This study was approved by the Georgetown University Oncology Institutional Review Board of Washington, DC, USA, and followed the Ethical Principles for Medical Research Involving Human Subjects from the Declaration of Helsinki. Written informed consent was obtained from subjects.

Patients with previously untreated locally advanced, unresectable pancreatic cancer were enrolled in a pilot study designed to demonstrate the feasibility and safety of administering hypo-fractionated SRS concurrently with full-dose gemcitabine (G). In an effort to decrease late duodenal toxicity, we examined the use of fractionated SBRT with full-dose gemcitabine (29). Patients received gemcitabine (1000 mg/m²) on days 1, 8, and 15 of every 28-day cycle, for up to 6 months. Patients were also treated with 5 Gy of SRS daily over five consecutive days (typically days 22–26) of Cycle 1 only. Patients also underwent serial endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) of pancreatic masses prior to therapy, after 2 months, and after 6 months (Figure 1).

Primary cultures of HTPSCs were generated from EUS-FNA aspirates obtained from patients with locally advanced, unresectable pancreatic adenocarcinoma before treatment (Pre) and at 6 months following treatment with gemcitabine (G) and SRS (Post) (Figure 1). Immediately upon collection, FNAs were placed in DMEM (high glucose) (Invitrogen, Carlsbad, CA, USA) with 1% fetal bovine serum (FBS), 10 mL/L of antibiotic–antimycotic (Invitrogen, Carlsbad, CA, USA) and moved to a laminar flow tissue culture hood in the laboratory. All subsequent manipulations were performed using standard sterile tissue culture techniques. After collection, 1 mL of dispase and 2 mL of Trypsin–EDTA were added to the media and the tubes were incubated for 30 min at 37°C with occasional vortexing. After incubation, the tissue was further dissociated by triturating with a sterile pipette. The pancreatic tumor cell suspensions were then subjected to gravity centrifugation for 5 min at 1800 rpm. The pellets were resuspended in 5 mL of FCM Lysing Solution (Santa Cruz Biotechnology, Santa Cruz, CA, USA) to lyse red blood cells and incubated at room temperature for 5 min with occasional trituration with sterile pipette. The pancreatic cell suspensions were again subjected to gravity centrifugation for 5 min at 1800 rpm. After the supernatant was removed, the pellets were resuspended in 2 mL of complete media containing DMEM (high glucose), 10% FBS, 10 ng/mL epidermal growth factor (EGF), 100 U of human insulin (per 500 mL of medium), and 10 mL/L of antibiotic–antimycotic. The cultures were plated in a single well of a six-well tissue culture treated plate and placed securely on humidified incubation chamber of inverted Nikon Eclipse TE300/PerkinElmer Spinning Disc Confocal microscope system (Waltham, MA, USA) at 37°C with 5% CO₂. Cells were allowed to attach for 24 h, wells were gently washed of debris with media, and replaced with fresh complete media previously equilibrated for CO₂ in the incubator and warmed at 37°C.

Images were obtained with inverted Nikon Eclipse TE300/PerkinElmer Spinning Disc Confocal microscope system (Waltham, MA, USA) with humidified incubation chamber using 10× objective lens in phase-contrast mode and Volocity software (v 5.3.1, PerkinElmer, Waltham, MA, USA) as previously described (30). Briefly, three to five sets of 812 μm × 875 μm contiguous fields were selected in a square (5 × 5) configuration in the middle of each well depending on cellular location. A focus map was created for each point and time-lapse image acquisition was set in 15-min intervals. Cells were imaged for 3–7 days. One to two milliliters of complete media were added daily.

Immediately after time-lapse imaging, cells were prepared in the respective well of a six-well plate and stained by immunofluorescence with vimentin 1:50 (sc-7557; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and fluorescently labeled secondary antibody FITC anti-goat (sc-2024; Santa Cruz Biotechnology, Santa Cruz, CA, USA). Sections (5 mm thick) of formalin-fixed, paraffin-embedded pancreatic adenocarcinoma FNAs were prepared and stained by immunohistochemical staining with GFAP (GA5) 1:50 (3670S; Cell Signaling, Danvers, MA, USA). Digital photographs taken with Nikon Eclipse E800M microscope with DMX1200 software; Nikon Instruments, Inc. (Melville, NY, USA) and Olympus IX71 with Olympus DP Capture Software (Center Valley, PA, USA).

For immunohistochemistry, heat induced epitope retrieval (HIER) was performed by immersing the tissue sections at 98°C for 20 min in 10 mM citrate buffer (pH 6.0) with 0.05% Tween. Immunohistochemical staining was performed using a horseradish peroxidase labeled polymer (K4001; Dako, Carpinteria, CA, USA) according to manufacturer’s instructions. Briefly, slides were treated with 3% hydrogen peroxide and 10% normal goat serum for 10 min each, and exposed to primary antibody for GFAP (1:50, 3670; Cell Signaling, Danvers, MA, USA) for 1 h at room temperature. Slides were exposed to anti-mouse labeled HRP for 30 min and DAB chromagen (Dako, Carpinteria, CA, USA) for 5 min. Slides were counterstained with Hematoxylin (Harris Modified Hematoxylin; Fisher, Waltham, MA, USA) at a 1:17 dilution for 2 min at room temperature, blued in 1% ammonium hydroxide for 1 min at room temperature, dehydrated, and mounted with Acrymount (McKinney, TX, USA). Sections with the omitted primary antibody were used as negative controls.

For immunofluorescence, cells in each well of a six-well plate were fixed for 15 min in 4% formaldehyde, rinsed with PBS, blocked for 20 min with 10% normal blocking serum, and incubated with primary antibody for vimentin (1:50, sc-7557; Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 60 min at 37°C. After washing, secondary antibody was applied for 45 min in a dark chamber. After a final washing, wells were covered with anti-fade reagent (#9071; Cell Signaling, Danvers, MA, USA) with DAPI (#8961; Cell Signaling, Danvers, MA, USA).

After 3–7 days of time-lapse image acquisition, Timm’s Tracking Tool (TTT) (31, 32) was used to track individual cell migration as previously described (30). Briefly, time-lapse images for each point were loaded into TTT; individual human tumor-derived pancreatic stellate cells (HTPSCs) were identified on the screen by morphology (a flattened angular appearance and long cytoplasmic processes giving them a typical “stellate” appearance) and the presence of multiple cytoplasmic lipid-droplet deposits. Each cell’s trajectory through the time-lapse images was recorded in TTT by following each cell’s nucleus.

As described above, time-lapse images for each point were loaded into TTT and individual HTPSCs were identified on the screen by morphology and cytoplasmic lipid-droplet deposits. Total cell distance traveled in the x,y axis was recorded via TTT tracking (Figure 2A) and cell speed was calculated by s = d/t where d is distance traveled and t is total time of travel. Additionally, each change of direction event was recorded manually by graphing the x,y coordinates of distance traveled and directional change (Figure 2B). For each cell, the base diameter was measured in pixels by recording the x,y coordinates of the cell boundary. The total pixels were converted to micrometers (0.73 μm/pixel). Cells were observed in the time-lapse images and membrane expansion and contraction events were noted whenever the cell membrane expanded by greater than 50% of total body length. During each membrane expansion, the x,y coordinates of the cell boundary were recorded at the cell’s longest points (Figure 2C). If the cell was diagonally positioned, the expansion length was calculated using the Pythagorean theorem a² + b² = c², where c is cell diameter (Figure 2D). The expansion and contraction ratio was calculated by x = b/e where b is the base cell diameter and e is the expanded cell diameter.

Differences and variance (σ) in mean cell speed, total distance traveled, and expansion and contraction ratios were compared with two-way ANOVA (GraphPad Prism version 6 for Mac, GraphPad Software, La Jolla, CA, USA). Significance was assigned at p < 0.05; n = 3 patients per cohort.

Human tumor-derived pancreatic stellate cells were identifiable by morphology in primary culture after 16–24 h of incubation and formed small colonies after 48–72 h (Figure 3A). HTPSCs were the predominant cell type identified during culture; the heterogeneous culture also included epithelial cells and non-stellate fibroblasts. HTPSCs were initially characterized by a flattened angular appearance, the presence of multiple cytoplasmic lipid droplets, long cytoplasmic processes giving them a typical “stellate” appearance, and cytoplasmic expansion and contraction behavior. HTPSC isolation correlated with diagnostically adequate EUA-FNA samples collected from patients with locally advanced, unresectable pancreatic adenocarcinoma before treatment (pre) and at 6 months following treatment with gemcitabine (G) and SRS (post) (Figure 3B).

Pancreatic adenocarcinoma sections showed positive staining for GFAP localized to stellate-shaped cells and their cytoplasmic processes. These stellate-shaped cells were localized in the interacinar space with cytoplasmic processes extending around the base of adjacent acinar cells (Figure 4A). After 72 h in culture, cells stained positive for vimentin showing a fibrillar staining pattern in the cell cytoplasm and had the presence of auto-fluorescent lipid droplets (Figure 4B). These markers of stellate cells were correlated with the morphological characteristics we saw in culture of live cells: a flattened angular appearance, the presence of multiple cytoplasmic lipid droplets, long cytoplasmic processes, and cytoplasmic expansion and contraction behavior.

To define the parameters of individual cell motility, we measured cell migration speed, total distance traveled, and expansion/contraction ratios in HTPSCs isolated from FNAs of patients with locally advanced, unresectable pancreatic adenocarcinoma. Behaviors of individual cells isolated at baseline before treatment (Pre) were compared to cells isolated at 6 months following treatment with G + SRS (Post) (Figures 4 and 5). Although the speed of HTPSC migration was not significantly different in the pre-treatment versus post-treatment samples (Figure 5A), the variance of speed was greater in cells from the pre-treatment samples [21–90 μm/h (σ = 291)] compared to cells from the post-treatment samples [24–60 μm/h (σ = 99)] (ANOVA, p < 0.05) and the mean total distance traveled was 1.9-fold greater in cells from post-treatment samples (mean 198 SEM ± 21 pre; mean 370 SEM ± 33 post; ANOVA, p < 0.05). While pre-treatment cells traveled a maximum of 350 μm, post-treatment cells traveled up to 597 μm over the same time frame (Figure 5B). Membrane extension was measured by calculating expansion/contraction ratios. On average, cells pre-treatment extended 2× their body length compared to a 2.8× extension in cells post-treatment, a 1.3-fold increase (ANOVA, p < 0.05) (Figure 5C). The number of times these expansion/contraction events took place was not different between groups (Figure 5D).