All experiments were performed according to the German regulations for animal experimentation. The study was approved by the Regierung von Unterfranken as the responsible authority (Permit Number 55.2-2531.01-76/10). All surgery was performed under esketamine/xylazine anesthesia, and all efforts were made to minimize suffering.

C57Bl/6 deficient for TNF (B6.129S-Tnf^tm1Gkl/J, short B6.TNF KO), TNFR1 (C57BL/6-Tnfrsf1a^tm1Imx/J, short B6.TNFR1 KO), TNFR2 (B6.129S7-Tnfrsf1b^tm1Imx/J, short B6.TNFR2 KO), TNFR1R2 (B6.129S-Tnfrsf1a^tm1ImxTnfrsf1b^tm1Imx/J, short B6.TNFR1R2 KO) were initially obtained from Jackson Laboratories (Bar Harbor, ME, USA) and backcrossed to the albino C57Bl/6 background (C57BL/6J-Tyr<c-2J mice, Jackson Laboratories) for improved in vivo bioluminescence imaging sensitivity (reduced light absorption due to lack of melanin in the skin) as described previously [16]. Genotypes of KO mice were routinely checked by PCR. Female mice were used for experiments between 8 and 12 weeks of age. All mice were bred within the specified pathogen-free animal facility of the Center for Experimental Molecular Medicine of the University Hospital, Würzburg receiving rodent chow and autoclaved drinking water ad libitum.

For lentiviral transduction of Panc02 cells [24], 293 T cells were transiently transfected with a standard calcium phosphate precipitation protocol in 10 cm dishes with 10 µg pMDL and 5 µg RSV-REV packaging plasmids, 6 µg VSV/G envelope plasmid and 20 µg of the eGFP and firefly luciferase encoding plasmid FUGLW. Two days later, the supernatant containing the lentiviral particles was aspirated, filtered through a 0.45-µm filter, 8 µg polybrene/ml were added and the mixture was used to transduce the tumor cells. The transduced cells were flow sorted twice for eGFP-expression, termed hereafter Panc02-FUGLW. Cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% antibiotics (penicillin, streptomycin), L-glutamine and 0.1% β-mercaptoethanol. Cells were trypsinized and passaged twice weekly. Cell culture medium and supplements were obtained from Invitrogen (Darmstadt, Germany), all plastic ware was from Greiner BioOne (Frickenhausen, Germany). Panc02-FUGLW cells are syngeneic to C57BL/6 mice.

Panc02-FUGLW cells were trypsinized, harvested and washed twice with PBS. Recipient mice were anesthetized with i.p. injection of 80 mg/kg body weight (bw) esketamine hydrochloride (Pfizer, Berlin, Germany) and 16 mg/kg bw xylazine (cp Pharma, Burgdorf, Germany) and placed on a 37°C heating plate. Panc02-FUGLW cells were injected orthotopically as described elsewhere [17], [25] with slight modifications. Briefly, the abdominal cavity was opened by a minimal invasive transverse laparotomy. The head of the pancreas was identified and externalized. 1×10⁴ viable Panc02-FUGLW cells were slowly injected in 30 µl PBS into the head of the pancreas using a 710 RN 100 µl Hamilton syringe with a Gauge 28, 10 mm, Point Style 4 needle (Hamilton Syringe, Bonaduz, Switzerland). The pancreas was placed back into the abdominal cavity and the peritoneum and the skin were closed by running single-layer of 6-0 polyglactin sutures (Johnson & Johnson, Norderstedt, Germany).

For TNF treatment, mice were injected i.p. with 5 µg human TNF in 200 µl PBS every other day starting on the day of tumor cell inoculation. For in vivo bioluminescence imaging [26], mice were anesthetized and co-injected with 300 mg/kg bw D-luciferin (Biosynth, Staad, Switzerland). Ten minutes later, bioluminescence signals of the anesthetized mice were assessed with an IVIS Spectrum imaging system (Caliper Life Sciences, Mainz, Germany). Pictures were taken from the lateral view in automatic mode with a maximum exposure time of five minutes per picture. Pictures were evaluated using Living Image 4.0 software (Caliper Life Sciences).

On day 23 or 30 after tumor cell inoculation, mice were injected with D-luciferin and 10 minutes later euthanized. Internal organs were removed and subjected to ex vivo bioluminescence imaging [26]. Tissue samples were embedded in Tissue Tek OCT (Sakura Finetek, Staufen, Germany) for further histological analysis.

Tissues were minced with a surgical blade and digested for 45 minutes at 37°C with 2 mg/ml collagenase D and 0.1 mg/ml DNase I (both from Roche, Mannheim, Germany). Tissue pieces were mashed through a 70 µm cell strainer and spun down. The cell pellet was resuspended in erythrocyte lysis buffer (168 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM ethylenediaminetetraacetic acid (EDTA)) and incubated for 2 minutes. Then 10 volumes of PBS were added and the cells were spun down again. The resulting pellet was resuspended in PBS and cells were used for flow cytometry. Spleens were directly filtered through a 70 µm cell strainer into erythrocyte lysis buffer and washed once with PBS.

Cryo-embedded tissues were cut into 3 µm thick sections on a Leica CM1900 cryostat (Leica Microsystems, Wetzlar, Germany) and mounted onto frosted slides. Slides were air-dried and fixed with acetone at room temperature for 7 minutes. Slides were washed and blocked with 2% FBS in PBS for 15 minutes. When biotin-conjugated antibodies were used, additional blocking using an Avidin/Biotin Blocking kit (Vector Laboratories, Burlingame, CA, USA) was performed. Slides were then incubated with the appropriate antibodies for 1 hour at room temperature. Between antibody-incubations, the slides were washed with PBS thrice. Slides were counterstained with DAPI and mounted with mounting medium (Vector Laboratories). Antibodies used were: CD11b-Alexa 647 (M1/70), CD31-Biotin (MEC13.3), CD4-Alexa 647 (GK1.5), CD8-Biotin (53-6.7), Foxp3-purified (FJK-16s) (eBioscience), donkey-anti-rat-Cy3 (Dianova, Hamburg, Germany), F4/80-Alexa 488 (CI:A3-1), GR-1-Biotin (RB6-8C5), streptavidin-Alexa 546 (Invitrogen). Images were obtained with a Zeiss Imager.Z1m fluorescence microscope (Carl Zeiss, Göttingen, Germany) and evaluated using Zeiss AxioVision software (Carl Zeiss). Immune cells were counted and given as cells/mm² or as pixels/150 000 µm² as assessed by Image J software (NIH, Bethesda, MD).

Cells were blocked with normal rat serum (1∶20 in PBS) and stained with appropriate antibodies at 4°C for 30 min. Following surface antigen staining, cells were washed once with PBS and labelled with propidium iodide. For intracellular staining, cells were stained with LIVE/DEAD fixable violet dead cell stain kit (Invitrogen) and further processed using the Mouse regulatory T cell staining kit #2 (eBioscience, Frankfurt, Germany) according to the manufacturer’s protocol. Antibodies used were from Biolegend (Uithoorn, The Netherlands) if not stated otherwise: α4β7-PE (DATK32), CCR4-APC (2G12), CCR5-PE (C34-3448), CCR7-APC (4B12), CD102-Biotin (3C4 (MIC2/4)), CD103-Pacific Blue (2E7), CD106-Alexa 488 (429), CD107a-FITC (1D4B), CD107b-Alexa 647 (M3184), CD11a-PE (M17/4) (BD), CD11b-Alexa 647 (M1/70), CD11b-PE/Cy7 (M1/70), CD11c-Alexa 647 (N418), CD25-PE (PC61.5) (eBioscience), CD24-PE (LG.7F9) (eBioscience), CD29-PE (HMβ1-1), CD4-FITC (RM4-5) (eBioscience), CD4-APC (RM4-5), CD44-PE (IM7), CD45.1-PE (A20), CD45.2-APC (104), CD49d-Alexa 647 (RI-2), CD49f-Alexa 488 (GoH3), CD54-PE (YN1/1.7.4), CD62E-PE (10E9.6) (BD), CD62L-APC/Cy7 (MEL-14), CD69-Pacific Blue (H1.2F3), CD8-PE/Cy7 (53-6.7), CXCR3-APC (CXCR3-173), CXCR4-Alexa 488 (2B11) (eBioscience), E-Selectin IgG fusion protein (R&D Systems, Wiesbaden, Germany), F4/80-Alexa 488 (C1:A3-1), Foxp3-APC (FJK-16s) (eBioscience), Ly6G-PE (1A8) (BD), P-Selectin IgG fusion protein (BD), TNFR1-APC (55R-286), TNFR2-PE (TR75-89), goat-anti-human IgG-PE (Jackson), streptavidin-Alexa 546 (Invitrogen). All experiments were performed on a BD FACS Canto II (BD) and sample data recorded using BD FACSDiva software and analyzed using FlowJo software (Tree Star, Ashland, OR, USA).

RNA was isolated from tissue samples using Qiashredder and RNeasy mini kit spin columns (Qiagen, Hilden, Germany). 1 µg total RNA was reverse transcribed using the QuantiTect Reverse Transcription Kit (Qiagen). Expression for genes of interest (primer sequences can be found in table 1) was analysed on an iCycler thermocycler (Bio-Rad, Munich, Germany) using iTaq Universal SYBR Green Supermix (Bio-Rad) and β-actin as reference gene. Expression levels were calculated using the ΔΔC_T method.

1×10⁵ Panc02 cells per well were seeded in 6 well plates and left overnight. Cells were treated with 1.67 nM human TNF or 1.67 nM murine TNF for 48 h, harvested by gently scraping the monolayer off the plastic surface, washed twice with PBS and assessed for the expression of adhesion molecules and chemokine receptors by flow cytometry.

To test the capability of Panc02 cells to attach to different extracellular matrix components, the CytoSelect 48-well Cell Adhesion Assay (Cell Biolabs, San Diego, CA, USA) was used according to the manufacturer’s instruction. To test the capability of Panc02 cells to invade the basement membrane, the CytoSelect 24-Well Cell Invasion Assay (Cell Biolabs) was used according to the manufacturer’s instruction.

Untreated and TNF-treated Panc02 cells were further assessed for the activity of matrix metalloproteinases MMP-2 and MMP-9 by gelatin zymography. For protein isolation, myocardium tissue of an infarcted mouse heart as well as Panc02-FUGLW cell pellets were homogenized with Ripa buffer and PMSF (Cell Signaling, Frankfurt, Germany). The samples were separated on a 10% polyacrylamide gel containing 2.5 mg/ml gelatin at a constant voltage of 120 V for 2 h at 4°C. After electrophoresis, the proteins were renaturated by incubation of the gels in 2.5% Triton X-100 for 90 min at room temperature. The gels were then incubated in activation buffer (50 mM Tris-HCl, pH 7.5, 5 mM CaCl2, 0.2 M NaCl, and 0.02% Brij-35) for 12 h at 37°C. Finally, the gels were stained for 1 h with 0.5% coomassie blue staining solution and then destained in 40% v/v methanol, 10% v/v acetic acid to reveal bands of clearing which indicate proteolytic activity. The band intensity was quantified using ImageJ (version 1.44p).

All graphs shown are combined data from at least two independent experiments; the number of animals is indicated in the figure legends. All data are shown as mean ± standard error of mean. Figures were prepared using GraphPad Prism 5 software (La Jolla, CA, USA) and Adobe Photoshop 7 (San Jose, CA, USA). All groups were compared to the wild type or untreated control group, respectively by two-tailed unpaired student’s t-test using GraphPad InStat 3 software. Data reaching statistical significance are indicated as: * p ≤ 0.05, ** p ≤ 0.01.

In order to follow tumor growth non-invasively in a syngeneic mouse model of PDA, we generated Panc02 cells expressing eGFP and firefly luciferase (Panc02-FUGLW) and injected 1×10⁴ of these tumor cells orthotopically into albino wild type C57Bl/6 mice and albino C57Bl/6 mice deficient for TNF, TNFR1, TNFR2 or both TNFRs. In all genotypes assessed, Panc02-FUGLW tumors initially grew for the first 7 days following tumor inoculation (Figure 1A and B). 3/8 B6 wild type mice, 1/8 B6.TNF KO mice, and 2/9 B6.TNFR2 KO mice spontaneously rejected the tumor within 14 days. In contrast 13/13 mice deficient for TNFR1 or TNFR1 and TNFR2 could not reject the tumor. In vivo BLI (Figure 1A and B) and ex vivo BLI one month after tumor inoculation (Figure 1C and D) also revealed significantly higher tumor burden in mice deficient for TNFR1 (or TNFR1 and TNFR2) than in wild type mice. Thus, TNFR1 appeared as an important factor for the control and rejection of Panc02 tumors.

Next, we analyzed immune cell infiltration into Panc02-FUGLW tumors grown in either C57Bl/6 wild type or C57Bl/6 mice deficient for TNF, TNFR1 or TNFR2 (Figure 2 and Table 2) to address whether the loss of tumor control might be related to changes in the immune cell infiltrate of the tumor. Loss of TNFR1 correlated with decreased numbers of infiltrating CD8⁺ T cells and increased numbers of infiltrating CD4⁺Foxp3⁺ T_regs. TNFR1 deficiency furthermore significantly increased vascular density as assessed by CD31-expression but did not significantly modify infiltration with innate immune cells of the myeloid lineage (CD11b⁺, F4/80⁺, Gr1⁺ cells). The altered T cell infiltration patterns of Panc02-FUGLW tumors from B6.TNFR1 KO one month after inoculation prompted us to assess the T cell phenotype in the early stage of tumor expansion. Therefore, we analyzed the expression of activation markers on CD4⁺ and CD8⁺ T cells 7 and 15 days after tumor cell injection (Figure 3). T cell activation markers only subtly differed. Yet, we observed a trend towards higher activation with time of CD8⁺ T cells (expression of CD25 and CD69) and lower activation of CD4⁺ T cells (expression CD27, CD54, and CD69). Tumor growth was strongly associated with myeloid inflammation. Flow cytometry revealed increased percentages of CD11b⁺ and Ly6G⁺ cells in the pancreas over time independent of the genetic background of tumor bearing mice. Interestingly, on a systemic level, loss of TNFR1 resulted in significantly decreased CD11b⁺ and Ly6G⁺ cell numbers in the spleens of tumor-bearing mice. Increased tumor growth within pancreata of TNFR1 KO mice was confirmed by significantly elevated numbers of eGFP⁺ Panc02-FUGLW cells in these mice two weeks after tumor inoculation (Figure 3).

To further assess the mechanistic differences of tumor control between wild type and TNFR1 KO mice, we assessed the expression of immunosuppressive genes and various cytokines in Panc02-FUGLW tumors (Figure 4). Here, we observed on a transcriptional level significantly increased expression of Galectin-9 and IL-4 while the expression of iNOS was significantly reduced.

TNF was shown to enhance tumor cell metastasis in several in vivo mouse models [27]–[29] including an orthotopic xenotransplantation model of PDA with pancreatectomy-induced metastasis [17]. To test whether TNF also influences Panc02-FUGLW tumor growth and metastasis, we treated tumor bearing mice with recombinant human TNF, which only binds to murine TNFR1, every other day for three weeks following tumor cell inoculation. This significantly increased overall tumor growth within pancreata and ablated spontaneous tumor rejection as assessed with in vivo and ex vivo BLI (Figure 5A and B) but did not induce metastasis to the liver, the kidneys, or the mesentery (data not shown). We also analyzed the infiltration of CD8⁺ and CD4⁺ T cells, and T_regs into tumors of untreated and TNF-treated animals and indeed observed that human TNF treatment significantly increase T_reg numbers within the tumors (Table 3). Since we and others have found previously that TNFR2 rather than TNFR1 on T_regs is required to drive their expansion [14]–[16], this points to an indirect mechanism by which administration of human TNF triggers T_reg expansion in our PDA model.

Due to the limited metastatic activity of Panc02 cells in vivo, we analyzed the metastatic capabilities of these cells in vitro particularly also upon TNF stimulation. Panc02 cells adhered to extracellular matrix proteins fibronectin, laminin I, and fibrinogen, but neither to collagen I nor IV. Neither human nor murine TNF modified adhesion of Panc02 cells to extracellular matrix components (Figure 6A) despite strong triggering of the classical NF-κB pathway (data not shown). Next, we analyzed Panc02 cells for their expression of proteins involved in adhesion and migration of cells (Figure 6B). We found the PDA tumor cells to express CD29 (integrin β1), CD49f (integrin α6), and CD107a and b, the expression of which was not modulated by TNF. CD106 (VCAM-1) expression, however, was induced by TNF. We also assessed the invasive capability of Panc02 cells by using an in vitro invasion assay and measuring gelatinolytic activities (Figure 6C). TNF did not significantly induce invasion and whereas infarcted mouse myocardium showed MMP-9 and -2 activities, Panc02 cells did not, irrespective of TNF stimulation. In sum, the results of the in vitro analysis of metastasis-related factors confirmed the limited metastatic activity observed with Pan02 cells in vivo.