Animal model and maintenance

The animal experiments in this study were approved by the state of Baden-Württemberg, Germany, and were in accordance with the federal law for welfare of animals.

Male adult mice (B6129PF2/J background, JAX Stock #100903, The Jackson Laboratory, Bar Harbor, ME, USA) were analyzed. No randomization method was used to assign the mice to designated experimental groups. Sample size was adjusted according to recent reports using male mice in HFD feeding models [21, 25]. No animals were excluded from analyses. Investigators were not blinded to group allocation during the experiments. To induce obesity, mice were fed a HFD containing 70% energy from fat (C1090-70, Altromin, Lage, Germany) for 12 weeks (HFD; n = 10). The control group received standard diet (CD; n = 10) with 13% energy from fat. For short-term feeding experiments, mice were fed either CD (CD; n = 5) or HFD (HFD; n = 4) for one week. Throughout the study, mice were housed in groups with free access to chow and water. After the feeding experiment, the animals were sacrificed by cervical dislocation under deep isoflurane anesthesia for tissue sampling.

Human gastric tissue sample acquisition

Patients with obesity scheduled for laparoscopic sleeve gastrectomy (LSG) in the obesity center Bad Cannstatt (Klinikum Stuttgart) were enrolled and compared to normal-weight subjects. They participated in a clinical trial (German Clinical Trials Register: DRKS00029161). Gastric biopsy specimens from our former study [34] of normal-weight subjects undergoing endoscopy for routine diagnosis served as controls. In total, 6 patients with obesity (3 men, 3 women, body mass index (BMI) 53.7 ± 1.5 ranging from 50.1-58.4, age 45.7 ± 5.3 years ranging from 24-58 years) and 6 normal-weight subjects (3 men, 3 women, BMI 24.5 ± 0.9 ranging from 21.6-27.4, age 41.8 ± 5.6 years ranging from 24-60 years) were analyzed. Exclusion criteria for all subjects were type 2 diabetes, infection with Helicobacter pylori, and gastrointestinal disorders.

Mouse tissue preparation

After removal of the fundus, rinsed stomachs were either used for polymerase chain reaction (PCR) analyses or fixed in 4% buffered paraformaldehyde for immunohistochemistry. After fixation for 2 h at 4 °C, the tissue was cryoprotected in 25% sucrose at 4 °C overnight, embedded in Tissue Freezing Medium (Leica Microsystems, Bensheim, Germany) and quickly frozen in liquid nitrogen. Sections (8 μm) were cut on a CM3000 cryostat (Leica Microsystems) and attached to Superfrost Plus microslides (Menzel Gläser, Braunschweig, Germany). For RNA isolation, narrow strips comprising the uppermost glandular circumference were carefully excised, immediately snap-frozen and stored at −70 °C until use.

Human tissue preparation

During LSG of patients with obesity, 1\(\times\)0.5 cm corpus specimens were excised and immediately snap-frozen and stored at −70 °C until use. For immunohistochemistry, samples were fixed in 4% formalin/formaldehyde solution (Roti-Histofix, Carl Roth, Karlsruhe, Germany) for 2 h at 4 °C and processed as described above. As controls, similarly prepared frozen tissue sections of normal-weight subjects were used [34].

Gastric pH determination

For intragastric pH measurements, a recently described method was adapted [22]. Briefly, stomachs were opened along the greater curvature and crude food was removed. Then, the inner side of the stomach was washed with 400 µl water and the resulting fluid collected. The pH was measured using a pH meter (Mettler Toledo, Albstadt, Germany).

Histological analysis and immunohistochemistry of the gastric mucosa

Cryosections were stained with periodic acid-Schiff (PAS) using 1% periodic acid (Carl Roth) and Schiff’s reagent (Merck, Darmstadt, Germany) to assess the distribution of neutral gastric mucus.

Prior to immunohistochemical analysis, citrate-antigen-retrieval was carried out (except for IL-1β) by incubating the frozen sections in sodium citrate buffer for 45 min at 4 °C followed by boiling for 1 min at 100 °C. After three washing steps with phosphate-buffered saline (PBS), sections were blocked in PBS containing 10% normal donkey serum and 0.3% Triton X-100 for 1 h at room temperature, followed by an overnight incubation with the primary antibody. For murine sections, the following antibodies at a dilution of 1:400 (except for IL-1β) were used: rat anti-fibroblast growth factor 23 (FGF23) (MAB26291, R&D systems, Minneapolis, MN, USA), mouse anti-hydrogen potassium-ATPase (H+/K+-ATPase) as parietal cell marker (sc-374094, Santa Cruz Biotechnology, Santa Cruz, CA, USA), mouse anti-platelet endothelial cell adhesion molecule (PECAM/CD31) for endothelial cells (sc-376764, Santa Cruz Biotechnology), rabbit anti-trefoil factor 1 (TFF1) for mucus-producing cells (GTX121461, GeneTex, Irvine, CA, USA), rabbit anti-α-smooth muscle actin (α-SMA) to visualize pericytes and smooth muscle cells (ab5694, Abcam, Cambridge, England), the leukocyte marker rat anti-CD45 to detect immune infiltrates (550539, BD Biosciences, Heidelberg, Germany), and rabbit anti-IL-1β for labeling of infiltrating and epithelial cells (at a dilution of 1:200; P420B, Thermo Fisher Scientific, Darmstadt, Germany).

For human sections, the above-listed rat anti-FGF23 antibody was used (1:100) and for detection of endothelial cells mouse anti-PECAM/CD31 antibody (1:200) (3528 (89C2), Cell signaling, Leiden, Netherlands). Appropriate secondary antibodies conjugated with either Alexa 488 and Cy3 (Dianova, Hamburg, Germany) or to Alexa 555 and Alexa 568 (Thermo Fisher) were diluted 1:500 in blocking solution. After incubation for 2 h at room temperature, sections were rinsed three times in PBS and counterstained with 4,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich, Schnelldorf, Germany) to visualize nuclei. Sections were processed without the respective primary antibody to validate immunohistochemical findings. No immunoreactivity was observed in these control experiments. Staining with Nile red (dissolved 1 mg/ml in acetone, Thermo Fisher) was performed using a 1:3000 solution in PBS to visualize lipid deposition.

Immunofluorescence was examined and documented on a Zeiss Axiophot microscope (Carl Zeiss MicroImaging, Jena, Germany). Images were captured using a SensiCam CCD camera (PCO Computer Optics), adjusted for contrast in AxioVision LE Rel. 4.3 (Carl Zeiss MicroImaging) and arranged in PowerPoint (Microsoft; version 16) or Adobe Photoshop (Adobe Systems; version 7.0).

Morphometric analyses

The proximal mucosal region of the stomach within 3 mm around the lesser curvature was analyzed by acquiring digital microscopic images of sections at 40× magnification. The number of FGF23-positive (FGF23+), mucus-like (TFF1+) and endothelial cells (PECAM+) as well as the quantity of blood vessels were expressed per mm2. Relative FGF23+/TFF1+ areas were determined using ImageJ 1.52p software and given as percentage per mm2.

Luminex assay

Serum concentrations of tumor necrosis factor receptor type 1 (TNFR1) and C-C motif chemokine ligand 2 (CCL2) were quantified using Luminex technology. Therefore, mouse-specific assays from R&D Systems were employed according to the instructions of the manufacturer.

Qualitative and quantitative polymerase chain reaction of murine samples

Total RNA was isolated from narrow strips of the proximal stomach using peqGOLD TriFast reagent (VWR, Bruchsal, Germany). For cDNA synthesis, 1.2 µg (or 3 µg for Fgf23 expression analysis) RNA with SuperScript III Reverse Transcriptase (Thermo Fisher Scientific), oligo(dT)12–18 and random primers (Promega, Mannheim, Germany) were used on a Biometra TAdvanced thermal cycler (Analytik Jena, Jena, Gemany).

For qualitative expression analysis, RNA from each group was pooled, and 150 ng cDNA used in PCRs with Titanium Taq DNA Polymerase (Takara, Frankfurt, Germany). The cycling profile was: initial incubation at 95 °C for 2 min, then 24 cycles: 30 s at 95 °C, 20 s at 57 °C, and 10 s at 68 °C. Samples processed with intron-spanning primers and without reverse transcriptase (-RT) served as controls for the absence of genomic DNA. PCR products were separated on an agarose gel and stained with ethidium bromide.

Quantitative polymerase chain reactions (qPCRs) were performed using the CFX Connect Real-Time PCR Detection System (Bio-Rad, Feldkirchen, Germany). Each qPCR sample (20 µl per reaction) consisted of a reaction mix containing 0.5 µM of forward and reverse primer, 10 µl GoTaq qPCR Master Mix (Promega), 300 ng cDNA of individual mice, and water. Conditions were: initial incubation at 95 °C for 2 min, then 45 cycles: 95 °C for 15 s, 60 °C (57 °C for Fgf23) for 15 s, and 72 °C for 15 s. Melting curve analysis and agarose gel electrophoresis were included to ensure that only a single, specific amplicon was produced. The oligonucleotide primer sequences (5’→3’) were: Dmp1 sense: GAA CAG TGA GTC ATC AGA AG, Dmp1 as: AAA GGT ATC ATC TCC ACT GTC (nt 490-685 from GenBank accession number NM_001359013.1), Fgf23 sense: TCG AAG GTT CCT TTG TAT GGA, Fgf23 as: AGT GAT GCT TCT GCG ACA AGT (nt 394–524, NM_022657.5), Il-6 sense: AAG AAA TGA TGG ATG CTA CC, Il-6 as: GAG TTT CTG TAT CTC TCT GAA G (nt 360-523, NM_001314054.1), Il-11 sense: TCT CCT AAC CCG ATC CCT CCT G, Il-11 as: TGC AAA GAT CCC AAT GTC CCA G (nt 284-421, NM_008350), Rpl8 sense: GTG CCT ACC ACA AGT ACA AGG C, Rpl8 as: CAG TTT TGG TTC CAC GCA GCC G (nt 600–803, NM_012053.2), Muc2 sense: AGC CTG GGG AGA TTC ACA AAA ACC, Muc2 as: ACG GAG ACA GCA GAG CAA GGG A (nt 6691-6907, NM_023566), and Sox2 sense: TGA CCA GCT CGC AGA CCT ACA TG, Sox2 as: CGG ACT TGA CCA CAG AGC CCA T (nt 668-778, AB108673), Tnf-α sense: GGA TGA GAA GTT CCC AAA TG, Tnf-α as: TGA GAA GAT GAT CTG AGT GTG (nt 347-421, NM_013693.3).

Western blot analysis

Narrow strips of the proximal stomach were lyzed in T-PER tissue protein extraction reagent (Thermo Fisher Scientific) supplemented with complete protease and phosphatase inhibitor cocktail and EDTA (Thermo Fisher Scientific). Proteins (30 µg per lane) were separated on 12% SDS polyacrylamide gels and transferred onto nitrocellulose membranes. For total protein normalization, membranes were stained with Ponceau S solution (AppliChem, Darmstadt, Germany). After destaining, membranes were blocked and subsequently probed with the above-listed rat anti-FGF23 antibody diluted 1:1000 in EveryBlot Blocking Buffer (Bio-Rad) overnight at 4 °C followed by incubation with a HRP-conjugated secondary antibody. Antibody binding was detected with the ECL detection reagent (Bio-Rad) using the ChemiDoc MP Imaging System (Bio-Rad). After normalization to total protein, data were presented as relative fold change of FGF23 expression in CD- versus HFD-fed mice using Image Lab software 6.1 (Bio-Rad).

Stimulation of cultured endothelial and epithelial cells and expression analysis

For human umbilical vein endothelial cell (HUVEC) cultures, cells were freshly isolated from umbilical cords provided by local hospitals under the approval of the Ethics Committee of the University Hospital Heidelberg (S-383/2013). Cultivation was carried out as previously described [35]. As a gastric epithelial model, the normal human gastric epithelial cell line GES-1 (Cytion, Eppelheim, Germany) was cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (all from Thermo Fisher Scientific) under standard conditions.

For stimulation of HUVECs, 1.0-1.2 × 106 cells were seeded into 6-well plates, subcultured upon 90-95% confluence and incubated with 100 ng/ml recombinant human leptin (Thermo Fisher Scientific) for 48 h. For stimulation of GES-1 cells, 1.5 × 105 cells of passages 17–21 were plated on 6-well plates. After 24 h, GES-1 cells were treated with 100 ng/ml recombinant human IL-1β (Miltenyi Biotec, Bergisch Gladbach, Germany), 50 µM BSA-palmitate saturated fatty acid complex (Cayman Chemicals, Ann Arbor, MI, USA), and 25 ng/ml recombinant human FGF23 (Thermo Fisher Scientific) for 24 h. Treatment with equal amounts of vehicle and BSA control for BSA-fatty acid complexes (Cayman Chemicals) were used as control.

Total RNA was isolated from cells as described above. After DNase treatment (Thermo Fisher Scientific), 2 µg RNA were reverse transcripted (GoScript Reverse Transcription System with random primers (Promega)). Expression of FGF23 normalized to the internal control encoding either RPL32 (HUVEC) or HPRT1 (GES-1) was analyzed by qPCR using the following human primers: FGF23 sense: GGA TGC TGG CTT TGT GGT GA, FGF23 as: TGC AGT TCT CCG GGT CGA AAT (nt 364-470, NM_020638.3), HPRT1 sense: ATA AGC CAG ACT TTG TTG G, HPRT1 as: ATA GGA CTC CAG ATG TTT CC (nt 606-784, HUMHPRT), and RPL32 sense: AGG CAT TGA CAA CAG GGT TC, RPL32 as: GTT GCA CAT CAG CAG CAC TT (nt 194-353, NM_000994.4). Conditions were: initial incubation at 95 °C for 2 min, then 45 cycles: 95 °C for 15 s, 60 °C (58 °C for GES-1) for 15 s, and 72 °C for 15 s.

Analysis of qPCR data

For analysis of murine samples, relative amounts of transcripts were normalized to Rpl8 that served as a non-regulated housekeeping gene. For analysis of expression levels in cell culture experiments, the expression of FGF23 was normalized to the non-regulated housekeeping gene RPL32 (HUVEC) or HPRT1 (GES-1). Data from mouse experiments were calculated using the 2−ΔCt method. Relative expression changes of HUVECs and GES-1 cells were calculated using the 2−ΔΔCt method. Data were expressed relative to levels in the vehicle-treated samples (control) and indicated as fold change. For calculations of FGF23 expression, samples with Cycle threshold (Ct) values ≥ 40, were arbitrarily assigned a Ct value of 40.

Statistical analysis

Statistical analyses were performed using JASP software (version 0.19.3.0, University of Amsterdam, Amsterdam, Netherlands) or GraphPad Prism 6 (version 6.01; GraphPad Software, Boston, MA, USA). Normality of data distribution was evaluated using Shapiro-Wilk test. Homogeneity of variance was assessed using Levene’s test. Two groups were analyzed with student’s t-test, Welch’s t test, or Mann-Whitney U test. Fold change was analyzed with one-sample Wilcoxon signed rank test. All data are presented as mean ± standard error of the mean (SEM), with n indicating the number of individuals per group or independent cell culture experiments. Statistical significance was set at P < 0.05.