The human tissue samples used in the present study were obtained from consecutive archival paraffin blocks of patients who were diagnosed with pancreatic ductal adenocarcinoma and underwent surgery in Huadong Hospital (Shanghai, China) between November 2018 and January 2019. The demographic and pathological features of the patients are listed in Table I. Written informed consent for the use of tissues in scientific research was obtained from all patients. The collection of the pathological data from patients with PC was approved by the Institutional Research Ethics Committee of Huadong Hospital, Fudan University (approval no. 2018K098). Immunohistochemical staining of paraffin-embedded tissues with rabbit anti-FABP4 polyclonal antibody (pAb; cat. no. 12802-1-AP; ProteinTech Group, Inc.) was performed in order to observe the characteristics of the adipocytes. For each slide, representative images of adipocytes surrounding the normal tissue and adipocytes in the vicinity of the PC cells were obtained. Adipocyte cell sizes were assessed using ImageJ software (National Institutes of Health) (23).

The human PC cell lines Panc-1 and Mia PaCa2 were purchased from the Cell Bank of the Type Culture Collection of Chinese Academy of Sciences and cultured according to standard protocols. In brief, Panc-1 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc.), containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Mia PaCa2 cells were cultured in DMEM, supplemented with 10% FBS and 2.5% horse serum. Murine 3T3-L1 preadipocytes were purchased from the American Type Culture Collection and cultured in DMEM supplemented with 10% newborn calf serum. All cells were cultured at 37°C (5% CO₂) in a humidified incubator. Two days after reaching 100% confluence (designated as Day 0), the 3T3-L1 preadipocytes were then induced to differentiate in DMEM containing 10% FBS, 0.5 mM 3-isobutyl-1-methylxanthine (M), 1 µM dexamethasone (D) and 1 µg/ml insulin (I). After 48 h (Day 2), these cells were sustained in DMEM supplemented with 10% FBS and 1 µg/ml insulin for another 2 days. Next, 3T3-L1 mature adipocytes were cultured in DMEM supplemented with 10% FBS. After induction (at Day 8), mature adipocytes were cocultured with PC cells using a Transwell indirect coculture system (0.4 µm pore size; Corning), and mature adipocytes cultured alone at the same time-points were used as controls.

Total RNA was extracted from 3T3-L1 mature adipocytes using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. RNA-seq libraries were prepared according to the TruSeq™ RNA Sample Preparation kit from Illumina. Briefly, 5 µg total RNA was isolated according to the polyA mRNA selection method by oligo(dT) beads and then fragmented. Double-stranded cDNA was synthesized using a SuperScript Double-stranded cDNA Synthesis kit (Invitrogen; Thermo Fisher Scientific, Inc.) with random primers (Illumina). Libraries were size selected for cDNA target fragments of 200–300 bp on 2% Low Range Ultra Agarose, and the cDNA library was amplified using Phusion DNA polymerase (NEB) for 15 PCR cycles. Quantified by TBS380, amplified cDNA fragments were sequenced with the Illumina HiSeq X10 platform (2×150 bp read length). The expression level of each gene was calculated using the fragments per kilobase of transcript per million (FPKM) base pairs sequenced method (24). In order to identify the differentially expressed genes (DEGs) between 2 groups (3 biological replicates per group), Cuffdiff was used (25). Genes with a fold change ≥2 and a false discovery rate (FDR) <0.05 were identified as significant DEGs. In order to comprehend the functions of these DEGs, Gene Ontology (GO) functional enrichment and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed using Goatools and KOBAS, respectively (26). Heat-map representations of the expression of selected genes expression were generated using the ‘pheatmap’ R package.

Following three washes with PBS, 3T3-L1 adipocytes were fixed with 3.7% formaldehyde for 20 min. Stock Oil Red O solution (0.5% in isopropanol) was diluted with distilled water (3:2). Fixed cells were subsequently stained with Oil Red O solution at room temperature for 1 h (27). The cells were then washed with water and visualized by light microscopy.

Following total RNA extraction, PrimeScript RT Master Mix (Takara Bio) was used for first-strand cDNA synthesis with random primers. The expression of the considered genes was determined via RT-qPCR, which was carried out with SYBR-Green PCR Master Mixture reagent (Applied Biosystems; Thermo Fisher Scientific, Inc.) using an ABI 7500 Real-Time PCR system (Applied Biosystems; Thermo Fisher Scientific, Inc.). Thermocycling conditions were as follows: 10 min denaturation at 95°C, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. Relative RNA quantity was calculated using the 2^−ΔΔCq method with 18S rRNA as an endogenous control (28). All reactions were performed in triplicate. The primer sequences used in the present study are listed in Table II.

Total lysates were extracted using lysis buffer containing 50 mM Tris-HCl (pH 6.8), 2% SDS, protease inhibitor mixture (all from Roche Diagnostics), 100 mmol/l NaF, and 1 mmol/l PMSF. The protein concentration was determined by the bicinchoninic acid assay. Equal amounts (20 µg) of protein were separated via 12% SDS-PAGE, transferred to polyvinylidene fluoride membranes (EMD Millipore), blocked with 5% skim milk at room temperature for 1 h and immunoblotted with the indicated primary antibodies (1:1,000) at 4°C overnight. Rabbit anti-PPAR-γ pAb (cat. no. 16643-1-AP), rabbit anti-FABP4 pAb (cat. no. 12802-1-AP), rabbit anti-HSL pAb (LIPE; cat. no. 17333-1-AP), rabbit anti-FASN pAb (cat. no. 10624-2-AP) and rabbit anti-α-tubulin pAb (cat. no. 11224-1-AP) were purchased from ProteinTech Group, Inc. Rabbit anti-phospho-Akt monoclonal antibody (mAb; cat. no. 4060) and rabbit anti-Akt mAb (cat. no. 4691) were purchased from Cell Signaling Technology, Inc. The following day, membranes were incubated with horseradish peroxidase-conjugated anti-rabbit IgG antibodies (1:5,000; cat. no. SA00001-2; ProteinTech Group, Inc.) at 37°C for 1 h. Blots were then visualized with enhanced chemiluminescence detection kit (Thermo Fisher Scientific, Inc.) using the ImageQuant LAS 4000 System (GE Healthcare). Semi-quantification analysis was conducted by Quantity One software, version 4.6 (Bio-Rad Laboratories).

PC cells (Panc-1 and Mia PaCa2) alone and cocultured with mature adipocytes were assessed for their ability to migrate through an 8 µm pore membrane (24-well insert; Corning). After 5 days of coculture with adipocytes, tumor cells were trypsinized and seeded (5×10⁴) into the upper chamber in serum-free DMEM. The lower chamber was filled with medium containing 10% FBS. After incubation for 36 h, cells on the bottom of the insert were fixed with 100% methanol and stained in 0.1% crystal violet at room temperature for 30 min. The number of migrating cells was counted separately in three randomly selected fields using a light microscope (IX71; Olympus).

Statistical analysis of all data was completed using GraphPad Prism software (version 5.0; GraphPad Software, Inc.). All experiments were performed in triplicate. The results are expressed as the mean ± standard deviation. Unpaired two-tailed Student's t-test was performed to determine significance of differences between two groups. P<0.05 was considered to indicate a statistically significant difference. P-values are presented in the figures (*P<0.05; **P<0.01; ***P<0.001).

A series of studies have demonstrated crosstalk between cancer cells and adipocytes, but these have largely focused on the postive effect of adiopocytes on tumor progression (29). In order to determine the direct impact of PC cells on mature adipocytes, the presence of adipocytes in human pancreatic tumors was first investigated. To address this issue, immunohistochemical staining of FABP4 (highly expressed in mature adipocytes) was performed in tumor samples. The presence of FABP4-stained adipocytes at the invasive front of pancreatic tumors was observed (Fig. 1A). Furthermore, tumor-neighboring adipocytes were smaller in size compared with those normal adipocytes further away (Fig. 1B). In order to assess the morphological changes of PC-related adipocytes, an adipocyte-PC cell indirect coculture system was established (Fig. 1C, left panel). First, 3T3-L1 cells were induced to differentiate into mature adipocytes, and human PC cells were then cultured on top (Fig. 1C, right panel). As indicated by Oil Red O staining, following coculture with Panc-1 for 5 days, mature adipocytes exhibited a decrease in lipid content (Fig. 1D) with a marked decrease in the size of lipid droplets (Fig. 1E) compared with those cultured alone, which is consistent with the clinical features of small adipocytes adjacent to the tumor tissues.

Given the great phenotypic changes of cancer-associated adipocytes in other types of tumor, such as breast and prostate cancer (13,16), we reasoned that PC-associated adipocytes may change their gene expression patterns. RNA-sequencing was performed on 3T3-L1 cells cultured with or without PC cells, and identified 1,081 genes that were differentially expressed in cancer-associated adipocytes, including 541 upregulated and 540 downregulated genes (Fig. 2A). Next, all DEGs were characterized through GO term mapping according to their major GO categories [i.e., the GO terms ‘biological processes’ (BP), ‘cellular components’ (CC), and ‘molecular functions’ (MF)] (Fig. 2B). The top BP GO annotation was ‘positive regulation of cell migration/motility/cellular component movement’. Within the CC of ontology, the most abundant term was ‘proteinaceous extracellular matrix’. Furthermore, ‘receptor regulator activity’ was primarily enriched in the MF group. DEGs were also mapped to the KEGG pathway database (Fig. 2C). Numerous metabolic pathways, such as ‘insulin resistance’, ‘glucagon signaling pathway’ and ‘cholesterol metabolism’, were enriched, indicating that diverse metabolic changes were occurring in the cancer-associated adipocytes. It was also revealed that ‘cytokine-cytokine receptor interaction’ was the most significantly enriched KEGG pathway.

The aforementioned data revealed that upon coculture with PC cells, a marked decrease in lipid accumulation and a profound impact on gene expression networks associated with metabolism were exhibited in adipocytes. Furthermore, a broad spectrum of genes involved in fatty acid and triglyceride synthesis for lipogenesis, and their upstream regulators, as well as triglyceride storage and degradation, were significantly decreased (Fig. 3A), revealing the dysregulated lipid homeostasis in cocultured adipocytes. RT-qPCR analysis of adipocytes cocultured with Panc-1 (Fig. 3B) or with another PC cell line Mia PaCa2 (Fig. 3C) was performed, which confirmed the substantial inhibition of lipid metabolism-associated genes, including FABP4, HSL, ATGL, CIDEA and FASN, and of two key adipogenic upstream factors, PPAR-γ and C/EBPα. In line with the RT-qPCR data, immunoblotting analysis also confirmed decreased levels of FABP4, HSL, FASN and PPAR-γ in cocultured adipocytes (Fig. 4A and B). The present study also investigated the effect of coculture on adipocyte glucose metabolism and revealed that there was a marked decrease in the insulin-induced phosphorylation of Akt (Fig. 4C), as well as the transcriptional expression of Glut4 and IRS1 (Fig. 4D).

Adipocytes are commonly regarded as terminally differentiated cells, but increasing evidence suggests they may have greater plasticity under certain circumstances (12,30). In the present study, the ‘fat cell differentiation’ pathway was highly enriched in cocultured adipocytes. A total of 25 DEGs were involved in this process (Table III) among which several mature adipocyte markers, such as leptin and resistin, were significantly downregulated. It was also observed that, following coculture with pancreatic cancer cells for 5 days, numerous adipocytes exhibited a stretched, fibroblastic shape (Fig. 5A). Further analysis using the GO BP demonstrated that the mesenchymal cell differentiation process was activated in cocultured adipocytes (Fig. 5B). There was also a marked increase in the expression of mesenchymal state-associated genes (31,32), indicating a shift towards more mesenchymal phenotypes in these cocultured adipocytes (Fig. 5C). The RT-qPCR results demonstrated that adipocytes cocultured with Panc-1 exhibited a loss in expression of mature adipocyte markers, including adiponectin (APN), leptin and resistin, and a gain in expression of fibroblast markers, including α-SMA, FSP-1, MMP-9, PAI-1 and MMP-11 (Fig. 5D). Similar results were obtained by coculturing adipocytes with Mia PaCa2 (Fig. 5E).

Previous research has demonstrated that adipocytes activated by tumor cells exerted an active role in tumor progression and TME remodeling (33). In the present study, bioinformatics analysis of the whole transcriptome of adipocytes revealed several enriched pathways associated with the remodeling of the pancreatic cancer stroma (Table IV), including the cytokine-mediated signaling pathway, angiogenesis and extracellular matrix organization. In order to directly assess the role of cancer-associated adipocytes in PC progression, the present study performed Transwell assays to assess the ability of the PC cells to traverse the membrane. PC cells were cocultured with or without adipocytes for 5 days, and it was revealed that both adipocyte-exposed Panc-1 and Mia PaCa2 tumor cells had a significantly increased ability to fully migrate through a Transwell membrane (Fig. 6). Overall, these results indicated that adipocytes can increase PC cell aggressiveness.