Diabetes promotes invasive pancreatic cancer by increasing systemic and tumour carbonyl stress in Kras^G12D/+ mice

The effect of diabetes on PaC progression was investigated in Pdx1-Cre;LSL-Kras^G12D/+ (KC) mice, which develop autochthonous PaC in a pattern recapitulating human pathology with high fidelity by developing the full spectrum of PaC progression, from preneoplastic lesions (PanINs) to adenocarcinoma and metastasis [26, 27]. KC mice were interbred with mitosis luciferase (MITO–Luc) reporter mice to obtain KC-Mito (KCM) mice [10, 28, 29]. The LSL-Kras^G12D/+ lineage was maintained in the heterozygous state. Mice were screened by polymerase chain reaction (PCR) using tail DNA amplified by specific primers to the Lox-P cassette flanking mutated Kras^G12D/+, wild type Kras, Cre recombinase and MITO genes, as previously reported [10, 29]. In the MITO-Luc mouse, an artificial minimal promoter derived from the cyclin B2 gene and induced by NF-Y drives the expression of the luciferase reporter specifically in replicating cells. Therefore, both normal (e.g., bone marrow) and tumour actively proliferating cells may be localized by a bioluminescence imaging (BLI)-based screen [10, 28, 29]. We have previously shown that KCM mice develop pre-invasive (PanINs) and invasive ductal PaC with the same penetrance, latency, and histological features as those described for KC mice [29]. According to the Ethics Committee recommendations, to limit the number of animals, the experiments were stopped when it was sufficient to confirm or reject the working hypothesis in a statistically and clinically meaningful manner.

Figure 1 shows the flowchart and timeline of study design. Thirty-three KCM mice were rendered diabetic with streptozotocin (STZ) and followed for 16 weeks (i.e., up to 22 weeks of age). After an overnight fast, 6-week-old mice were intraperitoneally injected with 190 mg·kg^− 1 STZ (Sigma-Aldrich, St. Louis, MO, USA). Success rate, defined as the percentage of STZ-injected mice with glucose levels > 250 mg/dL for the entire study period was 72.7% (24/33). Three days after injection, diabetic mice were randomized to receive no treatment (Diab, n = 12) or FL-926-16 (gift of Flamma S.p.A., Chignolo d’Isola, Italy) [30] at a dose of 30 mg·kg^− 1∙day^− 1 in the drinking water (Diab+FL, n = 12) and injected weekly with 1 IU of insulin glargine to prevent excessive weight loss and ketoacidosis. FL-926-16 was shown to have a suitable absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile, and the greatest potency and selectivity toward RCS among all other carnosine derivatives [18]. The FL-926-16 dose was chosen based on previous results from our group [22], showing high efficacy in preventing diabetes-induced renal injury, and from other investigators, indicating a good safety profile at the dose of 10–45 g·kg^− 1·day^− 1 [18, 30]. Neither histological abnormalities of the liver, kidney, lung, and heart, nor functional abnormalities attributable to toxicity on these tissues were observed in this study or in a previous one [22]. STZ-treated mice not fulfilling the criteria for diabetes diagnosis (STZ-non-Diab, n = 9) served as control for STZ effect on PaC; seven of these mice failed to develop hyperglycaemia, whereas two had spontaneous recovery from diabetes within 2 weeks. Vehicle (saline)-injected KCM mice were used as non-diabetic controls and either left untreated (Ctr; n = 12) or treated with FL-926-16 (Ctr + FL; n = 9) to check for any drug effect.

Mice were subjected to in vivo BLI every other week [10, 29] and daily manual palpation of the abdomen to check for tumour growth and avoid the loss of animals, along with the need to cope with the related ethical issues (i.e., compliance with the 3Rs principles). Briefly, 10 min after administration of D-luciferin (75 mg/kg body weight, intraperitoneal; Perkin Elmer, Hopkinton, MA, USA), photon emission from the different body areas was acquired for 5 min and analysed with a CCD camera (Xenogen IVIS Lumina System; Perkin Elmer). A specific region of interest (ROI) corresponding to the abdominal area occupied by the pancreas was manually selected and the total photon flux (p/s) from this ROI was evaluated with Living Image software (Caliper Life Sciences; Perkin Elmer) [10, 28, 29].

At the end of the study, mice were anaesthetized with ketamine (60 mg·kg^− 1 Imalgene i.p.) and xylazine (7.5 mg·kg^− 1 Rompum i.p.) and killed by cervical dislocation. According to the Ethics Committee recommendations, to avoid suffering, three Diab and one Diab+FL mice presenting with both positive BLI and a palpable abdominal mass or poor general condition were killed 5 and 3 weeks, respectively, before the end of the study. The lungs and the middle part of the gastrointestinal tract, including the pancreas and the liver, were dissected, and exposed to the CCD camera for 5 min for photon emission assessment. The pancreas was dissected, photographed and weighted; then one part was stored at − 80 °C for molecular analysis, whereas the other part was processed for histological/immunohistochemical analysis [10]. At time of collection, a technician (C.C., see Acknowledgements) recoded biological samples to allow blinded analysis.

Body weight and blood glucose were monitored weekly. At the end of the study, the levels of haemoglobin (Hb) A1c, an indicator of long-term glycaemic control, were assessed by using the Mouse HbA1c Assay Kit (80,310, Crystal Chem, Zaandam, Netherlands), and serum AGEs and total protein carbonyls (PCOs), two carbonyl stress markers, were measured by ELISA (OxiSelect™ Advanced Glycation End-Product Competitive ELISA Kit, no. STA-817 and OxiSelect™ Protein Carbonyl ELISA Kit, no. STA-310, respectively, Cell Biolabs, Inc., San Diego, CA, USA).

Six 4-μm-thick non-serial pancreatic sections stained with haematoxylin and eosin were examined to confirm the presence of invasive PaC. Pancreas without invasive PaC were analysed to grade dysplastic ducts (i.e., PanINs) according to previously established criteria [26]. The numbers of low-grade (PanIN-1A/B) and high-grade (PanIN-2/3) dysplastic ducts were counted and expressed as a percentage of total ducts in the specimen [10].

ERK 1/2 phosphrylation status, nuclear YAP, and its target gene connective tissue growth factor (CTGF). Levels of AGEs, p-ERK 1/2, and CTGF protein in homogenates and of active (non-phosphorylated) YAP1 in nuclear extracts of pancreas of mice were assessed by Western blot. Human PDA tissues (n = 14) were obtained from the Pathology Unit of Sant’Andrea Hospital, Rome, Italy, in agreement with the ethical guidelines established by the locally appointed Ethics Committee. Pancreatic tissue distribution of AGEs and activated YAP1 in mouse and human specimens were evaluated by dual label immunofluorescence and immunoperoxidase, respectively [10, 31]. For immunofluorescence, a goat polyclonal anti-AGE antibody and a rabbit monoclonal antibody to active (non-phosphorylated) YAP1 were used as primary antibodies, followed by appropriate secondary fluorescent antibodies (see Supplementary Table S1 for antibodies in Additional file 1). Sections were analysed at a fluorescence microscope (Zeiss AXIO A1), equipped with an Axiocam 503 color camera (Carl Zeiss Italy, Milan, Italy). For immunoperoxidase, formalin-fixed paraffin embedded sections (4-μm thick) were rehydrated and treated with 0.3% H₂O₂ in PBS for 30 min to block endogenous peroxidase activity. Heat mediated antigen retrieval was performed with “Antigen Unmasking Solution, Citric Acid Based” (H-3300, Vector Laboratories, Burlingame, CA, USA) for AGE staining, or Tris/EDTA buffer pH 9.0, for YAP staining, both for 20 min. Nonspecific binding was blocked by incubation in Protein block serum free (Agilent/Dako, Santa Clara, CA, USA) for 30 min at room temperature. Then, sections were incubated with Avidin/Biotin blocking Kit (SP-2002, Vector Laboratories) for 30 min, an anti-AGE antibody (Abcam, Cambridge, UK, ab23722) or an antibody directed to the active (non-phosphorylated) YAP1 (Abcam, ab205270) at 4 °C overnight, and the appropriate biotinylated secondary antibody at room temperature for 30 min (see Supplementary Table S1 for antibodies in Additional file 1). Finally, sections were stained with UltraTek Horseradish Peroxidase (ABL015, ScyTek Laboratories, UT, USA) for 10 min followed by 3,3-diaminobenzidine (DAB)/H₂O₂ Chromogen/Substrate Kit (High Contrast) (ACV500, ScyTek Laboratories) until the reaction product was visualized (~ 3 min), and counterstained with hematoxylin. AGE positive staining and nuclear expression of YAP were measured in 10 random fields of each section at a final magnification of 250X and 400X, respectively, by means of the interactive image analyzer Image-Pro Premier 9.2 (Immagini&Computer, Milan, Italy). AGE positivity was expressed as the mean percentage of field’s area occupied by the specific stain. Expression status of active YAP in tumor specimens was assessed by counting the number of nuclei positive for YAP and expressed as the mean ratio (%) of YAP-positive nuclei to total nuclei.

Human MIA PaCa-2 (Catalogue No. 85062806, Lot No. 14A02) and Panc-1 (Catalogue No. 87092802, Lot No. 10G011) cells (Sigma-Aldrich) were used for assessing the effects of HG and FL-926-16 on cell proliferation. Experiments aimed at investigating the molecular mechanisms underlying the glucose-mediated effects and the protection by FL-926-16 were conducted on MIA PaCa-2 cells. Mycoplasma contamination in cell cultures was regularly tested by PCR MycoSPY Kit (Biontex Laboratories GmbH, Munchen, Germany). Human PDA cells were maintained in DMEM supplemented with 10% FBS and incubated in different conditions for three days, i.e., (1) normoglycaemia (normal glucose, 5 mM); (2) hyperglycaemia (HG, 25 mM); treated with (3) MGO or GO (200 μM, Sigma-Aldrich), two RCS and AGE precursors, or (4) the preformed AGE N^ε-carboxymethyllysine (CML, 100 μg/mL), prepared as previously reported [10, 21], with or without FL-926-16 (20 mM); and (5) exposed to DMEM low glucose medium containing 10% of pooled sera from non-diabetic or diabetic individuals, before and after AGE removal from diabetic serum by an immunoadsorption method (see below), with or without FL-926-16 (20 mM). Informed consent was obtained from non-diabetic and diabetic individuals. Moreover, both YAP and Epidermal Growth Factor Receptor (EGFR) were silenced to assess the role of YAP and EGFR pathway in RCS and AGE-induced cell proliferation (see below).

AGEs were removed from diabetic serum using an immunoadsorption method. To immunoprecipitate AGE-modified proteins, 500 μl of diabetic serum was incubated for 1 h with 25 μl of Pierce NHS-activated magnetic beads (Thermofisher Scientific) covalently conjugated with 10 μg of anti-AGE antibody (Abcam, see Supplementary Table S1 for antibodies in Additional file 1), according to the manufacturer instruction. To confirm the efficiency of AGE depletion, AGE concentration in both treated (unbound serum fraction) and untreated diabetic serum was evaluated in triplicate by ELISA (OxiSelect™ Advanced Glycation End-Product Competitive ELISA Kit, no. STA-817, Cell Biolabs, Inc., San Diego, CA, USA). Following this procedure, the concentration of AGEs in diabetic serum was reduced by about 60%, reaching a concentration similar to that of the non-diabetic serum (see the “Results” section).

YAP and EGFR were silenced using small interfering RNAs (siRNAs) and irrelevant scrambled siRNAs as control (Thermo Fisher Scientific, Waltham, MA, USA). Validated predesigned siRNA oligonucleotides and related TaqMan assays are detailed in Supplementary Table S2 (see Additional file 1). Lipofectamine RNAiMAX (Thermo Fisher Scientific) transfections were performed using 20 nM of each siRNA.

Cell viability and proliferation were evaluated by Cytoselect WST-1 Cell Proliferation Assay (Cell Biolabs) following the manufacturer instructions.

YAP1, its upstream regulators large tumour suppressor Kinase 1(LATS1) and EGFR-ERK pathway, and its molecular targets CTGF, WTN5A and EMP2 in in human PDA cells. Cells were extracted in 1% SDS buffer containing protease and phosphatase inhibitors (Sigma Aldrich). Nuclear protein extracts were obtained from cell monolayers with the Nuclear Extract Kit (Active Motif Corp., Carlsbad, CA, USA). Protein concentrations were determined using the Bradford Assay Kit (Bio-Rad Hercules, CA, USA). Nuclear protein levels of YAP1 and cellular protein levels of total and EGFR phosphorylated at Tyr1068 (p-EGFR), total and p-ERK 1/2 and LATS1, a key kinase of the Hippo pathway [32], were assessed by Western blotting (see Supplementary Table S1 for antibodies in Additional file 1). KRAS activity was evaluated by the KRAS activation Assay Kit (no. STA-400-K Cell Biolabs, Inc.) according to the manufacturer’s protocol. Briefly, 1 mg of lysate was subjected to pull-down and 50 μg of lysate was used to measure total KRAS. Pull-down and total lysates were subjected to Western blotting procedure using the primary antibody against KRAS provided by the kit. The mRNA levels of CTGF/CCN2, WTN5A and EMP2, three recognized molecular targets of YAP [23, 33, 34], were assessed by real-time PCR (RT-PCR) using a StepOne Real-Time PCR System and TaqMan Gene Expression assays (Thermo Fisher Scientific) [10] listed in Supplementary Table S3 (see Additional file 1).

STZ-treated KCM mice developed hyperglycaemia starting about 72 h post-injection (Fig. 2a) and showed a slight decline in the growth curve vs Ctr mice, which reached statistical significance only at 8 and 12 weeks of age (Fig. 2b). Despite no difference in body weight (Fig. 2c), blood glucose (Fig. 2d), and HbA1c levels (Fig. 2e), FL-926-16 treatment prevented the diabetes-associated increase in circulating AGEs (Fig. 2f) and total PCOs (Fig. 2g), as assessed at the end of the study.

HG concentration mimicking diabetic hyperglycaemia promoted PDA cell growth and this effect was prevented by FL-926-16 (Fig. 5a-b). The AGE precursors RCS, MGO and GO, and the preformed AGE CML also stimulated PDA cell proliferation. FL-926-16 was able to inhibit cell proliferation induced by MGO and GO, but not CML (Fig. 5c). Treatment with CML, but not with MGO, induced ERK 1/2 activation, and FL-926-16 was ineffective in counteracting the effect of CML on ERK 1/2 phosphorylation status (Fig. 5d). However, the proliferating effect of both the RCS MGO and the AGE CML was associated with YAP1 nuclear persistence and activity. Again, FL-926-16 efficiently prevented the nuclear translocation of YAP1 induced by MGO, but failed to counteract the effect of CML (Fig. 5e). Consistently, FL-926-16 treatment reversed the MGO-induced upregulation of gene expression of CTGF, Wnt Family Member 5A (WNT-5A) and Epithelial Membrane Protein 2 (EMP2), three well-recognized YAP target genes [24, 33, 34]. Conversely, FL-926-16 was ineffective in preventing the modulatory effect of CML on the mRNA level of these genes (Supplementary Fig. S2 in Additional file 1).

Silencing of YAP1 using two independent siRNAs (siYAP#1 and #2) (Fig. 6a) significantly inhibited the transcription activity of YAP target genes induced by both MGO and CML in PDA cells (Fig. 6b). In MGO-treated cells, YAP induction was associated with a decrease in protein levels of LATS1, a well-established negative regulator of YAP activity [32], whereas CML treatment failed to modulate LATS1 (Fig. 6c). Instead, treatment with CML, but not with MGO, was found to induce EGFR phosphorylation (pEGFR) (Fig. 6d). EGFR silencing (Fig. 6e) almost completely reversed YAP1 nuclear translocation (Fig. 6f), KRAS activation, and ERK 1/2 phosphorylation (Supplementary Fig. S3A-B) induced by CML.

The levels of AGEs were 178.2 ± 13.2 μg/mL in the pooled sera from diabetic patients and 72.3 ± 4.54 μg/mL in pooled sera from non-diabetic individuals. The diabetic serum induced a 3-fold increase in PDA cell proliferation compared to the non-diabetic serum. This effect was greatly reduced by prior selective AGE removal from the diabetic serum (AGE levels, 95.5 ± 7.92 μg/mL) and almost completely reversed by combining AGE removal from serum and FL-926-16 treatment of PDA cells (Fig. 7).