Specimen collection was approved by the ethics committee of the University of Heidelberg (Decision no. S-134-2010), and written informed consent was obtained from the patients. The study was performed with tissue samples obtained from patients admitted to the Department of General, Visceral and Transplantation Surgery, University Hospital Heidelberg, for surgical treatment of PDAC or benign disease. Diagnoses were established by a pathologist according to the World Health Organization (WHO) classification.

The human pancreatic cancer cell lines PANC-1, MIA PaCa-2, and BxPC-3 were purchased from ATCC (LGC Standards, Wesel, Germany). They were cultured in a humidified atmosphere containing 5% CO₂ at 37°C. RPMI-1640 supplemented with 10% fetal calf serum (FCS) was used as the medium for PANC-1. MIA PaCa-2 cells were cultured in DMEM, with 10% FCS and 2.5% horse serum added.

The following antibodies were used: mouse anti-KRAS (sc-30, Santa Cruz Biotechnology, Heidelberg, Germany), rabbit anti-LC3A/B (#4108, Cell Signaling, Danvers, MA, USA), rabbit anti-Akt (#9272, Cell Signaling), rabbit anti-phospho-AKT (Ser473, #9271, Cell Signaling), rabbit anti-TFEB (#4240, Cell Signaling), anti-TFEB (V-17, Santa Cruz Biotechnology), mouse monoclonal anti-TFEB (clone 3E1-G6, Sigma-Aldrich, Taufkirchen, Germany), anti-PARP (#9542, Cell Signaling), and rabbit anti-GAPDH (#2118, Cell Signaling).

The chemotherapeutic agent gemcitabine (Hexal, Holzkirchen, Germany) was added to the cells after their treatment with siRNAs in indicated concentrations, and incubated for 48–72 h. Choloroquine (chloroquine disphophate salt, Sigma-Aldrich) was diluted in H₂O to the indicated concentrations.

The plasmid ptfLC3 was purchased from Addgene (#21074, Cambridge, MA, USA). It encodes for LC3 tagged with mRFP and EGFP to distinguish between autophagosomes and autophagolysosomes based on its loss of GFP in lysosomal acidic environment (17). For the transfection, Lipofectamine 2000 (Life Technologies, Darmstadt, Germany) was used according to the manufacturer's protocol.

Transfection of siRNAs was performed using HiPerfect Transfection reagent (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The following sequences were targeted by siRNAs: KRAS 5′-GGCTATATTTACATGCTACTA-3′ [Eurofins MWG Operon, Ebersberg, Germany (18)], TFEB 5′-CACAACTTAATTGAAAGGAGA-3′ (Qiagen), and KIF11 (Eg5) 5′-AACTGAAGACCTGAAGACAAT-3′ as the positive control (19); and nonsense siRNA (Allstars, Qiagen) served as the negative control. All transfections were performed with a concentration of 30 nM per target gene and an incubation time of 48–72 h.

RNA was isolated with the Nucleo Spin II kit (Macherey-Nagel, Düren, Germany) and converted to cDNA with the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturers' protocols. The quantitative RT-PCR was performed using the Power SYBR Green Master Mix (Life Technologies). ACTB served as a housekeeping gene for normalization of C_t-values. The Delta-Delta-C_t-method was used to quantify gene expression levels. The specific TFEB primers were obtained from Search-LC (Heidelberg, Germany). RT-PCR results in human tissues are presented as TFEB transcripts per 10,000 cyclophilin B copies (10 kCPB).

Cells were washed and lysed on ice by RIPA buffer containing protease and phosphatase inhibitors (Thermo Fisher Scientific, Rockford, IL, USA). For extracting proteins, cell lysates were centrifuged for 15 min at 11,000 × g followed by concentration measurement of the supernatant with the Pierce BCA Protein assay kit (Thermo Fisher Scientific). A total of 10–15 μg protein per sample was then mixed with 4X LDS sample buffer and Sample Reducing Agent (Thermo Fisher Scientific) and heated to 90°C for 10 min with gentle agitation according to the manufacturer's protocol of the NuPAGE SDS-PAGE Gel system (Thermo Fisher Scientific). Probes were separated electrophoretically in a 4–12% Bis-Tris NuPAGE Gel (Thermo Fisher Scientific) and blotted onto a nitrocellulose membrane using the semidry blot NuPAGE system. After blotting, the membrane was blocked for 1 h at room temperature in a blocking solution, containing 5% Slim Fast powder (Allpharm Vertriebs GmbH, Messel, Germany) dissolved in TBS-Tween 0.1% buffer. The blots were then incubated with primary antibodies (anti-Akt 1:1,000, anti-GAPDH 1:5,000, anti-KRAS 1:100, anti-LC3A/B 1:500, anti-phospho-Akt 1:1,000, anti-TFEB 1:500) in a blocking solution overnight at 4°C. After three washing steps with TBS-Tween 0.1% buffer, the membrane was incubated with HRP-tagged, host-specific secondary antibodies (Cell Signaling) for 1 h at room temperature. After three additional washing steps, chemiluminescence solution (Immobilon Western Chemi, Millipore, Billerica, MA, USA) was added for detection of protein bands using G:Box Chemi XT4 (Syngene, Cambridge, UK). GAPDH served as the loading control. Quantification of protein intensity was performed using ImageJ (Version 2.0.0; http://imagej.net/).

The paraffin-embedded tissue sections were immunostained with horseradish peroxidase (HRP) labeled polymer (Dako Deuschland GmbH, Hamburg, Germany). Shortly, consecutive tissue sections were deparaffinized and rehydrated in progressively decreasing concentrations of ethanol. After the antigens were retrieved by boiling the tissue sections in 10 mM citrate buffer for 3×5 min in the microwave oven, endogenous peroxidase activity was blocked with 3% hydrogen peroxide in methanol for 10 min. The slides were then placed in a washing buffer (10 mM Tris-HCl, 0.85% NaCl, 0.1% bovine serum albumin, pH 7.4), and the sections were incubated overnight at 4°C with the rabbit anti-TFEB antibody (#4240, Cell Signaling) diluted in antibody diluent (Dako) or rabbit immunoglobulin as negative control (Dako). Then, the slides were incubated with a secondary anti-rabbit-labeled polymer (Dako) for 45 min at room temperature. The immunoreactivity was visualized using a 3,3′-diaminobenzidine (DAB) chromogen/H₂O₂ system (Dako), and the sections were counterstained with Mayer's hematoxylin. Finally, the slides were viewed with an Axiophot microscope (Carl Zeiss, Göttingen, Germany) coupled with a charge-coupled device (CCD) camera (AxioVision, Zeiss, Jena, Germany).

Cells were grown on collagen-A (0.1 mg/ml) coated glass cover slips and treated with serum starvation or siRNA transfection. Immunostaining was performed as described previously (20). Briefly, cells were fixed with 2% paraformaldehyde, permeabilized by 0.3% Triton X-100 for 5 min, and incubated in a blocking solution (2% fetal calf serum, 2% bovine serum albumin, 0.2% fish gelatine resolved in PBS). Samples were then incubated with the primary antibody for 1 h. After several washing steps with PBS, cover slips were treated with Cy3- or Alexa Fluor 488-tagged secondary antibodies. Cell nuclei were stained with DAPI (1:200). Confocal imaging was performed on a Leica SP5 laser-scanning microscope using the 63× oil-immersion objectives (Leica Microsystems, Heidelberg, Germany). The fluorescence intensity of TFEB was quantified by calculating the ratio of the mean intensity in the nucleus and cytoplasm of the respective cells using ImageJ (Version 2.0.0; http://imagej.net/).

Cells including the supernatant were collected in a FACS tube. After centrifugation for 5 min at 4°C and 1,500 rpm, the supernatant was discarded and pellets were used for propidium iodide (PI)/Annexin V apoptosis assays. For the apoptosis assay, the samples were washed twice with FACS binding buffer (10 mM HEPES pH 7.4, 2.5 mM CaCl₂, 140 nM NaCl), and PI (50 μg/ml) (Sigma-Aldrich) and Annexin V (BD Biosciences, Heidelberg, Germany) were added. Measurements were performed with FACSCalibur (BD Biosciences) and analyzed using WinMDI (Scripps Research Institute, La Jolla, CA, USA). Apoptotic cells were defined as the sum of PI⁺/Annexin V⁺ (late apoptosis) and PI⁻/Annexin V⁺ (early apoptosis) cells.

The relative cell count was determined, which reflects proliferation inhibition and apoptosis. The treated cells were harvested by trypsinization and counted with the TC-20 Automated Cell Counter (Bio-Rad, München, Germany). Cell numbers were normalized to the media controls.

If not indicated otherwise, data are presented as mean ± standard error of the mean (SEM) and the two-sided, unpaired t-test was used as the statistical test. Results were analyzed and plotted using the Prism 5 software (GraphPad Software, La Jolla, CA, USA).

The expression of TFEB transcription was analyzed in human normal pancreas tissue, in PDAC samples and in liver metastases. The expression levels in primary tumor samples (n=45) were significantly elevated compared with normal pancreas tissue (P<0.001), and PDAC liver metastasis had also higher TFEB expression compared with normal pancreas (P=0.02, Fig. 1A). Immunohistochemistry revealed a weak to moderate cytoplasmic TFEB expression in normal ductal cells, but no nuclear staining; moderate or high TFEB staining and nuclear localization was found in 11 of 15 cancer specimens (P=0.02; Fisher's exact test) (Fig. 1B).

Likewise, TFEB expression was investigated in human PDAC cell lines. Immunoreactivity was compared using three different, specific antibodies directed against TFEB. The primary PDAC cell lines PANC-1 and MIA PaCa-2 showed both cytoplasmic and nuclear TFEB expression under normal culture conditions. Serum starvation led to minor enhancement of the nuclear staining (Fig. 1C and D). The ratio of nucleus and cytoplasmic fluorescence intensity was significantly increased by starvation (P=0.01, n=35 cells). These results indicate TFEB activation in PDAC cells even under basal conditions and functional sensing of the available nutrient supply.

To assess whether TFEB expression affected autophagy in PDAC cells, a transient TFEB RNA interference (siRNA) was established, which achieved a mean protein downregulation to 23 and 11% in PANC-1 and MIA PaCa-2 cells, respectively (Fig. 2A and B). Both PDAC cell lines responded to starvation with increased expression of TFEB and microtubule-associated protein 1A/1B-light chain 3 (LC3)-II. The starvation-induced upregulation of LC3-II was even maintained during TFEB silencing (Fig. 2C and D). We further explored whether TFEB silencing affected the autophagy route from autophago-to autophagolysosomes. To this end, a red fluorescent protein-green fluorescent protein (RFP-GFP) tandem tagged LC3 protein was exogenously expressed to differentiate autophagosomes (GFP and RFP-positive) from autophagolysosomes (GFP-negative, but RFP-positive) (17). However, there was neither a measurable difference in the LC3-positive vesicle number, nor in the ratio of autophago-to autophagolysosomes after TFEB silencing (Fig. 2E). Consequently, TFEB is an indicator of autophagy in PDAC cells, but low expression levels are sufficient to run the autophago-to-lysosome machinery. This was further supported by chemotherapeutic treatment of the cancer cells with gemcitabine, which caused a nearly dose-dependent LC3-II accumulation using gemcitabine at concentrations ≤100 μM, indicating increased autophagy, regardless of the amount of TFEB protein (Fig. 2F and G). Thus, the autophagy response of PDAC cells to gemcitabine did not solely depend on the TFEB expression levels.

Previous data have demonstrated that autophagy sustained PDAC cell survival and counteracted chemotherapeutic cytotoxicity, but it is unknown whether these effects are mediated by TFEB (5,7). Therefore, cell death (apoptosis) induced by gemcitabine treatment was examined after TFEB siRNA depletion. Only 20–30% of the cancer cells went into apoptosis at concentrations >1 μM gemcitabine, and there was no significant difference depending on the TFEB expression level (Fig. 3A and B). Because it has been shown that the known inhibitor of autophagosome degradation, chloroquine (CQ), can suppress PDAC growth, we further explored whether the cytotoxic effect of CQ depends on TFEB activity. Autophagy inhibition by CQ was seen in both cell lines (Fig. 3C and D). Induction of apoptosis was found in 15–50% of the cells using established concentrations of CQ, but there was no difference in the apoptosis rate when TFEB expression was silenced (Fig. 3E and F).

Because the results indicated that autophagy sustained tumor cell survival in PDAC, we further investigated the crosstalk between the PDAC driver oncogene KRAS and autophagy. Ras signaling can promote, but also inhibit, autophagy through use of different cascades including Rac1, Raf1, and phosphoinositide-3-kinase (PI3K) signaling (10). The latter PI3K downsignaling involves the Akt/mTOR1 pathway, and can negatively regulate autophagy activity (21). Silencing of KRAS expression increased cell death and apoptosis of the cancer cells independently of TFEB expression, and TFEB knockdown had no significant effect on cell death (Fig. 4A–C). Blockage of KRAS expression led to an unchanged or minor increase in the LC3-II fraction, which was not accompanied by changes in Akt phosphorylation (Fig. 4D). However, KRAS silencing resulted in reciprocally augmented levels of TFEB protein and transcripts (Fig. 4E and F), whereas TFEB knockdown did not enhance KRAS expression. If oncogenic KRAS activity instead of KRAS transcription levels would determine TFEB expression, wild-type PDAC cells (e.g., BxPC-3) were expected to harbor higher TFEB concentrations. However, TFEB expression was comparably low in the KRAS wild-type BxPC-3 cells (data not shown). These findings were supported by analysis of published data of an inducible KRAS mouse model; oncogenic KRAS activity status (on/off) did not influence TFEB expression, but KRAS expression levels negatively and significantly correlated with TFEB expression (Fig. 4G and H). Moreover, the inverse relationship of KRAS and TFEB expression levels could be confirmed in human PDAC samples, and again, there was no TFEB dependence on the KRAS mutation status (Fig. 4I).