The care and experimental use of all animals in the study were in accordance with EC directive 2010/63/EU and in compliance with the Association for Research in Vision and Ophthalmology statement. All animal experiments were approved by the local ethics committee (Regierungspräsidium Karlsruhe, Karlsruhe, Germany). Mice were kept in a specified pathogen-free environment in 12 h light–dark cycles with free access to food and water. Male C57BL/6J mice, purchased from Charles River (Wilmington, MA, USA) and homozygous Glp1r−/− mice (Glp1rtm1Ddr), kindly provided by Boehringer Ingelheim, were used for the OIR model (Boehringer Ingelheim, Biberach an der Riß, Germany) . Each mouse strain was divided into three groups: mice without OIR as controls; mice with OIR; and mice with OIR treated with subcutaneous linagliptin (Boehringer Ingelheim) from postnatal day 12 to 16 at a concentration of 10 mg/kg body weight (OIR + Lina). Animals of each litter were randomly numbered and numbers equally assigned from top to bottom to OIR and OIR + Lina. Every third litter was selected as control group without subjection to the OIR model. At postnatal days 12 and 17, both eyes of mice in randomly selected litters were isolated during isoflurane anaesthesia and the mice killed by cervical dislocation. The eyes were fixed in 4% (vol./vol.) formalin for analysis of neovascularisation or stored at −80°C for gene expression analysis. Plasma was obtained for the determination of active GLP-1 as an indicator of DPP-4 inhibition by linagliptin.
Glp1r−/− and C57BL/6J mice were investigated for quantification of preretinal neovascularisation using the established OIR model for mice in periodic acid Schiff-stained paraffin sections . In brief, newborn mouse litters were exposed from postnatal day 7 until postnatal day 12 in an incubation chamber with 75% (vol./vol.) oxygen and then returned to room air until postnatal day 17. Linagliptin was administrated from postnatal day 12 to postnatal day 16. OIR litters consisted of OIR control mice and linagliptin-treated animals, randomly assigned to the treatment group. Litters kept in room air for the entire time until postnatal day 17 served as controls.
Determination of the avascular zone
To evaluate vasoregression, eyes from C57BL/6 and Glp1r−/− mice at OIR postnatal day 12 were submitted to immunofluorescence staining with collagen IV (1:100; Acris, Herford, Germany) to visualise the vessel net in the superficial layer and to lectin-FITC (1:50; Sigma-Aldrich, Munich, Germany) to visualise capillaries in the superficial and deep layers. After overnight incubation, the following secondary antibody for collagen IV was used: swine anti-rabbit tetramethylrhodamine (1:50; Dako Cytomation, Hamburg, Germany). The avascular zone was measured using Cell-F imaging software (Olympus Opticals, Hamburg, Germany) at ×2 magnification.
Experimenters were blinded for the quantification of neovascularisations and avascular zones.
Determination of active GLP-1
Active GLP-1 concentrations (GLP-1 [7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36] amide and GLP-1 [7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37]) were determined in the plasma of the animals at the end of the study using commercially available ELISAs (K150JVC-1 and K150JWC-1, Meso Scale Discovery, Gaithersburg, MD, USA).
Receptor tyrosine kinase signalling reporter assay
Human umbilical vein endothelial cells (HUVECs; Thermo Fisher Scientific, Weiterstadt, Germany; authenticated and confirmed mycoplasma-negative by the supplier) were cultured to 90% confluence. The RTK Signaling 10-Pathway Reporter Array (Qiagen, Hilden, Germany) was used to measure the activity of the mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase (ERK), MAPK/c-Jun N-terminal kinase (JNK), phosphoinositide 3-kinase (PI3K)/Akt and NFκB pathways according to the manufacturer’s instructions. In brief, cells were reversely transfected with the corresponding DNA constructs using the Attractene transfection reagent (Qiagen) for 24 h in OPTI-MEM (Thermo Fisher Scientific). After transfection, cells were stimulated with 10 ng/ml VEGF (Sigma-Aldrich, Munich, Germany) with or without 50 nmol/l linagliptin (Boehringer Ingelheim) for 24 h. Firefly and Renilla luciferase signals were detected using a common microplate reader (Spark, Tecan Trading, Männedorf, Switzerland) using a dual-luciferase assay (Promega, Mannheim, Germany) according to the manufacturer’s protocol.
Western blot for p-ERK1/2, ERK1/2 and p-VEGFR2
The isolated retina was homogenised in 0.1% (wt/vol.) SDS lysis buffer and the protein concentration was determined using the Bradford assay (Bio-Rad, Munich, Germany). Samples were separated in a 4–20% (wt/vol.) gradient TGX Gel (Bio-Rad) and immunoblotted to a polyvinylidene difluoride membrane (Bio-Rad). Unspecific binding was blocked by incubation with 5% (wt/vol.) non-fat dry milk in Tris-buffered saline containing 0.1% (vol./vol.) Tween (Sigma-Aldrich, Darmstadt, Germany), followed by overnight incubation at 4°C with the antibodies mouse anti-mouse p-ERK1/p-ERK2 (1:10,000; Abcam, Cambridge, UK), rabbit anti-mouse ERK1/ERK2 (1:1500; Abcam), rabbit anti-mouse p-Tyr1175VEGFR2 (1:1300; Cell Signaling, Frankfurt, Germany) or rabbit anti-mouse β-tubulin (1:500; Abcam). For detection, horseradish-peroxidase conjugated antibodies against mouse for p-ERK1/p-ERK2 (1:1000; Dako Cytomation) and against rabbit for ERK1/ERK2 (1:2000, Cell Signaling), p-VEGFR (1:2000; Cell Signaling) and tubulin (1:3000; Dako Cytomation) were used. Antibodies were validated by the manufacturer and in previous studies. Immunoreactive bands were visualised by incubation in chemiluminescence reagent (PerkinElmer, Boston, MA, USA) and signals were detected using the Fusion SL (VWR, Darmstadt, Germany). Integrated densities were measured using ImageJ software v 1.50i .
Gene expression analysis
Retinal RNA was isolated and homogenised in TRIzol reagent (Invitrogen, Karlsruhe, Germany). RNA was reverse transcribed using the QuantiTect Reverse Transcription kit (Qiagen) and subjected to quantitative (q)PCR analysis using hydrolysis probes (TaqMan probes, Applied Biosystems, Weiterstadt, Germany). Gene expression was analysed by the comparative ΔΔCq method using Actb and Gapdh as reference genes. All primers and probes were purchased from Applied Biosystems (see electronic supplementary material [ESM] Table 1 for details).
Measurement of total SDF-1α
Systemic total SDF-1α was measured in EDTA plasma using a standard Luminex assay (Natural and Medical Sciences Institute, University of Tübingen, Tübingen, Germany). The protocol uses overnight incubation on Millipore filter plates (Merck Millipore, Darmstadt, Germany) and beads to measure total SDF-1α.
For cAMP measurement, eyes for retinal explants from adult C57BL/6J and Glp1r−/− mice were obtained under deep anaesthesia and immediately transferred to OPTI-MEM (Thermo Fisher Scientific) containing IBMX (3-isobutyl-1-methylxanthine; Sigma-Aldrich, Munich, Germany). The left retinal explant was treated with GLP-1 (7-36) amide (Bachem, Bubendorf, Switzerland) and lixisenatide (Sanofi, Frankfurt, Germany) as binding ligands for GLP-1R. The right retinal explant served as an untreated control. After explant lysis, cAMP was measured using a cAMP direct fluorimetric immunoassay kit (Abcam) according to the manufacturer’s instructions.
Quantification of neovascular nuclei, avascular zones and active GLP1 has been performed once with the number of animals stated in the figure legends. The receptor tyrosine kinase signalling reporter assay has been performed as four-time quantified duplicates, with the mean of two simultaneously measured duplicates for each group being n = 1 of 4. Western blots have been performed twice with the experiment showing the maximum spread of grey values being quantified. Gene expression analysis has been performed in duplicate. Data are presented as means ± SD. Differences between groups were analysed using ANOVA with the Bonferroni post hoc method for multiple comparisons. Statistical outliers were identified using the ROUT method with Q = 1%. Statistical analyses were performed using GraphPad Prism v6.01 (GraphPad Software, San Diego, CA, USA). For all comparisons, p < 0.05 was considered statistically significant.