Hypoxia Enhances Metastatic Efficiency in HT1080 Fibrosarcoma Cells by Increasing Cell Survival in Lungs, Not Cell Adhesion and Invasion

Tumor hypoxia has been strongly associated with tumor resistance to therapy and malignant progression such as increased metastatic potential (1–3). For tumor cells to form metastases, they have to intravasate into the circulatory system, survive the mechanical shearing force in the circulation, arrest and survive at the remote sites, and then initiate growth and continue to proliferate (4, 5). Although the mechanisms involved in hypoxia-increased tumor cell metastatic efficiency are still not completely understood, studies have suggested that exposure to hypoxia may have an effect on all these steps of the metastatic process. In addition, hypoxia may also promote angiogenesis at primary sites to produce new lymphatic and blood vessels that provide a route for tumor cells to enter the circulation system.

Studies have shown that metastasis is an inefficient event and that the mechanisms for the inefficiency can be cell type specific. For example, in vivo video microscopy and cell fate analysis using mouse melanoma B16F10 cells suggested that metastatic inefficiency is primarily due to regulation of the balance between tumor cell growth and death at secondary sites; the failure of growth initiation of arrested tumor cells plays an important role (4). Studies using transformed rat embryo fibroblasts have shown that apoptosis of lung-arrested tumor cells after i.v. injection is an early event involved in metastatic inefficiency (6). Both experimental and clinical studies have indicated that metastatic efficiency of tumor cells can be associated with their resistance to apoptosis (7–9). Because hypoxia can regulate cell growth through various pathways (10), it is possible that hypoxia may affect metastatic efficiency by modifying the balance between tumor cell death and growth. Although direct evidence to support this hypothesis is still needed, studies have shown that gene products, such as vascular endothelial growth factor (VEGF), a protein up-regulated by hypoxia, can act not only as an angiogenic factor but also as a survival factor for some human tumor cells (11), and blocking VEGF with neutralizing antibody has been reported to significantly decrease metastatic efficiency of some human melanoma cell lines (12, 13). Recently, lysyl oxidase (LOX) has been reported to be essential in hypoxia-induced metastasis, and one of the postulated mechanisms is to allow the growth and proliferation of tumor cells at metastatic sites (14, 15).

Hypoxia-induced LOX can also modify tumor cell motility and invasion; thus, it may increase tumor cell metastatic efficiency by affecting both intravasation and extravasation (14). Other studies have also reported that hypoxia may have an effect on intravasation and extravasation by modulating tumor cell motility and invasion potential through different mechanisms [e.g., through the regulation of protease activity on extracellular matrix (ECM) degradation; refs. 1, 2]. Hypoxia can also enhance the expression of autocrine motility factor, and by doing so increase motility of human pancreatic cancer cells (16). In addition, exposure to 3% oxygen was found to activate transcription of the Met proto-oncogene and consequently to promote the invasive growth of the tumor cells (17). Furthermore, hypoxia-inducible factor-1 (HIF-1), a protein that can be induced by hypoxia through the stabilization of its subunit HIF-1α, can regulate cell invasion in colon carcinoma (18). Recently, it has been reported that hypoxia can stimulate carcinoma invasion by stabilizing microtubules and by promoting the Rab11 trafficking of integrin α₆β₄ (19).

Hypoxia may also affect tumor cell arrest at metastatic sites by modifying their adhesion potential. Although it has been reported that the arrest of tumor cells is a passive process due to the size restriction of capillaries smaller than tumor cells, there is evidence suggesting that cancer cells may undergo adhesive arrest in the liver in precapillary vessels, when the endothelium has been activated by the cytokine interleukin-1α (20). In addition, it has been reported that certain types of tumor cells can be arrested in lungs by attaching to relatively larger pulmonary blood vessels, where the cells are unlikely to be trapped by size limitation (21). This attachment is mediated through the interaction between integrin α₃β₁ on tumor cells and laminin-5 in blood vessel basement membrane (22). Integrins are proteins that have been shown to play an essential role in cell adhesion to ECM and other cells and to be involved in tumor progression (23). The expression of some integrins can be regulated by hypoxia through the induction of HIF-1 or the activation of extracellular signal-regulated kinase (ERK) 1/2 and p38 mitogen-activated protein kinase, thereby potentially modifying cellular adhesion (24–26). These studies suggest that hypoxia might promote tumor cell arrest by increasing tumor cell adhesion potential through the induction of integrins to increase metastases.

In this study, we investigated mechanisms that may contribute to hypoxia-enhanced metastasis using a green fluorescent protein (GFP)–labeled human fibrosarcoma HT1080 cell line. We observed that the main effect of hypoxic exposure on this cell line was increased survival of cells in the lungs, suggesting that by increasing tumor cell survival alone, hypoxia is able to enhance tumor cell metastatic efficiency. This was further supported by a finding that treatment with farnesylthiosalicylic acid (FTS) both reduced metastatic efficiency in hypoxia preexposed cells and inhibited the effect of hypoxia on tumor cell survival in lungs.

Cell culture and hypoxic treatment. HT1080-GFP cells were kindly provided by Dr. Ruth Muschel (Radiobiology Research Institute, Churchill Hospital, Headington, Oxford, United Kingdom). The cells were cultured in α-MEM medium (Invitrogen) supplemented with penicillin/streptomycin and 10% fetal bovine serum (FBS). For all experiments, 2 × 10⁵ to 5 × 10⁵ cells were plated in 10-cm dishes and incubated in 5% CO₂ and air at 37°C. To render cells hypoxic, dishes were placed in a plastic chamber (Billups-Rothenberg) flushed with a gas mixture of 5% CO₂ and a designated percentage of O₂ balance N₂. It took 3 to 6 h to achieve severe hypoxia (≤5 mmHg) in medium when 0% O₂ was used. For reoxygenation, cells were incubated in a tissue culture incubator. The average rate of oxygen increase was at ∼3 mmHg/min during the first 30 min of reoxygenation. FTS was purchased from Toronto Research Chemicals, Inc.

Assays for experimental lung metastasis. Severe combined immunodeficient mice were housed in the specific pathogen-free colony of the Ontario Cancer Institute under conditions approved by the Canadian Council on Animal Care. For the lung colony assay, each mouse received 5 × 10⁴ cells through the tail vein and was killed 28 days later. The lungs were fixed in Bouin's solution. The pulmonary tumors on the surface were counted under a dissecting microscope. The number of injected viable cells was determined using clonogenic assays. The plating efficiencies for different treatment groups, which varied from 60% to 90%, were then used to calculate metastatic efficiency.

Clonogenic assays for viability of lung-arrested tumor cells. For the viability assay of the arrested tumor cells, 10⁶ or 10⁵ cells per mouse were injected i.v. To prepare lung single-cell suspensions, the lungs were removed, minced, and incubated in 4.5 mL PBS, 0.5 mL 0.05% trypsin-EDTA (Invitrogen), and 150 Kunitz units/mL DNase I (Sigma-Aldrich) for 30 min at 37°C with agitation, and then the tissue pieces were incubated in α-MEM medium containing 0.25% collagenase type 4 (Sigma-Aldrich) and 300 Kunitz units/mL DNase I for 1 h at 37°C with agitation. The cells were dispersed by shaking and then passed through a 40-μm cell strainer to collect a single-cell suspension. A total of 5 × 10⁴ lung-derived cells were plated into a 10-cm tissue culture dish. Colonies were counted 10 days later.

Plating efficiency assay. Tumor cells from the same population as i.v. injected cells or cells for other assays were plated at 100 tumor cells/10-cm tissue culture dish. The colonies formed on dishes were fixed, stained, and counted at 10 days later to determine plating efficiency.

Western blot analysis. Cells were washed with cold PBS and lysed in 0.5 to 1.0 mL lysis buffer [50 mmol/L HEPES (pH 8.0), 10% glycerol, 1% Triton X-100, 150 mmol/L NaCl, 1 mmol/L EDTA, 1.5 mmol/L MgCl₂, 100 mmol/L NaF, 10 mmol/L NaP₂O₇, freshly added protease inhibitors]. The lysates were scraped into tubes, incubated on ice for 1 h, and then centrifuged for 10 min at 10,000 rpm and 4°C to collect the supernatant. Whole-cell lysates were resolved on 10% SDS-polyacrylamide gels and transferred to a Hybond enhanced chemiluminescence nitrocellulose membrane (Amersham Biosciences). The antibodies used were polyclonal anti–integrin α₃, anti–integrin β₁, anti–β-actin (Santa Cruz Biotechnology, Inc.), anti–phospho-ERK1/2, anti–phospho-Akt (Cell Signal Technology), and anti-tubulin (Abcam) and monoclonal anti-MDM2 (Oncogene/Calbiochem), anti-p53 (Santa Cruz Biotechnology), and anti–HIF-1α (Abcam). The hybridization and detection were done according to the instructions from the antibody manufacturers.

Invasion assay. Invasion assays were done using a Cell Invasion Assay (fluorometric) kit (Chemicon). Cells were harvested and resuspended in medium (2% FBS) at 5 × 10⁵ cells/mL. An aliquot of the cell suspension (250 μL) was added to inserts coated with a matrix of reconstituted basement membrane proteins (ECMatrix). Complete medium was added as chemoattractant into a 24-well plate (500 μL/well), in which the coated inserts were placed. At 24 and 48 h of incubation, the cells that had invaded through the inserts were detached and lysed, and a CyQuant GR Dye was added. The fluorescence intensity was measured with a SpectraFluor Plus plate reader (Tecan) using 485/535 nm filter set.

Adhesion assay. Cell adhesion potential was tested using an Integrin-Mediated Cell Adhesion kit, a CytoMatrix Screen kit (Chemicon), or 96-well plates coated with laminin-5 (Biodesign) at 2 μg/mL overnight. Cells were harvested and resuspended in complete medium at 5 × 10⁵ cells/mL. An aliquot of the cell suspension was added to a coated well (100 μL/well of a 96-well plate). The cell number and incubation time were determined in preexperiments. After 2 h of incubation at 37°C in a cell culture incubator, the plate was rinsed three times with PBS containing 1 mmol/L Ca²⁺ (200 μL/well), stained with 0.2% crystal violet in 10% ethanol, and rinsed again, and the dye was extracted into a mixture of 50 μL of 0.1 mol/L NaH₂PO₄ (pH 4.5) and 50 μL of 50% ethanol. The absorbance at 560 nm was measured as aforementioned.

Proliferation (WST-1) assay. Cells (1 × 10³–5 × 10³) were seeded into a 96-well plate and cultured under normal or hypoxic conditions. On the day of the assay, 10 μL/well of the WST-1 labeling solution (EMD Biosciences) were added. After 2 h of incubation at 37°C in a cell culture incubator, the absorbance was measured as aforementioned at 450 nm with the reference wavelength 620 nm.

Immunohistochemistry. A total of 10⁶ cells per mouse were injected i.v. The lungs were removed and fixed in formalin and later embedded in paraffin. The paraffin-embedded lungs were cut into 4-μm-thick sections, dewaxed, dehydrated, and microwave heat treated for epitope retrieval. Sections were treated with a protein blocker (Signet Labs) before incubating with antibody to GFP (Abcam), Ki67 (DAKO), or cleaved caspase-3 (Cell Signaling). The sections were then incubated for 30 min each with biotinylated secondary antibody (Vector Laboratories) and horseradish peroxidase–labeling reagent (ID Labs). The incubations were all at room temperature. Reaction products were revealed by 3,3′-diaminobenzidine, counterstained, and mounted in Permount (Fisher). For double staining, the images were immediately captured using a ScanScope scanner (Aperio Technologies), and then the coverslips and mounting medium were removed to do second round staining. Alkaline phosphatase-streptavidin (Vector Laboratories) and a freshly prepared Vector red solution (Vector Laboratories) were used to reveal reaction products.

Real-time quantitative PCR. Total RNA was isolated using an RNeasy Miniprep kit (Qiagen). For quantitative PCR, 2 μg of total RNA were reverse transcribed using an OmniScript kit (Qiagen), and 1 μL of the reverse transcription products was mixed with PCR primers, double-distilled water, and SYBR Green PCR Master Mix (Applied Biosystems) to a total volume of 12 μL. The reaction condition was as follows: 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 60 s. The reactions were run and analyzed with an ABI Prism 9700 Sequence Detector (Applied Biosystems).

Statistics. The t test (two tailed) was done for comparisons between two groups. Kruskal-Wallis statistic or ANOVA was used to detect significant changes among multiple groups followed by Dunn's test for multiple comparisons against one single control group.

Hypoxia pretreatment increased experimental metastatic efficiency in HT1080-GFP cells. Severe chronic hypoxia (5% CO₂ and 95% N₂, 24 h) pretreatment significantly increased (P < 0.025) experimental lung metastatic efficiency in HT1080-GFP cells by 3- to 4-fold (Fig. 1A). Reoxygenation for 6 to 18 h showed a trend for further increase, but it was not statistically significant compared with that of groups without reoxygenation. After 24 h of reoxygenation, the metastatic efficiency started to return toward control level. At 48 h of reoxygenation, the metastatic efficiency was similar to that of normoxic control groups.

Because tumor cells exposed to chronic anoxia or severe hypoxia in a solid tumor may lie far from functional vasculature, and thus may be unable to contribute significantly to the development of metastasis, we examined the effect of hypoxia on tumor cell metastatic efficiency at higher oxygen concentrations. As shown in Fig. 1B, the lung experimental metastatic efficiency was increased in response to a wide range of low oxygen preexposure at concentrations from 0% to 2%, with the largest increase occurring at 0.2% oxygen (P < 0.01 for 0.2% versus 0% and P < 0.05 for 0.2% versus 2%), although the difference between 0.2% and 1% oxygen treatments was not statistically significant. The concentration of 0.2% O₂ was used in all subsequent experiments. Figure 1C shows that HT1080-GFP cell viability was not significantly changed after hypoxia treatment as tested by clonogenic assay.

Hypoxic exposure did not affect adhesion or invasion potential in HT1080-GFP cells. Previously, we showed that hypoxia increased metastasis in a mouse fibrosarcoma cell line (KHT-C) by up-regulating Mdm2 and suppressing p53 activity (27). However, this mechanism does not explain hypoxia-enhanced metastatic efficiency in the human fibrosarcoma cell line (HT1080). As shown in Fig. 2A, p53 protein level was increased by hypoxia. The accumulation of MDM2 protein, a target gene of p53 transactivity, was observed in HT1080-GFP cells after 5 Gy of ionizing radiation even with hypoxia pretreatment, suggesting that hypoxia may induce, rather than suppress, p53 activity in this cell line.

Because previous studies using this HT1080-GFP cell line have shown that lung metastasis can originate from the proliferation of endothelium-attached tumor cells and that the vascular attachment of these tumor cells rather than size limitation was observed as the mechanism of lung arrest (21), we investigated whether hypoxia was able to increase the adhesion potential of these cells. First, we examined the expression level of α₃β₁ integrin after hypoxia treatment, based on the observation by Wang et al. (22) that lung arrest of the HT1080-GFP cells was mediated through an interaction between α₃β₁ integrin on tumor cells and laminin-5 in exposed pulmonary base membranes. We did not detect any change of integrin α₃ or β₁ expression in HT1080-GFP cells after hypoxia treatment with or without reoxygenation (Fig. 2B).

In addition to integrin α₃β₁, other integrins can also play important roles in cell adhesion and may be regulated by hypoxia (25, 28). Furthermore, hypoxia can increase cell adhesion by activating integrins through a mechanism that does not involve the regulation of integrin expression at either mRNA or protein level (24). Thus, we did adhesion assays to test cell adhesion potential using antibodies to various integrins and using different ECM proteins, including fibronectin, vitronectin, laminin, collagen-1, and collagen-4. The tested integrins (integrin α_vβ₃, integrin α_vβ₅, and all β₁-containing integrins) have been shown to play important roles in mediating cell adhesion to ECM (22–25). Figure 2C, (top) shows that hypoxia pretreatment, with or without reoxygenation, did not change HT1080-GFP cell adhesion potential mediated by any one of the integrins. There was also no difference observed between normoxic and hypoxia-pretreated HT1080-GFP cells with or without reoxygenation in their adhesion to ECM proteins, including fibronectin, vitronectin, laminin, collagen-1, and collagen-4 (Fig. 2C , bottom). Finally, the adhesion of HT1080-GFP cells to laminin-5 was not affected by hypoxia (please see below). Taken together, these data suggested that hypoxia did not change adhesion potential of HT1080-GFP cells.

Extravasation is another step in the metastatic cascade that can be potentially modulated by hypoxia (5). Because at the time of extravasation, tumor cells are usually exposed to normal physiologic oxygen concentration, an invasion assay on tumor cells pretreated with hypoxia was done under normoxic conditions to mimic the effect of hypoxia on tumor cell extravasation. Normoxic or hypoxic cells that migrated through the ECMatrix-coated inserts were quantified at 24 and 48 h after seeding. The numbers of migrated cells were normalized against the plating efficiencies of oxic or hypoxia-pretreated cells seeded directly in tissue culture dishes and is presented as relative fluorescence units in Fig. 2D. In this assay, the number of the cells migrated through the coated inserts is dependent on both ECM degradation and cell motility, the combination of which determines the efficiency of tumor cell extravasation. Again, no difference was detected between control and hypoxia-pretreated HT1080-GFP cells at both time points, suggesting that hypoxia did not affect tumor cell invasion in this model. Although the oxic or hypoxia-pretreated cells had been grown for 24 and 48 h on ECM-coated inserts and might have proliferated, the proliferation rates of these cells are unlikely to cause any change in the number of migrated cells because no significant difference in proliferation rates was observed between oxic and hypoxia-pretreated HT1080-GFP cells (Fig. 3C).

Hypoxia pretreatment increased the survival of lung-arrested HT1080-GFP cells. Experimental and clinical studies indicate that metastatic potential can be associated with tumor cell resistance to apoptosis in certain types of tumors (6–8, 29–31). Furthermore, our previous studies showed that hypoxia can increase metastatic efficiency by inhibiting an apoptotic response in KHT-C and SCC VII cells (27). Thus, we investigated the effect of hypoxia on survival of HT1080-GFP cells arrested in mouse lungs. Similar to previous results in KHT-C cells, apoptotic cell death was observed in lung-arrested HT1080-GFP cells (Fig. 3A), suggesting a role of apoptosis in the death of HT1080-GFP cells after lung arrest. Clonogenic assays on cells derived from lungs of mice injected i.v. with tumor cells showed that more clonogenic tumor cells were recovered from the lungs of mice injected with hypoxia-pretreated tumor cells than from those injected with normoxic control cells (10⁵ or 10⁶ cells/mouse injected; Fig. 3B). In addition, this increase is likely due to a better survival rather than enhanced proliferation of hypoxia-pretreated cells because a large difference in proliferation rate would be required to explain the big difference observed. Furthermore, hypoxia did not promote cell proliferation in this cell line (Fig. 3C), and at 24 h after injection, the tumor cells are unlikely to have undergone significant proliferation. This was further confirmed by Ki67 immunohistochemistry staining on sections of lungs removed from mice at 24 h after tumor cell injection. As shown in Fig. 3D, Ki67-positive staining was not correlated with GFP-positive staining (for HT1080-GFP cells) in mouse lungs at 24 h after injection of oxic or hypoxia-preexposed HT1080-GFP cells, suggesting that the injected tumor cells did not initiate proliferation in mouse lungs during the first 24 h after injection.

FTS inhibited the effect of hypoxia on tumor cell survival and lung experimental metastasis. Blum et al. (32) reported that a Ras inhibitor FTS inhibited the phosphorylation but not the total amount of Akt or ERK protein and down-regulated HIF-1α at the protein level by inhibiting the Akt/mammalian target of rapamycin (mTOR) pathway, causing cell death in glioblastoma cells, suggesting that it could potentially be used as a cancer therapeutic agent targeting HIF-1α activation. We tested whether FTS was able to block the effects of hypoxia on the HT1080-GFP cells. We observed that FTS treatment was able to inhibit the effect of hypoxia on increased experimental metastatic efficiency (Fig. 4A). Surprisingly, FTS treatment of the cells during hypoxic exposure inhibited hypoxia-induced VEGF but not hypoxia-induced carbonic anhydrase 9 (CA9) or LOX (Fig. 4B). Furthermore, Fig. 4C shows that the accumulation of HIF-1α was not correlated with Akt activation. This suggests that the mechanism of the action of FTS is unlikely to be through inhibition of HIF-1α accumulation by targeting the Akt/mTOR pathway. Further investigation showed that the phosphorylation of Akt or ERK was not changed by hypoxic exposure alone or by FTS. Reoxygenation was required for Akt activation, and this activation was inhibited by FTS (Fig. 4C and D); however, these changes were not observed in ERK phosphorylation. Because HIF-1α has a very short half-life under oxic condition, the accumulation of HIF-1α was observed only in cells treated with hypoxia without reoxygenation (Fig. 4C) and it was not inhibited by FTS treatment (Fig. 4D), suggesting that other mechanisms and/or transcription factors may be involved in the inhibitory effect of the drug on hypoxia-increased metastases in HT1080-GFP cells.

To determine the role of tumor cell survival after lung arrest in hypoxia-increased metastasis, we further tested whether FTS could affect the survival and adhesion potential of HT1080-GFP cells. FTS treatment did not affect the in vitro viability of either oxic or hypoxic HT1080-GFP cells (Fig. 5A) nor did it inhibit the adhesion of oxic or hypoxic cells to laminin-5 (Fig. 5B,, top) or to some other ECM proteins (Fig. 5B,, bottom). However, if the cells were treated with FTS during hypoxic exposure and then injected i.v. into mice, the hypoxia-enhanced survival of lung-arrested cells was inhibited, whereas no effect was observed in FTS-treated normoxic cells (Fig. 5C). These data support the linkage between the effect of hypoxia on the survival of lung-arrested tumor cells and hypoxia-increased experimental metastatic efficiency.

It has been well shown that most solid tumors grow in a unique microenvironment characterized by an abnormal vascular structure, which leads to an insufficient supply of oxygen and nutrients to tumor cells. Tumor hypoxia is a consequence of this abnormal vasculature and promotes tumor malignant progression such as increased metastasis (1, 2). A better understanding of the mechanisms involved in the hypoxia-induced increase in metastatic disease is needed to improve the outcome of patients with hypoxic tumors. This study investigated the effect of hypoxia on various stages of metastatic process using a GFP-labeled human fibrosarcoma cell line HT1080. We found that increased survival of tumor cells under secondary stress conditions after hypoxic exposure may be enough to cause increased metastases. More importantly, by inhibiting the enhanced tumor cell survival, the hypoxia-increased metastatic efficiency was prevented.

Another important biological event involved in metastatic growth is angiogenesis. Angiogenesis can be stimulated by hypoxia through the induction of growth factors such as VEGF. In a primary tumor, angiogenesis may provide escape routes for dissociated tumor cells, thereby promoting intravasation. At later stages of metastatic progression, angiogenesis is required to sustain the growth of micrometastases to develop into clinically detectable macrometastases (4, 33). However, the effect of hypoxia on angiogenesis and intravasation is unlikely to play a role in the increased lung colonies observed in this study because tumor cells were injected i.v. into mice, and also the effect of hypoxia on lung metastasis in this cell line is transient and lost after 48 h of reoxygenation (Fig. 1A). Angiogenesis is unlikely to be required at this early stage. For the same reason, other mechanisms that affect intravasation, for example hypoxia-regulated expression of urokinase-type plasminogen activator or matrix metalloproteinases (1), are unlikely to be involved.

The results in this study are consistent with our previous findings in a murine fibrosarcoma cell line KHT-C (27), in which we showed that lung-arrested KHT-C cells underwent apoptosis and that the altered response of KHT-C cells to apoptosis, which involves hypoxia-induced up-regulation of Mdm2 and the consequent inhibition of p53 activity, can contribute to increased metastatic efficiency. Apoptosis was also observed to occur in the lung-arrested HT1080-GFP cells, and hypoxia pretreatment increased the survival of these cells (Fig. 3), indicating apoptotic potential as an important mechanism involved in the effect of hypoxia on tumor metastatic progression in this model. However, the molecular mechanisms involved are likely to be different in these two models because activation rather than suppression of p53 by hypoxia was observed in HT1080-GFP cells (Fig. 2A). Further studies are needed to understand the molecular mechanisms involved in the HT1080-GFP model. Equally, the importance of the hypoxia-induced increase in lung survival also needs to be investigated in different types of tumor cells to determine whether it is a cell type specific effect of hypoxia.

An interesting and clinically relevant aspect of these results is that the increased metastatic efficiency can be due to enhanced survival potential achieved by altered response to apoptosis, which may confer resistance to radiotherapy and some chemotherapies (34). More importantly, this can occur after up to 24 h of reoxygenation (Fig. 1A). The existence of hypoxic tumor cells thus may contribute to both the failure of local control and the increased incidence of metastatic disease. Moreover, these effects are transient, suggesting that long-term hypoxic exposure to select for mutants is not required. Thus, it could be easier for the tumor cells to acquire this transient more aggressive phenotype. This model therefore may provide a useful tool to study and test for the efficacy of new therapeutic regimens to target the malignant progression promoted by hypoxia.

In this context, the potential use of FTS to inhibit the effect of hypoxia on tumor cells was tested. FTS has been reported to be a Ras-dislodging antagonist through the inhibition of Ras methylation, which is one of the Ras post-translational modifications required for its membrane localization (35). FTS has been shown to inhibit cell growth and induce apoptosis in several different tumor types by blocking Ras downstream pathways (32, 36–39). As a promising agent for cancer therapy in tumors with mutant activated Ras, FTS may also have the additive benefit of controlling hypoxia-induced tumor progression due to the important role of Ras activation in hypoxia-regulated gene expression. For example, hypoxia-dependent up-regulation of osteopontin, which is linked to malignant progression in several tumor sites, has been reported to be mediated by a Ras-activated enhancer in NIH-3T3 cells (40, 41). The inhibition of Ras by FTS has been shown to cause cell death, possibly through the down-regulation of HIF-1α, in glioblastoma cells (32). Other hypoxia-inducible transcription factors have also been reported to be mediated by Ras downstream pathways, such as nuclear factor-κB (NF-κB; refs. 42, 43) and activator protein-1 (AP-1). Furthermore, some hypoxia-regulated genes that are important in tumor malignancy may be regulated by more than one transcription factor (44). One example is VEGF, which can be regulated by HIF-1 (45), NF-κB (46), Egr1 (47), and AP-1 (48). Therefore, to achieve a better control of hypoxia-induced tumor progression, Ras activation can potentially be an effective target. However, whether or how Ras activation may be involved in hypoxia-increased metastatic efficiency in HT1080-GFP cells still needs to be determined.

The data in Figs. 4 and 5 show that FTS successfully inhibited the hypoxia-induced increase in tumor cell survival and metastatic efficiency in HT1080-GFP cells, but no effect was observed on oxic cells, suggesting that this inhibitory effect of FTS on metastatic efficiency is by targeting hypoxia-induced proteome changes in this cell line, at least at the tested concentrations. Despite this finding, our initial studies show no evidence for an effect of FTS on down-regulation of HIF-1α in hypoxic HT1080 cells (Fig. 4D). Thus, a much more detailed investigation of the mechanisms of action of FTS in HT1080-GFP cells is needed. Currently, we are investigating the involvement of other Ras downstream pathways, such as p38 and c-Jun NH₂-terminal kinase pathways, which have also been shown to regulate cell apoptosis (49, 50). The FTS inhibition of hypoxia-upregulated VEGF also merits further study. The potential clinical use of FTS as an antimetastatic agent needs to be further studied to determine if it can inhibit metastases by oxic tumor cells at different concentrations.

Overall, although hypoxia has been reported to modify many aspects of tumor cell behavior that may be important in the metastatic process, we have shown that they are not all required for hypoxia to increase metastatic efficiency using human fibrosarcoma HT1080-GFP cell line. By promoting the survival of arrested cells alone, hypoxia is able to promote metastatic progression. These results provide important guidelines for developing new therapeutic strategies. For example, metastasis can originate intravascularly from the proliferation of these tumor cells attached to the endothelium without the need for extravasation (21); therefore, targeting tumor cell invasion alone may not be effective to control metastatic progression in patients with similar tumors. This study confirms that a transient alteration in the apoptotic response of cells can play an important role in metastasis formation. This aspect needs to be considered in the development of cancer therapies targeting metastatic development.

Grant support: National Cancer Institute of Canada with funds raised by the Terry Fox Run and by an Interdisciplinary Health Research Team grant from the Canadian Institutes for Health Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Ruth Muschel for kindly providing us the HT1080-GFP cell line and Bob Kuba for technical assistance.