Background and aims Metastasis accounts for the poor outcome of patients with pancreatic cancer. We recently discovered PRSS3 to be over-expressed in metastatic human pancreatic cancer cells. This study aimed to elucidate the role of PRSS3 in the growth and metastasis of human pancreatic cancer.
Methods PRSS3 expression in human pancreatic cancer cell lines was detected by qPCR and immunoblotting. The effect of PRSS3 on cancer cell proliferation, migration and invasion in vitro, tumour growth and metastasis in vivo were investigated by manipulation of PRSS3 expression in human pancreatic cancer cell lines. VEGF expression was detected by ELISA, and the pathway through which PRSS3 regulates VEGF expression was investigated. The therapeutic effect of targeting this pathway on metastasis was assessed in vivo. Immunohistochemistry was employed to detect PRSS3 expression in human pancreatic cancer tissues.
Results PRSS3 was over-expressed in the metastatic PaTu8988s cell line, but not in the non-metastatic PaTu8988t cell line. Over-expression of PRSS3 promoted pancreatic cancer cell proliferation as well as invasion in vitro, and tumour progression and metastasis in vivo. Stepwise investigations demonstrated that PRSS3 upregulates VEGF expression via the PAR1-mediated ERK pathway. ERK inhibitor significantly delayed the progression of metastases of pancreatic cancer and prolonged the survival of animals bearing metastatic pancreatic cancer (p<0.05). 40.54% of human pancreatic cancers (n=74) were positive for PRSS3 protein. A significant correlation was observed between PRSS3 expression and metastasis (p<0.01). Multivariate Cox regression analysis indicated that patients with PRSS3 expression in their tumours had a shorter survival time compared to those without PRSS3 expression (p<0.05).
Conclusion PRSS3 plays an important role in the progression, metastasis and prognosis of human pancreatic cancer. Targeting the PRSS3 signalling pathway may be an effective and feasible approach for treatment of this lethal cancer.
- Pancreatic cancer
- gene expression
- gene targeting
- pancreatic tumours
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
Pancreatic cancer is one of the most aggressive human cancers
Many pancreatic cancer patients at diagnosis are already in well-advanced stages of local invasion and metastasis, and even small (<2 cm diameter) pancreatic cancers are associated with remote metastases
Metastasis leads to a dismal median survival of 3–6 months for pancreatic cancer patients
PRSS3 has been reported to be involved in human cancer; however, the role of PRSS3 expression in human pancreatic cancer remained unknown before this study.
What are the new findings?
PRSS3 is over-expressed in human pancreatic cancer cell lines and promotes cell proliferation as well as invasion in vitro, and tumour progression as well as metastasis in vivo
PRSS3 upregulates VEGF expression via the PAR1-mediated ERK pathway in human pancreatic cancer
Blockade of ERK signalling triggered by the PRSS3 pathway can delay the progression of metastasis of human pancreatic cancer and increase the survival of animals bearing human pancreatic cancer in vivo
Over-expression of PRSS3 in human primary pancreatic cancer tissues is associated with metastasis and poor prognosis of patients with pancreatic cancer.
How might it impact on clinical practice in the foreseeable future?
PRSS3 may be a useful biomarker to predict metastasis and prognosis of human pancreatic cancer. Targeting the downstream signalling pathway instead of PRSS3 itself should be considered as a feasible and effective approach for treatment of pancreatic cancer.
Pancreatic cancer is one of the most aggressive human cancers. It is currently the fourth leading cause of cancer death in the USA and the fifth leading cause of cancer death in Europe.1 2 Poor survival is likely because only 15% of patients have a potentially resectable tumour when diagnosed, and even in operable cases 16% of patients have macroscopically incomplete resections.3 Moreover, even if the primary cancer is radically removed, local recurrence occurs in 50–80% of all patients within 1 year.3 4 In fact, many cases at diagnosis are already in well-advanced stages of local invasion and metastasis, and even small (<2 cm diameter) pancreatic cancers are associated with remote metastases.5 This leads to a dismal median survival of 3–6 months.3 6 Therefore, a better understanding of the factors that contribute to pancreatic cancer progression and metastasis is critical before we can proceed to conquer this disease.
Tumour metastasis has been revealed as an exceedingly complex process, which occurs through a series of sequential steps including dissemination of malignant cells, intravasation, transport via the circulatory system, extravasation, settlement and growth of metastatic tumour cells in distant organs.7 8 At the molecular level, an increasing number of genetic and epigenetic alterations have been discovered, with a particular focus on growth factors and related pathways. For example, VEGF has been thought to play a pivotal role by promoting the proliferation, survival and migration of endothelial cells as well as cancer metastasis.9 Over-expression of the vascular endothelial growth factor/receptor system (VEGF/VEGFR) has been described in pancreatic cancer in association with increased microvessel density and liver metastasis.10 However, the exact mechanism of action is only partly understood. In contrast to early pancreatic carcinogenesis, the molecular features of advanced stage pancreatic cancers are relatively unexplored,11 although molecular alterations have been identified that enable the cancer cells to invade the perineurium and the retroperitoneal space, thus explaining at least, in part, the high rate of local recurrence.12 A better understanding of the molecular features of metastatic pancreatic cancer will provide new avenues for therapeutic intervention. To this end, gene expression profiles of advanced stage pancreatic cancers associated with locally destructive behaviour and/or metastatic spread have been performed by Campagna et al.13 Their comparison of primary carcinomas and their matched metastases failed to show commonly deregulated genes. Given the lack of availability of high-quality metastatic cancer tissues for expression profiling, cancer cell lines that have differential potentials for metastasis might represent a tractable model to discover metastasis-associated genes in human pancreatic cancer. Two human pancreatic cancer cell lines derived from the same patient (PaTu8988t and PaTu8988s) are attractive for this purpose because PaTu8988s develops metastases localised exclusively in the lung after intravenous injection into the tail vein, whereas PaTu8988t produces no metastasis in any organ.14 We recently examined the differential gene expression between these two cell lines by Affymetrix array analysis.15 Interestingly, we found that PRSS3 is over-expressed in the metastatic PaTu8988s cell line, but not in the non-metastatic PaTu8988t cell line. In the present study, we built upon our previous finding and demonstrated that PRSS3 promotes tumour cell growth and metastases by upregulation of VEGF via the PAR1-mediated extracellular signal-related kinase (ERK) pathway. Treatment with ERK inhibitor can delay the progression of metastases and prolong the survival of animals bearing metastases of pancreatic cancer cells. Finally, we found that that PRSS3 expression is associated with metastasis and poor prognosis of human pancreatic cancer.
Materials and methods
The study population consisted of 74 patients with pancreatic cancer; 58 cases were from the First Teaching Hospital of Zhejiang University, China and 16 cases of primary pancreatic and matched lymph node metastases were obtained from the Department of Pathology, General Hospital Osijek, Croatia. The mean age was 58 years (range, 32–79 years). None of the patients in this study had undergone chemotherapy, radiotherapy or immunomodulatory therapy before surgery. The characteristics of the study population are shown in table 1.
All surgical specimens were immediately fixed in 10% neutral-buffered formalin. Routine pathological diagnosis was made by light microscopic examination of haematoxylin and eosin (H&E)-stained sections. Histological classification and staging of pancreatic cancer were performed using the TNM criteria of the International Union Against Cancer and Guidelines for the Clinical and Pathological studies on pancreatic cancer. All the specimens were obtained with full ethical approval from the host institutions.
Cell lines and cell culture
The human pancreatic cancer cell lines PaTu8988s, PaTu8988t, Capan1 and MiaPaca2 were obtained from the Cancer Research UK Central Cell Service (Clare Hall Laboratories, Potters Bar, UK). All cell lines were grown at 37°C with 5% CO2 and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 50 μg/ml streptomycin, and 50 μg/ml penicillin.
Construction of plasmid expressing human PRSS3
Total RNA was extracted from PaTu8988s cells using TRIzol reagent (Invitrogen, Paisley, UK) according to the manufacture's protocol. The cDNAs were synthesised using a MultiScribe kit (Applied Biosystems, Foster City, California, USA) with random hexamers. PRSS3 cDNA was cloned by PCR using 300 ng cDNA as template with the following primers containing HindIII and XbaI restriction enzyme sites: 5′-TAGAAGCTTATGAATCCATTCCTGATCCTTGCCTTTGTG-3′ (forward), 5′-TAGTCTAGAACCGGTACGGCTGTTGGCAGCGATGGTGTC-3′ (reverse), corresponding to the sequence 14-757 in human PRSS3 cDNA (sequence identification number NM_002771 in GenBank). The PCR products were ligated to pcDNA3.1 (Invitrogen) to create the plasmid pcDNA-PRSS3.
Establishment of PRSS3-expressing stable cell lines
PaTu8988t and MiaPaca2 cells were transfected with pcDNA-PRSS3 and pcDNA3.1 vector respectively, using Effectene reagent (QIAGEN, Hilden, Germany) according to the manufacturer's protocol. The stable cell lines, PaTu8988t-PRSS3, MiaPaca2-PRSS3 and their counterparts, PaTu8988t-pcDNA3.1 and MiaPaca2-pcDNA3.1 were established following selection with 1200 μg/ml (PaTu8988t) or 800 μg/ml (MiaPaca2) of G418 (Invitrogen) and maintained with 600 μg/ml G418.
Cell invasion assay
The in vitro cell invasion assay was performed with 24-well plate Matrigel chambers (BD Biosciences, Franklin, Laskes, NJ, USA, 8 μm) according to the manufacturer's procedure.
Real-time PCR for quantification of PRSS3 expression
qPCR was carried out using the Applied Biosystems 7500 Real-Time PCR System and the TaqMan 5′ nuclease assay (Applied Biosystems), as used previously except primers and probes were different (Hs00366967_m1 for PRSS3 (99-bp amplicon) and 4319413E for 18S endogenous control).15 Data were analysed using Sequence Detection Software version 1.3 (Applied Biosystems). The relative expression was normalised to PaTu8988t values.
Enzyme-linked immunosorbent assay for detection of VEGF
VEGF protein in the supernatant medium of different cells after various treatments were detected by ELISA assay using VEGF human Quantikine colorimetric sandwich ELISA (R&D Systems, Minneapolis, MN, USA) in accordance with the manufacturer's instructions.
Immunohistochemistry was performed on 4 μm thick paraffin-embedded tissue sections as previously described.15 Briefly, the processed sections were incubated with either primary Goat anti-PRSS3 (R&D Systems, at 1:50 dilution (1 mg/ml), or mouse anti-VEGF antibodies (Santa Cruz Biotechnology, Santa Cruz, California, USA) at 1:50 dilution (1 mg/ml). The biotinylated donkey anti-goat IgG or linker antibody (rabbit monoclonal, anti-mouse IgG1 and IgG2 Fc-specific, from Epitomics, (Burlingame, CA, USA) was used as the secondary antibody. The DAB Detection Kit was used according to the protocols provided for the Ventana Molecular Discovery System (Illkirch, France). The slides were counterstained with haematoxylin. Negative control experiments were performed by replacing the primary antibodies with isotype serum. All sections were analysed by light microscopy. Although qualitative differences in the staining intensity were observed, all positive cases showed strong unequivocal staining in the cytoplasm. The staining scores were categorised as 0 (no staining), 1+ (cytoplasmic staining in less than 20% of cancer cells), 2+ (cytoplasmic staining in 21–50% of cancer cells), and 3+ (cytoplasmic staining in greater than 50% cancer cells).
In vivo experiment
To observe the effect of PRSS3 on tumour growth in vivo, 1×107 PaTu8988t-pcDNA3.1 or PaTu8988t-PRSS3 cells were subcutaneously implanted into one flank of 4–6-week-old BALB/c nu/nu female mice. The presence of tumours and their size (length × width2 ×π/6) were monitored twice weekly. When the animals died or were killed due to tumour burden, the tumours, draining lymph nodes, liver and lungs were harvested and then subjected to histopathological examination. For the in vivo experiment of anti-metastasis by ERK inhibitor, 1×106 of PaTu8988s cells were injected intravenously and then 25 μmol/kg of U0126 in 50 μl of DMSO was administered intraperitoneally twice weekly for 3 weeks. The control groups were treated with DMSO only. The animal survival was monitored and autopsies were performed, in particular examination of the entirety of the lungs by H&E staining.
Statistical analysis was performed using the Student two-tailed t test or as indicated. Data were represented as mean ± standard deviation (SD) or ± standard error of the mean (SEM). Differences were considered statistically significant when the p value was less then 0.05.
PRSS3 is differentially expressed between metastatic and non-metastatic pancreatic cancer cell lines
In our previous study,15 it was found that the metastatic PaTu8988s cell line showed significantly (p<0.001) higher expression of PRSS3 compared with the non-metastatic PaTu8988t cell line. In order to validate the differential expression of PRSS3 between the two cell lines, PRSS3 expression was detected by quantitative real-time PCR (qRT-PCR). As shown in figure 1A, PRSS3 was expressed 505-fold higher in PaTu8988s cells than in PaTu8988t cells. It was also over-expressed in another human pancreatic cancer cell line, Capan1, but not in MiaPaca2. The PRSS3 protein was undetectable in PaTu8988t and MiaPaca2 cells, but expressed at high levels in PaTu8988s and Capan1 cells (figure 1B), consistent with mRNA expression. In addition, PRSS3 expression was detectable from the supernatant of PaTu8988s and Capan1 cells (figure 1B), indicating that PRSS3 is secreted by human pancreatic cancer cells.
Manipulation of PRSS3 expression in pancreatic cancer cells affects tumour cell proliferation and invasion in vitro
To assess the functional significance of PRSS3 expression in pancreatic cancer cells, we investigated whether stable over-expression of PRSS3 could alter the biological behaviour of human pancreatic cancer cells. To this end, the stable PRSS3-expressing cell lines PaTu8988t-PRSS3, MiaPaca2-PRSS3 and empty vector-transfected control cell lines PaTu8988t-pcDNA3.1 and MiaPaca2-pcDNA3.1 were established by transfection with pcDNA3.1-PRSS3 or pcDNA3.1 vectors, respectively. Over-expression of PRSS3 in PaTu8988t and MiaPaca2 cells, verified by PCR (data not shown) and immunoblotting (figure 2A), led to a significantly higher cell proliferation compared to the vector-transfected cell lines PaTu8988t-pcDNA3.1 and MiaPaca2-pcDNA3.1 (figure 2B, C). To further validate the effect of PRSS3 on the proliferation of pancreatic cancer cell lines, PaTu8988s and Capan1 cells, both of which endogenously express PRSS3, were transfected with PRSS3-specific siRNAs and non-targeting control siRNAs. Downregulation of PRSS3 expression in both cell lines, as shown in figure 2D, significantly reduced the proliferation of PaTu8988s and Capan1 cells (figure 2E). To demonstrate that the observed effect was not an off-target effect of a particular siRNA oligonucleotide, we confirmed that two different oligonucleotides targeting different regions of the PRSS3 mRNA were able to decrease PRSS3 expression (data not shown). To examine whether PRSS3 also affects invasion and migration, the PRSS3-expressing cell lines, PaTu8988t-PRSS3 and PaTu8988t-pcDNA3.1, as well as Capan1 cells transfected with PRSS3-specific siRNA or non-targeting control siRNA, were used for in vitro migration and invasion assays as described in the Materials and Methods. We found that PRSS3 over-expression in pancreatic cancer cells enhanced tumour cell invasion whereas downregulation of PRSS3 expression significantly decreased cell invasion (figure 2F and supplementary figure 1). However, there was no effect of PRSS3 on cell migration (data not shown).
Manipulation of PRSS3 expression in pancreatic cancer cells affects tumour progression and metastasis in vivo
Next we investigated the effect of PRSS3 on tumour growth and metastasis in vivo. PaTu8988t-PRSS3 and PaTu8988t-pcDNA cells respectively were subcutaneously injected into the right flank of nude mice, and the tumour growth measured twice weekly. On day 105 after injection of tumour cells, all mice were killed for a comprehensive examination. On day 35 and 105, there was a significant difference (p<0.05) in tumour prevalence between animals injected with PaTu8988t-PRSS3 (50% (6/12) of mice having tumour formation at day 35, and 91.7% (10/12) of mice having tumour formation at day 105) and those with PaTu8988t-pcDNA3.1 cells (no mice having tumour formation at day 35, 33.3% (4/12) mice having tumour formation at day 105). The volumes of PaTu8988t-PRSS3 tumours were significantly higher than those of PaTu8988t-pcDNA3.1 over the time course (figure 3A). At the end of this experiment (day 105), one mouse, which was bearing PaTu8988t-PRSS3 tumour cells, had liver metastasis, while none of the mice bearing PaTu8988t-pcDNA tumour cells had metastasis. In a separate experiment over a longer time period (6 months), as expected, the PaTu8988t-PRSS3 xenografts were still significantly larger than the PaTu8988t-pcDNA3.1 xenografts over the time course (supplementary figure 2). Interestingly, after this longer time interval, a higher percentage of mice (33.3%, 2/6) bearing PaTu8988t-PRSS3 tumour cells developed liver metastases compared to the mice bearing PaTu8988t-pcDNA3.1 tumour cells (0%, 0/7) although there was no statistical difference between the two groups. Given that PRSS3 was over-expressed in PaTu8988s cells, which typically develop metastases in the lung after intravenous (IV) injection, PaTu8988s cells were employed to clarify the effect of PRSS3 on metastasis of pancreatic cancer in vivo. As shown in figure 3B, 58.33% (7/12) of mice injected intravenously with PaTu8988s cells in which PRSS3 was downregulated by PRSS3-specific siRNA developed lung metastases whereas 100% (12/12) of mice injected intravenously with PaTu8988s cells pre-transfected with control siRNA had lung metastases (p<0.05). In addition, downregulation of PRSS3 expression in PaTu8988s cells significantly reduced the number of micro-metastatic nodules in the lungs (figure 3B and supplementary figure 3). To demonstrate that these observations are not specific to PaTu8988s cells, we further tested the effect of PRSS3 on the metastasis of Capan1 cells that also endogenously express PRSS3 in an intraperitoneal (IP) model. Intraperitoneal injection of Capan1 cells treated with PRSS3-specific siRNA did not result in visible macroscopic tumour nodules in the peritoneum whereas injected Capan1 cells treated with control-siRNA formed obvious tumour nodules (supplementary figure 4A) by day 21. Most interestingly, only 37.5% (3/8) of mice injected with PRSS3-siRNA-treated Capan1 cells developed liver metastases whereas 100% (6/6) of mice injected with control-siRNA-treated Capan1 cells had liver metastases. Also, downregulation of PRSS3 expression by siRNA in Capan1 cells dramatically reduced the number of micro-metastatic nodules in the liver (figure 3C, supplementary figure 4B) compared to the control siRNA-treated Capan1 cells (p<0.001). These data from PaTu8988s and Capan1 cells strongly demonstrate that PRSS3 promotes the progression and metastasis of human pancreatic cancer in vivo.
PRSS3 upregulates VEGF expression in human pancreatic cancer cells
Given that high expression of VEGF is associated with liver metastases in human primary pancreatic cancer,10 we next investigated whether the PaTu8988t-PRSS3 tumours over-expressed VEGF. As shown in figure 3D, tumours that over-expressed PRSS3 showed very strong and uniform VEGF expression compared to the PRSS3-negative tumours. These results suggest that over-expression of PRSS3 in human pancreatic cancer cells may promote tumour proliferation, progression and metastasis by upregulation of VEGF.
To confirm that PRSS3 upregulates VEGF in human pancreatic cancer cells, expression of VEGF was examined by ELISA in PaTu8988t and MiaPaca2 cells after transfection of pcDNA-PRSS3 plasmid or empty pcDNA3.1. As shown in figure 4A, VEGF expression into the supernatant of the cells after transfection with pcDNA-PRSS3 plasmid was dramatically increased compared to empty vector-transfected cells. Given that PRSS3 is secreted from cells expressing this gene (figure 1B), we tested whether the active PRSS3 protein could upregulate VEGF expression in human pancreatic cancer cells and found that this was indeed the case and was dose-dependent. In cells incubated with 10ng/ml PRSS3 (R & D Systems, USA), the level of VEGF protein was more than 2-fold higher than in those treated with vehicle buffer (figure 4B). In addition, when 10 ng/ml PRSS3 was used to stimulate PaTu8988t cells, the increase of VEGF protein levels was time-dependent (figure 4C). We next investigated whether downregulation of PRSS3 by siRNA could decrease VEGF expression. As shown in figure 4D, VEGF expression in PaTu8988s and Capan1 cells was significantly decreased after transfection of PRSS3-specific siRNA compared to control siRNA.
PRSS3 upregulates VEGF expression by a PAR1-mediated ERK pathway
Proteinase-activated receptors (PARs), a subfamily of G protein-coupled receptors, which are activated by serine proteases, such as trypsin, play pivotal roles in the central nervous system and in cancer.16 17 Mesotrypsin, encoded by the PRSS3 gene, selectively activates PAR-1, but not PAR-2 in rat astrocytes, although there were controversial reports that PRSS3-activated cell signalling involves the cleavage of PAR1.17–20 We hypothesised that PRSS3 might activate PAR-1 and upregulate VEGF expression. To confirm this, we first examined whether PRSS3 is a co-activating factor for PAR-1. We assayed the functional activation of PAR-1 by measurement of Ca2+ flux in PaTu8988t cells (PAR1 is over-expressed in both PaTu8988t and PaTu8988s cell lines, data not shown) after treatment with active PRSS3 protein. Addition of active PRSS3 protein (10 ng/ml) to PaTu8988t cells stimulated a biphasic Ca2+ response with a sharp increase within the first 100 s, followed by a slow decrease (figure 5A). This robust Ca2+ response was completely blocked by two different function-blocking anti-PAR1 antibodies,21 but not by isoform IgGs (figure 5A, middle panel). VEGF expression in PaTu8988t cells after treatment with 10 ng/ml PRSS3 protein was also significantly reduced by pretreatment with function-blocking anti-PAR-1 antibodies (figure 5C, bottom panel). These results suggest that PRSS3 upregulates VEGF expression through the PAR-1 pathway.
Since it is known that PAR1 can signal through activation of ERK1/2,22 we next investigated whether this pathway is involved in the upregulation of VEGF after PRSS3 stimulation. As shown in figure 5B, incubation of PaTu8988t cells with active PRSS3 protein (10 ng/ml) resulted in an increase of phosphorylated ERK while total ERK was unchanged. Moreover, the PRSS3-induced ERK phosphorylation was inhibited when the cells were pre-treated with PAR1-blocking antibodies (figure 5C). When PaTu8988t cells were pre-treated with the ERK-specific inhibitor U0126, ERK phosphorylation and VEGF upregulation after treatment with active PRSS3 protein (10 ng/ml) were completely inhibited (figure 5C).
ERK-selective inhibitor delays the progression of metastases and prolongs the survival of mice bearing PRSS3-positive human pancreatic tumours
Although PRSS3 is unlikely to be tractable as a clinical target due to its universal distribution, the signalling pathway downstream of PRSS3 might be a potential therapeutic target for human pancreatic cancer. As shown in figure 1, PaTu8988s cells inherently have a very high level of PRSS3, as well as over-expressing VEGF as a consequence (data not shown, figure 4D and figure 6A). We found that the ERK-selective inhibitor U0126 dramatically reduced VEGF production in PaTu8988s cells compared to vehicle control in vitro (figure 6A). To exploit the potential of U0126 as a therapeutic agent for PRSS3-positive pancreatic cancer in vivo, 1×106 of PRSS3-expressing PaTu8988s cells, which typically develop metastases in the lung after intravenous injection, were intravenously injected into the tail vein of nude mice, and then the animals were then immediately treated with U0126 or vehicle buffer at 25 μmol/kg of U0126 in 50 μl of DMSO twice weekly for 3 weeks. As shown in figure 6B, 83% of the mice treated with vehicle buffer developed lung metastases and were dead by 85 days, whereas all mice treated with U0126 were still alive at the same time point. At the end of the experiment, 33.3% of mice treated with U0126 were still alive, despite having a few micrometastatic nodules in the lung at autopsy whereas the mice treated with DMSO showed diffuse lung metastatic nodules taking up more than 60% of the whole lung (figure 6C). This suggests that the regime of U0126 used in the present study prolonged the survival of animals by delaying the progression of metastatic tumours.
PRSS3 is over-expressed in human pancreatic cancer and associated with metastasis and shortened survival
Finally, we investigated PRSS3 expression in human pancreatic cancer tissues and its relationship with clinicopathological factors. As expected, PRSS3 was expressed in pancreatic acinar cells, but not in normal duct epithelial cells (figure 7A). Of the 74 ductal pancreatic adenocarcinomas, 30 (40.54%) were positive and 44 (59.46%) were negative for PRSS3 (figure 7B and table 1). Interestingly, the cancer cells that penetrated into blood (figure 7C) or lymph vessels (figure 7D) showed stronger PRSS3 expression. There was no significant association between PRSS3 expression and tumour stage, tumour differentiation, gender and age (table 1). However, there was a significant association between PRSS3 expression and lymph node metastases (p=0.004) as well as distant metastases (p=0.009). In this study, we had 16 cases of primary pancreatic and matched lymph node metastases. There was a good correlation of PRSS3 expression in primary tumours and metastatic tumours (supplementary figure 5). Detailed information on PRSS3 expression in 16 cases of primary pancreatic and matched lymph node metastases is shown in supplementary table 2. The survival times for 58 of the 74 cancer patients were available and were found to be inversely correlated to PRSS3 expression (figure 8). The median survival time of pancreatic cancer patients with PRSS3 expression (6 months) was significantly shorter than the patients without PRSS3 expression (12 months) (p=0.04). In an attempt to clarify the significance of PRSS3 expression related to the prognosis of human pancreatic cancer, we performed multivariate analysis. As shown in table 2, multivariate Cox regression analysis demonstrated that PRSS3 expression in primary pancreatic tumours and distant metastases were two factors significantly associated with shorter patient survival. These results support our findings that PRSS3 promotes progression as well as metastasis of human pancreatic cancer. PRSS3 may be a useful biomarker to predict the prognosis of human pancreatic cancer.
PRSS3, also named mesotrypsinogen, is one of three major isoforms of trypsinogen that is a serine protease synthesised in the acinar cells of the pancreas, and then secreted into the small intestine to facilitate the digestive processes.23–25 Unlike PRSS1 and PRSS2, which are major digestive enzymes, PRSS3 is only a minor component of pancreatic exocrine secretions, representing 3–10% of the total trypsinogen content in normal pancreatic juice.26 The biological function of human mesotrypsin is digestive degradation of trypsin inhibitors.27 An inappropriate activation of mesotrypsinogen in the pancreas might lower levels of the protective factor SPINK1 and contribute to the development of human pancreatitis. It was recently reported that trypsin IV or mesotrypsin can cleave and activate PARs, causing PAR1- and PAR2-dependent inflammation and PAR2-dependent hyperalgesia.20
There are controversial reports about the role of PRSS3 in human carcinogenesis. Diederichs et al demonstrated that expression of PRSS3 is upregulated in metastatic non-small cell lung cancer.28 However, there have been several reports that epigenetic silencing of the PRSS3 promoter by hypermethylation occurs in bladder, lung, oesophageal and gastric cancers, supporting a case for PRSS3 as a tumour suppressor gene.29–31 In the present study, we, for the first time, demonstrated that PRSS3 promotes human pancreatic cancer cell proliferation and invasion in vitro, and tumour growth as well as metastasis in vivo when cancer cells that over-express PRSS3, or PRSS3-downregulated tumour cells were subcutaneously, intravenously or intraperitoneally injected into nude mice (figure 3, table 1, supplementary figures 2, 3 and 4). Further investigations of PRSS3 expression in human primary cancer demonstrated that intermediate levels of PRSS3 expression are present in pancreatic acinar cells, whereas ductal epithelial cells did not express PRSS3 (figure 7A). However, PRSS3 was over-expressed in invasive human pancreatic cancer, in particular in the cancer cells that penetrated into blood and lymph vessels (figure 7B–D). Most importantly, PRSS3 is associated with metastases of human pancreatic cancer, but not other factors such as age, gender, tumour differentiation, nor stage etc. (table 2, supplementary figure 5 and supplementary table 2). Most strikingly, the PRSS3 expression in tumour tissues is inversely correlated to the survival of cancer patients with pancreatic cancer (figure 8). These findings indicate an important role for PRSS3 in the progression and metastasis of human pancreatic cancer. Our stepwise investigations demonstrated further that PRSS3 can specifically activate PAR1 and trigger ERK phosphorylation and upregulation of VEGF in human pancreatic cancer cells (figure 4 and figure 5). An ERK-selective inhibitor can block VEGF upregulation induced by PRSS3 in vitro and delay the progression of metastases of human pancreatic cancer in vivo (figure 6). The discovery of a connection between PRSS3, PAR1, ERK and the VEGF signalling pathway provides a molecular explanation for the ability of PRSS3 to enhance tumour progression and metastasis, and represents a new therapeutic target for human pancreatic cancer. However, it is worth pointing out that PRSS3 is not the exclusive stimulus for activating the ERK pathway, and an ERK-selective inhibitor may still be effective on PRSS3-negative human pancreatic cancers. In addition, once PAR1 is activated by PRSS3, the activated G protein may in turn trigger a cascade of downstream events, leading to diverse cellular outcomes such as calcium signalling, engagement of integrins, cell adhesion, gene expression and mitogenesis. It is conceivable that PRSS3 may play more roles in the progression of human pancreatic cancer than our current observations, which warrants further investigation.
In the present study, we also found that PRSS3 is secreted whether it is expressed endogenously in PaT8988s cells and Capan1 cells, or exogenously in PaT8988t and MiaPaca2 cells after transfection of a PRSS3-expressing plasmid. This is consistent with our observation that PRSS3 immunoreactivity is also present in the stroma of some human pancreatic cancer tissues (data not shown). In this regard, there are two interesting points that we should consider. One is that the secreted PRSS3 may also function to promote tumour progression directly by degrading extracellular matrix proteins, supported by our invasion assay data (figure 2F), or indirectly by inhibiting and digesting trypsin inhibitors27 32 33 to promote tumour progression. Secondly, additional studies are required to determine whether endogenously released mesotrypsinogen from acinar cells contributes to the aggressive progression of human pancreatic cancer. It is noteworthy that under certain conditions the well-known pathological trypsinogen activator cathepsin B exhibits a preference for the activation of mesotrypsinogen amongst the three human trypsinogen isoforms.27 A higher level of cathepsin B protein expression has been found in the serum and malignant cells of human pancreatic cancer patients.34 35 Cathepsin B expression in cancer cells is an independent prognostic marker for cancer recurrence.35 Cathepsin B activity progressively increases and is positively correlated with metastatic potential.36 Taking into account our finding that active PRSS3 protein (mesotrypsin) can upregulate VEGF expression, it is conceivable that mesotrypsinogen released from acinar cells could be locally activated by cathepsin B from the malignant cells and contribute to the progression of disease, which might be one reason why pancreatic cancer is so aggressive.
In summary, our results suggest that PRSS3 plays an important role in the progression and metastasis of human pancreatic cancer by upregulation of VEGF expression via a PAR1-mediated pathway. Targeting this pathway instead of PRSS3 itself may be a feasible and effective approach for treatment of this dismal cancer. PRSS3 may be a useful biomarker to predict metastasis of human pancreatic cancer and the prognosis of patients with this disease.
Many thanks go to Dr Ming Yuan for his critical comments on this manuscript. We appreciate the help of Keyur Trivedi and Mohamed Ikram for immunohistochemistry.
Web Only Data gut.2009.200105
Files in this Data Supplement:
Funding This project is supported by the Nature Sciences Foundation of China (30530800), Cancer Research UK (C633-A6253/A6251) and Barts & The London Charity.
Competing interests None.
Ethics approval This study was conducted with the approval of the host institutions.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.