Trametinib and dactolisib but not regorafenib exert antiproliferative effects on rat pancreatic stellate cells
Laura Witteck and Robert Jaster
Rostock, Germany

BACKGROUND: Modulation of the stroma response is con- sidered a promising approach for the treatment of chronic pancreatitis and pancreatic cancer. The aim of this study was to evaluate the effects of three clinically available small mol- ecule kinase inhibitors, regorafenib, trametinib and dactolisib, on effector functions of activated pancreatic stellate cells (PSCs), which play a key role in pancreatic fibrosis.
METHODS: Cultured rat PSCs were exposed to small mol- ecule kinase inhibitors. Proliferation and cell death were as- sessed by measuring the incorporation of 5-bromo-2′-deoxy- uridine and cytotoxicity, respectively. Levels of mRNA were determined by real-time PCR, while protein expression and phosphorylation were analyzed by immunoblotting. Interleu- kin-6 levels in culture supernatants were quantified by ELISA. Zymography assays were performed to monitor collagenase activity in culture supernatants.
RESULTS: The MEK inhibitor trametinib and the dual phos- phatidylinositol 3-kinase/mTOR inhibitor dactolisib, but not the multi-kinase inhibitor regorafenib, efficiently inhibited PSC proliferation. Trametinib as well as regorafenib suppressed the expression of two autocrine mediators of PSC activation, interleukin-6 and transforming growth factor-β1. Dactolisib- treated cells expressed less α1 type I collagen and lower levels of α-smooth muscle actin, a marker of the myofibroblastic PSC phenotype. Simultaneous application of dactolisib and trametinib displayed additive inhibitory effects on cell growth without statistically significant cytotoxicity. Activity of matrix metalloproteinase-2 was not affected by any of the drugs.

CONCLUSION: We suggest the combination of two drugs, that specifically target two key signaling pathways in PSC, Ras-Raf-MEK-ERK (trametinib) and phosphatidylinositol 3-kinase-AKT-mTOR (dactolisib), as a concept to modulate the activation state of the cells in the context of fibrosis.
(Hepatobiliary Pancreat Dis Int 2015;14:642-650) KEY WORDS: stellate cell biology;
pancreatic cancer; chronic pancreatitis; trametinib; dactolisib

Introduction
ancreatic stellate cells (PSCs) are the main source of extracellular matrix (ECM) proteins in the diseased pancreas. In response to profibrogenic mediators such as ethanol metabolites and various cyto- kines/growth factors, the cells undergo a transition from a quiescent to an activated stage.[1, 2] Quiescent PSCs (roughly 4% of all pancreatic cells[3]) are phenotypically characterized by the presence of vitamin A-containing fat droplets[3, 4] and have recently been suggested to ex- hibit stem cell properties.[5] Activated PSCs proliferate at a high rate, synthesize large amounts of collagen type I and other ECM proteins and express the myofibroblastic marker α-smooth muscle actin (α-SMA).[3, 4] PSC activa- tion is part of a physiological wound healing response

after pancreatic injury and not pathological per se. Under

Author Affiliations: Department of Medicine II, Division of Gastroen- terology, Rostock University Medical Center, E.-Heydemann-Str. 6, 18057 Rostock, Germany (Witteck L and Jaster R)
Corresponding Author: Robert Jaster, MD, Department of Medicine II, Division of Gastroenterology, Rostock University Medical Center, E.- Heydemann-Str. 6, 18057 Rostock, Germany (Tel: +49-381-4947349; Fax:
+49-381-4947482; Email: [email protected])
© 2015, Hepatobiliary Pancreat Dis Int. All rights reserved. doi: 10.1016/S1499-3872(15)60032-7
Published online November 9, 2015.

persistent pathophysiological conditions, however, PSC activation is sustained and not terminated, resulting in an overshooting matrix deposition. This is typically the case in chronic pancreatitis and pancreatic cancer (PC), where pancreatic tissue is replaced by connective tissue, leading to organ fibrosis.[1, 2] Although the pathophysi- ological mechanisms are not entirely understood yet, it is well established that paracrine mediators (e.g., derived from inflammatory cells and tumor cells) as well as au-

tocrine loops play a key role in the maintenance of PSC activation in chronic pancreatitis and PC.[1, 2] Impor- tant examples of such mediators include transforming growth factor β1 (TGF-β1), the main stimulator of ECM synthesis in PSC, and the proinflammatory cytokine in- terleukin-6 (IL-6), which both may act in an autocrine and paracrine manner.[6-9]
Numerous studies have shown that the stroma re- sponse not only accompanies PC but plays an active role in its progression (reviewed in[10]). Thus, the connective tissue forms a fibrotic wall around tumor cells, thereby creating a barrier against chemotherapeutics and im- mune cells. Cytokines, chemokines and growth factors which are secreted by stroma cells have also been impli- cated in chemoresistance, suppression of apoptosis and accelerated proliferation of PC cells.[11-15] Targeting the stroma, therefore, has been suggested as a complemen- tary strategy to treat PC, one of the deadliest human malignancies with very limited therapeutic options. Ac- cordingly, efforts have been made to develop specific antifibrotic therapies,[10] which, however, are not estab- lished in the clinics yet. Moreover, two recent studies[16, 17] have put the simple concept of depleting the stroma for an improved cancer treatment into question. In a genetic mouse model of PC, reduction of the stroma through a gene deletion of the profibrogenic ligand sonic hedge- hog was unexpectedly associated with more aggressive tumors and increased tumor angiogenesis.[16] In a trans- genic mouse model, a more invasive tumor growth was observed when α-SMA-positive myofibroblasts were specifically deleted.[17] Together, both studies indicated that at least some components of the tumor stroma can act to restrain tumor growth. Most recently, the origi- nal concept has therefore been refined by suggesting a stroma modulation (rather than a stroma depletion), in a way that pro-tumorigenic interactions between stroma and cancer cells are blocked and anti-tumorigenic effects of the stroma remain preserved.[18] One prerequisite to achieve this goal is to gain more mechanistic insights into key effector functions of activated PSCs.
We and others have previously shown that two in- tracellular signaling cascades, Ras-Raf-MEK (mitogen- activated protein kinase kinase)-ERK (extracellular sig- nal-regulated kinase) and phosphatidylinositol 3-kinase (PI3K)-AKT (protein kinase B)-mTOR (mammalian tar- get of rapamycin), play a crucial role in PSC activation in response to mitogens and profibrogenic mediators.[19-21] Furthermore, we have also tested clinically available small molecule kinase inhibitors (SMIs) and found that two anticancer drugs, the multi-kinase inhibitors sorafenib and sunitinib, efficiently blocked activation of the AKT pathway and displayed distinct antifibrogenic effects.[22] Here, we have advanced our studies by investigating the

effects of two more recent and specific drugs, the MEK inhibitor trametinib[23] and the dual PI3K/mTOR inhibi- tor dactolisib,[24] on cultured rat PSC. For comparison, an additional multi-kinase inhibitor, the potent antineo- plastic agent regorafenib,[25] was included into the inves- tigations.

Methods
Reagents
Regorafenib, trametinib and dactolisib were purchased from Selleckchem (Houston, TX, USA). Unless stated otherwise, all other reagents were obtained from Sigma- Aldrich (Deisenhofen, Germany).

Cell culture
PSCs were isolated from the pancreas of healthy male Lewis inbred rats (Charles River Laboratories, Sulz- bach, Germany) essentially as described before.[19] Briefly, the organ was digested with collagenase, and PSC were separated by Nycodenz® (Nycomed, Oslo, Norway) den- sity gradient centrifugation. Immediately after isolation, PSCs were resuspended in cryopreservation medium [fe- tal calf serum (FCS) supplemented with 10% dimethyl sulfoxide (DMSO)] and stored at -150 ℃ until required. After thawing, the cells were cultured in Iscove’s modi- fied Dulbecco’s medium (IMDM; Biochrom, Berlin, Ger- many) supplemented with 17% FCS, 1% non-essential amino acids (dilution of a 100× stock solution), 105 U/L penicillin and 100 mg/L streptomycin (all reagents from PAA Laboratories, Pasching, Austria). Compared to our previous studies where freshly isolated PSCs were cul- tured immediately,[19] the cycle of freezing and thawing only insignificantly affected the number of vital cells for the experimental studies. On the other hand, it proved very useful to standardize experimental conditions, e.g., when comparing different preparations of PSCs. After approximately 5 days in primary culture, proliferat- ing PSCs reached subconfluency and were harvested by trypsination. Unless indicated otherwise, the cells were subsequently plated in 12-well plates or 24-well plates at average seeding densities of 6×104 and 3×104 cells per well, respectively, and further grown and treated accord- ing to the experimental requirements. All experiments were performed with cells passaged no more than two times. Trypan blue staining was used to distinguish live from dead cells and to determine absolute cell counts.

Quantification of DNA synthesis
Cell proliferation was assessed by quantifying incor- poration of 5-bromo-2′-deoxyuridine (BrdU) into newly

synthesized DNA, employing the BrdU labelling and detection enzyme-linked immunosorbent assay (ELISA) kit (Roche Diagnostics, Mannheim, Germany). There- fore, cells were plated in 96-well plates at equal seeding densities (2×103 cells per well) and allowed to adhere overnight. Subsequently, regorafenib, trametinib and dactolisib (solved in DMSO; final solvent concentration here and in all further experiments ≤0.1%) were added as indicated. After 24 hours, BrdU labelling was initiated by adding labelling solution at a final concentration of 10 µmol/L. Another 24 hours later, labelling was stopped, and BrdU uptake was measured according to the manu- facturer’s instructions.

Quantification of cytotoxicity
Cytotoxic effects of the investigated SMI were ana- lyzed using the CytoTox-Glo Cytotoxicity Assay accord- ing to the instructions of the manufacturer (Promega, Madison, WI, USA). Therefore, PSCs were seeded into 96-well plates at equal seeding densities (2×103 cells per well) and allowed to adhere overnight, before regorafenib, trametinib and dactolisib were added as indicated. After 48 hours, CytoTox-Glo Cytotoxicity Assay reagent was added to all wells, and incubation continued for 15 min- utes at room temperature. Luminescence (originating from dead cells) was measured using a Glomax Multi De- tection System (Promega) for 96-well plates. Afterwards, the remaining viable cells were lysed by incubation with lysis reagent for another 15 minutes, and luminescence was recorded again. Since the second luminescence signal corresponds to all cells, the ratio of the first and the sec- ond luminescence signals indicates the rate of cell death in the original sample.

Immunoblotting
Protein extracts of PSCs (growing in 24-well plates and pretreated as indicated) were prepared and sub- jected to immunoblotting analysis as described before,[19] using polyvinylidene fluoride membrane (Merck Mil- lipore, Darmstadt, Germany) for protein transfer. Unless specified otherwise, primary antibodies were obtained from New England BioLabs (Frankfurt, Germany). The following antibodies were employed: anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH; #2118), anti- phospho-AKT (P-AKT; #4060), anti-phospho-ERK1/2 (P-ERK1/2; #4370), anti-AKT protein (#4691), and anti- ERK1/2 (#06-182, Merck Millipore). The blots were de- veloped using LI-COR reagents for an Odyssey® Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA) as described previously.[26] Signal intensities were quanti- fied by means of the Odyssey® software version 3.0 and the raw data processed as described in the corresponding

figure legend.

IL-6 ELISA
Quantitative determinations of IL-6 protein levels were performed using a rat IL-6-specific ELISA (Bio- Legend, San Diego, CA, USA). Therefore, PSC growing in 12-well plates were pretreated by SMI application as indicated. Cell culture supernatants were collected and stored at -80 ℃ until use. The ELISA was performed ac- cording to the manufacturer’s instructions.

Zymography
PSCs were plated in 24-well plates, grown to subcon- fluency, washed free of FCS and cultured for 24 hours in serum-free culture medium supplemented with SMI as indicated. Afterwards, the supernatants were collected and mixed 3:1 with 4× zymography sample buffer (250
mmol/L Tris-HCl, pH 6.8, 10% SDS, 40% glycerol, 0.1% bromophenol blue). Subsequently, samples and matrix metalloproteinase-2 (MMP2) zymography standard were loaded onto 10% precast polyacrylamide gel with gelatin (Bio-Rad Laboratories, Hercules, CA, USA). Af- ter electrophoresis, the gels were soaked in zymogram renaturation buffer (Bio-Rad Laboratories) with gentle shaking (60 minutes at room temperature; one change of detergent solution), before they were incubated over- night at 37 ℃ in zymogram development buffer (Bio-Rad Laboratories). Afterwards, lysis bands were visualized using Coomassie G-250 SimplyBlue SafeStain according to the instructions of the manufacturer (Thermo Fisher Scientific, Darmstadt, Germany). Finally, band intensi- ties were quantitated with the Odyssey® Infrared Imag- ing System (software version 3.0), taking advantage of the fluorophore properties of protein-bound Coomassie stain.

Quantitative RT-PCR using real-time TaqMan tech- nology
PSC growing in 12-well plates were treated with SMI for 24 hours as indicated, and total RNA was isolated using TriFast reagent (PEQLAB Biotechnologie, Erlan- gen, Germany). All further steps were performed with reagents from Thermo Fisher Scientific. First, any traces of genomic DNA were removed employing the DNA- free kit. Afterwards, 1 µg of RNA was reverse transcribed into cDNA by means of TaqMan Reverse Transcription Reagents and random hexamer priming. Relative quan- tification of target cDNA levels by real-time PCR was performed in a Viia 7 sequence detection system (Thermo Fisher Scientific). Therefore, TaqMan Universal PCR Master Mix and rat-specific TaqMan Gene Expression

Assays with fluorescently labelled MGB probes were used. The following assays were employed: Rn01759928_g1 (Acta2; α-smooth muscle actin; α-SMA), Rn00432360_m1 (cyclin D1; Ccnd1), Rn00589996_m1 (cyclin-dependent kinase inhibitor 1A; Cdkn1A), Rn00561420_m1 (Il-6), Rn00572010_m1 (Tgf-β1), Rn01463848_m1 (α1 type I collagen, Col1a1), and Rn01527840_m1 (hypoxanthine guanine phosphoribosyl transferase; Hprt; housekeeping gene control). PCR conditions were: 95 ℃ for 10 minutes, followed by 40 cycles of 15 seconds at 95 ℃/1 minute at 60 ℃. The relative expression of each mRNA (n≥5 inde- pendent samples per experimental condition) compared with HPRT was calculated according to the equation
∆Ct=Cttarget-CtHPRT. The relative amount of target mRNA in control cells and cells treated with SMI as indicated

Fig. 1B (subgroup analyses), staurosporine was the only substance that displayed a significant cytotoxic effect. We noticed that application of dactolisib at concentrations
≥30 nmol/L (alone and in combination with trametinib) also was associated with increased rates of cell death, but this effect was not statistically significant.
Together, these data suggest that inhibition of cell proliferation, rather than induction of cell death, reduces BrdU incorporation in response to trametinib and dac- tolisib treatment.

was expressed as 2-∆∆Ct, where ∆∆Ct

=∆Ct

-∆Ct .

treatment

SMI

control

Statistical analysis
All data were stored and analyzed using the IBM SPSS Statistics 20.0. Values were expressed as mean± standard error (SE) of the mean for the indicated num- ber of separate cultures per experimental protocol. Sta- tistical significance was checked using one-way ANOVA followed by Bonferroni’s post-hoc test. If data did not meet the assumptions for ANOVA, analysis of variance was performed employing the Kruskal-Wallis test, before subgroups were tested using the Mann-Whitney U test. P<0.05 (Bonferroni-adjusted as indicated in the figure legends) was considered to be statistically significant.

Results
Effects of regorafenib, trametinib and dactolisib on DNA synthesis and cell death
SMI application was associated with significant effects on DNA synthesis (PKruskal-Wallis<0.001). The subsequent anal- ysis of subgroups revealed a dose-dependent inhibition of BrdU incorporation by trametinib and dactolisib at nanomolar concentrations (Fig. 1A). Furthermore, two combinations of trametinib and dactolisib were found to be more effective than each substance at the correspond- ing concentration alone, suggesting additive effects of MEK and PI3K inhibition. Unexpectedly, micromolar concentrations of the multi-kinase inhibitor regorafenib did not diminish but even enhance DNA synthesis.
To distinguish between SMI effects on cell prolifera- tion and cell death (which both may affect BrdU incor- poration), cytoxicity of the drugs was evaluated. As a positive control, the pro-apoptotic agent staurosporine was included. Indeed, drug treatment of PSC affected cell survival (PKruskal-Wallis=0.013). However, as shown in

Fig. 1. Effects of SMI on DNA synthesis and survival of PSCs. PSCs growing in primary culture were harvested, replated at equal seeding densities and treated with drugs as indicated for 24 hours. Control cultures were exposed to culture medium supplemented with the solvent of the three SMIs, DMSO, only. A: The pretreated cells were labeled with BrdU for another 24 hours and prolifera- tion was assessed with the BrdU DNA incorporation assay. One hundred percent BrdU incorporation corresponds to solvent- treated PSC. Data were presented as mean±SE (n≥18 separate cul- tures); *: PU-test<0.0033, versus control cultures with Bonferroni- adjusted α=0.0033; #: PU-test<0.0033, versus cultures treated with each of the single substances alone with Bonferroni-adjusted α=0.0033. B: Incubation was continued for another 24 hours, before dead cells were quantified using the CytoTox-Glo Cytotox- icity Assay. One hundred percent cell death corresponds to the lu- minescence signal measured after cell lysis. Data were presented as mean±SE (n≥5 separate cultures). *: PU-test<0.0045, versus control cultures with Bonferroni-adjusted α=0.0045.

Ras-Raf-MEK-ERK and PI3K-AKT signaling in SMI- treated PSC
To gain insights into the molecular action of rego- rafenib, trametinib and dactolisib, we focused on the in- tracellular signal transduction pathways Ras-Raf-MEK- ERK and PI3K-AKT, which both play a key role in PSC activation and represent targets of the SMI tested in this study.[19-25] Specifically, we asked how the three drugs af- fect activation of AKT and ERK in PSC, using levels of

P-AKT and P-ERK1/2 as surrogate markers (Fig. 2A-D). In accordance with previously published data,[22] serum restimulation of serum-starved PSCs induced a rapid phosphorylation of both AKT and ERK1/2. Global analysis of variance revealed that both AKT and ERK activation were significantly affected in SMI-treated PSC at any time point of FCS stimulation (AKT: PKruskal-Wallis≤ 0.009; ERK: PKruskal-Wallis≤0.03). Further analysis showed that dactolisib significantly inhibited phosphorylation of AKT but not ERK, while trametinib strongly inhibited

Fig. 2. Distinct effects of regorafenib, trametinib and dactolisib on phosphorylation of AKT and ERK1/2. Cultured PSCs were grown to subconfluency before the standard culture medium was substituted by FCS-free medium. Fifteen hours later, culture medium was supplemented with regorafenib, trametinib and dactolisib at the indicated concentrations and incubation continued for another hour. Afterwards, FCSs (at 17%) were added for the indicated periods of time. Protein extracts from equal amounts of the cells were sub- jected to Western blotting analysis. P-AKT, P-ERK1/2, the respective total proteins and GAPDH (for loading control) were detected using fluorescein (IRDye)-labelled secondary antibodies. For each inhibitor (A, regorafenib; B, trametinib and C, dactolisib), one rep- resentative Western blotting is shown. D: Fluorescence signal intensities of phospho-proteins (P-AKT and P-ERK1/2, respectively) and corresponding total proteins were quantified using Odyssey® software. Subsequently, the ratios P-AKT/AKT protein (upper panel) and P-ERK/ERK protein (lower panel) were determined. A ratio of 1 corresponds to cells that were cultured without SMI and stimulated with FCS for 5 minutes. E: Dose-dependent effects of dactolisib on the ratio P-AKT/AKT protein were analyzed in serum-starved PSC treated with FCS (at 17%) for 5 minutes, following the protocol described above. A ratio of one hundred percent corresponds to cells cultured without dactolisib. Data in (D) and (E) are expressed as mean±SE (n≥5 separate cultures); *: PU-test<0.0166, versus control cul- tures (no SMI; identical time of FCS stimulation) with Bonferroni-adjusted α=0.0166.

activation of ERK but even enhanced phosphorylation of AKT. Interestingly, the multi-kinase inhibitor regorafenib blocked AKT activation as efficient as dactolisib, but did not affect the levels of P-ERK.
Since lower doses of dactolisib were used in the biological assays as well (Fig. 1), we tested the dose- dependency of the dactolisib effect. At the time point of 5 minutes after FCS-restimulation, PSCs pretreated with 100 and 30 nmol/L dactolisib displayed significantly low- er P-AKT levels than untreated controls (Fig. 2E). The effect of the higher dose of the drug, however, was much stronger and corresponded to the more pronounced bio- logical activity (Fig. 1A).

SMI target genes in PSC
Employing real-time PCR, we characterized the gene expression profiles of PSCs after 24 hours of SMI treat- ment. Fig. 3 shows the results for selected genes that were chosen based on their established role in PSC function and ECM metabolism, and the results of pilot experi- ments. In all cases, global analysis of variance indicated the presence of significant differences among the groups (PKruskal-Wallis≤0.001). Further analysis revealed for each of the three inhibitors a specific action profile.
Despite its stimulatory effect on BrdU incorpora- tion, regorafenib inhibited expression of cyclin D1 and induced expression of the cell cycle inhibitor CDKN1A.

Furthermore, regorafenib-treated PSCs displayed in- creased mRNA levels of the PSC activation marker α-SMA and α1 type I collagen. In contrast, expression of the profibrogenic mediator TGF-β1 was diminished and expression of IL-6 was virtually blocked. Trametinib dis- played similar effects on PSC expression as regorafenib, except for that CDKN1A and α1 type I collagen mRNA levels were not significantly affected by the drug. The ac- tion profile of dactolisib differed strongly from the other two SMIs: The drug reduced the mRNA levels of α-SMA and α1 type I collagen, but triggered the expression of TGF-β1. Cyclin D1, CDKN1A and IL-6 levels remained unchanged. In summary, each of the three drugs simul- taneously displayed effects on gene expression that favor PSC activity (proliferation, synthesis of ECM) and that inhibit the functions of these cells. When combined, however, trametinib and dactolisib displayed a more homogenous action profile. Expression of cyclin D1, α1 type I collagen and IL-6 was reduced, while the mRNA levels of the other three genes remained unaffected.

SMI effects on IL-6 secretion
To verify the striking effects of regorafenib and dactolisib, IL-6 was also assayed on the protein level by measuring the cytokine concentration in supernatants of PSC cultures (Fig. 4). Indeed, lower IL-6 levels (compared to control cells) were detected when PSCs were exposed

Fig. 3. Gene expression profiles of SMI-treated PSCs. Cultured PSCs were exposed to SMI at the indicated concentrations for 24 hours. The mRNA expression of (A) cyclin D1, (B) CDKN1A, (C) α-SMA, (D) collagen type I, (E) TGF-β1, (F) IL-6, and the housekeeping gene HPRT was analyzed by real-time PCR, and relative amounts of target mRNA were calculated as described in the Methods section. One hundred percent mRNA expression of each gene corresponds to cells exposed to the SMI solvent, DMSO, only. Data of n≥5 inde- pendent cultures were used to calculate mean±SE. *: P<0.0083, versus control cultures with Bonferroni-adjusted α=0.0083.

Fig. 4. Effects of SMI on IL-6 secretion. Subconfluent cultures of PSCs were exposed to SMI at the indicated concentrations. Twenty-four hours later, the supernatants were collected, and IL-6 protein levels in cell culture supernatants were analyzed by ELISA. Mean values and SE were calculated from n≥3 independent sam- ples. Statistical significance was checked using one-way ANOVA.
*: P<0.0083, versus control cultures with Bonferroni-adjusted
α=0.0083.

Fig. 5. Collagenase activities in supernatants of SMI-treated PSCs. PSCs were grown to subconfluency before they were exposed to SMI at the indicated concentrations under serum-free conditions. A: After 24 hours, supernatants were harvested and subjected to zymography as described in the Methods section. In the first lane, a MMP2 standard (5 ng) was loaded. B: Lysis band intensities were quantitated by scanning densitometry. One hundred percent intensity corresponds to DMSO-treated control cultures. Data were expressed as mean±SE (n≥5 separate cultures).

to regorafenib, trametinib or a combination of trametinib and dactolisib. These inhibitory effects, however, were less pronounced than on the level of mRNA.

Analysis of collagenase activities
Supernatants of SMI-treated PSCs were also assayed for collagenase activity by zymography. In accordance with previously published data,[27] strong lytic bands were detected in the molecular range of MMP2 (Fig. 5A), the dominant gelatinase secreted by activated PSCs. A quantitative assessment (Fig. 5B) showed no significant differences between the samples, suggesting that MMP2 secretion/activity was not affected by the SMI treatment.

Fig. 6. Mechanisms of trametinib and dactolisib action in PSCs. Serum growth factors stimulate PSC proliferation and fibrogen- esis through the binding to their cognate membrane receptors and the activation of intracellular signal transduction pathways, such as Ras-Raf-MEK-ERK and PI3K-AKT. The protein kinase mTOR represents one of the downstream effector molecules of AKT. The signaling cascades are depicted in a highly simplified manner. Crosstalk and feedback loops are not considered. Dactolisib inhib- its PI3K and mTOR, whereas trametinib targets MEK. Together, both drugs block two key signaling pathways of PSC activation and therefore exert synergistic effects in the cells.

Discussion
Stroma cells are increasingly acknowledged as an im- portant target for the treatment of stroma-rich tumors, such as PC.[10, 28] Specifically, PSCs have been implicated in the progression of PC through paracrine effects favor- ing tumor growth and metastasis as well as through the secretion of ECM proteins that mediate chemoresistance and provide protection against anti-tumor immune re- sponses.[11, 15] On the other hand, stroma depletion alone may be deleterious and result in more aggressive tumors with increased vascularity and even heightened prolif- eration.[16, 17] Antifibrotic therapy will, therefore, have to consider both sides of the coin, the stroma and the tu- mor cells themselves.
In this study, we took advantage of three SMIs that have been developed as antineoplastic agents. None of these substances has been shown to be beneficial in the treatment of PC, and we do not claim direct clinical rel- evance of our in vitro findings. The results of our stud- ies, however, do show that two of the drugs, the MEK inhibitor trametinib and the dual PI3K/mTOR inhibitor dactolisib, are effective inhibitors of PSC proliferation and may combine antineoplastic with antifibrotic effects. Each of the two substances targets a key signaling path- way in PSC activation, Ras-Raf-MEK-ERK (trametinib)

and PI3K-AKT-mTOR (dactolisib), respectively.[19-21, 23, 24] Consistently, a combination of trametinib and dactolisib was particularly effective regarding the inhibition of DNA synthesis. On the other hand, the rate of cell death was increased by trend only. At the level of gene expres- sion, the same combination diminished the levels of cy- clin D1, α1 type I collagen and IL-6, which is also in line with an antifibrotic efficiency.
An unexpected result was the transient increase of P-AKT levels in trametinib-treated PSC. Although simi- lar findings have previously been made in a colorectal cancer cell line,[29] such an effect cannot immediately be explained from the known action profile of trametinib as an MEK inhibitor, suggesting molecular follow-up studies.
The failure of regorafenib to inhibit PSC prolifera- tion was surprising for several reasons. Thus, we have previously shown that two other multi-kinase inhibitors, sorafenib and sunitinib, display dose-dependent antip- roliferative effects in cultured PSCs.[22] Furthermore, the growth-stimulatory effect of regorafenib was restricted to primary PSCs, since cells of an immortalized rat PSC line[30, 31] responded to the drug by growth retardation (data not shown). Finally, regorafenib blocked AKT sig- naling as effective as dactolisib. The latter observation suggests that regorafenib may also inhibit additional signaling pathways that, in primary PSC, transduce an- timitotic signals, thereby overriding the antiproliferative effect of AKT inhibition. Interestingly, regorafenib never- theless also displayed effects which we interpret as antifi- brotic, such as inhibition of IL-6 and TGF-β1 expression.
Taken together, the results of this study show that clinically available inhibitors of Ras-Raf-MEK-ERK and PI3K-AKT-mTOR signaling exert specific antifibrotic effects on cultured rat PSC. A combination of two anti- neoplastic agents, trametinib and dactolisib, displayed strong growth-inhibitory effects without significant cy- totoxicity. These data suggest a particular antifibrotic ef- ficiency of drug combinations that simultaneously target the two key signaling pathways in PSC proliferation, Ras- Raf-MEK-ERK and PI3K-AKT-mTOR (Fig. 6). When
combined, trametinib and dactolisib efficiently inhibited DNA synthesis already at relatively low doses, an aspect that is potentially interesting with respect to side effects of the drugs in vivo. We conclude that the antifibrotic efficiency of trametinib and dactolisib should be further evaluated in the context of PC-associated fibrosis, using animal models of the disease.

Acknowledgements: We gratefully acknowledge the excellent tech- nical assistance of Mrs. Katja Bergmann.
Contributors: WL and JR designed the research, performed the experiments and analyzed the data. JR wrote the manuscript. Both authors reviewed and contributed to the final manuscript. JR is

the guarantor.
Funding: None.
Ethical approval: Not needed.
Competing interest: No benefits in any form have been received or will be received from a commercial party related directly or in- directly to the subject of this article.

References
1 Erkan M, Adler G, Apte MV, Bachem MG, Buchholz M, Detlef- sen S, et al. StellaTUM: current consensus and discussion on pancreatic stellate cell research. Gut 2012;61:172-178.
2 Jaster R, Emmrich J. Crucial role of fibrogenesis in pancreatic diseases. Best Pract Res Clin Gastroenterol 2008;22:17-29.
3 Apte MV, Haber PS, Applegate TL, Norton ID, McCaughan GW, Korsten MA, et al. Periacinar stellate shaped cells in rat pancreas: identification, isolation, and culture. Gut 1998;43: 128-133.
4 Bachem MG, Schneider E, Gross H, Weidenbach H, Schmid RM, Menke A, et al. Identification, culture, and characteriza- tion of pancreatic stellate cells in rats and humans. Gastroen- terology 1998;115:421-432.
5 Kordes C, Sawitza I, Götze S, Häussinger D. Stellate cells from rat pancreas are stem cells and can contribute to liver regen- eration. PLoS One 2012;7:e51878.
6 Schneider E, Schmid-Kotsas A, Zhao J, Weidenbach H, Schmid RM, Menke A, et al. Identification of mediators stimulating proliferation and matrix synthesis of rat pancreatic stellate cells. Am J Physiol Cell Physiol 2001;281:C532-543.
7 Shek FW, Benyon RC, Walker FM, McCrudden PR, Pender SL, Williams EJ, et al. Expression of transforming growth factor- beta 1 by pancreatic stellate cells and its implications for matrix secretion and turnover in chronic pancreatitis. Am J Pathol 2002;160:1787-1798.
8 Aoki H, Ohnishi H, Hama K, Shinozaki S, Kita H, Yamamoto H, et al. Existence of autocrine loop between interleukin-6 and transforming growth factor-beta1 in activated rat pancreatic stellate cells. J Cell Biochem 2006;99:221-228.
9 Mews P, Phillips P, Fahmy R, Korsten M, Pirola R, Wilson J, et al. Pancreatic stellate cells respond to inflammatory cytokines: potential role in chronic pancreatitis. Gut 2002;50:535-541.
10 Haqq J, Howells LM, Garcea G, Metcalfe MS, Steward WP, Dennison AR. Pancreatic stellate cells and pancreas can- cer: current perspectives and future strategies. Eur J Cancer 2014;50:2570-2582.
11 Müerköster S, Wegehenkel K, Arlt A, Witt M, Sipos B, Kruse ML, et al. Tumor stroma interactions induce chemoresistance in pancreatic ductal carcinoma cells involving increased secre- tion and paracrine effects of nitric oxide and interleukin-1beta. Cancer Res 2004;64:1331-1337.
12 Bachem MG, Schünemann M, Ramadani M, Siech M, Beger H, Buck A, et al. Pancreatic carcinoma cells induce fibrosis by stimulating proliferation and matrix synthesis of stellate cells. Gastroenterology 2005;128:907-921.
13 Qian LW, Mizumoto K, Maehara N, Ohuchida K, Inadome N, Saimura M, et al. Co-cultivation of pancreatic cancer cells with orthotopic tumor-derived fibroblasts: fibroblasts stimu- late tumor cell invasion via HGF secretion whereas cancer cells exert a minor regulative effect on fibroblasts HGF production. Cancer Lett 2003;190:105-112.

14 Armstrong T, Packham G, Murphy LB, Bateman AC, Conti JA, Fine DR, et al. Type I collagen promotes the malignant phe- notype of pancreatic ductal adenocarcinoma. Clin Cancer Res 2004;10:7427-7437.
15 Schneiderhan W, Diaz F, Fundel M, Zhou S, Siech M, Hasel C, et al. Pancreatic stellate cells are an important source of MMP-2 in human pancreatic cancer and accelerate tumor pro- gression in a murine xenograft model and CAM assay. J Cell Sci 2007;120:512-519.
16 Rhim AD, Oberstein PE, Thomas DH, Mirek ET, Palermo CF, Sastra SA, et al. Stromal elements act to restrain, rather than support, pancreatic ductal adenocarcinoma. Cancer Cell 2014;25:735-747.
17 Özdemir BC, Pentcheva-Hoang T, Carstens JL, Zheng X, Wu CC, Simpson TR, et al. Depletion of carcinoma-associated fibroblasts and fibrosis induces immunosuppression and ac- celerates pancreas cancer with reduced survival. Cancer Cell 2014;25:719-734.
18 Wang Z, Li J, Chen X, Duan W, Ma Q, Li X. Disrupting the bal- ance between tumor epithelia and stroma is a possible thera- peutic approach for pancreatic cancer. Med Sci Monit 2014;20: 2002-2006.
19 Jaster R, Sparmann G, Emmrich J, Liebe S. Extracellular signal regulated kinases are key mediators of mitogenic signals in rat pancreatic stellate cells. Gut 2002;51:579-584.
20 Masamune A, Kikuta K, Satoh M, Kume K, Shimosegawa T. Differential roles of signaling pathways for proliferation and migration of rat pancreatic stellate cells. Tohoku J Exp Med 2003;199:69-84.
21 McCarroll JA, Phillips PA, Kumar RK, Park S, Pirola RC, Wil- son JS, et al. Pancreatic stellate cell migration: role of the phosphatidylinositol 3-kinase(PI3-kinase) pathway. Biochem Pharmacol 2004;67:1215-1225.
22 Elsner A, Lange F, Fitzner B, Heuschkel M, Krause BJ, Jaster
R. Distinct antifibrogenic effects of erlotinib, sunitinib and sorafenib on rat pancreatic stellate cells. World J Gastroenterol 2014;20:7914-7925.
23 Gilmartin AG, Bleam MR, Groy A, Moss KG, Minthorn EA, Kulkarni SG, et al. GSK1120212 (JTP-74057) is an inhibitor of

MEK activity and activation with favorable pharmacokinetic properties for sustained in vivo pathway inhibition. Clin Can- cer Res 2011;17:989-1000.
24 Maira SM, Stauffer F, Brueggen J, Furet P, Schnell C, Fritsch C, et al. Identification and characterization of NVP-BEZ235, a new orally available dual phosphatidylinositol 3-kinase/mam- malian target of rapamycin inhibitor with potent in vivo anti- tumor activity. Mol Cancer Ther 2008;7:1851-1863.
25 Wilhelm SM, Dumas J, Adnane L, Lynch M, Carter CA, Schütz G, et al. Regorafenib (BAY 73-4506): a new oral multikinase in- hibitor of angiogenic, stromal and oncogenic receptor tyrosine kinases with potent preclinical antitumor activity. Int J Cancer 2011;129:245-255.
26 Rateitschak K, Karger A, Fitzner B, Lange F, Wolkenhauer O, Jaster R. Mathematical modelling of interferon-gamma signal- ling in pancreatic stellate cells reflects and predicts the dynam- ics of STAT1 pathway activity. Cell Signal 2010;22:97-105.
27 Jaster R, Hilgendorf I, Fitzner B, Brock P, Sparmann G, Emmrich J, et al. Regulation of pancreatic stellate cell function in vitro: biological and molecular effects of all-trans retinoic acid. Biochem Pharmacol 2003;66:633-641.
28 Rucki AA, Zheng L. Pancreatic cancer stroma: understanding biology leads to new therapeutic strategies. World J Gastroen- terol 2014;20:2237-2246.
29 Lange F, Franz B, Maletzki C, Linnebacher M, Hühns M, Jaster
R. Biological and molecular effects of small molecule kinase inhibitors on low-passage human colorectal cancer cell lines. Biomed Res Int 2014;2014:568693.
30 Jaster R, Lichte P, Fitzner B, Brock P, Glass A, Karopka T, et al. Peroxisome proliferator-activated receptor gamma overexpres- sion inhibits pro-fibrogenic activities of immortalised rat pan- creatic stellate cells. J Cell Mol Med 2005;9:670-682.
31 Sparmann G, Hohenadl C, Tornøe J, Jaster R, Fitzner B, Koc- zan D, et al. Generation and characterization of immortalized rat pancreatic stellate cells. Am J Physiol Gastrointest Liver Physiol 2004;287:G211-219.

Received July 20, 2015
Accepted after revision October 8, 2015