Downregulation of TRAIL-Receptor 1 Increases TGFβ Type II Receptor Expression and TGFβ Signalling Via MicroRNA-370-3p in Pancreatic Cancer Cells

The accumulation of perturbations in signalling pathways resulting in an apoptosis-insensitive phenotype is largely responsible for the desperate prognosis of patients with pancreatic ductal adenocarcinoma (PDAC). Accumulating evidence suggests that the death receptors TRAIL-R1 and TRAIL-R2 play important roles in PDAC biology by acting as either tumour suppressors through induction of cell death or tumour promoters through induction of pro-inflammatory signalling, invasion and metastasis. TRAIL-R2 can also associate with nuclear proteins and alter the maturation of micro RNAs (miRs). By genome-wide miR profiling and quantitative PCR analyses we now demonstrate that knockdown of TRAIL-R1 in PDAC cells decreased the level of mature miR-370 and led to an increased abundance of the type II receptor for transforming growth factor β (TGFβ). Transfection of cells with an artificial miR-370-3p decreased the levels of TGFβ-RII. We further show that transient expression of the miR-370 mimic decreased TGFβ1-induced expression of SERPINE1 encoding plasminogen activator-inhibitor 1 and partially relieved TGFβ1-induced growth inhibition. Moreover, stable TRAIL-R1 knockdown in Colo357 cells increased TGFβ1-induced SERPINE1 expression and this effect was partially reversed by transient expression of the miR-370 mimic. Finally, after transient knockdown of TRAIL-R1 in Panc1 cells there was a tendency towards enhanced activation of Smad2 and JNK1/2 signalling by exogenous TGFβ1. Taken together, our study reveals that TRAIL-R1 through regulation of miR-370 can decrease the sensitivity of PDAC cells to TGFβ and therefore represents a potential tumour suppressor in late-stage PDAC.


Introduction
The tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) is a member of the tumour necrosis factor (TNF)-family of ligands [1]. Once bound to its receptors, signalling cascades are initiated

TRAIL-R1 Regulates the Expression of miR-370
To analyse the possible impact of TRAIL-R1 on the expression of miRNAs, we performed genome-wide miR-profiling in Panc1 cells with and without knockdown of TRAIL-R1. Interestingly, miR-370 was among the most strongly downregulated miRs in response to TRAIL-R1 depletion ( Figure S1). To verify the array data, we again transfected the Panc1 cells with siRNA against TRAIL-R1 and determined the levels of the mature miR-370-3p by quantitative real-time PCR (qPCR) analysis. As shown in Figure 1A, downregulation of TRAIL-R1 but not TRAIL-R2 ( Figure S2), resulted in significantly reduced levels of miR-370-3p. In an attempt to elucidate the underlying mechanism, we first asked whether TRAIL impacts miR-370 expression. Therefore, control siRNA-transfected Panc1 cells were treated with either recombinant TRAIL or a neutralizing antibody against TRAIL (anti-TRAIL). The abundance of miR-370-3p was not affected by treatment with TRAIL ( Figure 1B) as no significant changes in expression of mature miR-370 was observed. However, antibody-mediated deprivation of endogenous TRAIL reduced miR-370 levels significantly albeit slightly. These results suggest that mature miR-370-3p is positively regulated by TRAIL-R1 in a manner independent of rec. human TRAIL ligand. Here, we demonstrate that knockdown of TRAIL-R1 in PDAC cells results in decreased amounts of mature miR-370-3p associated with higher expression of TGFβ-RII and enhanced TGF-β target gene expression and growth inhibition.

TRAIL-R1 Regulates the Expression of miR-370
To analyse the possible impact of TRAIL-R1 on the expression of miRNAs, we performed genome-wide miR-profiling in Panc1 cells with and without knockdown of TRAIL-R1. Interestingly, miR-370 was among the most strongly downregulated miRs in response to TRAIL-R1 depletion ( Figure S1). To verify the array data, we again transfected the Panc1 cells with siRNA against TRAIL-R1 and determined the levels of the mature miR-370-3p by quantitative real-time PCR (qPCR) analysis. As shown in Figure 1A, downregulation of TRAIL-R1 but not TRAIL-R2 ( Figure S2), resulted in significantly reduced levels of miR-370-3p. In an attempt to elucidate the underlying mechanism, we first asked whether TRAIL impacts miR-370 expression. Therefore, control siRNAtransfected Panc1 cells were treated with either recombinant TRAIL or a neutralizing antibody against TRAIL (anti-TRAIL). The abundance of miR-370-3p was not affected by treatment with TRAIL ( Figure 1B) as no significant changes in expression of mature miR-370 was observed. However, antibody-mediated deprivation of endogenous TRAIL reduced miR-370 levels significantly albeit slightly. These results suggest that mature miR-370-3p is positively regulated by TRAIL-R1 in a manner independent of rec. human TRAIL ligand.  -si were stimulated with anti-TRAIL (10 µg/mL), TRAIL (10 ng/mL) or left untreated. Levels of mature miR-370-3p were quantified by qPCR. (C,D) QPCR analyses of pri-miR-370 levels in Panc1 cells transiently transfected with ctrl.-si or TRAIL-R1-si (C) or in ctrl.-si transfected cells with and without treatment with TRAIL (10 ng/mL) or anti-TRAIL (1 µg/mL) (D). Shown are the mean ± SD of five biological replicates (n = 5), with each one analysed in technical duplicates. The asterisks (*) indicate significance (p < 0.05); n.s.: not significant.
Next, we addressed the question whether TRAIL-R1 regulates miR-370-3p expression at the transcriptional level. For this purpose, we compared the levels of pri-miR-370 in cells with and without knockdown of TRAIL-R1 using qPCR. Although the levels of pri-miR-370 appeared reduced, differences missed statistical significance ( Figure 1C). Likewise, neither treatment with anti-TRAIL nor with recombinant TRAIL affected the abundance of pri-miR-370 relative to control siRNA ( Figure 1D). These results suggest that neither TRAIL-R1 nor TRAIL (in its exogenous or endogenous form) affects miR-370-3p expression at the transcriptional level.

MiR-370-3p Negatively Controls TGFβ-RII in PDAC Cells
Although the regulation of TGFβ-RII by miR-370-3p has been shown in gastric carcinoma [44], data on pancreatic carcinoma are not available so far. To examine if TGFβ-RII is subject to regulation by miR-370-3p in PDAC-derived cells, we transfected Panc1 cells with an artificial miR-370-3p (miR-370-3p mimic) and performed Western blot analysis of TGFβ-RII. As shown in Figure 2, abundance of TGFβ-RII was decreased in miR-370-3p mimic transfected cells relative to control cells at 48 and 72 h after the start of transfection. This indicates that expression of TGFβ-RII protein is inhibited by miR-370-3p. five biological replicates (n = 5), with each one analysed in technical duplicates. The asterisks (*) indicate significance (p < 0.05); n.s.: not significant.
Next, we addressed the question whether TRAIL-R1 regulates miR-370-3p expression at the transcriptional level. For this purpose, we compared the levels of pri-miR-370 in cells with and without knockdown of TRAIL-R1 using qPCR. Although the levels of pri-miR-370 appeared reduced, differences missed statistical significance ( Figure 1C). Likewise, neither treatment with anti-TRAIL nor with recombinant TRAIL affected the abundance of pri-miR-370 relative to control siRNA ( Figure  1D). These results suggest that neither TRAIL-R1 nor TRAIL (in its exogenous or endogenous form) affects miR-370-3p expression at the transcriptional level.

MiR-370-3p Negatively Controls TGFβ-RII in PDAC Cells
Although the regulation of TGFβ-RII by miR-370-3p has been shown in gastric carcinoma [44], data on pancreatic carcinoma are not available so far. To examine if TGFβ-RII is subject to regulation by miR-370-3p in PDAC-derived cells, we transfected Panc1 cells with an artificial miR-370-3p (miR-370-3p mimic) and performed Western blot analysis of TGFβ-RII. As shown in Figure 2, abundance of TGFβ-RII was decreased in miR-370-3p mimic transfected cells relative to control cells at 48 and 72 h after the start of transfection. This indicates that expression of TGFβ-RII protein is inhibited by miR-370-3p. The graph underneath the blot shows results from densitometric quantification of band intensities from three independent experiments (mean ± SD, n = 3). The asterisks (*) indicate significance (p < 0.05) relative to respective untreated control.

TRAIL-R1 Knockdown Increases the Abundance of TGFβ-RII
Since TGFβ-RII is a target of miR-370 ( Figure 2) and knockdown of TRAIL-R1 decreases the cellular levels of miR-370 (Figure 1), we hypothesized that TRAIL-R1 might impact the levels of TGFβ-RII in PDAC cells. To validate this hypothesis, we downregulated the expression of TRAIL-R1 in two PDAC cell lines and analysed the levels of TGFβ-RII by Western blot. As demonstrated in Figure 3A, inhibition of TRAIL-R1 expression via siRNA in Panc1 cells was associated with considerably increased levels of TGFβ-RII. Similar results were obtained with Colo357 cells, which The graph underneath the blot shows results from densitometric quantification of band intensities from three independent experiments (mean ± SD, n = 3). The asterisks (*) indicate significance (p < 0.05) relative to respective untreated control.

TRAIL-R1 Knockdown Increases the Abundance of TGFβ-RII
Since TGFβ-RII is a target of miR-370 ( Figure 2) and knockdown of TRAIL-R1 decreases the cellular levels of miR-370 (Figure 1), we hypothesized that TRAIL-R1 might impact the levels of TGFβ-RII in PDAC cells. To validate this hypothesis, we downregulated the expression of TRAIL-R1 in two PDAC cell lines and analysed the levels of TGFβ-RII by Western blot. As demonstrated in Figure 3A, inhibition of TRAIL-R1 expression via siRNA in Panc1 cells was associated with considerably increased levels of TGFβ-RII. Similar results were obtained with Colo357 cells, which were either transiently transfected with the same siRNA sequences or cells stably transduced with a short-hairpin-RNA (shRNA, sequence different from that of the siRNA) against TRAIL-R1 ( Figure S3). This confirms the presence of a functional axis of TRAIL-R1, miR-370 and TGFβ-RII. were either transiently transfected with the same siRNA sequences or cells stably transduced with a short-hairpin-RNA (shRNA, sequence different from that of the siRNA) against TRAIL-R1 ( Figure  S3). This confirms the presence of a functional axis of TRAIL-R1, miR-370 and TGFβ-RII.  To examine a possible ligand dependency of this novel TRAIL-R1 function, Panc1 cells were either stimulated with TRAIL or incubated with anti-TRAIL ( Figure 3B). Interestingly, the abundance of TGFβ-RII remained unchanged after incubation with TRAIL or anti-TRAIL. We thus conclude that TRAIL-R1 functions independently of its ligand TRAIL in the regulation of TGFβ-RII.

Knockdown of TRAIL-R1 Enhances Activation of Smad and Non-Smad Pathways after TGFβ Stimulation
As TGFβ-RII is indispensable for TGFβ signalling, we addressed the question whether an increase in abundance of this protein after TRAIL-R1 knockdown was associated with enhanced TGFβ signalling activity. This was studied by measuring C-terminal phosphorylation of Smad2 (p-Smad2C), a marker for activation of the canonical TGFβ/Smad signalling pathway as well as by assessing the phosphorylation state of JNK1/2 as indicators of non-Smad signalling. We depleted Panc1 cells of TRAIL-R1 by siRNA transfection and 48 h later treated the cells with TGFβ1. As shown in Figure 3 To examine a possible ligand dependency of this novel TRAIL-R1 function, Panc1 cells were either stimulated with TRAIL or incubated with anti-TRAIL ( Figure 3B). Interestingly, the abundance of TGFβ-RII remained unchanged after incubation with TRAIL or anti-TRAIL. We thus conclude that TRAIL-R1 functions independently of its ligand TRAIL in the regulation of TGFβ-RII.

Knockdown of TRAIL-R1 Enhances Activation of Smad and Non-Smad Pathways after TGFβ Stimulation
As TGFβ-RII is indispensable for TGFβ signalling, we addressed the question whether an increase in abundance of this protein after TRAIL-R1 knockdown was associated with enhanced TGFβ signalling activity. This was studied by measuring C-terminal phosphorylation of Smad2 (p-Smad2C), a marker for activation of the canonical TGFβ/Smad signalling pathway as well as by assessing the phosphorylation state of JNK1/2 as indicators of non-Smad signalling. We depleted Panc1 cells of TRAIL-R1 by siRNA transfection and 48 h later treated the cells with TGFβ1. As shown in Figure 3  We also analysed the non-phosphorylated and phosphorylated forms-the ratio of which reflects the activation state-of the mitogen-activated protein kinase JNK1/2. Depletion of TRAIL-R1 in Panc1 cells was associated with a tendency towards increased levels of p-JNK1/2 when compared to the irrelevant control siRNA ( Figure 4B, lane 2 vs. 1 and lane 4 vs. 3, again statistical significance was tightly missed in three independent experiments). These results suggest that TRAIL-R1 might not only inhibit Smad but also non-Smad, for example, JNK signalling secondary to inhibition of TGFβ-RII.

MiR-370-3p Reduces TGFβ-induced Expression of PAI-1 and SLUG
Above, we have shown that miR-370 controls the expression of TGFβ-RII in a negative fashion. To examine whether alterations in miR-370 levels affect the expression of established TGFβ target genes, we transfected Panc1 cells with the miR-370-3p mimic, or control miR and subsequently treated the cells with rec. TGFβ1 for 24 or 72 h. Interestingly, we noted a lower sensitivity of SERPINE1 (encoding plasminogen activator-inhibitor 1, PAI-1) and SNAI2 (encoding SNAIL2/SLUG) to TGFβ1 stimulation as measured by qPCR ( Figure 5).  We also analysed the non-phosphorylated and phosphorylated forms-the ratio of which reflects the activation state-of the mitogen-activated protein kinase JNK1/2. Depletion of TRAIL-R1 in Panc1 cells was associated with a tendency towards increased levels of p-JNK1/2 when compared to the irrelevant control siRNA ( Figure 4B, lane 2 vs. 1 and lane 4 vs. 3, again statistical significance was tightly missed in three independent experiments). These results suggest that TRAIL-R1 might not only inhibit Smad but also non-Smad, for example, JNK signalling secondary to inhibition of TGFβ-RII.

MiR-370-3p Reduces TGFβ-induced Expression of PAI-1 and SLUG
Above, we have shown that miR-370 controls the expression of TGFβ-RII in a negative fashion. To examine whether alterations in miR-370 levels affect the expression of established TGFβ target genes, we transfected Panc1 cells with the miR-370-3p mimic, or control miR and subsequently treated the cells with rec. TGFβ1 for 24 or 72 h. Interestingly, we noted a lower sensitivity of SERPINE1 (encoding plasminogen activator-inhibitor 1, PAI-1) and SNAI2 (encoding SNAIL2/SLUG) to TGFβ1 stimulation as measured by qPCR ( Figure 5).  We also analysed the non-phosphorylated and phosphorylated forms-the ratio of which reflects the activation state-of the mitogen-activated protein kinase JNK1/2. Depletion of TRAIL-R1 in Panc1 cells was associated with a tendency towards increased levels of p-JNK1/2 when compared to the irrelevant control siRNA ( Figure 4B, lane 2 vs. 1 and lane 4 vs. 3, again statistical significance was tightly missed in three independent experiments). These results suggest that TRAIL-R1 might not only inhibit Smad but also non-Smad, for example, JNK signalling secondary to inhibition of TGFβ-RII.

MiR-370-3p Reduces TGFβ-induced Expression of PAI-1 and SLUG
Above, we have shown that miR-370 controls the expression of TGFβ-RII in a negative fashion. To examine whether alterations in miR-370 levels affect the expression of established TGFβ target genes, we transfected Panc1 cells with the miR-370-3p mimic, or control miR and subsequently treated the cells with rec. TGFβ1 for 24 or 72 h. Interestingly, we noted a lower sensitivity of SERPINE1 (encoding plasminogen activator-inhibitor 1, PAI-1) and SNAI2 (encoding SNAIL2/SLUG) to TGFβ1 stimulation as measured by qPCR ( Figure 5).

MiR-370 Reduces TGFβ-induced Growth Inhibition
In order to test whether miR-370 can also affect more complex cellular responses to TGFβ, for example, growth inhibition, we again transfected Panc1 cells with the miR-370-3p mimic, or control miR and subsequently treated the cells with rec. TGFβ1 for 72 h. Intriguingly, the percentage of viable cells in TGFβ1-treated cultures relative to non-treated control cultures was higher in cultures that had been transfected with the miR-370-3p relative to ctrl.-miR transfected cells ( Figure 6). These data clearly show that miR-370-3p by downregulating TGFβ-RII not only suppresses the TGFβ response of individual genes but also impacts growth arrest, a hallmark feature of TGFβ's tumour suppressor function. represent the mean ± SD of three independent experiments (n = 3). The asterisks (*) indicate significance relative to control (p < 0.05).

MiR-370 Reduces TGFβ-induced Growth Inhibition
In order to test whether miR-370 can also affect more complex cellular responses to TGFβ, for example, growth inhibition, we again transfected Panc1 cells with the miR-370-3p mimic, or control miR and subsequently treated the cells with rec. TGFβ1 for 72 h. Intriguingly, the percentage of viable cells in TGFβ1-treated cultures relative to non-treated control cultures was higher in cultures that had been transfected with the miR-370-3p relative to ctrl.-miR transfected cells ( Figure 6). These data clearly show that miR-370-3p by downregulating TGFβ-RII not only suppresses the TGFβ response of individual genes but also impacts growth arrest, a hallmark feature of TGFβ's tumour suppressor function.

Ectopic Expression of a miR-370-3p Mimic Partially Reverses the TRAIL-R1 shRNA-induced TGFβ Hyperstimulation of PAI-1 Expression
If TRAIL-R1 inhibits TGFβ1-dependent PAI-1 expression through miR-370-3p-mediated downregulation of TGFβ-RII, then the enhancement of the TGFβ1 effect on PAI-1 expression following TRAIL-R1 depletion (see Figure 7) should be rescued by transfection of a miR-370-3p mimic. To this end, transfection of Colo357-shTRAIL-R1 cells with miR-370-3p mimic partially reversed the TGFβ1induced hyperstimulation of PAI-1 expression when compared to a control miR (Figure 8). Together with the results shown in Figure 7, these data confirm the functional link between miR-370-3p, TRAIL-R1 and sensitivity to TGFβ being largely determined by expression of TGFβ-RII.  The asterisk (*) indicates significance relative to control (p < 0.05).

Ectopic Expression of a miR-370-3p Mimic Partially Reverses the TRAIL-R1 shRNA-induced TGFβ Hyperstimulation of PAI-1 Expression
If TRAIL-R1 inhibits TGFβ1-dependent PAI-1 expression through miR-370-3p-mediated down-regulation of TGFβ-RII, then the enhancement of the TGFβ1 effect on PAI-1 expression following TRAIL-R1 depletion (see Figure 7) should be rescued by transfection of a miR-370-3p mimic. To this end, transfection of Colo357-shTRAIL-R1 cells with miR-370-3p mimic partially reversed the TGFβ1-induced hyperstimulation of PAI-1 expression when compared to a control miR (Figure 8). Together with the results shown in Figure 7, these data confirm the functional link between miR-370-3p, TRAIL-R1 and sensitivity to TGFβ being largely determined by expression of TGFβ-RII.

Ectopic Expression of a miR-370-3p Mimic Partially Reverses the TRAIL-R1 shRNA-induced TGFβ Hyperstimulation of PAI-1 Expression
If TRAIL-R1 inhibits TGFβ1-dependent PAI-1 expression through miR-370-3p-mediated downregulation of TGFβ-RII, then the enhancement of the TGFβ1 effect on PAI-1 expression following TRAIL-R1 depletion (see Figure 7) should be rescued by transfection of a miR-370-3p mimic. To this end, transfection of Colo357-shTRAIL-R1 cells with miR-370-3p mimic partially reversed the TGFβ1induced hyperstimulation of PAI-1 expression when compared to a control miR (Figure 8). Together with the results shown in Figure 7, these data confirm the functional link between miR-370-3p, TRAIL-R1 and sensitivity to TGFβ being largely determined by expression of TGFβ-RII.

Discussion
The role of miR-370-3p in malignant cells seems to vary depending on the tumour entity. Whereas some studies found a tumour-promoting function, others conclude that miR-370-3p works as a tumour suppressor [45][46][47]. According to the latter scenario, levels of miR-370 in gastrointestinal stromal tumours, bladder carcinomas and neuroblastomas were reported to be low [48][49][50]. As shown by Xu and co-workers, amounts of mature miR-370-3p declined during the development from pre-malignant liver lesions to end-stage HCC [47]. Furthermore, these authors demonstrated that inhibition of miR-370 resulted in a larger tumour mass, increased invasive potential and higher tumour stage. Moreover, in tissue samples from patients, low levels of miR-370 correlated with reduced survival [47].
Previous studies indicated that miR-370-3p negatively controls expression of TGFβ-RII in gastric carcinoma cells [45], however, equivalent data for PDAC cells are not available. High levels of miR-370-3p would therefore be expected to be associated with lower levels of TGFβ-RII and presumably attenuated TGFβ signalling activity and target gene expression. In agreement with this assumption we found that transfection of cells with a miR-370-3p mimic reduced the sensitivity of SERPINE1 and SNAI2 to TGFβ1 stimulation (see Figure 5) and partially relieved TGFβ1-induced growth arrest (see Figure 6).
Little is known on the mechanisms that control mature miR-370 expression. In addition to their well-known function as membrane-bound death-inducing receptors, TRAIL-R1 and -2 were also found to localize in the nuclei of malignant cells [19]. Recently, it was uncovered that TRAIL-R2, by associating with the nuclear proteins p68, NF45 and hnRNPA1, negatively regulated the maturation of miR let-7, while expression of let-7 target proteins were elevated resulting in a more malignant phenotype [20]. Here, we report that TRAIL-R1, too, engages in modulating miR expression since knockdown of this receptor was accompanied by downregulation of miR-370-3p representing a novel mechanism of positive regulation of miR-370 by TRAIL-R1. The observation that TRAIL-R1 but not treatment of the cells with exogenous TRAIL, impact the expression of miR-370-3p argues in favour of a ligand independent function of TRAIL-R1. Interestingly, however, antibody-mediated deprivation of endogenous TRAIL slightly reduced miR-370-3p levels, the reason of which is not clear at present.
Given the negative association of miR-370 with TGFβ-RII expression on the one hand and its positive association with TRAIL-R1 expression on the other hand, we hypothesized that TRAIL-R1 via modulation of miR-370 should also affect the expression of TGFβ-RII in PDAC cells. Regulation of TGFβ-RII via TRAIL/TRAIL-R system has not been described so far. As predicted, we observed that TRAIL-R1 negatively regulates abundance of TGFβ-RII. We assume that the extra TGFβ-RII protein that appears after TRAIL-R1 knockdown, is cell surface-associated as the TGFβ receptor complex which is central to the regulation of TGFβ signalling is localized at the plasma membrane. This is in line with the changes seen in the sensitivity of cells to TGFβ stimulation with respect to gene expression and growth inhibition. We also analysed whether stimulation of PDAC cells with TRAIL would affect expression of TGFβ-RII, however, TRAIL-R1 appears to function independently of its ligand TRAIL here.
To gain further insight into a possible functional relevance of TGFβ-RII upregulation upon TRAIL-R1 knockdown for the activation of TGFβ signalling, analysis of the canonical Smad signalling pathway revealed a tendency towards an activation after stimulation with TGFβ1. The same tendency was noted for non-Smad TGFβ signalling pathways, for example, JNK1/2, which appeared to be activated after TRAIL-R1 knockdown under both non-stimulated and TGF β1-stimulated conditions. An attempt to quantify the increase in p-Smad2C and p-JNK levels in TRAIL-R1-depleted cells from three experiments, however, yielded p-values slightly greater than 0.05. We are currently performing additional experiments to increase the probability that differences become statistically significant. In addition, we are monitoring activation of Smad3 to further support inhibition of canonical Smad signalling by TRAIL-R1. A possible increase in JNK activation following TRAIL-R1 knockdown independent of TGFβ1 stimulation (see Figure 4, lane 1 vs. 2) may be due to the binding of endogenous TRAIL to TRAIL-R2. This explanation is supported by the fact that TRAIL-R2 but not TRAIL-R1 signals through JNK activation [51]. Finally, we show that TRAIL-R1 knockdown-induced upregulation of TGFβ-RII led to a dramatic increase in the expression of the (Smad-dependent) TGFβ target gene SERPINE1 in Colo357 cells (see Figure 7). Intriguingly, this effect could be partially blocked by transient transfection of the cells with the miR-370-3p mimic (see Figure 8).
Recently, we have uncovered an interdependence of TGFβ and TRAIL-induced signalling in PDAC cells. We found that TGFβ1 decreased the levels of TRAIL-R1 and negatively impacted TRAIL-R1-mediated apoptosis and non-apoptotic signalling [52]. Together, our previous and current data create interesting scenarios depending on the level of expression of TGFβ. Since PDAC cells show high expression levels of TRAIL-R1 [19,20], which positively controls miR-370 expression, we propose that miR-370 is abundant in early-stage PDAC when TGFβ is low. In this miR-370-rich environment, mRNA transcribed from TGFBR2 will be degraded and, hence, TGFβ signalling and TGFβ-driven tumourigenesis inhibited. In line with this assumption, high levels of cytoplasmic TRAIL-R1 expression in tumour cells have been shown to correlate with better prognosis of patients with PDAC [53] and other tumour entities (reviewed in Ref. 19).
In contrast, in later stages of PDAC when TGFβ1 levels are high in tumours, this scenario will change dramatically. TGFβ1 is then expected-via inhibiting TRAIL-R1 [52]-to downregulate also miR-370. Interestingly, the fraction of pancreatic cancer samples with positive membrane staining for TRAIL-R1 was found to be lower than that of cells from surrounding non-tumour tissues [54]. Likewise, miR-370 expression, too, is underexpressed in both pancreatic cancer precursor lesions [37] and cancer cells [55] compared to normal healthy tissue and cells, respectively. A more direct link between TGFβ and miR-370-3p recently came from an ovarian cancer cell model [56]. Here, TGFβ treatment downregulated miR-370-3p and a miR-370-3p knockdown or miR-370-3p overexpression promoted or suppressed, respectively, TGFβ-induced epithelial-mesenchymal transition [56] the latter process of which is considered a crucial event in TGFβ's protumourigenic function. Downregulation of miR-370 may relieve TGFβ-RII expression from inhibition and eventually further stimulate TGFβ expression [57] resulting in hyperactivation of TGFβ signalling. This vicious cycle of TGFβ signalling [57] may be responsible for driving aggressive behaviour, invasive potential and metastasis in late-stage PDAC and other gastrointestinal stroma-rich tumours.

Conclusions
In summary, we present here a novel mechanism of TRAIL-R1-dependent, miR-370-mediated regulation of the expression of TGFβ-RII that finally resulted in corresponding alterations in TGFβ-dependent target gene expression and growth inhibition. Many miRs were found to be deregulated in PDAC cells correlating with worsened patient outcome, chemoresistance and invasive behaviour [58]. Although selective modulation of miRs is still not trivial [59], pharmacological intervention to decrease or increase distinct miRs to improve therapy outcomes in patients with pancreatic carcinomas is nevertheless a promising strategy. Eventually, miR-370 may emerge as a promising candidate among other TGFβ signalling inhibitors that are currently evaluated in anticancer therapy [60][61][62].

Culture Conditions, Cell Stimulation and Counting
The human pancreatic cancer cell lines Panc1 and Colo357 were cultured in RPMI 1640 media supplemented with 10% FCS, 1 mM GlutaMAX and 1 mM sodium pyruvate (Life Technologies, Darmstadt, Germany). For selection of stably retrovirally transduced Colo357 cells the antibiotic puromycin was added to the culture media (2.5 µg/mL). Cells were grown for 24 h in six-well-plates before transfection and/or stimulation. Cell-stimulating agents used were TGFβ1 (ReliaTech, Wolfenbüttel, Germany), recombinant human TRAIL (10 ng/mL; PeproTech, Hamburg, Germany) and neutralizing TRAIL antibody (anti-TRAIL) (10 µg/mL; R&D Systems). For cell counting, cells were detached by trypsinization and viable cells (as determined by trypan blue exclusion) counted in a Neubauer chamber.

Transfection of siRNA, shRNA and miR-370
1.7 × 10 5 cells were seeded in six-well-plates the day before transfection. Control siRNA and siRNA against TRAIL-R1 were used in a concentration of 2 nM, both of them purchased from GE Healthcare (ON-TARGETplus ® SMARTpool non-targeting pool D-001810-10, L-008090-00; GE Healthcare, Freiburg, Germany). To obtain cells with a stable knockdown of TRAIL-R1, Colo357 cells were transduced with GIPZ lentiviral shRNAmir vectors for TRAIL-R1 or with a non-silencing control (Open Biosystems; CloneID: V3LHS_383714). MirVana ® miR mimic hsa-miR-370-3p and the respective control were purchased from Ambion, Life Technologies, Darmstadt, Germany. DharmaFECT ® 1 (Thermo Scientific, Schwerte, Germany) or Lipofectamin 2000 (Life Technologies/ThermoFisher Scientific, Darmstadt, Germany served as transfection reagents and were used according to the manufacturers' instructions. Forty-eight h after transfection, cells were stimulated with the aforementioned factors and then subjected to protein extraction or RNA isolation.

Statistical Analysis
All data were normally distributed and analysed using two-tailed student's t-test. A p value < 0.05 was considered statistically significant. Data are shown as mean ± standard deviation (SD). Statistical analysis was performed using Sigma Plot 12.5 software (Systat Software, Erkrath, Germany).