Cancer Associated Fibroblasts Derived from Pancreatic Adenocarcinoma and Their Role in Cell Migration

Background/Aim: Pancreatic ductal adenocarcinoma (PDAC) shows poor survival and early systemic dissemination. Cancer associated fibroblasts (CAFs) enhance migration and invasion of cancer cells. We aimed to investigate the role of CAFs in cell migration and their underlying paracrine effects. Materials and Methods: Using Transwell® migration assays, PDAC cells (PANC-1) and three distinct types of fibroblasts were analyzed: CAFs, genetically transformed human foreskin-fibroblasts (BJeLR), and non-transformed human foreskin-fibroblasts (VH7). IL6 in the culture supernatant was measured to investigate paracrine communication in monocultures and direct/indirect cocultures. Results: CAFs showed a significantly higher capacity to migrate in vitro when compared to benign fibroblasts (p=0.009). They also facilitated the migration of PDAC cells in coculture (p=0.001). Neither BJeLR, nor VH7 displayed such features. This was accompanied by a significant increase in IL-6 when CAFs were cocultured with PANC-1 (p=0.009). Conclusion: CAFs are a key element of intra-tumoral migration and should be further investigated as a potential therapeutic target.

migration as a key mechanism of metastasis in the PDAC microenvironment (32). We used a Transwell ® migration assay in which pancreatic cancer cells (PANC-1) and fibroblasts were spatially separated but able to communicate in a paracrine manner.
In search of the molecular mechanisms by which CAFs potentially impact and somehow maintain their procancerous effects during in vitro cultures, genetic alterations were also investigated (33,34). Notably, down-regulation of p53 may distinguish CAFs from other fibroblasts (35). Additionally, conditioned medium from fibroblasts with down-regulated p53 expression was shown to increase tumor cell migration in vitro (36).
In order to investigate the pro-invasive effects of CAFs, we compared CAFs isolated directly from resected pancreatic cancer specimens with the genetically transformed human foreskin derived fibroblasts BJeLR (BJ fibroblasts transformed with hTERT, genomic SV40 LT and ST oncoproteins, and oncogenic HRASV12). These fibroblasts were immortalized via an hTERT vector by lentiviral transfection, but also equipped with the oncogenic SV40T and HRASV12, altogether suppressing the cell cycle control protein p53 (37). Non-transformed human foreskin fibroblasts (VH7) served as a non-malignant control.

Materials and Methods
Patient derived cell lines. Three patients who suffered from resectable PDAC and gave informed consent before operation served as tissue donors. Exclusion criteria were pregnancy and age under 18 years. Analyses were conducted following approval from the institutional review board (EA1/292/16) and in accordance with the Helsinki Declaration of 1975.
Cell culture. Three different types of fibroblasts; primary pancreatic CAFs, BJeLR, VH7 and a pancreatic cancer cell line (PANC-1) were used for the experiments. In addition, CAFs were derived from the tumor tissues of three individuals suffering from PDAC (Table I). Derivation protocols were described previously (38). In brief, the tissue was dissociated mechanically, rinsed with cell culture medium and seeded into cell culture flasks. CAFs were characterized using immunohistochemical staining following established protocols (39), which used vimentin-rabbit polyclonal IgG antibody (Santa Cruz Biotechnology, CA, USA), monoclonal mouse anti-glial fibrillary acidic protein (GFAP) antibody (Sigma Aldrich, St. Louis, MO, USA) and monoclonal mouse anti-actin α-smooth muscle (α-SMA) antibody (Sigma Aldrich). Primary fibroblast cultures were continued when mesenchymal origin with high purity had been validated. VH7 is a non-transformed human foreskin cell line that served as a nonmalignant control, as previously described (40). BJeLR cells are human foreskin fibroblasts transformed with a vector containing the human telomerase reverse transcriptase (hTERT), genomic SV40 LT and ST oncoproteins and oncogenic HRASV12. They were obtained directly from Hahn et al. and kindly provided by Prof. Dr. rer. nat. Reinhold Schäfer (Charité Universitätsmedizin Berlin) (37). PANC-1 is a commercially available cell line of pancreatic adenocarcinoma. All cell lines were cultured in RPMI 1640 supplemented with 10% fetal bovine serum, penicillin and streptomycin and incubated at 37°i n humidified air containing 5% CO 2 .
For verification of the correct identity and origin of the cell lines, cell authentication was carried out by DNA fingerprinting, which used 8 different and highly polymorphic short tandem repeat (STR) loci. In addition, all samples were tested for the presence of mitochondrial DNA sequences from rodent cells. Analyses were carried out by Leibniz-Institute German Collection of Microorganisms and Cell Cultures.
Migration assay. To analyze the migration of cells, standardized cell migration assay was performed, using 8 μm pore Transwell ® cell culture inserts (Falcon ® , Cat #353097) in a 24-well plate. For the monoculture setup, cells were trypsinized and 10 4 cells were seeded into the upper chamber of the Transwell ® cell culture inserts in fetal bovine serum (FBS) free cell culture medium. The bottom chamber was filled with cell culture medium containing FBS. In an indirect co-culture setup, 5×10 4 cells were additionally seeded into the lower chamber. After 24 h incubation, the Transwell ® membranes were washed and stained with 0.05 % Crystal violet for 20 min.
The membrane was captured at 5× magnification by analyzing four visual fields per sample. The results were displayed as mean pixels/area and counted via Fiji using a macro for the threshold and pixel analysis function as described previously (41,42).
ELISA. Cell culture supernatant was collected from a CAF and PANC-1 monoculture and co-culture set-up in a standardized fashion, as previously described (38). The levels of IL6 in the medium were measured using a Human IL-6 ELISA set (BD OptEIA™, San Jose, CA, USA, Cat# 555220). The ELISAs were performed according to the manufacturer's instructions and the levels of IL6 were normalized to the number of cells used in each culture.   Passages used for analysis   1  53  m  Adenocarcinoma  2  2  G3  2, 3, 5  2  57  w  Adenocarcinoma  2  1  G2  2, 3, 4  3  52  w  Adenocarcinoma  3b  1  G3  1, 2 Diego, CA, USA). For the statistical analysis all data obtained from the migration assays and the cell culture supernatant analysis was tested without following the non-Gaussian distribution. To test more than two groups, the Kruskal-Wallis test was performed followed by a post-hoc Dunn's multiple comparisons test. When significance correlation was tested, results were followed by a Mann-Whitney Utest for testing independent variables in a pairwise comparison. For matching pairs, a Wilcoxon signed-rank test was conducted. In all tests performed, p<0.05 characterized significant correlations.

Results
Characterization of CAFs. CAFs of three patients suffering from PDAC were isolated (Table I). Before conducting additional studies, we ensured that the isolated fibroblasts from the three individuals could be defined as CAFs, all coexpressing α-SMA, Vimentin and GFAP ( Figure 1).

CAFs show significant capacity for migration.
To study the migration capacity of fibroblasts, non-patient derived fibroblast cell lines VH7 and BJeLR as well as the primary pancreatic CAF were analyzed by conducting a standardized migration assay on monocultures. All three types of fibroblasts maintained a reproducible degree of migration ( Figure 2). However, there was a significant difference in the capacity of self-induced cell migration. Multiple comparison analysis revealed a remarkable difference regarding the migration capacity in monoculture between VH7 and CAF (p=0.018, Table II) and was confirmed by pairwise comparison (p=0.009, Table II).  established PDAC cell line PANC-1 (Figure 3). We were not able to detect any significant effect of PANC-1 cocultures on the migration of studied fibroblasts when compared to migration in monoculture. In pairwise comparison however, we observed a clear trend of increasing CAF migration when cocultured with PANC-1 when compared to the monoculture (p=0.098, Table II).

CAFs increase migration of PANC-1 in indirect co-culture.
We subsequently analyzed the effect of fibroblasts on the migration of PANC-1. For this, PANC-1 cells were studied either in monoculture or in indirect coculture with VH7, BJeLR and CAFs, respectively ( Figure 4). PANC-1 cells showed a consistent and reproducible migration in monoculture ( Figure 5). When cultured in indirect coculture with the two non-patient derived tumor associated cell-lines VH7 and BJeLR, we did not detect any change in cell migration of PANC-1 cells. After coculturing with CAFs however, the migration of PANC-1 cells increased significantly, when tested with multiple comparison (p=0.025, Table II) and in pairwise correlation (p=0.001, Table II).

Significant increase of IL6 in CAF-PANC-1 co-culture.
To further understand the underlying paracrine effects on tumor cell migration, we analyzed the IL6 concentration in the supernatant of PANC-1-and CAF monocultures as well as in direct and indirect co-cultures ( Figure 6).

Discussion
In this study, we aimed to better understand the impact of CAFs on pancreatic carcinoma cells using PANC-1 carcinoma cell line, as well as primary and established cell culture models of human fibroblasts, that mimic the tumor microenvironment of PDAC (43). There is an ongoing debate about whether CAFs actually possess an innate malignant potential (44). Recent studies by our group have already demonstrated that CAFs can confer chemoresistance when co-cultured in vitro (38). In the present study, we compared different types of human fibroblasts: patient derived malignant CAFs, fibroblasts with activated oncogenes and inhibited tumor suppressors (BJeLR), and benign fibroblasts. We observed that CAFs exhibit malignant features as they migrate unlike other fibroblasts and also, exclusively promote the migration of PDAC cells. Thus, they evidently play an active role in cell migration and we would therefore refute their former characterization as 'bystander cells' (45). Genetic alterations in the RAS family and p53 have been discovered to be critical in pancreatic cancer development and progression (46). This has been mainly established on the duct/acinar cell fraction. Stromal cells have not yet been analyzed regarding their RAS and p53 characteristics. Lately, it has been theorized that these alterations may play a role in CAF function as well. Several studies have detected a p53 alteration in CAFs that modify their functioning and enhance the malignant features of their respective cancers (47). This non-autonomous effect is initiated via regulation of cancerpromoting genes and secreted proteins (35,48), also in line with a different study, which revealed a direct link between cancer cells and CAFs in chemotherapy response of lung cancers based on p53 dependence (49).
RAS mutations (particularly KRAS) are known to be oncogenic, and in combination with p53 alterations suffice to develop tumors in both, human and genetically engineered mouse models (e.g. KPC mice) (50)(51)(52). Moreover, HRAStransformed mouse fibroblasts show increased genomic instability and altered chromosomal protein phosphorylation (53). Murine RAS mutated fibroblasts generated a high local level of proteins involved in tumor growth and angiogenesis (54). Studies suggest that a RAS induced activation of CAF might be one important underlying mechanism causing this (55). We therefore evaluated the effect of HRAS mutated and p53 suppressed fibroblasts (BJELR) in our experiments. Remarkably, we were unable to detect an increased capacity for cell migration by BJeLR fibroblasts, nor found evidence they could enhance the migration of PDAC cells in indirect co-cultures.
Several limitations of this study should be noted: First of all, we chose a two-dimensional cell culture setup in our experiments. For this reason, important factors that might influence cell migration, including fiber composition and alignment of the extracellular matrix scaffold, were not considered. However, the setup essentially facilitated high reproducibility of in vitro assays, while considering the Von Hoerschelmann et al: Pancreatic Cancer CAF and Cell Migration  limited number of available patient derived fibroblasts. This was in part due to the fastidious derivation process with a relatively slow doubling time and early senescence of mesenchymal cells in the primary tumor. However, we believe that working with patient derived primary CAFs is essential, in order to investigate and model the biology of the pancreatic tumor microenvironment. This could be supported by our findings that neither benign, nor genetically modified fibroblasts show significant effect on the migration capacity of PDAC cells. In addition, we focused solely on CAFs, while other stromal components like immune cells or endothelial cells were not included in the analysis. Also, the interaction with vascular structures may play an important role in the early steps of metastasis. For this reason, experiments that include the investigation of other cell components of the pancreatic cancer microenvironment are warranted.
CAFs in PDAC are known to express α-SMA, a microfilament forming protein particularly active in activated myofibroblasts (56). All CAFs analyzed in the present study were verified to express α-SMA (Figure 1). Activated CAFs are known to produce excessive amounts of ECM in the presence of carcinoma cells (57,58). However, the increase in migration detected in our experiments must be an additional paracrine effect as there was no direct cell-cell contact with cancer cells. In the densely packed desmoplastic stroma, systemic hormonal mediators and cytokines are unlikely to reach the tumor cells via the blood stream. Thus, a more local crosstalk must play a predominant role in intercellular communication. To this end, we investigated IL-6, a pro-inflammatory cytokine known to promote migration and metastasis (30). We could verify that CAFs are the major source of IL-6 in the PDAC-CAF axis. PDAC cells did not show IL-6 secretion, while enhanced the capacity of CAFs to produce IL-6, most eminent in direct co-cultures.
The crosstalk between stromal cells and tumor cells has recently been discovered as a potential source of novel therapeutic targets. Preclinical studies investigating the depletion of CAFs revealed promising anti-cancerous effects (59). Yet, this targeted approach remains challenging as the CAFs role in the tumor microenvironment is complex and still under investigation. Depletion even leads to an increase in EMT (60). Therefore, the focus must remain on CAFdependent pro cancerous pathways, such as IL6/STAT3 signaling. The anti-fibrotic tyrosine kinase inhibitor nintedanib was shown to reduce tumor aggressiveness in intrahepatic cholangiocarcinoma via suppression of IL6 and other cytokines (61). Moreover, several novel inhibitors of the IL6/STAT3 pathway, as Stattic (6-Nitrobenzo(b)thiophene-1,1-dioxide) and AZD9150, [a signal transducer and activator of transcription 3 (STAT3) antisense] are being investigated in both, preclinical and early phase clinical studies. In combination with other established chemotherapeutics, a reversed effect of STAT3mediated chemoresistance was documented (62). Napabucasin (2-acetylfuro-1,4-naphthoquinone or BBI-608), an inihibitor of STAT3 signaling, is currently being investigated in a phase III trial together with nab-paclitaxel and gemcitabine for metastatic pancreatic cancer (63). In addition to chemotherapeutics, targeting the IL6/STAT3 pathway is promoted as a feasible therapeutic approach in the advanced tumor stage. However, to specifically target the invasive properties supported by CAFs, further studies investigating the underlying signaling pathways are needed.

Conclusion
CAFs show the capacity to migrate in vitro and also enhance the migration of PDAC cells. This feature is accompanied by an increase in IL-6 production. Non-cancer associated fibroblasts from established cell lines display none of these properties. We therefore conclude that CAFs are a key element of intra-tumoral migration and should, in this context, be investigated as a potential therapeutic target.

Conflicts of Interest
The Authors have no conflicts of interest to declare in relation to this study.