Ni et al., 2000, 2002; Dhir et al., 2002; Mora et al., 2002), melanoma (Florenes et al., 1999; Kirkwood et al., 1999), leukemia (Gouilleux-Gruart et al., 1996; Shuai et al., 1996a; Schuringa et al., 2000; Benekli et al., 2002, 2003), and EBV-transformed cells (Weber-Nordt et al., 1996). In breast cancer, constitutive activation of Stat1 and Stat3 has been identified in cell lines (Garcia et al., 1997, 2001) and primary tumor nuclear extracts (Garcia et al., 2001), but not in mammary epithelial cells (Garcia et al., 2001) or healthy breast tissues (Garcia et al., 2001). This activation has been linked to EGFR and Src overexpression and overactivation (Shuai et al., 1996a; Garcia et al., 2001). In non-small-cell lung cancer, constitutive Stat3 activation has been linked to an autocrine ErbB-1/TGF-alpha loop mediated by ErbB-2 (HER2/neu) receptor tyrosine kinase (Fernandes et al., 1999) and to tumor survival (Song et al., 2003). Activation of Stat3 in prostate cancer (Dhir et al., 2002; Ni et al., 2002) has been shown to potentiate malignant phenotypes and induce cell growth in prostate cancer cells (Ni et al., 2000).
STAT in Head and Neck Cancer
In head and neck SCC, there is compelling evidence that STAT constitutive activation is linked to cancer development and growth (Grandis et al., 2000a; Song and Grandis, 2000; Kijima et al., 2002).
(a
) In vitro
studies
In squamous epithelial cells, there is evidence that overactivation of EGFR and its ligand (TGF-α) results in constitutive activation of both Stat1 and Stat3 (Song and Grandis, 2000). Accordingly, SIF-A (Stat3 homodimer) and SIF-B (Stat3/Stat1 heterodimer) are the predominant protein-DNA complexes formed in response to treatment of head and neck SCC cell lines with TGF-α (Grandis et al., 2000a). The resulting activation of Stat1 and Stat3 molecules, however, is associated with different signaling events: Antisense targeting of Stat3, but not Stat1, caused cell growth inhibition in vitro, revealing the critical role of Stat3 in head and neck SCC growth (Grandis et al., 1998). Moreover, transfection of dominant-negative constructs of Stat3 or antisense oligonucleotide targeting of the Stat3 translational start site resulted in significant growth inhibition in vitro (Grandis et al., 1998, 2000a). Interestingly, similar to Stat3, constitutive activation of Stat5b, but not Stat5a, also contributes to head and neck SCC growth (Leong et al., 2002).
(b ) In vivo
studies
Stat3 antisense gene therapy delivered to head and neck SCC xenograft models also leads to increased tumor apoptosis in vivo (Grandis et al., 2000a; Song and Grandis, 2000). In tumors and adjacent normal mucosa from head and neck SCC patients, Stat3 protein expression, tyrosine phosphorylation, and DNA-binding activity are constitutively amplified compared with levels in the normal mucosa from healthy individuals (Grandis et al., 2000a). Interestingly, tissue distribution of active Stat3 in normal mucosa differs significantly between healthy individuals and patients with head and neck cancer. While phosphorylated Stat3 is expressed in only basal epithelial layers of normal mucosa from healthy individuals, phosphorylated Stat3 is detected throughout the entire epithelium in normal mucosa from head and neck cancer patients, implicating a role for Stat3 activation in early carcinogenesis (Grandis et al., 2000a). In addition, Stat1 and Stat3 expression are inversely correlated with respect to tumor differentiation in head and neck SCC. While Stat1 is detected to a higher degree in well-differentiated tumors, Stat3 is up-regulated in poorly differentiated tumors (Arany et al., 2003).
Regulators of Oncogenic STAT Signaling in Head and Neck Cancer
Deregulation of various upstream pathways involving overexpression or overactivity of specific cytokines, growth factors and their respective receptors has been implicated in the constitutive activation of STAT proteins (Bowman et al., 2000; Bromberg, 2001, 2002).
(a ) TGF-α and EGFR
In head and neck cancer, there is significant evidence suggesting that aberrant TGF-α/EGFR signaling leads to constitutive activation of Stat3 (Grandis et al., 1998, 2000a; Song and Grandis, 2000; Leong et al., 2002). EGFR has been shown to interact directly with Stat-1, -3, -5 in A431 cells (Olayioye et al., 1999). Moreover, targeting of TGF-α and EGFR with either TGF-α neutralizing antibodies or EGFR-specific tyrosine kinase inhibitors abrogates Stat3 activation in vitro (Grandis et al., 1998). Furthermore, repression of Stat3 constitutive activity, with an antisense gene therapy approach targeting the EGFR gene, has been documented in vivo (Grandis et al., 2000b; Song and Grandis, 2000). Similarly, Stat3 targeting indicates the critical role of Stat3 in TGF-α/EGFR-mediated growth in head and neck SCC cells (Grandis et al., 1998, 2000b).
(b ) Src kinases
Recent evidence has unveiled the role of Src kinases in EGFR-mediated STAT activation (Xi et al., 2003). Activation of Src kinases correlates with constitutive activation of Stat3 and Stat5 in head and neck SCC. In addition, the direct interaction among EGFR, Src, and STAT proteins suggests that Src kinases contribute to oncogenic STAT signaling by facilitating STAT interactions with EGFR (Xi et al., 2003). The importance of Src kinases in constitutive STAT activation is not exclusive to head and neck cancer, in that recent reports have indicated an essential role of Src in transformed cells and breast cancer cells (Bowman et al., 2001; Garcia et al., 2001).
(c ) Cytokines and JAKs
Despite the strong association between aberrant TGF-α/EGFR signaling and activation of STAT proteins in head and neck cancer, the requirement of EGFR stimulation for mediating constitutive Stat3 activation remains unclear. The ability of constitutively active Stat3 to serve as a growth-promoting signal independent of EGFR signaling supports the existence of other pathways contributing to oncogenic Stat3 signaling in head and neck cancer (Kijima et al., 2002). Consistent with this view, autocrine/paracrine stimulation of IL-6-mediated gp130/JAK signaling has been linked to EGFR-independent constitutive activation of Stat3 in head and neck SCC cells (Sriuranpong et al., 2003). Head and neck SCC cells express high levels of several pro-inflammatory and pro-angiogenic cytokines, including IL-1α, IL-6, IL-8, granulocyte-macrophage colony-stimulating factor, and VEGF in vitro and in vivo, which may also play a possible role in autocrine and/or paracrine signaling leading to oncogenic STAT activation (Chen et al., 1998, 1999; Ondrey et al., 1999). Importantly, targeting the gp130 receptor or inhibiting JAK kinases abrogates constitutive Stat3 activation and inhibits cell growth in head and neck cancer (Sriuranpong et al., 2003). However, the role of JAK kinases in oncogenic STAT signaling remains controversial, since JAK kinases were shown not to be required for STAT activation and cell growth in head and neck SCC (Xi et al., 2003). Such discrepancies may be the result of differences in levels of JAK kinase inhibition.
(d ) Other regulators
In addition to signaling effects by cytokines and growth factors, ligand-independent activation of Stat3 linked to inhibition of cyclin-dependent kinase-2 has also been described, implying that cell-cycle inhibitors may contribute to the regulation of persistent Stat3 activation in head and neck SCC (Steinman et al., 2003). Finally, aberrations in negative regulators of STAT proteins may also result in constitutive STAT activation. In this regard, we recently observed that SOCS1 expression may be subject to down-regulation in oral SCC cells. Constitutive activation of STAT proteins in head and neck cancer most likely involves multiple upstream regulatory elements (depicted in Fig. 2), which may differ among different cell lines, resulting in variations in tumor phenotypes with respect to constitutive activation of STAT proteins.
Targeting of STATs
STAT latent transcription factors, along with several other transcription factors, are critical for mediating deregulation of oncogenic pathways in cancer (Darnell, 2002). The central role of STAT proteins (in particular, Stat3 and Stat5) in malignant transformation and cancer progression presents STATs as a potentially effective target for cancer therapy. In light of that, several strategies for the design of specific inhibitors that hamper STAT signaling have been developed (Turkson and Jove, 2000; Buettner et al., 2002).
(a ) Inhibition of upstream kinases
Because tyrosine phosphorylation is critical for STAT activation, kinase inhibitors have become the focus of intensive investigation. Inhibition of specific upstream tyrosine kinases, including JAKs, Src, and EGFR, has been effective in inhibiting oncogenic STAT signaling and growth in myeloma, breast, and prostate cancer cells (Fry et al., 1994; Catlett-Falcone et al., 1999; Ni et al., 2000; Turkson and Jove, 2000; Garcia et al., 2001; Buettner et al., 2002; Mora et al., 2002). Recent findings indicate that a selective JAK inhibitor, AG490, represses Stat3 activation and cell growth in head and neck SCC (Sriuranpong et al., 2003). Similarly, our recent findings suggest that targeting of JAK kinase in oral SCC effectively reduces cell growth and inhibits Stat3 signaling. In addition to JAK inhibition, inhibitors of Src kinases also abrogate Stat3 and Stat5 activation and induce cell growth inhibition in head and neck SCC (Xi et al., 2003). Finally, targeting of EGFR has a potent growth-inhibitory effect in head and neck cancer (Ford and Grandis, 2003). Consistent with this view, EGFR antisense therapy inhibits constitutive Stat3 signaling and induces cell growth inhibition (Grandis et al., 2000b). Interestingly, delivery of phosphopeptides, spanning critical tyrosine residues within the EGFR C-terminal domain, to head and neck SCC cells disrupts EGFR-mediated Stat3 DNA-binding activity and cell growth (Shao et al., 2003). Similarly, EGFR binding peptide aptamers inhibit EGFR-mediated Stat3 activation in A431 and breast cancer cells (Buerger et al., 2003).
(b ) Direct targeting
Direct targeting of STAT proteins provides the most specific therapeutic approach. Targeting of Stat3, through transfection of dominant-negative constructs or application of antisense oligonucleotide treatment, results in growth arrest and apoptosis in head and neck SCC (Grandis et al., 1998), breast cancer (Li and Shaw, 2002), prostate cancer (Mora et al., 2002), non-small-cell carcinoma (Song et al., 2003), and leukemia cells (Nakajima et al., 1996; Epling-Burnette et al., 2001). Furthermore, Stat3 antisense gene therapy in head and neck xenograft models leads to increased tumor apoptosis in vivo (Grandis et al., 2000a; Song and Grandis, 2000; Kijima et al., 2002). Importantly, a decoy oligonucleotide corresponding to the Stat3 response element induces cell growth inhibition in head and neck SCC without affecting the proliferation of normal oral keratinocytes (Leong et al., 2003), highlighting the dependence of head and neck SCC cells on persistent Stat3 activity. Finally, the development of short peptides that target Stat3 dimerization and inhibit DNA-binding suppresses gene regulation and transformation mediated by oncogenic Stat3 in vivo (Turkson et al., 2001).
(c ) Alternative strategies
Alternatively, a combination of retinoic acid (RA) and IFN signaling suggests a potentially useful strategy in the treatment of cancer including head and neck cancer (Toma et al., 1994; Lingen et al., 1998; Gonçalves et al., 2001). Importantly, the up-regulation of Grim-19, shown to be critical for IFN/RA-induced cell death, leads to inhibition of Stat3 proteins in breast cancer cells (Zhang et al., 2003). In view of the fact that targeting of Stat3 severely hinders tumor survival in head and neck cancer, we have recently reported that a non-steroidal anti-inflammatory drug, sulindac, abrogates Stat3 signaling in oral cancer cells (Nikitakis et al., 2002a). Similarly, our findings suggest that cyclopentenone prostaglandins, especially 15-deoxy-Δ12,14-PGJ2, also target Stat3 signaling and induce apoptosis in oral cancer cells (Nikitakis et al., 2002b). Elucidation of mechanisms of STAT inhibition by novel inhibitors may help us identify other critical regulators of STAT proteins in head and neck cancer.
Conclusions and Perspectives
Compelling evidence has accumulated revealing a critical role of STAT proteins in carcinogenesis. Although constitutive activation of STAT proteins is not the only contributor to transformation and cancer progression, its crucial role has been clearly demonstrated in head and neck cancer. The mechanisms responsible for aberrant STAT activation in head and neck cancer remain uncertain and need further exploration. Moreover, knowledge of the cross-interaction of STAT molecules with other critical cellular proteins involved in growth regulation and survival may better serve to explain head and neck carcinogenesis. Importantly, the association between transcriptional co-activators and co-repressors with STAT molecules in regulating transcription of target genes will prove valuable in our understanding of the molecular basis of head and neck cancer.
Malignancies in which STAT activation is linked to tumor progression and survival are preferentially susceptible to STAT-targeted therapy. The observed dependence of such malignancies, but not normal cells, on aberrant STAT activation for growth and survival has wide implications for cancer therapeutics (Turkson and Jove, 2000; Buettner et al., 2002). That STAT activation is transient in normal physiological conditions may also help explain the enhanced sensitivity to cell death for cancer cells compared with normal cells. However, the consequence of targeting STAT molecules in therapeutics requires more vigorous investigation, especially given the importance of STAT proteins in normal cellular processes. More challenging is to elucidate selective inhibition of transcriptional regulation for STAT-regulated genes related to tumor growth and survival. Elucidation of mechanisms that further define the dependence of head and neck cancer cells on STAT proteins will also provide important insights for the development of more effective therapeutic strategies in head and neck cancer.
Functional domains of STAT proteins. Model of persistent Stat3 activation and possible modes of regulation in oral squamous carcinoma cells. Mechanisms involving aberrant Stat3 activation may include overexpression of receptors, kinases, and ligands, and de-regulation of positive or negative regulators. Overexpression and overactivation of EGFR and its ligand (TGF-α) result in aberrant activation of Stat3, possibly involving constitutive activity of Src kinases. Production of IL-6 and overactivation of gp130 and associated JAK kinases can lead to persistent Stat3 activity. Constitutive Stat3 signaling leads to activation of downstream genes, including anti-apoptotic protein Bcl-XL and cell cycle regulator Cyclin D1. Possible negative regulators of activated Stat3 are also depicted, including Stat3 inducible suppressors of cytokine signaling (SOCS) and protein inhibitors of activated Stat3 (PIAS3). Activation of MAPKs (ERK, JNK, and p38) by MAPK kinases (MEK1/2, MEK4/7, MEK3/6) may also regulate constitutive Stat3 activity through cross-interactions.

Footnotes
Acknowledgements
This work was supported by grants from PHS/NIH DE13118, DE12606, and DE014935.
