|Home | About | Journals | Submit | Contact Us | Français|
The MST1R gene is overexpressed in pancreatic cancer producing elevated levels of the RON tyrosine kinase receptor protein. While mutations in MST1R are rare, alternative splice variants have been previously reported in epithelial cancers. We report the discovery of a novel RON isoform discovered in human pancreatic cancer. Partial splicing of exons 5 and 6 (P5P6) produces a RON isoform that lacks the first extracellular immunoglobulin-plexin-transcription (IPT) domain. The splice variant is detected in 73% of pancreatic adenocarcinoma patient derived xenografts and 71% of pancreatic cancer cell lines. Peptides specific to RON P5P6 detected in human pancreatic cancer specimens by mass spectrometry confirms translation of the protein isoform. The P5P6 isoform is found to be constitutively phosphorylated, present in the cytoplasm, and it traffics to the plasma membrane. Expression of P5P6 in immortalized human pancreatic duct epithelial (HPDE) cells activates downstream AKT, and in human pancreatic epithelial nestin-expressing (HPNE) cells activates both the AKT and MAPK pathways. Inhibiting RON P5P6 in HPDE cells using a small molecule inhibitor BMS-777607 blocked constitutive activation and decreased AKT signaling. P5P6 transforms NIH3T3 cells and induces tumorigenicity in HPDE cells. Resultant HPDE-P5P6 tumors develop a dense stromal compartment similar to that seen in pancreatic cancer. In summary, we have identified a novel and constitutively active isoform of the RON tyrosine kinase receptor that has transforming activity and is expressed in human pancreatic cancer. These findings provide additional insight into the biology of the RON receptor in pancreatic cancer and are clinically relevant to the study of RON as a potential therapeutic target.
While there have been recent advances in the use of cytotoxic therapy to treat pancreatic cancer, molecularly targeted treatments have yet to make an impact in the clinic. To develop effective therapeutics for a disease with few treatment options, the molecular structure and function of proteins expressed in pancreatic cancer must be identified and understood. RON is a growth factor receptor belonging to the MET family which has been suggested as a putative target for cancer therapeutics 1, 2. RON signaling impacts not only the cancer cell but is a regulator of the tumor microenvironment thru its effects on tumor-associated macrophages 3, 4. Currently, a monoclonal antibody (RON8, ImClone) and small molecule inhibitors are in Phase 1 trials and various stages of pre-clinical development, respectively 5, 6.
RON is expressed at minimal levels in normal pancreatic tissue, but is overexpressed in pancreatic and epithelial cancers of the breast, colon, lung, and prostate2. During multistep carcinogenesis from low to high grade pancreatic intraepithelial neoplasia (PanIN), RON is increasingly expressed 7. Overall the protein is expressed in 69-96% of pancreatic cancer specimens 7, 8. Upon receptor activation, the MAPK and AKT signaling pathways are but two of the downstream targets activated to promote survival, motility, tumor growth and invasion of cancer cells in vivo 9-11.
The full length RON protein is 1400 amino acids and is translated as a single pro-protein which is cleaved into an α and β subunit to form the extracellular sema domain12. The sema domain recognizes the receptor ligand, Macrophage Stimulating Protein (MSP), and binding subsequently activates the intracellular tyrosine kinase domain 13. RON is activated when two receptor molecules form a homodimer and bind one MSP ligand molecule 14. When the receptor is overexpressed, interactions between the extracellular portions of the receptor cause dimerization and is hypothesized to be the mechanism of ligand independent activation.
RON is rarely mutated, but alternative splicing events are common as the gene transcript is increasingly overexpressed. Approximately eight different RON isoforms have been described in various epithelial cancer types 15. Exon skipping is the most common splicing event, creating RONΔ165 and RONΔ160 where exon 11 and exons 5+6 are skipped, respectively 16, 17. A short form of RON is produced by an alternative promoter and transcription start site in intron 10, producing a protein that lacks the entire extracellular domain 18. In breast cancer, short form RON overexpression specifically activates the PI3K signaling pathway which results in increased tumor growth, epithelial to mesenchymal transition (EMT), and increased metastasis 19. Each of these three isoforms is constitutively phosphorylated and contributes to an increasingly invasive phenotype.
Several preclinical studies have demonstrated robust antineoplastic activity following RON inhibition and form the basis for development of therapeutics targeted to the receptor 9, 20, 21. Understanding the biology of RON isoform splice variants will allow us to more effectively evaluate current therapeutics and develop new, targeted therapies. In this study we identified a novel RON isoform produced by alternative splicing of exons 5 and 6. Alternative 5’ and 3’ splice sites produce a transcript and protein product that omits a portion of exons 5 and 6 (P5P6). We demonstrate that P5P6 protein is expressed in human pancreatic cancer specimens and it transforms immortalized human pancreatic ductal epithelial cells such that they become tumorigenic in vivo.
Previous studies in colon cancer and glioma revealed that RON isoforms are often generated based on alternatively spliced transcripts between exons 4 and 7 17, 22, 23. As no study to date has explored alternative splicing of RON in human pancreatic cancer, we sought to determine if those alternative splicing events previously identified were present. Initially, we analyzed RNA from a human pancreatic cancer by end point PCR to amplify a region of RON between exons 4 and 7 (Figure 1A). Sequencing the resulting bands confirmed that they were specific for the full length protein, exon 6 skipping (RONΔ90), partial exon 5 and partial exon 6 skipping (P5P6), and exon 5 and 6 skipping (RONΔ160). To our knowledge, the RON P5P6 splice variant has not previously been identified in any tumor type. This novel isoform is produced by an alternative 5’ splice site within exon 5 and an alternative 3’ splice site within exon 6. Splicing is in frame and compared to the full length protein produces a transcript which has skipped 195 nucleotides (Figure 1B).
The P5P6 protein product is 1335 amino acids, yielding a molecular weight of 145 kDa for the pro-P5P6 and 112 kDa for the cleaved β-subunit. Omission of this 65 amino acid sequence corresponds to loss of the first Immunoglobulin-Plexin-Transcription (IPT) domain (Figure 1C). A similar region is also deleted in the RON variant Δ160 that is expressed in colon and breast cancer 17, 24. Three cysteine residues are removed by deletion of these amino acids that function to produce intermolecular disulfide bonds. Deletion of these cysteine residues is functionally significant and in RONΔ165 produces unbalanced bonds and RON oligomer formation 16.
We next determined whether the P5P6 isoform could be identified in normal pancreatic tissue and what the prevalence of P5P6 expression is in pancreatic cancer. First we performed PCR using primers specific for the P5P6 junction as well as total RON transcripts using primers directed at the C-terminus (Figure 2A). These studies revealed that P5P6 transcript could not be identified in normal pancreas, but was detected in 73% of PDX and 71% of cell lines (Figure 2B).
Next we quantified the amount of RON mRNA being produced using absolute quantitative Real Time PCR in both patient-derived xenografts (PDX) and pancreatic cancer cell lines. There are no commercial antibodies specific to isoforms of RON therefore we could not quantitate isoform protein expression using traditional methods. Plasmids containing full length RON and P5P6 transcripts were used to generate standard curves (Supplementary Figure S1). Efficiency was calculated at 106% for RON c-terminal primers and 91% for the P5P6 primer pair. We found that whole normal pancreas expresses 67 copies of total RON transcripts per 73 ng of RNA and reaffirmed the absence of P5P6 expression (Figure 2C). We demonstrated quantitatively for the first time that the magnitude of RON overexpression at the RNA level is 150-400 times that present in normal pancreatic tissue. Similarly, pancreatic cancers also generated high copy numbers of P5P6 transcript ranging from 1304 – 4215 copies per 73 ng of RNA or 20-70 fold greater abundance than total RON transcript seen in normal tissue. Cell lines that overexpress RON (Bx-PC3, AsPC1, 79E, 34E) also generated high copy numbers of P5P6 transcript.
While we have demonstrated that the P5P6 transcript was abundant in pancreatic cancer specimens, any putative biologic consequences of P5P6 are predicated on translation to a functional protein. We first utilized a non-specific RON antibody (C-20) to the c-terminal portion of the β chain to determine RON expression pattern in PDXs. Although some PDXs give a western blot signal around 115 kDa (Red arrow in Figure 3A), there is no certainty that this corresponds to the P5P6 protein (Figure 3A).
We therefore utilized mass spectrometry to investigate the presence of P5P6 protein in pancreatic cancer based on the finding that RON isoforms produce unique peptide fragments after digestion with trypsin (Figure 3B). Each isoform protein sequence was analyzed by the ExPASy PeptideCutter tool that predicts peptides generated after trypsin digestion 25. Immunoprecipitation of two PDXs (1713 and 1444) was performed with an antibody specific for the C-terminal portion of RON (E9) and the N-terminal portion (RON8). There was adequate protein pulled down as seen on a stained gel and RON peptides were the most common in the sample (Supplementary Figure S2). Ultimately, PDX 1713 contained a peptide unique for the P5P6 protein (Figure 3C). None of the remaining peptides that were identified were isoform specific (Figure 3D). This proves that generation of the P5P6 transcript results in translation to P5P6 protein in pancreatic cancer. P5P6 protein identification suggests it is present in relatively high abundance compared to other isoforms whose unique peptides could not be detected using this methodology.
To understand its effects on cancer cell biology we cloned the RON P5P6 cDNA into full length RON mRNA and attached it to enhanced Green Fluorescent Protein (eGFP). This recreated the partial exon 5 and 6 splicing and created the P5P6-eGFP (P5P6e) fusion protein. This was transfected into HPDE, Human Pancreatic Nestin-expressing Epithelial (HPNE), COS-1 (Cos), and MiaPaca2 (MP2) cell lines using a lentiviral vector. HPDE expresses low levels of full length RON while HPNE, Cos, and MP2 were chosen as they do not express detectable RON protein. RON P5P6e expression and phosphorylation levels were stable throughout all experiments. HPDE P5P6e expressing cell lines displayed no obvious change in morphology in culture (Figure 4A). The P5P6e protein primarily localizes to the cytoplasm in a reticular pattern surrounding the nucleus and to the plasma membrane (arrows, Figure 4B). We also created an untagged version RON P5P6 cDNA and expressed it in HPDE cells to rule out any effects of the fusion protein on oncogenic phenotypes (Supplementary Figure 3).
Western blot of the stably transduced cell lines show that the P5P6 protein is detectable by conventional RON β antibodies (Figure 4B). In all cell lines, it produces a similar pattern of two bands which are an uncleaved pro-form (172 kDa) and a cleaved β chain (139 kDa). The HPDE P5P6e cell line expresses somewhat higher levels of the fusion protein and subsequent degradation products can be seen (shown is a less exposed blot, red arrows Figure 4C). Immunoblot for the α subunit of RON confirms that P5P6 is cleaved into separate subunits similar to the full length protein. Immunoprecipitation with RON β antibody and immunoblot with phosphotyrosine (pY) shows that the RON P5P6 protein is constitutively phosphorylated. In the HPDE cell line the degradation products are also still phosphorylated as the N terminal part of the β chain is cleaved (red arrows). Importantly, the level of P5P6e protein expressed in the HPDE and HPNE cell lines is similar to the overexpression of total RON in 3 pancreatic cancer cell lines and 3 PDXs (Figure 4D). This demonstrates these cell lines are producing physiologic levels of RON protein comparable to that seen in human pancreatic cancers and cell lines.
In HPDE cells AKT is active at baseline and RON P5P6 increased phosphorylation by 2 fold. However there were no significant effects on ERK phosphorylation (Figure 5A). Parental HPNE cells do not express RON, however when RON P5P6e is expressed both the AKT and MAPK pathways are activated. In contrast, in MP2 cells there was a modest increase in MAPK signaling while minimal changes were noted in Cos cells, strongly suggesting the effects of P5P6 are cell and context specific. We also noted that in HPDE cells, MSP stimulation did not change phosphorylation of P5P6 and appeared to alter downstream signaling only through the native full-length RON receptor (Figure 5B).
RON is under investigation as a therapeutic target and BMS-777607 is a MET family inhibitor which effectively inhibits RON signaling in pancreatic cancer 30. Treatment with BMS-777607 at 3 μmol/L completely inhibited phosphorylation in the HPDE P5P6e line that could not be rescued with MSP stimulation (Figure 5C). Signaling through AKT was significantly reduced and there was no effect on MAPK. Since BMS-777607 is not a highly specific RON inhibitor, the effects on signaling are obviously not necessarily secondary to P5P6 inhibition alone.
The P5P6 isoform is most similar in structure to RONΔ160 that results from complete skipping of exons 5 and 6. The mechanism of RONΔ160 induced oncogenesis involves accumulation of β-catenin, which is required for isoform mediated cell transforming activities 26. We observed no accumulation of total β-catenin in the P5P6e expressing cell lines (Figure 5D). STAT3 is a transcription factor implicated in multiple stages of KRAS driven cancer progression 27. STAT3 is activated when full length RON is stimulated with its ligand MSP, and high levels of STAT3 phosphorylation were correlated with high levels of RON expression in gastroesophageal adenocarcinoma 28. In contrast, we found that RON P5P6 expression does not activate STAT3 signaling (Figure 5E). Conversely, in the Cos and MP2 lines there is a loss of total STAT3 and a concordant decrease in phosphorylation. These experiments suggest that P5P6 signaling is mechanistically distinct from both RON Δ160 and the full length protein as the β-catenin and STAT3 pathways are not downstream targets.
We next performed functional assays to determine if P5P6e signaling induced an oncogenic phenotype. We found no differences in proliferation between the HPDE P5P6e, MP2 P5P6e and their parental lines, a finding consistent with prior studies of RON signaling in pancreatic cancer cells (Supplementary Figure S4A). Wound closure assay showed HPDE P5P6e expressing cells induced significantly greater wound closure at all time points. At 72 hours the extent of wound closure was 93% and 66% (P < .001) for the HPDE P5P6e expressing and parental HPDE line respectively (Figure 6A). Interestingly, in the poorly differentiated MP2 cell line, P5P6e cell line slightly decreased wound closure at all times points again suggesting the contextual relevance of P5P6 activity (Supplementary Figure S4B). We have previously demonstrated that RON signaling activates VEGF production by pancreatic cancer cells via the MAPK pathway 29. While we noted a 36% increase (263 pg/mL vs 194 pg/mL, P = 0.011) in VEGF secretion in HPDE P5P6e compared to the parental line (Supplementary Figure 5), this increase is small and likely is not biologically significant.
Finally, we utilized a colony forming assay to determine whether P5P6 expression could transform NIH3T3 cells (Figure 6B). P5P6 expression significantly increased colonies formation compared to empty vector, yet the effect was more modest than expression of mutant KRAS. Collectively, these experiments reveal that P5P6 has transforming activity and enhances migration in human pancreatic duct cells. The effects on migration appear to be specific to the context as they are not observed either in a non-pancreatic epithelial cell line (Cos) or in the poorly differentiated pancreatic cancer cell line (MP2).
The previous series of assays suggested that expression of P5P6e in HPDE cells induced both molecular and phenotypic alterations. To determine if these alterations reached a threshold sufficient to alter in vivo biology, we used an orthotopic cell-line derived xenograft model. The HPDE cell line has previously been shown to be non-tumorigenic in SCID mice 31. We injected one million HPDE or HPDE P5P6e cells into murine pancreata as an orthotopic xenograft. While the parental HPDE-derived xenografts failed to induce tumors in ten mice after 16 weeks, the HPDE P5P6e orthotopic injections resulted in tumor formation in six of eight mice that were palpable after 4 weeks (p-value 0.02, Figure 7A). Expression of the P5P6 eGFP fusion protein was evidenced by bright tumor fluorescence. Subcutaneous injection assays were also performed and produced identical results (Figure 7B). Orthotopic tumors had a mean weight and volume of 178 mg and 156 mm3 (p-value 0.03, Figure 5C). One orthotopic tumor locally invaded the abdominal wall but we found no gross evidence of metastatic disease.
A hallmark of pancreatic cancer is the dense stromal component present within the tumors. We performed H&E staining and immunohistochemistry (IHC) on the P5P6e tumor sections to characterize tumor histology and protein expression (Figure 7D). Merged confocal images show the majority of the tumor was HPDE P5P6e expressing cells (Green, GFP), but that there was a significant stromal component to the tumors (Blue, DAPI). Ductal structures which express E cadherin (arrows, Figure 7D) but not the GFP fusion protein represent residual clusters of host mouse pancreatic acinar cells. RON isoforms have been reported to promote epithelial to mesenchymal transition (EMT) but we did not observe significant morphologic evidence of P5P6 induced EMT either in vivo or in vitro. Cells maintained a round shape and expressed e-cadherin homogenously (Supplementary Figure S6).
The macrophage stimulating protein receptor 1 (MST1R) gene locus contains the gene for RON and its splice variants on chromosome 3p21 32. This region of the genome contains many oncogenes and is frequently altered in many epithelial malignancies 33. During carcinogenesis as RON is increasingly expressed in pancreatic cancer there is subsequent generation of splice variants. In this study, we report the discovery of a novel RON isoform that is produced by partial skipping of exons 5 and 6 and is expressed in the majority of pancreatic cancers and pancreatic cancer cell lines. No previous reports have quantified levels of RON isoform production for comparison, however the amount of P5P6 transcript is 20-70 fold times the amount of total RON expressed in normal pancreas. Quantification of splice variants will be important in future studies to determine if there is isoform switching between cancer types or during oncogenic progression.
Detection of P5P6 protein is difficult as there are no specific antibodies and the size is similar to the wild type protein. Despite this, we were able to confirm expression of P5P6 protein in pancreatic cancer by detection of a specific peptide using mass spectrometry. To our knowledge this is the first demonstration of RON isoform protein expression in human cancer to be documented by a method other than immunoblotting. The P5P6 protein isoform has a truncation of the first IPT domain which may be a cause of its deregulation. RON shares domain structures with plexins, a family of semaphorin protein receptors 34. Class B plexins can couple with RON and MET in the absence of ligand to activate the receptor and promote invasive activity 35. Deletion of this domain will alter the proteins molecular structure leading to increased or decreased affinity for conventional RON binding partners. IPT domain deregulation may also contribute to constitutive activation of the protein through unbalanced cysteine residues. The P5P6 isoform traffics to the plasma membrane similar to the full length protein which allows for binding of ligand, interaction with cell surface receptors, and interaction with cytoplasmic binding partners. It is clear that P5P6 protein is phosphorylated in the absence of ligand and may be due to accumulation of P5P6 dimer formation and subsequent auto-phosphorylation.
Overexpression of P5P6 in the immortalized HPDE cell line resulted in increased AKT signaling, cell migration, and transformed the cell line to become oncogenic. The HPDE cell line has no mutations in the most commonly mutated genes in pancreatic cancer - p53, KRAS, CDKN2A, or SMAD4, though p53 is nonfunctional and there is loss of the Rb gene 31. The HPDE and HPNE lines are the closest representation of a normal pancreatic epithelium available and is a limitation to this study. The context of P5P6 overexpression clearly matters as the biologic changes we observed in HPDE and HPNE differ in an unrelated cell line (Cos) and a mesenchymal pancreatic cell line (MP2). This raises the possibility that P5P6 expression may exert biologic effects early in oncogenesis rather than after the accumulation of multiple mutational events. This study has limitations in that functional studies rely on P5P6 overexpression. However, total RON levels in P5P6 lines are similar to pancreatic cancer cell lines and human samples. Furthermore, our qPCR transcript quantification is an average of all cell types contained within the tumor. In pancreatic cancer stromal cells represent 50-90% of the tumor volume and physiologic levels of RON P5P6 transcripts in the malignant epithelial cells are likely much higher than we have detected 36.
The discovery of novel RON protein isoforms has potential translational impact. Full length RON expression has been shown to be a prognostic marker for several cancers, but not for resectable pancreatic cancer 28, 37. Use of RON as a predictive or prognostic marker may require consideration of its isoforms to make valid correlations. RON P5P6 is constitutively active and does not need ligand to interact with other proteins or signal. Monoclonal antibody therapies that prevent binding of ligand to the receptor and subsequent activation may therefore be ineffective against some RON isoforms. Small molecules may therefore be more broadly effective in inhibiting RON and its isoforms. In fact, we found that BMS-777607 inhibited P5P6 phosphorylation and decreased AKT signaling. Our findings for the P5P6 isoform are consistent with previous studies that suggest alteration of the IPT domain structure effects RON function in profound ways. Therapies that target these interactions may be of interest as modulators of full length RON and RON isoform signaling.
In summary, this study identified a new RON isoform, P5P6, which is commonly expressed in pancreatic cancer. Overexpression of the isoform transforms NIH3T3 cells and immortalizes, non-tumorigenic human pancreatic ductal epithelial cells. Initial studies into the mechanism of its action imply signaling mechanisms distinct from the full length protein and a structurally similar isoform. Ultimately discovery of this isoform supports previous studies suggesting that RON isoform expression is important in cancer initiation and tumorigenesis. Therapeutics directed at the RON receptor will be most effectively evaluated if isoform activity is considered. Future work will be directed at further characterizing expression of RON isoforms in pancreatic cancer to understanding their functional significance and susceptibility to therapeutic agents.
All human tissues were collected and utilized in accordance with IRB approved and IACUC approved protocols at the University of California, San Diego. Tumor tissue was implanted into the pancreas of NOD SCID Gamma (NSG) mice as described previously 38, 39. Patient Derived Xenografts (PDX) were established and passaged after they were at least 10mm in size. Low passage (<5) tumors were used for analysis.
Trizol (Invitrogen) was used to extract total RNA from PDXs. A cDNA library was created using SuperScript III Reverse Transcriptase (Invitrogen) and PCR used RedTaq DNA Polymerase (Sigma). The following primers were used:
RON Exon 4 to 7: Forward – GGCATGGCATTTCATGGGCT, Reverse - AGAGTCTGTCTGTAGGCACC
RON C-Terminal : Forward – TAGTGTCTGCACTGCTTGGG, Reverse - GCTGTTCTGGACGCACATTC
RON P5P6 specific: Forward – GGCAGTACAAGGCCTACCAA, Reverse - TCAGTCCCATTGACCAGCAC
All reactions used SsoFast EvaGreen Supermix (Bio-Rad), primers (375nM final concentration), 1 μL of cDNA (corresponding to 73 ng of RNA). Absolute quantification used methods previously described 40. A plasmid containing the full length RON and RON P5P6e mRNA was used to generate a standard curve using RON c-terminal and P5P6 specific primers respectively. The equation DNA MW (g) = ((# of bp) × (660 g/mol/bp))/ 6.022×1023 particles/mol was used to convert plasmid size to weight. Weight was used to construct a serial dilution of 107 – 102 plasmid copies. Standards were run in duplicate on each plate used to analyze unknown samples and linear standard curve was generated (Supplementary Figure S1). The slope of the lines corresponds to the efficiency of the reaction using equation Efficiency = (10−1/slope−1) × 100%.
Pancreatic tumors of approximately 100mg or cultured cells at approximately 80 % confluency were used to make protein lysates in RIPA buffer. Immunoprecipitation (IP) was performed using RON C-terminal (E9, Santa Cruz) or RON N-terminal (RON8, ImClone). For western blot primary antibodies used are RON β (C20, Santa Cruz), RON α (EP1132Y, Abcam), Phosphotyrosine (4G10, Millipore), Actin (A2066, Sigma), Phospho-AKT (9271,Cell Signaling), AKT (9272, Cell Signaling), Phospho- p44/42 MAPK (9101, Cell Signaling), p44/42 MAPK (9102, Cell Signaling), Phospho-β-catenin (Tyr-654, ECM Biosciences), β-catenin (8480, Cell Signaling), Phospho-Stat3 (9138, Cell Signaling), Stat3 (4904, Cell Signaling). For immunohistochemistry the antibodies used were GFP (NB100-1770, Novus Biologicals) and E-cadherin (610182, BD). ImageJ program was used for densitometry.
Immunoprecipitation was performed on 3 mg of protein with RON c-terminal (E9, Santa Cruz) and RON N-terminal (RON8, Imclone) then beads were submitted and sample preparation performed as previously described 41. Trypsin-digested peptides were analyzed by high pressure liquid chromatography (HPLC) coupled with tandem mass spectroscopy (LC-MS/MS) using nano-spray ionization 42. The nanospray ionization experiments were performed using a TripleTof 5600 hybrid mass spectrometer (ABSCIEX) interfaced with nano-scale reversed-phase HPLC (Tempo). The collected data were analyzed using MASCOT® (Matrix Sciences) and Protein Pilot 4.0 (ABSCIEX) for peptide identifications. RON isoform protein sequences are not normally included in the Protein Pilot analysis software and were supplemented in order to determine their detection.
The P5P6 protein transcript was created by recreating the P5P6 splicing event in the full length RON cDNA which had previously been attached to GFP. The region between exons 3 and 8 was amplified using the primers (F – GGGACCAGGTTTTCCAGGTACC and R – GGTACCTGGTTCCTGGACCTTCCAG) from a primary pancreatic cancer sample. The PCR amplicon was TA cloned using a PCR Cloning Kit (Qiagen) and competent DH5α E. Coli (Invitrogen). The full length RON+eGFP vector and the P5P6 containing vector were cut with specific endonucleases HindII and KpnI (NEB) and ligated using T4 DNA Ligase (NEB). DNA sequencing confirmed correct P5P6 sequence and the cDNA was then transferred into a lentiviral plasmid for transfection. After transfection selective puromycin media was used and the TE300 inverted fluorescence microscope (Nikon) was used to screen for green fluorescence.
For scratch wound assay, 1×106 cells were plated in a 6-well dish and allowed to attach and grow to confluency. A 1 mL aspirating pipette tip was used to create 3 vertical scratches in each well. Images were taken at 4× magnification with Nikon TE300 inverted fluorescence microscope. Metamorph (Molecular Devices) analysis software was used to outline and determine wound area. Scratch assay was performed three separate times and all data combined. NIH 3T3 cells were stably infected with pCDH CMV empty vector, P5P6-eGFP, or KRAS G12V plasmid and plated. After 2 weeks cells were fixed, stained with crystal violet and a Keyence BZ-X710 microscope scanned the well and counted purple colonies.
The Human Pancreatic Ductal Epithelial (HPDE) cell line was a generous gift from Dr. Ming Tsao and the University Health Network and cultured in Keratinocyte-SFM (Gibco) with epidermal growth factor and bovine pituitary extract, and antibiotic-antimycotic (Gibco)43. MiaPaca2 and Cos1 cell lines were obtained from ATCC and cultured in DMEM (Corning) supplemented with 10% fetal bovine serum (Sigma). HPNE cells were obtained from Dr. Klemke at UCSD and cultured in DMEM/Base F media (InCell) supplemented with 5% FBS. For cell injections 1×106 HPDE and HPDE P5P6e cells were resuspended in 10 μL of PBS and mixed with 10 μL of Matrigel (BD). NSG mice were injected orthotopically using a 30 gauge needle. Mice were monitored for signs of distress and disability throughout the experiment. After 12-16 weeks mice were sacrificed and tumors were harvested. Tumor volume was measure as: Volume = 1/2(length × width 2). Mice and tumors were imaged using MVX10 (Olympus) camera and cellSens (Olympus) software.
All data was analyzed using SPSS Statistics v.20 (IBM) and Prism v.6 (GraphPad Software). Functional assay values were averaged from each cell line per time point and compared using one way ANOVA.
We gratefully acknowledge the generosity of our patients who supported this work through both monetary contributions and/or by allowing tumor tissue to be used for research. Without their contributions, this work would not have been possible. We thank the Taso and Klemke labs for providing HPDE and HPNE ells, respectively. Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number R01 CA155620 (AML) and T32 CA121938 (JC). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. DNA sequencing was performed by the DNA Sequencing Shared Resource, UCSD Moore’s Cancer Center, which is funded in part by NCI Cancer Center Support Grant # 2 P30 CA023100-23.
Conflicts of Interest
The authors have no conflicts of interest to disclose.
Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)