Cancer Cell Biology
Tumor–stromal cell interaction under hypoxia increases the invasiveness of pancreatic cancer cells through the hepatocyte growth factor/c-Met pathway
Article first published online: 22 SEP 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 12, pages 2750–2759, 15 December 2006
How to Cite
Ide, T., Kitajima, Y., Miyoshi, A., Ohtsuka, T., Mitsuno, M., Ohtaka, K., Koga, Y. and Miyazaki, K. (2006), Tumor–stromal cell interaction under hypoxia increases the invasiveness of pancreatic cancer cells through the hepatocyte growth factor/c-Met pathway. Int. J. Cancer, 119: 2750–2759. doi: 10.1002/ijc.22178
- Issue published online: 26 OCT 2006
- Article first published online: 22 SEP 2006
- Manuscript Accepted: 8 JUN 2006
- Manuscript Received: 21 APR 2006
- tumor–stromal interaction;
- pancreatic cancer;
- hepatocyte growth factor;
The hypoxic environment in tumor is reported to play an important role in pancreatic cancer progression. The interaction between stromal and cancer cells also contributes to the malignant behavior of pancreatic cancer. In the present study, we investigated whether hypoxic stimulation affects stromal as well as pancreatic cancer cells. Our findings demonstrated that hypoxia remarkably elevated the HIF-1α expression in both pancreatic cancer (PK8) and fibroblast cells (MRC5). Hypoxic stimulation accelerated the invasive activity of PK8 cells, and invasiveness was thus further accelerated when the hypoxic PK8 cells were cultured with conditioned medium prepared from hypoxic MRC5 cells (hypoxic conditioned medium). MMP-2, MMP-7, MT1-MMP and c-Met expressions were increased in PK8 cells under hypoxia. Hypoxic stimulation also increased the hepatocyte growth factor (HGF) secretion from MRC5 cells, which led to an elevation of c-Met phosphorylation in PK8 cells. Conversely, the elevated cancer invasion, MMP activity and c-Met phosphorylation of PK8 cells were reduced by the removal of HGF from hypoxic conditioned medium. In immunohistochemical study, the HIF-1α expression was observed in surrounding stromal as well as pancreatic cancer cells, thus indicating hypoxia exists in both of cancer and stromal cells. Moreover, the stromal HGF expression was found to significantly correlate with not only the stromal HIF-1α expression but also the c-Met expression in cancer cells. These results indicate that the hypoxic environment within stromal as well as cancer cells activates the HGF/c-Met system, thereby contributing to the aggressive invasive features of pancreatic cancer. © 2006 Wiley-Liss, Inc.
Most aggressive tumors have acquired the ability to develop their own blood vessels as they grow. However, the tumor cells grow faster than the endothelial cells, so the structure and function of the tumor vasculature is highly disorganized in comparison to normal tissues.1, 2 In this microenvironment, within tumor tissues, cancer cells are exposed to hypoxia.3 Recent studies have revealed that hypoxia-inducible factor-1 (HIF-1) plays a key role in tumor progression under hypoxia.4, 5 HIF-1 is a heterodimeric basic helix-loop-helix PER/ARNT/SIM (HLH-PAS) transcription factor that consists of HIF-1α and a constitutively expressed aryl hydrocarbon receptor nuclear translocator (ARNT) known as HIF-1β. Under normoxic condition, HIF-1α is bound to the tumor suppressor Von Hippel-Lindau (VHL) protein. This protein complex causes HIF-1α to be targeted by proteasomes, thus leading to rapid protein degradation. Conversely, under hypoxic conditions, HIF-1α is stabilized and dimerized with HIF-1β, translocates to the nucleus and transactivates various gene expressions. HIF-1 is reported to regulate various gene expressions by binding to the hypoxia-responsive element on the target genes. More than 40 genes that are important for glucose transport, angiogenesis, erythropoiesis, vasomotor regulation and the survival of cancer cells6, 7, 8 are known to be targets for HIF-1α. In an immunohistochemical study using several cancer tissue specimens, association of the HIF-1α overexpression with a poor prognosis of the patients was reported.8
Pancreatic cancer is one of the most aggressive malignancies in industrialized countries.9, 10 Most patients with pancreatic cancer have a poor outcome due to difficulties in its early diagnosis, its highly invasive and metastatic features, and the limited efficiency of conventional therapeutics such as a surgical resection, chemotherapy and radiotherapy.11 A noteworthy report has indicated that tumor hypoxia exists within pancreatic cancer. Using xenografted tumors of nude mice, the studies have shown pancreatic tumors to contain regions of extremely low pO2, in comparison to normal pancreatic tissue.12 Clinically, it is also accepted that hypoxic tumor regions are responsible for resistance against chemotherapy and radiation therapy.13, 14 Moreover, interactions between cancer cells and the surrounding stromal fibroblasts have been reported to play a critical role in tumor invasion and metastasis.15 An infiltrating adenocarcinoma of the pancreas is often characterized by abundant desmoplastic stroma16 and some evidence has suggested that the stromal-cell reaction favors the tumor progression.17, 18 Berger et al. showed that the production of several growth factors to increase in human prostatic stromal cells by hypoxia, thus playing an important role in the development of prostate hypertrophy.19
In the present study, we developed a unique culture system to evaluate tumor–stromal cell interaction under hypoxia. Using this system, we analyzed whether hypoxia enhances the invasive activity of pancreatic cancer cells, while also increasing some secretions from fibroblasts, thus leading to the activation of cancer invasion. Furthermore, we attempted to isolate which molecules play an important role in activating cancer invasion through tumor–stromal cell interaction under hypoxia. Through these studies, we elucidated the implications of the hepatocyte growth factor (HGF)/c-Met signal in the hypoxic cancer–stromal action. In immunohistochemical study, the HIF-1α expression was evaluated to confirm whether or not a hypoxic environment exists in cancer and stromal cells. We also tried to clarify whether or not HGF/c-Met expression is specifically elevated in the hypoxic region of pancreatic cancer and the surrounding stromal cells.
Material and methods
Cell culture and exposure to hypoxia
Three human pancreatic cancer cell lines (PK1, PK8, KP4) were purchased from the Institute of Development, Aging and Cancer, Tohoku University (Sendai, Japan). The MRC5 human fibroblast cell line was purchased from the RIKEN Cell Bank (Ibaragi, Japan). The cells were cultured in RPMI-1640 medium (Sigma, St. Louis, MO) supplemented with 10% heat-inactivated fetal bovine serum (Sigma) and 100 μg/ml kanamycin (Meiji, Tokyo, Japan) and incubated at 37°C in a humidified atmosphere containing 20% O2 and 5% CO2 in air. When the culture reached semiconfluence, the medium was removed and replaced by fresh medium with 1% fetal bovine serum; half of the cells remained in the same condition (referred to as normoxia), whereas the remaining half was moved to a hypoxic chamber (ASTEC, Fukuoka, Japan) containing 1% O2, 5% CO2 and 94% N2, and this hypoxia chamber was maintained at 37°C (referred to as hypoxia). The serum-free conditioned media from MRC5 fibroblasts were collected following the 24-hr cultivation under normoxia and hypoxia and thus were used in experiments as a normoxic conditioned medium (NCM) and hypoxic conditioned medium (HCM).
In vitro invasion assay
The in vitro invasion activities were examined as reported previously20 using a gel matrix (Matrigel®; Beckton Dickinson, Franklin Lakes, NJ) in 24-well plates. Briefly, 6.5-mm diameter polycarbonate filters (8-μm pore size) of the Falcon Transwell™ chemotaxis chambers (Beckton Dickinson) were coated with 50 μl (0.25 mg/ml) of Matrigel® biomatrix in cold RPMI-1640 medium and dried overnight. Suspensions of 5 × 105 cells in 200 μl of complete RPMI-1640 medium were placed in the upper compartments of the chamber, whereas the lower compartments were each filled with 800 μl NCM or HCM from MRC5 fibroblasts. These culture units were incubated for 24 hr at 37°C under normoxia and hypoxia. Noninvasive cells on the upper surface of the filters were then removed completely with a cotton swab. Any viable invasive cells, which infiltrated onto the lower surface of the filter, were fixed using 70% ethanol and the nuclei were stained using hematoxylin. Next, the number of invasive cells was counted. These experiments were carried out in triplicate and independently repeated at least 3 times.
Cell proliferation assay
Cell proliferation was analyzed by the MTT quantitative colorimetric assay. In brief, 5 × 103 cells/well were seeded in triplicate onto 96-well plates and incubated under normoxia and hypoxia at 37°C in a humidified atmosphere. After 24 hr, the numbers of viable cells were measured in triplicate every day for 3 days using the CellTiter96™ nonradioactive cell proliferation assay kit (Promega, Madison, WI). The proliferation curves were constructed by calculating the mean value of the optical density at 430 nm using a 96-well plate reader.
Reverse transcription-polymerase chain reaction assay
The PK8 pancreatic cancer cells and MRC5 fibroblasts were incubated under normoxia and hypoxia up to 24 hr and total RNA was isolated from each cell line using an ISOGEN® RNA extraction kit (Nippongene, Toyama, Japan). RT-PCR was carried out using the RNA LA PCR Kit (AMV) Version 1.1 (Takara Biochemicals, Shiga, Japan). Samples of RNA (1 μg) were converted into cDNA reverse transcriptase. Amplification by PCR for 21 candidate genes (Table I) was then carried out according to the manufacturer's instructions. The amplification was composed of 1 cycle of denaturing at 94°C for 2 min, 25–36 cycles of denaturing at 94°C for 30 sec, annealing at 60°C for 30 sec, extension at 72°C for 90 sec and 1 cycle of extension at 72°C for 10 min. The primer sequences, the fragment size and the number of cycles for amplification of each candidate gene are shown in Table I. Regarding the MRC5 fibroblasts, genes marked by an asterisk were examined.
|Gene||Sequence||Fragment size (bp)||cycle|
|Invasion candidate genes|
|Motility candidate genes|
|Positive control genes|
Quantitative reverse transcription-polymerase chain reaction assay
To precisely estimate the mRNAs of MMP-2, MMP-7, MT1-MMP, c-Met and VEGF-A in PK8 cells and the mRNAs of HGF and VEGF-A in MRC5 cells, quantitative RT-PCR was performed on a Light-Cycler™ instrument system (Roche, Mannheim, Germany) using the Light-Cycler-FastStart DNA Master™ SYBR green I Kit (Roche, Mannheim, Germany) according to the manufacturer's instructions. The primer used for each gene was the same as shown in Table I. After a denaturing step at 95°C for 3 min, PCR amplification was performed with 50 cycles of 15 sec denaturing at 95°C, 5 sec annealing at 60°C and 10 sec extension at 72°C. Melting curves were obtained according to the protocol under the following conditions: 0 sec denaturation period at 95°C, starting temperature of 65°C, ending temperature of 95°C and a rate of temperature increase of 0.1°C/sec. The quantitative value was normalized by β-actin expression used as an internal control. These experiments were carried out in triplicate and then independently repeated at least 3 times.
Western blot analysis and immunoprecipitation
The PK8 and MRC5 cells cultured under normoxia and hypoxia were lysed in lysis buffer composed of 150 mM NaCl, 50 mM Tris-HCl (pH 7.6), 0.5% Triton X-100 and a protease inhibitor cocktail mix (Roche Diagnostics GmbH, Mannheim, Germany). For the Western blot analysis of HIF-1α, the cells were treated with 100 μM CoCl2, which is reported to induce HIF-1α expression, as a positive control. Aliquots of each cell extract containing 30 μg of protein were subjected to 4–12% SDS-PAGE and the separated extracts were electrophoretically transferred onto Hybond™ nitrocellulose enhanced chemiluminescence membranes (Amersham Pharmacia Biotech, Buckinghamshire, UK) in a transfer buffer. Primary antibodies used in the Western blot analysis were anti-MMP-2 anti body (Clone H-76, 1:100 dilution; Santa Cruz Biotechnology, Santa Cruz, CA), anti-MMP-7 antibody (Clone M-8683, 1:500 dilution; Sigma St. Louis, MO), anti-MT1-MMP antibody (Clone M-5808, 1:100 dilution; Sigma), anti-c-Met antibody (Clone C-28, 1:200 dilution; Santa Cruz), anti-HIF-1α antibody (Clone HI-67, 1:1,000 dilution; Novus Biologicals, Littleton, CO), anti-HGF antibody (Clone BAF294, 1:500 dilution; R&D Systems, Oxford, UK) and anti-β- actin antibody (Clone AC-15, 1:10,000 dilution; Sigma). After incubation with the corresponding secondary antibodies, the signals were finally developed using an enhanced chemiluminescence detection kit (Amersham Pharmacia Biotech). To detect HGF expression in culture supernatants, 100 μl of protein A-agarose beads along with 5 μg/ml of anti-human HGF capture antibody (Clone 24516, R&D Systems) was added to 10 ml of NCM or HCM and incubated for 12 hr at 4°C. Next, the supernatants were discarded by centrifugation at 2,000 rpm and then the beads were washed, followed by boiling at 95°C for elution. Finally, the eluted samples were subjected to a Western blot analysis. For immunoprecipitation, cell lysates were prepared in lysis buffer composed of 20 mM Tris-HCl (pH 7.4), 5 mM EDTA, 10 mM Na4P2O7, 100 mM NaF, 2 mM Na3VO4, 1% NP-40 and protease inhibitor cocktail mix (Roche Diagnostics GmbH). Immunoprecipitation was performed using 500 μg of cell extracts incubated with anti-c-Met polyclonal antibody (clone C-28, Santa Cruz) with 20 μl of protein A-agarose beads (Sigma). After the beads were washed and boiled at 95°C for elution, the sample was subsequently subjected to a Western blot analysis using anti-c-Met monoclonal antibody (Clone B-2, 1:100 dilution; Santa Cruz) and anti-p-Tyr antibody (Clone PY99, 1:1,000 dilution; Santa Cruz).
Enzyme-linked immunosorbent assay
An enzyme-linked immunosorbent assay (ELISA) kit for HGF was purchased from R&D Systems (Oxford, UK). The concentrations of HGF in NCM and HCM from MRC5 fibroblasts were determined by an optical density at 450 nm using a 96-well plate reader, according to the manufacturer's instructions. These experiments were carried out in duplicate and were independently repeated at least 3 times.
Removal of hepatocyte growth factor from the fibroblast-conditioned medium
For the removal of HGF, 100 μl of protein A-agarose beads along with 5 μg/ml of anti-human HGF antibody (R&D Systems, Oxford, UK) or normal rabbit serum was added to 10 ml of HCM and incubated for 12 hr at 4°C. The supernatants were collected by centrifugation at 2,000 rpm and designated as HCM pretreated (+) or not pretreated (−) by an HGF antibody column. These supernatants were used in the immunoprecipitation of c-Met, a western blot analysis of p-Tyr c-Met, zymography of MMP-2 and in an in vitro invasion assay.
Prior to zymography, we prepared serum-free NCM and HCM from MRC5 fibroblasts. Subconfluent PK8 cells were washed 3 times with phosphate-buffered saline and cultured under normoxia and hypoxia for 24 hr in serum-free NCM and HCM that had been pretreated or not pretreated by an HGF antibody column. After 24 hr, the culture medium was collected and concentrated 10-fold using Centricon-10™ concentrators (Amicon, Beverly, MA). For the measurement of gelatinase activity, samples were mixed with SDS sample buffer, and subjected to SDS-PAGE, using 10% polyacrylamide gel containing 1 mg/ml gelatin. After electrophoresis, the gel was washed in 2.5% Triton X-100 and incubated in 50 mM Tris-HCl buffer (pH 8.0) containing 0.5 mM CaCl2 and 1 mM ZnCl2 for 12 hr at 37°C. The gel was thereafter stained with 1% Coomassie brilliant blue and photographed.
Thirty-eight patients, between January 1994 and April 2005, with pancreatic cancer underwent resection at the Department of Surgery, Saga University Faculty of Medicine, Japan, and a histological examination confirmed the diagnosis of ductal adenocarcinoma of the pancreas in all cases. Immunohistochemical staining was performed according to the procedures of a previous report with slight modifications. The primary anti-HIF-1α antibody (Clone HI-67, 1:500 dilution; Novus Biologicals), anti-c-Met antibody (Clone B-2, 1:100 dilution; Santa Cruz), anti-HGF antibody (Clone H55, 1:50 dilution; IBL, Gunma, Japan) was placed onto the slides and the slides were incubated at 4°C overnight. After washing in phosphate-buffered saline, the slides were incubated in a prediluted solution containing biotinylated anti-mouse, anti-rabbit antibody conjugated to a peroxidase-labeled dextran polymer (Dako EnVision+™, Carpinteria, CA) for 30 min at room temperature. The slides were then washed in phosphate-buffered saline, followed by incubation for 3 min at room temperature in a prediluted chromogen solution from a liquid DAB (3,3-diaminobenzidine) substrate kit (Nichirei Co., Tokyo, Japan). The immunoreactivity for HIF-1α was estimated as follows: −;, no staining; +, nuclear staining in up to 10% of tumor cells and/or with cytoplasmic staining; ++, nuclear staining 10–30% of tumor cells with cytoplasmic staining; +++, nuclear staining in >30% of tumor cells with or without cytoplasmic staining. Tumors with ++ and +++ staining were considered to show a positive HIF-1α expression in cancer cells. The immunoreactivity for c-Met and HGF in cancer cells was classified as follows: negative; cytoplasmic staining in <30% of tumor cells, positive; cytoplasmic staining in >30% of tumor cells. When weak staining of HIF-1α or HGF was observed in stromal fibroblasts adjacent to cancer cells, the stromal expression of these proteins was thus judged to be positive. The results of immunohistochemistry were evaluated independently by 3 investigators (T.I., Y.K., D.M.) without any prior knowledge of the clinical and pathological data.
The values were expressed as the mean ± SD. Comparisons between the 2 groups were analyzed by Student's t-test and Fisher's exact test. p values less than 0.05 were considered to be statistically significant.
Effects on the invasive activity and proliferation of pancreatic cancer cells by hypoxia
As shown in Figure 1a, the invasive activity of PK8 pancreatic cancer cells was significantly increased under hypoxia, compared with normoxia. We further analyzed the invasive activity of PK8 cells under hypoxia, where 2 types of conditioned media from MRC5 fibroblasts cultured under normoxia and hypoxia were placed in the lower compartments of the Transwell™ chemotaxis chambers. As a result, the invasive activity of PK8 cells was the highest when PK8 cells under hypoxia were treated with HCM (Fig. 1a). Conversely, there was no significant difference in the cell proliferation of PK8 cells between normoxic and hypoxic conditions (Fig. 1b). Two other pancreatic cancer cells (PK1, KP4) consistently showed the same results (data not shown).
Expression of HIF-1α under hypoxia
A Western blot analysis was performed to examine the induction of HIF-1α protein by hypoxic stimulation using pancreatic cancer cells and fibroblasts. As shown in Figure 2, HIF-1α protein was not expressed under 20% O2 condition (normoxia) in PK8 pancreatic cancer cells or in MRC5 fibroblasts. However, HIF-1α protein in both of the cell lines was induced by exposure to 1% O2 (hypoxia) for 4–24 hr. The same effects were confirmed in PK1 and KP4 cells (data not shown).
Examination of candidate genes relating to invasive activity affected by hypoxia in pancreatic cancer cells
To identify the genes involved in the elevated invasion of PK8 cells under hypoxia, we evaluated an alteration in gene expression between normoxia and hypoxia among the candidate gene profiles, which had reportedly been related with cancer invasion and motility (Table I). An analysis by RT-PCR was carried out using primer sets for the corresponding genes. Among the invasion candidate genes, the mRNAs of matrix metalloproteinase family members MMP-2, MMP-7 and MT1-MMP were increased under hypoxia compared with normoxia (data not shown). The mRNA levels of these genes were also quantitatively estimated with a Light-Cycler-assisted approach using the SYBR green I system (Fig. 3a). There were no differences in the other invasion candidates E-cadherin, occludin, snail, SIP1, ETS-1, MMP-1, MMP-9, TIMP-2, uPA, uPAR and TGFβ1 (data not shown). For the motility candidate genes, the expression of c-Met mRNA was remarkably elevated under hypoxia (data not shown). In a quantitative RT-PCR, the c-Met mRNA level was elevated in a time-dependent fashion under hypoxia (Fig. 3a). However, there were no differences between normoxia and hypoxia in the expression of other motility candidates RhoA, CD44, integrinβ4 and integrinα4. No expression of HGF and mts1 mRNAs was found in PK8 cells (data not shown). The expression of VEGF-A, which is known to be a target for HIF-1α, increased under hypoxia (Fig. 3a). Regarding another HIF-1α-controlled gene, erythropoietin mRNA was also elevated under hypoxia (data not shown). We performed a Western blot analysis to confirm the gene expressions on the protein level. The protein expressions of MMP-2, MMP-7, MT1-MMP and c-Met were elevated under hypoxia, compared with normoxia (Fig. 3b). Regarding MMP-2, the signal of the active form at 62 kDa (arrowhead) was more strongly visible, compared with the signal of the precursor form (upper band: arrow).
HGF expression and production in fibroblast is affected by hypoxia
The in vitro invasion assay suggested that hypoxia increased the secretion of some factor from MRC5 fibroblasts and acted on accelerating invasion of pancreatic cancer cells. An analysis by RT-PCR of growth factors HGF, VEGF and TGFβ1 was examined using total RNA from MRC5 fibroblasts (Table I). Not only HIF-1α-controlled gene VEGF, but also c-Met ligand, HGF mRNAs were increased under hypoxia (data not shown). Regarding quantification, the levels of these mRNAs were also elevated under hypoxia (Fig. 4a). The expression of TGFβ1 did not change by hypoxic stimulation in the MRC5 cells (data not shown). We next performed ELISA to measure the concentration of HGF in NCM and HCM from MRC5 fibroblasts after incubation for 4, 8, 12 and 24 hr. A marked increase in the HGF concentration was observed in HCM relative to that in NCM (Fig. 4b). The difference at 24 hr was statistically significant (p < 0.05). In Western blot analysis, not only pro-HGF (92 kDa) but also mature HGF (62 kDa) was expressed higher in HCM than NCM (Fig. 4c).
Phosphorylation of c-Met protein under hypoxia
To analyze whether the HGF/c-Met pathway is activated under hypoxia, we investigated the tyrosine phosphorylation of c-Met. Immunoprecipitation with anti-c-Met antibody, followed by a Western blot analysis with anti-phosphotyrosine antibody (PY-99) was performed using lysate from PK8 cells. The expression level and phosphorylation of c-Met protein from PK8 cells that were cultured with NCM were analyzed under normoxic and hypoxic conditions. In addition, PK8 cells in HCM were cultured under hypoxia and the c-Met status was evaluated. As a result, the expression and tyrosine phosphorylation of c-Met in PK8 cells were increased by hypoxic stimulation (Fig. 5). Interestingly, the degree of phosphorylation showed marked elevation when PK8 cells under hypoxia were cultured in HCM from hypoxic MRC5.
Effect on the invasive activity and tyrosine phosphorylation of c-Met protein by removing hepatocyte growth factor from the hypoxic fibroblast-conditioned medium
To determine whether the activation of the HGF/c-Met pathway plays an important role in the elevated invasiveness of PK8 cells under hypoxia, an invasion assay was carried out using HCM with or without pretreatment on an HGF antibody column. As shown in Figure 6a, the invasive activity of PK8 cells dramatically decreased when PK8 cells were cultured in HCM pretreated on an HGF antibody column. Furthermore, tyrosine phosphorylation of c-Met protein was completely abolished in the case of HCM pretreated on an HGF antibody column (Fig. 6b). These results indicated that the removal of HGF from HCM resulted in the suppression of c-Met phosphorylation, which thus led to a decrease in cancer invasion of PK8 cells under hypoxia.
Effect of hepatocyte growth factor signal on MMP-2 activity in PK8 cells under hypoxia
We analyzed the effect of hypoxia on the activity of MMP from PK8 cells by measuring the gelatinase activity. For gelatin zymograms, hypoxic supernatant was prepared from PK8 cells cultured under hypoxia with serum-free NCM or HCM. The PK8 cell line was cultured under normoxic condition with NCM and the supernatant was used for a control. In the control supernatant, a positive signal for MMP-2 was weakly observed (Fig. 7). The activities of MMP-2 moderately increased when PK8 cells were cultured with NCM under hypoxia. When hypoxic PK8 cells were cultured with HCM, the MMP-2 activity was strongly increased (Fig. 7). Moreover, we demonstrated the removal of HGF from HCM by an HGF antibody column to reduce the MMP-2 activity from hypoxic PK8 cells (Fig. 7). This finding indicated that HGF secretion from MRC5 fibroblasts further augments MMP-2 activities from PK8 cells cultured under hypoxic conditions.
Hypoxic environment exists within cancer and stromal cells in pancreatic cancer tissues
To reveal whether a hypoxic environment exists within cancer and stromal cells in pancreatic cancer tissue specimens, the expression of hypoxic marker HIF-1α was immunohistochemically evaluated using 38 cancer tissue sections. The expression pattern of HIF-1α protein in cancer cells was divided into 3 groups: nuclear (the staining was predominantly observed in the nuclei) (Fig. 8a), cytoplasmic (mainly stained in the cytoplasm) (Fig. 8b) and no staining (the staining was hardly observed). HIF-1α is known to be a transcription factor that exerts DNA binding activity in cell nuclei. In fact, the nuclear staining was observed in PK8 and MRC5 cells under hypoxic stimuli (data not shown). Higher nuclear staining was estimated as positive for HIF-1α expression as described in Materials and Methods. In cancer lesions, a positive HIF-1α expression was found in 14 of 38 (36.8%) tumor tissue specimens. In the stroma including fibroblasts, it was found in 12 of 38 (31.6%) tissue specimens (Fig. 8c). In Fisher's exact test, a significant correlation between HIF-1α immunoreactivity in cancer and stromal cells was thus observed (p < 0.05) (Table II), thus indicating the concomitant expression of HIF-1α in cancer and stromal cells within pancreatic cancer tissues. We next performed immunohistochemistry for c-Met and HGF in the 38 tissue sections. The expression of c-Met was observed in cytoplasmic and/or membrane of cancer cells (Fig. 8d). Cytoplasmic staining of HGF was seen in both cancer and surrounding stromal cells (Figs. 8e and 8f). Immunoreactivity for c-Met and HGF in cancer cells was observed in 17 (44.7%) and 27 (71.1%) of 38 cases, respectively (Table III). The expression of HIF-1α did not reveal any significant correlation with either c-Met or HGF in the cancer cells. However, the trend of positive correlation between HIF-1α and c-Met expression was found (p = 0.23). On the other hand, stromal cells adjacent to the cancer cells were positive for HGF in 18 (47.3%) of 38 cases (Table IV). A significant correlation between HIF-1α and HGF immunoreactivity in stromal cells was observed (p < 0.01). When the c-Met expression in cancer cells was finally compared with HGF in cancer and stromal cells, a positive correlation between HGF and c-Met was significantly observed in cancer as well as stromal cells (Table V).
|HIF-1||c-Met||p value||HGF||p value|
Several studies have recently revealed that a hypoxic condition might select for more malignant cell clones.21 Koong et al. intraoperatively measured tumor oxygenation in cases of pancreatic cancer and revealed that a significant population of tumor cells is hypoxic.12 Besides, pancreatic cancer often has abundant stromas and the tumor–stromal cell interactions play a critical role in tumor progression.16, 22 In the present study, we first hypothesized that a hypoxic environment might thus be present in cancer and stromal cells within pancreatic cancer tissue and that the tumor–stromal cell interaction under hypoxia might increase the malignant behavior of this cancer. The biological interaction between stromal and cancer cells under hypoxia has not yet been studied so far.
To verify this hypothesis, we established a unique culture system for studying whether some soluble factor from fibroblasts increase under hypoxia, thus affecting the invasiveness of pancreatic cancer, which is also under hypoxic conditions. At first, we carried out an in vitro invasion assay to investigate the effect of hypoxia on the invasive activity of PK8 cells. The invasive activity of PK8 cells increased significantly under hypoxia. Using other cancer cell lines, several reports have also demonstrated the increased cancer invasion under hypoxia.23 The most important finding of our study is that PK8 cells under hypoxia showed a higher invasive property when the cells were cultured with HCM in comparison with NCM from MRC5 fibroblasts. This result indicated that elevated secretion of some soluble factor occurred in hypoxic fibroblasts and subsequently conveyed further invasiveness to hypoxic PK8 cells.
To elucidate which genes were involved in cancer invasion under hypoxia, RT-PCR was performed to analyze the expressions of 21 candidate genes between normoxic and hypoxic conditions. The candidate genes in Table I were previously described regarding their relationship with cancer invasion and motility.2, 24, 25 Among these candidates, the expressions of MMP-2, MMP-7, MT1-MMP and c-Met were significantly upregulated in hypoxic PK8 cells at the mRNA and protein level. On the other hand, in MRC5 cells, we found that the expression of HGF mRNA was elevated under hypoxia. In ELISA, a higher concentration of HGF was observed in HCM compared with NCM. A Western blot analysis further revealed a higher expression of pro and mature type HGF in HCM than NCM. We also confirmed the induction of HIF-1α expression in MRC5 as well as PK8 cells under hypoxia. Together, it is possible that MMP-2, MMP-7, MT1-MMP and c-Met in PK8 cells and HGF in MRC5 cells might be upregulated by HIF-1α when the cells are cultured under hypoxia. HIF-1α has been reported to directly upregulate various genes by binding to the hypoxia-responsive element on the promoter. The expressions of MMP-2 and c-Met are demonstrated to be directly upregulated by HIF-1α in human cancers.8, 26 The MT1-MMP mRNA level was also reported to increase within a few hours of exposure to hypoxia.27 Petrella et al. showed the direct regulation of MT1-MMP by HIF-2α using metastatic renal-cell carcinoma.28 These reports support our findings of MMP-2, MT1-MMP and c-Met elevation in hypoxic PK8 cells. We first demonstrated that hypoxia elevated HGF in MRC5 fibroblasts as well as MMP-7 expression in PK8 cells. Especially, gene expression of mesenchymal cells under hypoxia has been rarely investigated. Two reports have demonstrated the controversial aspects that hypoxia mediates upregulation or downregulation of the HGF expression in aortic smooth muscle cells.29, 30 Thus it might be noteworthy aspect that HGF secretion from fibroblasts was stimulated under hypoxia. Although we also speculated that the expression of HGF is possibly regulated by HIF-1α. HGF was reported to stimulate HIF-1 activity in HCC cell lines,31 showing an inverted signaling pathway from our scenario. At present, the direct regulation of HGF by HIF-1α remains unclear.
Hepatocyte growth factor (HGF), which was originally identified as a potent mitogen for hepatocytes, is a stromal cell-derived cytokine that induces a spectrum of biological activities, including mitogenesis, motogenesis, morphogenesis, angiogenesis and the promotion of anti-apoptosis.32, 33, 34, 35, 36 The multiple effects of HGF are mediated through binding to the cognate receptor, c-Met receptor tyrosine kinase, which is generally known the expression on cell surfaces of epithelial origin. Various reports have implicated the HGF/c-Met system in cancer progression through a tumor–stromal cell interaction.37 Qian et al. sketched out the interaction between cancerous and stromal cell compartments with an emphasis on the HGF/c-Met pathway in pancreatic cancer using cocultivation of cancer cells with orthotopic tumor-derived fibroblasts.38 In the present study, we focused on the elevations of both HGF and c-Met expressions under hypoxia. Regarding the activation of HGF/c-Met signal, the degree of c-Met tyrosine phosphorylation was significantly increased in hypoxic PK8 cells; furthermore, the degree of phosphorylation was more enhanced when hypoxic PK8 cells were cultured with HCM. In gelatin zymography, hypoxia increased MMP-2 activity of PK8 cells, and the activity was further augmented by HCM treatment. Conversely, the elevated cancer invasion, c-Met phosphorylation and MMP-2 activity were dramatically reduced by the removal of HGF from HCM. These results strongly suggest that the activity of MMP-2 from PK8 cells is regulated by HGF/c-Met signaling. However, the invasion and MMP-2 activity were not completely abolished by interrupting this signaling, and another pathway except HGF/c-Met might also be important for the invasion of PK8 cells. Under hypoxia, the activated HGF/c-Met system causes MMP activation, thus leading to an acceleration of the invasive features of pancreatic cancer cells. Date et al. previously showed that HGF stimulated MMP activity in gallbladder cancer cells.39 Recent studies also revealed that HGF promoted cancer invasion by activating MMP.40, 41 These studies support our findings, although the experiments were conducted under normoxic conditions.
Another important finding in our study is that hypoxia exists within stromal cells as well as cancer cells in pancreatic cancer tissue. We analyzed the HIF-1α expression immunohistochemically using 38 pancreatic cancer tissue specimens. In more than 30% of the cancer samples, HIF-1α positive staining was found in both pancreatic and surrounding stromal cells. In addition, a significant correlation in the HIF-1α expression was observed between stromal and cancer cells. This result suggests that a hypoxic environment concomitantly exists in cancer and surrounding stromal cells. In previous reports, nuclear expression of HIF-1α was frequently demonstrated in tumor cells.25 Other reports also suggested correlation between HIF-1α expression and hypoxic status in cancer cells.42, 43 Current studies demonstrated that the expression of HIF-1α was also detected in some stromal cells within the tumor.44, 45 These reports support our notion that both cancer and stromal cells with HIF-1α expression are possibly exposed to hypoxia. On the other hand, the expression of HGF/c-Met previously revealed an association with a poor prognosis in the patients with meningioma, lympoma, and high grade salivary gland carcinoma.46, 47, 48 We subsequently analyzed HGF/c-Met expression in the pancreatic cancer tissues and investigate the correlation between HIF-1α and HGF/c-Met expressions. As a result, we demonstrated a significant association between the HIF-1α and HGF expressions in the stromal regions of 38 pancreatic carcinoma tissue specimens (p < 0.01). This association might support our in vitro data that hypoxic fibroblast cells increased HGF production, which led to the activation of c-Met on hypoxic cancer cells. When HIF-1α expression was compared with HGF and c-Met expression in cancer cells, we found the trend of positive correlation between HIF-1α and c-Met expression, thus suggesting our in vitro finding of c-Met upregulation by HIF-1α. However, the correlation was not statistically significant (p = 0.23). The upregulation of c-Met expression might not be induced by HIF-1α alone in pancreatic cancer tissues. We finally observed that c-Met expression in cancer lesion was significantly correlated with HGF in cancer as well as stromal cells. In pancreatic cancer tissue specimens, HGF might activate c-Met on cancer cells, which is mediated by both paracrine and autocrine system.
In conclusion, our findings provide the novel insight that a hypoxic environment is present in cancer and stromal cells within pancreatic cancer. The tumor–stromal cell interaction under hypoxia thus increases the malignant behavior of this cancer, where a paracrine loop between HGF/c-Met plays a crucial role. Effective therapy targeting the HGF/c-Met signal or HIF-1α is therefore desired to overcome the aggressive invasion of pancreatic cancer in a hypoxic environment.
We thank Dr. D. Mori, Dr. K. Kai and Dr. K. Tsukinoki for the advice on immunohistochemical analysis.
- 16Histological and fine structural features of pancreatic ductal adenocarcinomas in relation to growth and prognosis: studies in xenografted tumours and clinico-histopathological correlation in a series of 75 cases. Histopathology 1985; 9: 841–56., , , .
- 41Effects of hepatocyte growth factor on the expression of matrix metalloproteinases and their tissue inhibitors during the endometrial cancer invasion in a three-dimensional coculture. Int J Gynecol Cancer 2003; 13: 53–60., , , , , .