Hepatocyte Growth Factor Inhibits Anoikis by Induction of Activator Protein 1-dependent Cyclooxygenase-2

Anoikis, also called suspension-induced apoptosis, plays an important role in tumor development, progression, and metastasis. Recently we found that hepatocyte growth factor (HGF) inhibited anoikis of human head and neck squamous cell carcinoma (HNSCC) cells by activating the extracellular signal-regulated kinase (ERK)-signaling pathway. However, the anti-apoptotic effectors that were regulated by the ERK-signaling pathway were unknown. Here we report that HGF-mediated inhibition of anoikis was dependent on activator protein-1 activity through the activation of the ERK-signaling pathway. Using a combination of microarray analysis and Northern blot analysis, we found that an anti-apoptotic gene cyclooxygenase-2 (cox-2) was induced by HGF in an activator protein-1-dependent fashion. Inhibition of Cox-2 activity partially abolished HGF-mediated cell survival, and overexpression of Cox-2 in HNSCC cells provided resistance against anoikis. Moreover, HNSCC cells stably expressing Cox-2 had aggressive tumor growth in a nude mouse model compared with control cells. Taken together, our results demonstrate that Cox-2 plays an important role in HGF-mediated anoikis resistance. HGF may stimulate the progression and growth of HNSCC in vivo by induction of Cox-2.

HGF, 1 also known as a scatter factor, is a mesenchymalderived cytokine that stimulates cell proliferation, migration, survival, and invasion (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12). These diverse biological functions of HGF are mediated through its high affinity tyrosine kinase receptor, c-met proto-oncogene (8). Importantly, a number of studies demonstrate that abnormal expression of HGF and/or c-Met is associated with tumor progression and metas-tasis. c-Met/HGF is found to be overexpressed in renal, breast, and prostate cancer (4,5,11). Recently, epidemiological and clinical investigations have suggested that c-Met/HGF also plays a critical role in the development and metastasis of head and neck squamous cell carcinoma (13)(14)(15)(16)(17). SCC is the most common epithelial tumor occurring in the oral cavity, head, and neck. It represents the sixth most common cancer in the developed world and is often associated with low survival and high morbidity rates (18 -23). c-Met was highly expressed in lymph node metastases of HNSCC compared with primary tumors. Serum HGF in head and neck cancer patients was significantly increased compared with healthy control subjects (13)(14)(15)(16)(17). Consistent with these clinical findings, Dong et al. (14) demonstrate that HGF induced expression of angiogenic factors interleukin-8 and vascular endothelial growth factor in HNSCC cells, suggesting that HGF promotes tumor development by induction of angiogenesis (14). Matsumoto et al. (24) find that HGF induced tyrosine phosphorylation of focal adhesion kinase and promoted migration and invasion by oral SCC cells.
In addition to stimulating cell invasion and angiogenesis, HGF is also a potent anti-apoptotic factor (3,4,25). Rosen and co-workers (25) find that HGF protected various epithelial and carcinoma cell types against apoptosis induced by DNA-damaging agents, including the chemotherapeutic drug adriamycin, X-rays, and UV light. HGF helped to maintain the level of the survival-promoting protein Bcl-X L (25). Recently, they reported that HGF functioned upstream of the mitochondria to block mitochondrial apoptosis signaling and to prevent activation of multiple caspases (3). To explore the molecular mechanism by which HGF mediated tumor invasion and metastasis of HNSCC, we recently established an anoikis model system (26). Anoikis is a term used to describe apoptosis of epithelial cells induced by a loss of matrix attachment. Because tumor cells lose matrix attachment during metastasis, conceptually anoikis resistance plays an important role in tumor progression and metastasis (27)(28)(29)(30). We found that HGF strongly suppressed anoikis in HNSCC cells, which provided a new explanation for growth and metastasis of HNSCC mediated by HGF (26). Furthermore, we identified that HGF-induced anoikis resistance was dependent on both ERK-and Akt-signaling pathways. Inhibition of either ERK or Akt activity with a specific chemical inhibitor was sufficient to abolish HGF-mediated anoikis resistance (26).
In the current study, we further examined the anti-apoptotic effectors, regulated by the ERK-signaling pathway, to determine which were responsible for HGF-mediated anoikis resistance. The ERK-stimulated AP-1 activity was found to be important in HGF-mediated survival. Inhibition of AP-1 activity with the dominant negative mutant of c-Jun abolished HGF-mediated survival function. Furthermore, using microarray analysis, we found that the anti-apoptotic gene Cox-2 was induced by HGF in an AP-1-dependent manner. Inhibition of Cox-2 activity with a specific chemical inhibitor NS398 attenuated HGF-mediated survival function. To extend our findings in vivo, we xenografted HNSCC cells stably expressing Cox-2 into nude mice. We found that, compared with control cells, HNSCC cells expressing Cox-2 had aggressive tumor growth in vivo. These results provide new insights into the molecular mechanism by which HGF promotes growth and progression of HNSCC.

EXPERIMENTAL APPROACHES
Cell Culture and Retrovirus Transduction-Human HNSCC cell lines UMSCC1 and other UMSCC cells were derived from SCC of the head, neck, and oral cavity at the University of Michigan (Ann Arbor, MI). UMSCC1 cells, c-fosϪ/Ϫ and c-fosϩ/ϩ mouse neonatal fibroblasts (MFs; Ref. 31) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen), penicillin (100 units/ml), and streptomycin (100 g/ml). A chemical inhibitor used in this study for MEK1/2 (U0126) was purchased from Promega. The cDNA encoding TAM67 was subcloned into the retroviral vector pBabepuro. Retroviral vector encoding the full-length Cox-2 cDNA was constructed as described previously (29). Retroviruses were generated by transfecting the retroviral plasmids encoding TAM67 or Cox-2 into 293T cells by the calcium phosphate method. Retrovirus-containing supernatants were harvested 48 h after transfection and stored at Ϫ70°C. To stably express TAM67 or Cox-2, cells were infected with retroviruses in the presence of 6 g/ml Polybrene. For control, cells were infected with retroviruses expressing empty vector. Forty-eight hours after infection, cells were selected with puromycin (1.5 g/ml) for 1 week. The resistant clones were pooled, and cells expressing Cox-2 or TAM67 were confirmed by the Western blot analysis.
Anoikis Induction and Trypan Blue Exclusion Assay-To induce anoikis, cells were plated on 0.6% soft agar with or without HGF (40 ng/ml) (R&D Systems) in the presence of growth medium, as described previously (26). Forty-eight to seventy-two hours after cell suspension, cells were harvested, and any cell aggregates were dispersed by trypsinization. Cell viability was determined by the trypan blue exclusion analysis.
Western Blot Analysis-Cells were harvested, washed with ice-cold phosphate-buffered saline, and pelleted. Whole cell lysates were prepared with lysis buffer containing 1% Nonidet P-40, 5% sodium deoxycholate, 1 mM phenylmethylsulfonyl fluoride, 100 mM sodium orthovanadate, and 1:100 protease inhibitors mixture (Sigma). Protein concentrations were determined using the Bradford protein assay (Bio-Rad). Protein extracts were subjected to 10 -15% SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membrane by electroblotting (Bio-Rad). The membranes were blocked with 5% nonfat dry milk, 1ϫ TBST (25 mM Tris-HCl, 125 mM NaCl, 0.1% Tween 20) overnight at 4°C and probed with primary antibodies for 1 h and then horseradish peroxidase-conjugated secondary antibodies for 1 h. The immune complexes were visualized using the SuperSignal kit (Pierce) according to the manufacturer's protocol. For internal control, the blots were stripped with Tris buffer (62.5 mM, pH 8.0) containing 100 mM 2-mercaptoethanol and 2% SDS at 60°C for 1 h and re-probed with an anti-␣-tubulin monoclonal antibody. Primary antibodies were purchased from the following commercial sources: polyclonal antibodies against Cox-2 from Santa Cruz; polyclonal antibodies against c-Jun from Calbiochem; monoclonal antibodies against ␣-tubulin (1:7500) from Sigma; and secondary antibodies against rabbit or mouse IgG (1:7500) from Promega.
Microarray Analysis and Northern Blot Analysis-UMSCC1 cells were treated with human HGF (40 ng/ml, R&D Systems) for 1 h. Total RNA was extracted using Trizol reagents (Invitrogen) and further purified using RNeasy columns (Qiagen). Twenty-microgram aliquots of total RNAs were subsequently labeled with biotin as recommended by Affymetrix. cRNA was fragmented and used for hybridization to Affymetrix human U133 genechips. Data were analyzed using Affymetrix Genechip software. To confirm the results from microarray analysis, Northern blot analysis was performed. Briefly, 15-g aliquots of total RNA were separated on a 1.4% agarose-formaldehyde gel, transferred onto a nylon membrane, and cross-linked with a UV cross-linker. The membranes were hybridized overnight with random-primed 32 P-labeled probe at 42°C in 50% formamide, 5ϫ SSC (1ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate), 50 mM Tris-HCl, 0.1% sodium pyrophosphate, 1% SDS, 0.25% polyvinylpyrrolidone, 0.25% Ficoll, 5 mM EDTA, and 150 g of denatured salmon sperm DNA. Probes were made with a random-primed labeling kit (Amersham Biosciences) in the presence of [␣-32 P]dCTP (ICN) and purified with a micro G-50 Sephadex column (Amersham Biosciences). After hybridization, the blots were washed twice in 2ϫ SSC, 0.1% SDS for 10 min at 42°C and twice in 0.1ϫ SSC, 0.1% SDS for 20 min at 42°C. For the internal control, the membrane was stripped and reprobed with the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase.
In Vivo Tumor Growth-To determine tumor growth, both UMSCC1Cox-2 and UMSCC1V cells (2 ϫ 10 6 cells) were injected subcutaneously into the back of 8-week-old male nude mice (Taconic). Tumor growth was monitored every 3-4 days for 40 days (32). These procedures were approved by the University of Michigan Committee on Use and Care of Animals (Ann Arbor, MI).

Anoikis Resistance Induced by HGF Was Dependent on AP-1
Transcription-Recently, we found that HGF inhibited anoikis in HNSCC cells through the activation of the ERK-signaling pathway (26). We were interested in identifying the downstream anti-apoptotic effectors that were regulated by the ERK-signaling pathway. Because the AP-1 transcription factor is a direct target of the ERK-signaling pathway (33-36), we hypothesized that AP-1 activity might play a role in HGFmediated survival. Because the ERK activation can directly stimulate AP-1 activity via induction of c-fos (33), we performed Northern blot analysis to examine whether HGF induced c-fos expression in HNSCC cells. As shown in Fig. 1, c-fos mRNAs were rapidly induced 1 h after HGF stimulation in UMSCC1 cells. Consistent with previous findings, pretreatment with a MEK inhibitor, U0126, significantly inhibited c-fos expression induced by HGF. Additionally, we also found that c-fos mRNA was induced in other HNSCC cell lines (data not shown).
It is well known that c-Fos interacts with c-Jun to form a stable heterodimer that binds to AP-1 sites of the promoter region of target genes to stimulate transcription (33). Next, we sought to determine whether inhibition of AP-1 activity interfered with HGF-mediated cell survival using the dominant negative mutant of c-Jun TAM67. TAM67 is a deletion mutant that lacks the transactivation domain of c-Jun. It retains DNA binding activity but fails to activate transcription (33). UM-SCC1 cells were stably transduced with retroviruses expressing TAM67 or control viruses. Cells were selected with puromycin for 2 weeks, and the resistant clones were pooled. As shown in Fig. 2A, UMSCC1 cells expressing TAM67 (UMSCC1TAM67) and control cells (UMSCC1V) were obtained as confirmed by Western blot analysis. To determine whether HGF-mediated inhibition of anoikis was dependent on AP-1 transcription, both UMSCC1TAM67 and UMSCC1V cells were cultured in suspension and treated with or without HGF for 72 h. As shown in Fig. 2B, similarly to parent cells in our FIG. 1. HGF induces c-fos expression in UMSCC1 cells. UM-SCC1 cells were pretreated with MEK inhibitor U0126 or vehicle control and then treated with HGF (40 ng/ml) for the indicated time periods. Cells were harvested, and total RNA was extracted using Trizol reagents. Ten-microgram aliquots of total RNA were separated on 1.4% agarose-formaldehyde gel and hybridized with 32 P-labeled human c-fos cDNA probe. For the loading control, the membrane was stripped and rehybridized with 32 P-labeled glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe.
previous studies (26), more than 70% of UMSCC1V cells were dead 72 h after deprivation of matrix contact, whereas only 30% of cells were dead in the presence of HGF. In contrast, regardless of HGF treatment, the majority of UMSCC1TAM67 cells were dead, indicating that HGF-mediated survival was dependent on AP-1 transcription.
The role of AP-1 in apoptosis has been controversial (33). In contrast to our findings mentioned above, some studies have reported that AP-1 is involved in induction of apoptosis by stress stimuli. For example, c-Fos was persistently induced in the brains of mice treated with kainic acid, a potent activator of glutamate receptors that induces apoptosis of hippocampal neurons (37). To rule out the nonspecific effect of TAM67, c-fosϪ/Ϫ MFs were utilized. As shown in Fig. 2C, more than 80% of c-fosϩ/ϩ or c-fosϪ/Ϫ MFs were dead 48 h after a loss of matrix contact. However, cell death in c-fosϩ/ϩ MFs was significantly suppressed in the presence of HGF, whereas HGF could not provide protection against anoikis in c-fosϪ/Ϫ cells. Taken together, our genetic and dominant negative approaches suggest that AP-1 played an essential role in anoikis resistance induced by HGF.

HGF Induced the Anti-apoptotic Gene cox-2 to Suppress
Anoikis-Next, we were interested in identifying AP-1-dependent anti-apoptotic genes induced by HGF. Because Bcl-X L has been implicated in HGF-mediated protection against chemotherapy-induced apoptosis (25), we performed Western blot analysis to determine whether HGF modulated the expression of Bcl-2 family proteins. As shown in Fig. 3, the protein level of Bcl-2, Bcl-X L and A1 remained unchanged after HGF stimulation. Additionally, the pro-apoptotic protein Bid was also unmodified by HGF stimulation in HNSCC cells. To further identify anti-apoptotic genes induced by HGF, the microarray analysis was performed. Interestingly, we found that the antiapoptotic gene cox-2 increased 4-fold upon HGF stimulation in UMSCC1 cells (Fig. 4A). To further confirm this finding, Northern blot analysis was performed. As shown in Fig. 4B, Cox-2 mRNA in UMSCC1V cells was rapidly induced 1 h after HGF stimulation. However, HGF could not induce Cox-2 expression in UMSCC1TAM67 cells, indicating that Cox-2 expression was dependent on AP-1 transcription. Additionally, MEK inhibitor U0126 also inhibited Cox-2 expression induced by HGF (data not shown). Consistently, Western blot analysis also demonstrated that Cox-2 expression was induced by HGF in UMSCC1V cells but not in UMSCC1TAM67 cells (Fig. 4C).
Cox-2 has been shown to inhibit apoptosis induced by a variety of stimuli (38). To determine whether Cox-2 was involved in HGF-mediated inhibition of anoikis, the specific Cox-2 inhibitor NS398 was utilized. As shown in Fig. 5A, pretreatment of NS398 significantly attenuated HGF-mediated protection against anoikis. To rule out the nonspecific effect of NS398, we also determined whether overexpression of Cox-2 inhibited anoikis. UMSCC1 cells were stably transduced with retroviruses expressing Cox-2 or control viruses. Cells were subjected to puromycin selection for 1 week, and the stable clones were pooled to prevent clone variation. As shown in Fig.  5B, both UMSCC1Cox-2 and UMSCC1V cell lines were established, as confirmed by Western blot analysis. Subsequently,

FIG. 2. HGF-mediated anoikis resistance is dependent on AP-1 transcription.
A, establishment of UMSCC1 cells stably expressing TAM67. UMSCC1 cells were transduced with retroviruses expressing TAM67 or control vector. Cells were selected with puromycin (1.5 g/ ml) for 1 week, and the resistant clones were pooled. Whole cell extracts were prepared with lysis buffer as described under "Experimental Procedures." Fifty-microgram aliquots of cell extracts were separated on 10% SDS-PAGE gels, and membranes were probed with polyclonal antibodies against c-Jun. For the loading control, the membrane was stripped and reprobed with monoclonal antibody against ␣-tubulin. B, TAM67 abolished HGF-mediated anoikis resistance. Both UMSCC1TAM67 and UMSCC1V cells were cultured in suspension and treated with or without HGF (40 ng/ml) for 72 h. Cell viability was determined with the trypan blue exclusion assay. The assays were performed in duplicate, and the results represent average values from three independent experiments. Statistical differences between each group were determined by the Student's t test. *, p Ͻ 0.01. C, c-Fos was essential for HGF-mediated survival. Both c-fosϪ/Ϫ and c-fosϩ/ϩ MFs were cultured in suspension and treated with or without HGF (40 ng/ml) for 48 h. Cell viability was determined with the trypan blue exclusion assay. The assays were performed in duplicate, and the results represent average values from three independent experiments. Statistical differences between each group were determined by the Student's t test. *, p Ͻ 0.01.

FIG. 3. HGF does not modify the expression of Bcl-2,
Bcl-X L , A1, and Bid. UMSCC1 cells were treated with HGF (40 ng/ml) for the indicated times. The whole cell extracts were prepared with lysis buffer, and 50-g aliquots of protein extracts were separated on a 12% SDS-PAGE gel. The blots were probed with polyclonal antibodies against Bcl-2 (A), A1 (B), Bcl-X L (C), and Bid (D). To assess the equivalency of loading, the blots were stripped and reprobed with monoclonal antibodies against ␣-tubulin (1:7500).
both UMSCC1Cox-2 cells and UMSCC1V cells were deprived of matrix contact for 48 h. As shown in Fig. 5C, more than 70% of UMSCC1V cells were dead after loss of adhesion, whereas only 40% of UMSCC1Cox-2 cells were dead, indicating that Cox-2 indeed promoted anoikis resistance. The addition of NS398 totally abolished Cox-2-mediated survival. However, the protective level of Cox-2 was relatively weaker than that of HGF, indicating that other unidentified anti-apoptotic genes induced by HGF are likely involved in HGF-mediated survival function. As listed in Table I, several other genes, some of which had previously been found to play a role in cell survival (39,40), were induced by HGF according to our microarray analysis. We are determining whether these genes are responsible for HGFmediated inhibition of anoikis.
Cox-2 Promotes HNSCC Growth in Vivo-Cox-2 has been found to be overexpressed in human HNSCC and, thus, has been considered as an important target for chemoprevention of HNSCC (41,42). Recently, Bol et al. (43) generated transgenic mice that overexpressed Cox-2 under control of the human keratin 14 promoter, which allowed for expression in epidermis and some other epithelia. Unexpectedly, they found that Cox-2 overexpression protected, rather than sensitized, mice to skin tumor development induced by tumor promoters. Therefore, whether HGF-induced Cox-2 played a role in tumor develop-ment of HNSCC cells in vivo was determined. UMSCC1Cox2 cells and control UMSCC1V cells were subcutaneously xenografted into nude mice. As shown in Fig. 6A, UMSCC1Cox-2 cells had more aggressive tumor growth in mice than control UMSCC1V cells 40 days after injection. Although both cells formed equivalent numbers of tumors at injection sites, the average size of tumors from UMSCC1Cox-2 cells was three to four times larger than that from UMSCC1V cells (Fig. 6B). These results suggest that HGF-induced Cox-2 plays an important role in the development and progression of HNSCC. DISCUSSION Emerging evidence indicates that c-Met/HGF plays a pivotal role in tumor invasion, progression, and metastasis of HNSCC (13)(14)(15)(16)(17)26). To explore the molecular mechanism by which c-Met/HGF promotes development and progression of HNSCC, we recently established an in vitro anoikis model (26). We identified that the ERK-signaling pathway activated by HGF provided protection against anoikis. In the current study, we extended our previous observation and found that the ERK-dependent AP-1 activity was critical for HGF-mediated survival. Importantly, we found that AP-1-dependent anti-apoptotic Cox-2 was induced by HGF. According to our studies using a nude mouse model, our results suggest that HGF may promote HNSCC growth and development by induction of Cox-2.
AP-1 is a stress-responsive transcription factor composed of members of the Jun and Fos family proteins. It can be activated by multiple extracellular stimuli including cytokines, growth factors, and UV irradiation. The role of AP-1 in apoptosis has been controversial. Both c-Jun and c-Fos have been found to be FIG. 4. Cox-2 expression induced by HGF is dependent on AP-1 transcription. A, Cox-2 expression in UMSCC1 cells was induced by HGF, as detected by microarray analysis. UMSCC1 cells were treated with HGF (40 ng/ml) for 1 h, and total RNA was extracted with Trizol reagents and further purified with RNeasy column. The profile of gene expression was systemically analyzed with Affymetrix U133 genechips. B, induction of Cox-2 mRNA by HGF was dependent on AP-1, as detected by Northern blot analysis. Both UMSCC1V cells and UMSCC1TAM67 cells were treated with HGF (40 ng/ml) for the indicated time periods, and total RNAs were isolated using Trizol reagents. Ten-microgram aliquots of total RNA were separated on 1.4% agaroseformaldehyde gels, and blots were hybridized with the 32 P-labeled human Cox-2 cDNA probe. For loading control, the membrane was stripped and rehybridized with the 32 P-labeled human glyceraldehyde-3-phosphate dehydrogenase (gapdh) probe. C, induction of Cox-2 by HGF was dependent on AP-1 transcription, as detected by Western blot analysis. Both UMSCC1V and UMSCC1TAM67 cells were treated with HGF (40 ng/ml) for the indicated time periods, and whole cell extracts were prepared with lysis buffer. Fifty-microgram aliquots of extracts were probed polyclonal antibodies against Cox-2 (1:1000). For internal control, the membrane was stripped and reprobed with monoclonal antibodies against ␣-tubulin (1:7500).

FIG. 5. HGF-induced Cox-2 inhibits anoikis in UMSCC1 cells.
A, the Cox-2 inhibitor NS398 abolished HGF-mediated anoikis resistance. UMSCC1 cells were pretreated with NS398 or vehicle control for 30 min and then treated with HGF (40 ng/ml) for 72 h. Cell viability was determined with the trypan blue exclusion assay. The assays were performed in duplicate, and the results represent average values from three independent experiments. Statistical differences between each group were determined by the Student's t test. *, p Ͻ 0.01. B, establishment of UMSCC1 cells stably expressing Cox-2. UMSCC1 cells were transduced with retroviruses expressing Cox-2 or control vector and selected with puromycin for 1 week. The resistant clones were pooled, and the whole cell extracts were prepared with lysis buffer. Fiftymicrogram aliquots of proteins were probed with polyclonal antibody against Cox-2 (1:500). For loading control, the membrane was stripped and reprobed with monoclonal antibodies against ␣-tubulin. C, Cox-2 inhibited anoikis of UMSCC1 cells. Both UMSCC1V and UMSCC1TAM67 cells were cultured in suspension and treated with or without NS398 (10 M). Cell viability was determined with trypan blue exclusion analysis. The assays were performed in duplicate, and the results represent the average values from three independent experiments. Statistical differences between each group were determined by the Student's t test. *, p Ͻ 0.01. induced by multiple apoptotic stimuli. Inhibition of c-Jun or c-Fos expression has been found to protect neuronal cells and lymphoid cells from apoptosis induced by growth factor withdrawal or genotoxic stimuli, indicating that AP-1 is a proapoptotic factor (33). Paradoxically, there are some studies reporting that AP-1 is required for cell survival. For example, c-Jun was found to play an essential role in hepatocyte survival, and hepatocytes derived c-junϪ/Ϫ embryos underwent massive apoptosis (33). AP-1 activity was also involved in protection in tumor necrosis factor (TNF)-mediated apoptosis and c-junϪ/Ϫ mouse embryonic fibroblasts were sensitive to TNF killing (44). These results suggest that the role of AP-1 in apoptosis may be cell type-or stimulus-dependent. Because the inhibition of AP-1 activity with TAM67 abolished HGF-mediated anoikis resistance, our work suggests that AP-1 plays an anti-apoptotic role in anoikis. Additionally, our results from c-fosϪ/Ϫ MFs also support this conclusion. AP-1 has been implicated in tumor development and metastasis by inducing collagenase expression or promoting cell cycle progression (33). Our findings suggest that the AP-1-mediated anti-apoptotic function is also important for the development and progression of HNSCC.
Using microarray analysis, unexpectedly, Cox-2, but not the Bcl-2 proteins, was identified as an anti-apoptotic gene induced by HGF. Because Cox-2 is involved in development of many human cancers, the regulation of Cox-2 expression is under intensive investigation (38,45). The promoter region of Cox-2 contains multiple NF-B-binding sites. A great number of studies have demonstrated that the induction of Cox-2 by proinflammatory cytokines and growth factors is dependent on NF-B transcription in monocytes and keratinocytes (38). Studies by Chen et al. (46) demonstrate that p38 kinase, but not ERK kinase, played a critical role in the regulation of Cox-2 expression induced by UVB. According to our previous study, HGF did not induce the nuclear translocation of NF-B or NF-B transcription in HNSCC cells (26). We also found that HGF did not induce the activation of p38 in HNSCC cells (data not shown). Thus, it is unlikely that Cox-2 expression induced by HGF is dependent on NF-B or p38 in HNSCC cells. Our dominant negative approach demonstrated that Cox-2 expres- sion induced by HGF was dependent on AP-1 transcription. Additionally, we found that inhibition of the ERK-signaling pathway also blocked the induction of Cox-2 by HGF (data not shown). These results indicate that the regulation of Cox-2 is complex and that Cox-2 can be regulated by different transcription factors and signaling pathways, which may be dependent on the tumor types. Also, this may be a reason why Cox-2 is found to be overexpressed in a wide range of human cancers.
In contrast to other human cancer cell lines, many types of human HNSCC cell lines develop tumors very slowly in the ectopic subcutis compared with the orthotopic site (23). It is possible that the tumor environment and paracrine factors produced by stromal cells in vivo may be critical for the survival and progression of HNSCC. In the current study, overexpression of Cox-2 significantly promoted HNSCC growth in the ectopic subcutis, indicating that HGF-induced genes play an important role in tumor development and progression. According to this study, the inhibition of anoikis by Cox-2 may be one of the potential mechanisms. There are many studies reporting that Cox-2 expression is associated with the progression and development of HNSCC. For example, Gallo et al. (41) evaluate the role of Cox-2 in head and neck cancers by analyzing Cox-2 expression in relation to tumor angiogenesis and lymph node metastasis. They found that Cox-2 activity was correlated with vascular endothelial growth factor mRNA and protein expression. Cox-2 mRNA and protein expression was higher in tumor samples than in normal mucosa. Specimens from patients with lymph node metastasis exhibited higher levels of Cox-2 protein than specimens from patients without metastasis. Mestre et al. (42) also find that Cox-2 was overexpressed in HNSCC compared with adjacent control tissue, and retinoids exhibited their chemopreventive effect by inhibiting Cox-2 expression. Despite these studies, the molecular mechanism that up-regulates Cox-2 expression in HNSCC is unknown. According to the current study, the constitutive activation of the c-Met/HGFsignaling pathway may be responsible for the abnormal expression of Cox-2 in HNSCC. In this regard, the inhibition of the c-Met/HGF-signaling pathway with biological or pharmacological inhibitors may have a beneficial effect on suppression of Cox-2 activity in the treatment of HNSCC (38,45).