The study was approved by the Institutional Review Board (IRB) of Nanjing Medical University. All patients provided written informed consent prior to their enrolment.

The diagnosis of PC was based on the National Comprehensive Cancer Network (NCCN) clinical practice guidelines 2016 (7). CP was diagnosed based on the American Pancreatic Association's practice guidelines for chronic pancreatitis, 2014 (8). Blood samples from 20 healthy controls and serum samples of 18 CP and 16 PC patients were collected at the Wuxi People's Hospital. All body fluid specimens were flash-frozen upon collection and stored at −80°C until further use.

Three serum samples from each group were randomly selected based on the shortest sampling time for miRNA microarray analysis. Total RNA was isolated from these frozen body fluids using RNeasy Mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The quality and quantity of the isolated RNA was then assessed using agarose gel electrophoresis and a spectrophotometer (Sangon Biotech, Shanghai, China), and then labeled with biotin and hybridized to a GeneChip miRNA 3.0 Array from Affymetrix (Santa Clara, CA, USA), according to the manufacturer's protocol. Following hybridization, the images were digitized and analyzed using a laser scanner interfaced with Affymetrix GeneChip Command Console (AGCC). The most differentially expressed miRNAs in the CCA serum samples were identified and compared to those from normal control and PC samples.

miRNA microarray-identified miRNA expression was further confirmed with qRT-PCR using total RNA from serum samples and exosomes. In brief, RNA samples were converted into cDNA using 0.5 µg RNA sample and a reverse transcription kit (Takara, Ohtsu, Shiga, Japan). qRT-PCR was performed using the iQ™ SYBR^® Green Supermix (Bio-Rad) in an ABI 7500 qPCR system (Applied Biosystems, Foster City, CA, USA). The optimal dilution and melting curves were utilized to quantify each of the amplified production using specific primer sets. Bulge-loop™ miRNA qRT-PCR primer sets (one RT primer and a pair of qRT-PCR primers for each miRNA) were designed and produced by RiboBio (Guangzhou, China) and ribosome RNA 5S was used as an internal control for miR-23b-3p level in sera and bile. The relative quantification of each gene was measured using the 2^−∆∆CT method. We performed qRT-PCR in serum samples from 20 healthy controls, 18 patients with CP and 16 patients with PC; all measurements were performed in triplicate.

Human PC cell lines pancreatic cancer cells (PANC-1) cultured in DMEM containing 10% fetal bovine sera, 100 g/ml streptomycin, and 100 U/ml penicillin. The synthesized miR-23b-3p mimic (miR-23b-3pM) and non-specific mimic (NSM) were purchased from RiboBio. Here, 70% of confluent cells were transfected using 20 nM of miR-23b-3pM or NSM with Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA, USA). Cells were allowed to incubate for 48 h after transfection and then harvested for further studies.

Ten thousand cells were transfected and placed in 24-well plates (Corning, USA). The number of cells was determined every other day with a hemocytometer under an inverted-light microscope.

We also performed the 5-ethynyl-20-deoxyuridine-Apollo 488 (EdU-Apollo 488) incorporation assay to assess the cell proliferation capacity. CCA cells were seeded into 96-well plates at a density of 3×10⁴ cells/ml and 24 h later, cells were transfected with 20 nM of NSM or 23b-3pM for 6 h. The cells were stained with EdU and Hoechst 33342 separately from a kit from RiboBio, according to the manufacturer's instructions. The proliferating cells were checked under the fluorescence microscope and the cell proliferation rate calculated by the formula: (number of proliferating cells/total cells) × 100%.

This established a two-chamber system divided by a cell-permeable membrane coated with Matrigel (R&D Systems, USA). Cells were allowed to incubate for 24 h, fixed, and stained with crystal violet. Then invasive cells were counted and images of 10 random fields under an inverted-light microscope were captured.

For the wound healing assay, 2.5×10⁴ transfected cells were cultured in 24-well plates. A line was drawn when cells reached ~90% confluence. At 0, 24, and 48 h, images of the same 10 fields were recaptured.

Serum as well as supernatant of PANC-1 cells were transferred to ultracentrifuge tubes, and centrifuged at 16,500 × g for 20 min at 4°C to further remove cells and cell debris; the supernatant was filtered through a 0.2-µm filter to remove particles sized >200 nm. The filtered supernatant was transferred to new ultracentrifuge tubes and the tubes sealed prior to ultracentrifugation at 120,000 × g for 70 min at 4°C to obtain exosome pellets. The supernatant was discarded, and the exosomes were collected. For maximal exosome retrieval, the exosome enriched pellet was resuspended repeatedly in a small volume tube (9). Then total RNA was isolated from these exosomes using TRIzol as described elsewhere (10).

For the cell uptake experiment, the exosomes were stained with oil red, and washed using an ultra-filtration membrane (10 kDa, Merck Millipore, Darmstadt, Germany) to remove any free nucleotides. 1,1′-dioctadecyl-3,3,3′,3–tetramethylindocarbocyanineperchlorate (Dil) was used to label the membrane component of the exosomes. The exosomes were incubated with 10 µg/ml DiI (Beyotime, Nantong, China) for 15 min at 37°C and then washed twice with cold phosphate-buffered saline (PBS). The labeled exosomes were added to PANC-1 cells grown on chamber slides of a 35-mm dish with sterile small circular glass and incubated at 37°C. After 60 min, the cells were harvested and washed twice with PBS. Then, the cells were stained with 4′-6-diamidino-2-phenylindole (DAPI; Invitrogen) to visualize the nuclei and were examined using an UltraView VoX confocal fluorescence microscope (Perkin-Elmer, Waltham, MA, USA) (11).

All data are presented as mean ± SEM. One-way or two-way ANOVA was performed for statistical analysis, using GraphPad Prism 5. Differences between group means with p-values <0.05 were regarded as statistically significant. The correlations were assessed by calculation of Pearson correlation coefficients.

miRNA microarrays were used to assess the miRNA expression profiles of patients with CP and PC in order to assess changes in body fluid miRNAs from chronic inflammation to cancers. Thirty-five dysregulated serum miRNAs were found in CP, which were not observed in healthy controls (p<0.01). Of these, 12 were upregulated and 23 were downregulated (Fig. 1A). There were 42 dysregulated serum miRNAs in PC that were not found in healthy controls (p<0.01). Of these, 14 were upregulated and 28 were downregulated (Fig. 1B). Four miRNAs were verified altered in both CP and PC; among these, miR-23b-3p was upregulated, while 3 miRNAs, miR-4505, 4530, and 150-5p were downregulated (Fig. 1C). qRT-PCR was performed to determine the expression of miR-23b-3p in the serum of CP and PC patients. miR-23b-3p was found upregulated in sera of both CP and PC, and the expression level in PC was higher than that in CP (p<0.01) (data pertaining to 3 downregulated miRNAs is not shown). This finding was consistent with the microarray data (Fig. 1D).

Cell counting and EdU incorporation assays showed that transfection of PANC-1 cells with 23b-3pM increase tumor cell growth and proliferation rate as compared to those of NSM-transfected cells (p<0.05; Fig. 2).

Transwell and wound healing assays were used to analyze the effect of miR-23b-3p on in vitro invasion and metastasis of PANC-1 cells, respectively. The overexpression of miR-23b-3p in PANC-1 cells caused substantially greater cell invasion through the Transwell chamber at 48 h, as compared to that observed among cells transfected with NSM (p<0.05). The wound healing scratch test revealed that introduction of miR-23b-3p also accelerated the mobility of PANC-1 cells at 48 h after scratching, relative to cells transfected with NSM (p<0.05) (Fig. 3).

The exosomes in supernatant of PC patients sera as well as in pancreatic cell lines were isolated; the expression of exosomic miR-23b-3p was checked by RT-PCR. The results showed overexpression of exosomic miR-23b-3p in sera of patients with PC when compared with that in sera of healthy controls (Fig. 4A). Moreover, the exosomic miR-23b-3p expression in the supernatant of PANC-1 was higher than that in the pancreatic ductal epithelial cells (HPDE6c-7) (Fig. 4B). Furthermore, exosomes were labeled with the membrane dye Dil and the labeled exosomes were added to the culture medium of the PANC-1 cells. After co-culture for 60 min with unlabeled recipient cells, the labeled exosomes as well as PANC-1 cells were observed using a confocal microscope; it was detected that PANC-1 cells can take up exosomes (Fig. 4C). Our results demonstrated that the pancreatic cancer cells can release exosomes into the circulation and that the serum exosomes containing miRNAs can in turn be taken up by pancreatic cancer cells and may subsequently mediate cellular functions.

In order to investigate the relationship between miR-23b-3p and CA-19-9, we examined the correlation between the two. miR-23b-3p in sera of PC patients showed a significant positive correlation with CA-19-9 (R=0.6511; Fig. 5A). Moreover, the exosomic miR-23b-3p expression in the sera of PC patients also showed a positive correlation with that of CA-19-9 (R=0.7142; Fig. 5B).