Animals. Adult TgBAC (gfap:gfap-GFP) zebrafish (> 6 month of age; AB strain; European Zebrafish Resource Center, Karlsruhe, Germany) have been used in this study . They were kept under standard conditions in tank water and raised in a 14/10 h light/dark cycle. All new generations were monitored for GFP using a stereoscopic microscope . Female and male B6-Tg (Rlbp1-GFP) mice (4–8 weeks old; originally provided by Prof. Dr. Christian Grimm) were kept in standard conditions with a 12-light/12-h dark cycle with food and water available ad libitum. Genotyping of Rlbp1-GFP mice was performed as previously described . All animal experiments were approved by the local Animal Ethics Committee of the Canton Bern (Switzerland; BE34/19 and BE33/18) and conform to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.
Human Donor Eyes. Retinal tissue from eight human donors (70-90-year-old) was used.
Retinal laser focal Injury. For both animal models, laser focal injury was induced as previously described . Briefly, zebrafish were anesthetized with 0.16 mg/mL ethyl 3-aminobenzoate methanesulfonate salt (Tricaine; Sigma-Aldrich, St. Louis, MO, USA) dissolved in tank water. A 532 nm diode laser (Visulas 532 s, Carl Zeiss Meditec AG, Oberkochen, Germany) was used to create lesions at the region of the posterior pole around the optic nerve. Four laser burns were applied to both eyes and the surrounding intact tissue was used as a negative control. For the RNAseq analysis, 20 laser burns were created. Each burn was produced with 70 mW of power for 100 ms and aimed to have a diameter of 50 µm. Mice were anesthetized by injecting subcutaneously 45 mg/kg ketamine (Ketalar 50 mg/ml; Orion Pharma AG, Zug, Zurich, Switzerland) and 0.75 mg/kg medetomidine hydrochloride (Domitor, 1 mg/ml; Orion Pharma AG). The same diode laser was used to create six lesions on the both eyes. For the RNAseq analysis, 50 laser burns were created in both eyes. Each burn was produced with 120 mW of power for 60 ms and aimed to be 100 µm in diameter.
Spectral domain-optical coherence tomography (SD-OCT) and quantification. In vivo imaging of the murine retina was performed as previously described . After anesthesia, pupils were dilated with a drop of tropicamide 0.5% phenylephrine 2.5% (ISPI, Bern, Switzerland), and Methocel (OmniVision AG, Neuhausen, Switzerland) applied to each eye during imaging to keep them hydrated. Standard confocal laser scanning ophthalmoscope (Spectralis HRA + OCT; Heidelberg Engineering GmbH, Heidelberg, Germany) was used to image the murine retina . After examination of both eyes, SD-OCT was performed using a 55° lens at a high resolution of 1008 × 596 pixels in grid mode. After imaging, atipamezole (2.3 mg/kg, Antisedan 5 mg/ml, Provet AG, Lyssach, Switzerland) was used to antagonize the anesthesia. The area of each lesion was measured by using the Heidelberg Eye Explorer software (Heidelberg Engineering GmbH).
Pharmacological cell-cycle arrest in zebrafish. Male and female zebrafish were randomly selected to be treated with palbociclib (PD0332991, Selleck Chemicals, Houston, TX, USA), a selective inhibitor of cyclin-dependent kinase (CDK) 4/6. The final concentration of 2 µM in tank water was based on a previous report . Zebrafish were immersed at different timepoints (Day 4, 5 and 6) after injury induction and euthanized after 24 h (Day 5, 6 and 7). Injection paradigms are included in Figure S3. The negative control group was kept in tank water. Animals showing behavioral and/or morphological changes during treatment were excluded from the study.
Pharmacological treatment in mice. Both male and female Rlbp1-GFP mice were randomly divided into three groups. The first group was treated with a γ-secretase inhibitor, (2S)-N-[(3,5-Difluorophenyl) acetyl]-L-alanyl-2-phenyl]glycine 1,1-dimethylethyl ester (DAPT; Tocris, Zug, Switzerland). DAPT powder was dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich, Buchs, Switzerland) and injected intraperitoneally (8 mg/kg body weight) either at 3 h before injury, at day 2, or at day 6. One day after injury, mice were euthanized (Day 1, 3 and 7) . The second and the third groups were treated with either 5-methyl-1-phenylpyridin-2-one (Pirfenidone; Selleckchem, Houston, TX, USA), which decreases the expression of TGFβ1/2/3 cytokines , or (E)-1-(6,7-dimethoxy-3,4-dihydro-1H-isoquinolin-2-yl)-3-(1-methyl-2-phenylpyrrolo[2,3-b]pyridin-3-yl)prop-2-en-1-one (SIS3, Selleckchem), a novel specific inhibitor of p-Smad3. Pirfenidone and SIS3 solutions were prepared according to previous studies , both drugs were dissolved in PBS containing 2% DMSO. Solutions were sonicated at 45 °C until transparent. Then 30% of polyethylene glycol (PEG)-300 (Med Lab Supply, Miami, USA) was added to both. We added 2% Tween80 (Sigma-Aldrich) only to the SIS3 solution. Both mixtures were diluted with double distilled water (ddH2O) to 100 ml (5 mg/ml). Intraperitoneal injection dose was 50 mg/kg for pirferidone and 2.5 mg/kg for SIS3 at either 3 h before injury, at day 2 or at day 6. One day after injury, mice were euthanized (Day 1, 3 and 7, respectively). Injection paradigms are included in each figure (Fig. 5, S5, 6). Long-term treatment with SIS3 or control vehicle (PBS) was performed by intraperitoneal injection (2.5 mg/kg) at 3 h before injury, daily during the first three days after injury, and then every other day until day 14.
Tissue processing and immunohistochemical studies. The eyes in both animal models were enucleated at different timepoints (Day 1, 3, 7 and 14) after injury and fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) overnight. Human retina was fixed 1 h in 4% PFA in PBS. Paraffin sections (5 µm) were stained with Mayer’s hemalum and eosin (H&E; Roth, Karlsruhe, Germany;  or used for immunofluorescence. Antigen retrieval was achieved by incubation in either Tris-EDTA (pH 9.0) or Citrate buffer (pH 6.0) with 0.05% Tween-20 for 20 min and then cooled at room temperature (~ 30 min). All sections were blocked for 1 h in Tris-buffered saline (TBS) + 5% goat normal serum + 1% bovine serum albumin (pH 7.6) and incubated with primary antibodies overnight at 4 °C. Primary antibodies used in this study were: mouse anti-glutamine synthetase (GS; 1:200; MAB302; Millipore, Billerica, MA, USA), rabbit anti-glutamine synthetase (GS; 1:200; ab210107; Abcam, Cambridge, UK), rabbit anti-glial fibrillary acidic protein (GFAP; 1∶200; OPA1‐06100; ThermoFisher Scientific, Basel, Switzerland), rabbit anti‐phospho extracellular signal‐regulated kinase (Erk1/2; 1:100; 9101; Cell Signaling Technology, Danvers, MA, USA), mouse anti‐proliferating cell nuclear antigen (PCNA; 1∶500; ab29; Abcam), rabbit anti‐gamma histone H2A variant H2A.X (γH2A.X; 1∶200; ab228655; Abcam), rat anti-H2A.Z (1∶150; ab228655; Abcam), rabbit anti-phospho-histone 3 (pH3; 1:150; 9713T; Cell Signaling Technology), mouse anti-E Cadherin (1:200; ab76055; Abcam), rabbit anti-N Cadherin (1:500; ab18203; Abcam), rabbit anti‐Notch homolog 1 (Notch1; 1:200; ab52627; Abcam), rabbit anti‐Notch homolog 2 (Notch2; 1:100; D76A6; Cell Signaling Technology), rabbit anti-transforming growth factor beta 1 (Tgfβ1; 1:200 dilution; ab215715; Abcam), mouse anti-transforming growth factor beta 2 (Tgfβ2; 1:50 dilution; ab36495; Abcam), rabbit anti-transforming growth factor beta 3 (Tgfβ3; 1:100 dilution; ab15537; Abcam), rabbit anti-phosphorylated mothers against decapentaplegic homolog 3 (p-Smad3; 1:50 dilution; ab52903; Abcam), rabbit anti-orthodenticle homeobox 2 (Otx2; 1:200 dilution; ab183951; Abcam) and rabbit anti-paired box 6 (Pax6; 1:200 dilution; ab195045; Abcam). As secondary antibodies, goat anti-rabbit/anti-mouse Alexa 488 nm/594 nm (1∶500; ThermoFisher Scientific, Basel, Switzerland) was diluted in TBS with 1% BSA for 1 h at room temperature. The cell nuclei were counterstained using Vectashield with 4′, 6-diamidino-2-phenylindole (DAPI; Vector Labs, Burlingame, CA, USA).
Flow cytometry analysis. At different timepoints after injury (Day 1, 3 and 7), retinas of gfap:gfap-GFP zebrafish and Rlbp1:GFP mice were used for flow cytometry analysis. Both retinas of each mouse were analyzed as one sample. Before antibody staining, single cells suspensions were incubated with Hoechst 33342 Ready Flow™ Reagent (ThermoFisher Scientific) in Hank's Balanced Salt Solution (HBSS; ThermoFisher Scientific) with DNase I (200 U/ml; Roche, Rotkreuz, Switzerland) to exclude dead cells. For antibody staining, the samples were washed, re-suspended in HBSS with 20% fetal bovine serum (FBS; ThermoFisher Scientific) and 200 U/ml DNase I. Reactive MCs were subsequently stained with fluorescent-labeled antibodies against GFAP (Alexa Fluor® 488 anti-GFAP antibody, 2E1.E9; Biolegend, San Diego, CA, USA), PCNA (PE anti-human/mouse/rat PCNA antibody, 307908; Biolegend), Notch1 (Brilliant Violet 421™ anti-mouse Notch 1 antibody, 130615; Biolegend) and with Notch2 (APC anti-mouse Notch 2 antibody, 130713; Biolegend) at 4 °C in the dark for 40 min. Samples were washed again and re-suspended in 0.1% PFA (pH 7.4) at 4 °C in the dark for 10 min. Samples were washed twice, re-suspended in flow cytometry buffer, and analyzed. All washing steps involved addition of 1 ml HBSS with 0.01% DNase to each sample and centrifugation at 300 g at 4 °C for 3 min. Data were acquired with an LSR II Cytometer System and the BD FACSDiva software (BD Biosciences, Allschwil, Switzerland). The data were analyzed with the Flowjo Single Cell Analysis Software V10 (TreeStar, Ashland, OR, USA).
Retinal dissociation, sorting, and RNA-Seq library production. Both retinas of three gfap:gfap-GFP zebrafish per timepoint (Day 1, 3, and 7) were dissected and dissociated in 0.05% trypsin (ThermoFisher Scientific) for 10 min and then suspended in DEPC-PBS with 10% FBS (ThermoFisher Scientific) and DNase I (200 U/ml; Roche). Cell suspension was filtered and collected in Falcon® Round-Bottom Tubes with CellStrainer Cap (12 × 75 mm; Costar Corning, Cambridge, MA). Hoechst 33342 Ready Flow™ Reagent (ThermoFisher Scientific) was added as a DNA dye for cell-cycle analysis. Cells from gfap:gfap-GFP negative littermates were used to determine background fluorescence levels. 100 cells/µl were collected from gfap:gfap-GFP positive zebrafish using Moflo Astrias EQ Cell Sorter (Beckman Coulter, Brea, CA, USA) into 4 µl of Buffer TCL (Qiagen, Venlo, The Netherlands) with 1% 2-mercaptoethanol (63689; Sigma-Aldrich). Both retinas of three Rlbp1:GFP mice were dissected at different timepoints (Day 1, 3, and 7) and incubated with papain (Worthington Biochemical, Freehold, NJ, USA) for 15 minutes as previously described . After dissociation, cell suspension in HBSS with 0.4% BSA (ThermoFisher Scientific) and DNase I (200 U/ml; Roche) was filtered with a 35 µm cell strainer. Hoechst 33342 Ready Flow™ Reagent (ThermoFisher Scientific) was added as a DNA dye for cell-cycle analysis. Cells from Rlbp1:GFP negative littermates were used to determine background fluorescence levels. 100 cells/µl were collected from Rlbp1:GFP positive mice using Moflo Astrias into 4 µl Buffer TCL (1031576; Qiagen) plus 1% 2-mercaptoethanol (63689; Sigma-Aldrich). After cell sorting, all samples were processed using the published Smart-seq2 protocol to generate the cDNA libraries . The libraries were sequenced in an Illumina HiSeq4000 (Illumina, San Diego, CA, USA) with a depth of around 20 Mio reads per sample. Sequencing data are available in the Gene Expression Omnibus database (NCBI tracking system #19961614).
RNA-Sequencing Analysis. The raw reads were first cleaned by removing adapter sequences, trimming low quality ends, and filtering reads with low quality (phred quality < 20) using Trimmomatic (Version 0.36). The read alignment was done with STAR (v2.6.0c). As reference the Ensembl zebrafish genome build GRCz10 from 2017-06-07 (release 89) and respectively the Ensembl murine genome build GRCm38.p5 with the gene annotations downloaded on 2018-02-26 from Ensembl (release 91) were used. The STAR alignment options were "--outFilterType BySJout --outFilterMatchNmin 30 --outFilterMismatchNmax 10 --outFilterMismatchNoverLmax 0.05 --alignSJDBoverhangMin 1 --alignSJoverhangMin 8 --alignIntronMax 1000000 --alignMatesGapMax 1000000 --outFilterMultimapNmax 50". Gene expression values were computed with the function featureCounts from the R package Rsubread (v1.26.0). The options for feature Counts were: - min mapping quality 10 - min feature overlap 10 bp - count multi-mapping reads - count only primary alignments - count reads also if they overlap multiple genes. To detect differentially expressed genes, we applied a count based negative binomial model implemented in the software package DESeq2 (R version: 3.5.0, DESeq2 version: 1.20.0). The differential expression was assessed using an exact test adapted for over-dispersed data. Genes showing altered expression with an adjusted p-value < 0.05 (Benjamini and Hochberg method) were considered differentially expressed. Heatmaps were generated for selected subsets of genes in R v. 3.5.1 using the heatmap.2 function from package gplots v. 3.0.1. The data displayed the log2 fold-changes between two experimental groups. Rows are reordered based on a dendrogram from hierarchical clustering. Subsets of genes identified as interesting were explored using QIAGEN’s Ingenuity® Pathways Analysis suite (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity) for pathways, networks, and functional analyses.
Image analyses and quantification. Immunofluorescence imaging was performed at 40x magnification with a scanning laser microscope (Zeiss LSM710; Carl Zeiss Microscopy, Jena, Germany). Sagittally oriented zebrafish and murine retinal sections at the level of the laser burns were used to quantify positive cells. The analyzed length of the retina was 50 µm in zebrafish or 100 µm in mouse, corresponding to the induced injury size. Arbitrary quantification of the central (fovea), mid-peripheral, and peripheral zones of each human retina was 606 µm in length (microscope's visual field at 40x). The number of positive cells was normalized to the total number of MCs (cytoplasmatic GS+ or nuclear SOX9+). Cells were manually determined. Ratios between positive cells on the total of MCs in the injured area were expressed as percentages. High-throughput and high-quality brightfield H&E-stained images of the human retina at 40x or 63x total magnification were acquired with a motorized Pannoramic 250 Flash II microscope (3DHISTECH Ltd., Budapest, Hungary). Randomized quantification of the central (fovea), mid-peripheral, and peripheral zones of each human retina was 950 µm in length (microscope's visual field at 40x). Human samples were divided in two groups based on H&E and immunofluorescence data: control group (ctrl) and retina presenting drusen accumulation (drusen pos). Drusen were identified as accumulations of extracellular material that build up between Bruch's membrane and the retinal pigment epithelium and manually counted.
Statistical analysis. Statistical analysis was performed using GraphPad Prism (version 7.0, GraphPad Software, La Jolla, USA). Intergroup comparisons were based on a non-parametric one-/two-way analysis of variance (ANOVA) and the Bonferroni multiple comparison post hoc test. For the pharmacologically treated animals, comparison between uninjured and treatment groups was performed with two-tailed t-test. Quantifications were performed on three laser burns performed in the left eye in four different animals for all timepoints (n = 12). All results are expressed as the mean ± standard deviation (SD). The level for statistical significance was set at a p value ≤ 0.05.