Experimental Eye Research 212 (2021) 108801 Contents lists available at ScienceDirect Experimental Eye Research journal homepage: www.elsevier.com/locate/yexer Decreased endostatin in db/db retinas is associated with optic disc intravitreal vascularization Aina Bonet a, b, Andreia Valença c, d, Luísa Mendes-Jorge d, e, Alba Casellas a, f, g, Alfonso Rodríguez-Baeza h, Víctor Nacher a, b, g, David Ramos a, d, Judit Pampalona a, Rafael Simó g, i, Jesús Ruberte a, b, g, * a Centre for Animal Biotechnology and Gene Therapy (CBATEG), Universitat Autònoma de Barcelona, Bellaterra, Spain b Department of Animal Health and Anatomy, Faculty of Veterinary Medicine, Universitat Autònoma de Barcelona, Bellaterra, Spain c Faculty of Veterinary Medicine, Universidade Lusófona de Humanidades e Tecnologias, Lisbon, Portugal d Centre for Interdisciplinary Research in Animal Health (CIISA), Faculty of Veterinary Medicine, Universidade de Lisboa, Lisbon, Portugal e Department of Morphology and Function, Faculty of Veterinary Medicine, Universidade de Lisboa, Lisbon, Portugal f Department of Biochemistry and Molecular Biology, Faculty of Veterinary Medicine, Universitat Autònoma de Barcelona, Bellaterra, Spain g Centro de Investigación Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), Instituto de Salud Carlos III, Madrid, Spain h Department of Morphological Sciences, Faculty of Medicine, Universitat Autònoma de Barcelona, Bellaterra, Spain i Diabetes and Metabolism Research Unit, Vall d’Hebron Research Institute, Universitat Autònoma de Barcelona, Barcelona, Spain A R T I C L E I N F O A B S T R A C T Keywords: Endostatin, a naturally cleaved fragment of type XVIII collagen with antiangiogenic activity, has been involved in Endostatin the regulation of neovascularization during diabetic retinopathy. Here, the intracellular distribution of endo­ Intravitreal vessels statin in healthy mouse and human neuroretinas has been analyzed. In addition, to study the effect of experi­ Diabetic retinopathy mental hyperglycemia on retinal endostatin, the db/db mouse model has been used. Endostatin protein db/db mice expression in mouse and human retinas was studied by immunofluorescence and Western blot, and compared with db/db mice. Eye fundus angiography, histology, and immunofluorescence were used to visualize mouse retinal and intravitreal vessels. For the first time, our results revealed the presence of endostatin in neurons of mouse and human retinas. Endostatin was mainly expressed in bipolar cells and photoreceptors, in contrast to the optic disc, where endostatin expression was undetectable. Diabetic mice showed a reduction of endostatin in their retinas asso­ ciated with the appearance of intravitreal vessels at the optic disc in 50% of db/db mice. Intravitreal vessels showed GFAP positive neuroglia sheath, basement membrane thickening by collagen IV deposition, and presence of MMP-2 and MMP-9 in the vascular wall. All together, these results point that decreased retinal endostatin during experimental diabetes is associated with optic disc intravitreal vascularization. Based on their phenotype, these intravitreal vessels could be neovessels. However, it cannot be ruled out the possibility that they may also represent persistent hyaloid vessels. 1. Introduction can inhibit endothelial cell proliferation, migration, and invasion have been proposed. Endostatin binds the vascular endothelial growth factor Endostatin was first discovered in the laboratory of Judah Folkman (VEGF) receptor KDR/Flk-1 and, thereby, interferes with VEGF signal­ in 1997 and described as an inhibitor of angiogenesis in murine ling leading to the inhibition of proliferation and migration of endo­ hemangioendothelioma cells (O’Reilly et al., 1997). It is a 20 kDa thelial cells (Kim et al., 2002). Endostatin can also promote ß-catenin fragment activated by proteolytic cleavage of the C-terminal region of degradation, inducing cell cycle arrest through inhibition of the collagen XVIII by cathepsin L, elastases, and matrix metalloproteinases cyclin-D1 promoter (Hanai et al., 2002). Furthermore, antimigratory (MMPs) (Ferreras et al., 2000). Several mechanisms by which endostatin effect of endostatin can also be mediated through its binding with a5ß1 * Corresponding author. CBATEG - Center for Animal Biotechnology and Gene Therapy, Autonomous University of Barcelona, C/ de la Vall Morona, 08193, Bellaterra (Cerdanyola del Vallès), Spain. E-mail address: jesus.ruberte@uab.es (J. Ruberte). https://doi.org/10.1016/j.exer.2021.108801 Received 22 January 2021; Received in revised form 1 September 2021; Accepted 18 October 2021 Available online 21 October 2021 0014-4835/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 1. Endostatin staining and antibody specificity in mouse retina. (A) Immunofluorescence with a rabbit polyclonal anti-endostatin antibody in mice retina showed staining in the ganglion cell layer, outer plexiform layer, and in the outer segments of photoreceptors (left panel). Blocking peptide assay (Blocking PA) using recombinant mouse endostatin was performed to assess antibody specificity. Anti-endostatin antibody neutralization was complete and non-specific signal was observed in the retina (middle panel). Negative control was done by omission of the primary antibody (right panel). (B) FcR blocking assay confirmed the specificity of endostatin staining (left panel) when compared to retinas incubated with Mouse Seroblock FcR (middle panel). Negative control was done by omission of the primary antibody (right panel). (C) Isotype control assay using an anti-collagen I antibody from the same isotype and host as the anti-endostatin antibody was performed. Collagen I is absent from retina and very abundant in cornea. In comparison with the anti-endostatin antibody retinal signal (left panel), non-signal was observed in the retina when using anti-collagen I antibody (middle panel). Positive control for collagen I was performed in the cornea of the same eye in a consecutive section (right panel). Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer; Co: cornea. Scale bars: 37.03 μm. 2 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 including hyaloid vessels (Bhutto et al., 2004; Ohlmann et al., 2004; May et al., 2006; Määttä et al., 2007). It is believed that this staining pattern is due to cross-reactivity between endostatin and its precursor, collagen XVIII (Takahashi et al., 2003), which is a very well known component of basement membranes (Pehrsson et al., 2019). Endostatin is not expressed in neuroectodermal cells during development, (Miosge et al., 2003), however, adult neurons containing endostatin have been found in physiological conditions in human cortex (Deininger et al., 2002). Likewise, increased neuronal deposition of endostatin is found during Alzheimer’s disease (Deininger et al., 2002) and cerebral ischemia (Hou et al., 2010). To date, no studies about intraneuronal endostatin in the Fig. 2. Western blot analysis of mouse retinal endostatin. Retinal endostatin retina have been reported. shows a main band between 15 and 20 kD (lane 1), very close to the 21 kDa Proliferative diabetic retinopathy is still a leading cause of visual band from the recombinant endostatin (lane 2), confirming the antibody impairment in the working-age population in developed countries specificity for cleaved endostatin. α-tubulin was used as a loading control. (Nawaz et al., 2019). It is characterized by proliferation of fibrovascular tissue formed by the extension of retinal angiogenesis into the vitreous integrin (Wickström et al., 2002) and the inhibition of MMP-2 activity cavity, which ultimately leads to vitreous traction and retinal detach­ (Kim et al., 2000). MMP-2 regulates angiogenesis by degrading colla­ ment (Boulton et al., 1988). Prevailing evidence suggests that changes in gens, laminins, and glycosaminoglycans in the blood basement mem­ the relative balance between inducers and inhibitors of angiogenesis brane allowing the sprout of new vessels (Waszczykowska et al., 2020). may activate the angiogenic switch in the diabetic retina (Simó et al., In mouse and human neuroretinas, endostatin has only been 2006). The inhibitor of angiogenesis endostatin has been involved in the immunolocalized at the internal limiting membrane (Takahashi et al., regulation of neovascularization during diabetic retinopathy. The con­ 2003; Bhutto et al., 2004; Ohlmann et al., 2004; Ramesh et al., 2004; centration of endostatin in the vitreous fluid of patients with diabetic May et al., 2006; Määttä et al., 2007) and at the wall of retinal vessels, retinopathy significantly correlates with the severity of the disease. It Fig. 3. Endostatin in mouse neuroretina. (A) Immunostaining with anti-endostatin antibody revealed that endostatin is accumulated in the inner part of neuroretina, however, it was not observable at the optic disc (B) No endostatin signal was detected in blood vessel basement membranes or in the internal limiting membrane. Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer; BBM: blood vessel basement membrane; ILM: internal limiting membrane. Scale bars: 43.48 μm (A); 33.33 μm (B). 3 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 4. Endostatin is accumulated in bipolar cells of mouse retina. Double immunostaining with anti-PKCα antibody confirmed that endostatin was accumulated mainly in bipolar cells. Endostatin was found in their axon terminals (arrowheads), but not in their axons themselves (arrows) (A) and in the perikaryon and dendrites (arrowheads) (B). Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; IPL: inner plexiform layer; INL: inner nuclear layer; OPL: outer plexiform layer; ONL: outer nuclear layer. Scale bar: 25.42 μm; 9.41 μm (A); 10.17 μm (B). has been shown that eyes with low levels of endostatin and high levels of represent persistent hyaloid vessels. VEGF in the vitreous have a significantly greater risk of progression to the proliferative phase of diabetic retinopathy (Noma et al., 2002; 2. Materials and methods Funatsu et al., 2003). Furthermore, it has been observed that endostatin in age-related macular degeneration is highly decreased in angiogenic 2.1. Human eyes areas of the choroid (Bhutto et al., 2004). In this study, our results revealed for the first time the presence of Human eyes were obtained from voluntary donations to the Faculty intracellular endostatin in neurons of mouse and human retinas. Endo­ of Medicine of the UAB, following the Catalan law (Decret 297/1997) statin distribution was homogeneous throughout the neuroretina except about human samples for research purposes. The procedure was for the optic disc where endostatin was undetectable. Diabetic mice approved by the Ethics Committee in Animal and Human Experimen­ showed a reduction of endostatin in their neuroretinas that was asso­ tation of the UAB and carried out in accordance with The Code of Ethics ciated with the appearance of intravitreal vessels protruding from the of the World Medical Association (Declaration of Helsinki) for experi­ optic disc. Intravitreal vessels phenotype showed similarities to the ments involving humans. Two eyes from two different donors were used human diabetic papillovitreal neovessels, suggesting that neo­ in this study. Only biological data (sex, age, brief clinical history and vascularization may occur at the optic disc in diabetic mice. However, it cause of death) were provided. None of them suffered from diabetis cannot be ruled out that the intravitreal vessels in db/db mice may mellitus. Eyes were fixed in a solution of paraformaldehyde 4% with 4 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 5. Endostatin is present in cone photoreceptors of mice retina. Double immunostaining with anti-cone arrestin antibody showed endostatin in the inner seg­ ments of cone photoreceptors (arrows). Cone synaptic terminals did not show endostatin expression (arrowheads). Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; INL: inner nuclear layer; OPL: outer plexiform layer; ONL: outer nuclear layer; IS: inner segments of photoreceptors. Scale bar: 25.88 μm; 17.25 μm (inset). picric acid in 0.01 M PBS for 24 h. After several washes in PBS, the eyes develop obesity and hyperglycemia at 4–8 weeks of age, becoming a were partially dehydrated, and maintained in a solution of 70% alcohol widely accepted model of diabetes type II (Robinson et al., 2012). In this at 4 ◦ C. For immunohistochemical analysis, 3 μm paraffin retinal sec­ regard, significant statistical differences were found in 8-week-old mice tions were used. body weight (25.533 ± 0.383 g in db/+ vs. 36.067 ± 1.144 g in db/db; p < 0.0001; n = 6) and glycemia (150.167 ± 9.026 mg/dl in db/+ vs. 2.2. Mice 436.167 ± 14.597 mg/dl in db/db; p < 0.0001; n = 6) and also in 12-week-old mice body weight (25.65 ± 0.6462 g in db/+ vs. 43.73 ± A total of 14 male diabetic db/db (BKS.Cg-Dock7m +/+ Leprdb/J) 0.5474 g in db/db; p < 0.0001; n = 8) and glycemia (138.5 ± 3.674 and 14 non-diabetic db/+ (BKS.Cg-Dock7m + Leprdb/+) control mice mg/dl in db/+ vs. 563.0 ± 19.80 mg/dl in db/db; p < 0.0001; n = 8). aged 8 and 12 weeks were used in this study. Mice were maintained These data confirmed that db/db mice used during this study were under controlled conditions (20 ◦ C temperature, 60% humidity and 12 h diabetic. light/dark cycles) and provided with free access to water and food (2018S TEKLAD Global, Harlan Teklad, Madison, WI, USA). 2.4. Immunohistochemistry All mice were euthanized with an overdose of isoflurane inhalation followed by cervical dislocation. Animal care and experimental pro­ Human and mouse eyes embedded in paraffin were sectioned (3 μm) cedures were permormed according to the ARVO Statement for the Use along the eye axis, deparaffinized and rehydrated. For antigen retrieval, of Animals in Ohthalmic and Vision Research and Animal care ARRIVE a solution of citrate was used to unmask the epitopes of the retinal guidelines and were approved by the Ethics Committee in Animal and sections. After several washes in phosphate buffered saline (PBS), retinal Human Experimentation of the Universitat Autònoma de Barcelona sections were incubated overnight at 4 ◦ C with the following primary (UAB) (FUE-2018-00717286). antibodies: rabbit anti-endostatin (Invitrogen, Carlsbad, CA, USA, PA1- 601) at 1:100 dilution, obtained by immunizing a rabbit with two 2.3. Glycemia measurement peptides corresponding to amino acids 129–142 from mouse and human endostatin; goat anti-mouse collagen IV (Merck Millipore, Billerica, MA, Blood was collected from the tail vein every week to monitor gly­ USA) at 1:20 dilution; mouse anti-mouse PKCα (Sigma-Aldrich, St Louis, cemia using a glucometer (Bayer, Leverkusen, Germany). Mice were MO, USA) at 1:100 dilution; goat anti-mouse cone arrestin (Santa Cruz analyzed in non-fasted conditions and were considered diabetic when Biotechnology, Heidelberg, Germany) at 1:100 dilution; rabbit anti- glycemia exceeded 250 mg/dl in two consecutive measurements. The mouse GFAP (DAKO, Glostrup, Denmark) at 1:1000 dilution; rabbit last measurement was performed before euthanasia. db/db mice anti-mouse MMP-2 (Abcam, Cambridge, UK) at 1:500 dilution; rabbit 5 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 6. Human optic disc was practically devoid of endostatin. Double immunostainig with anti- endostatin and anti-collagen IV antibodies showed a markedly endostatin accumulation in the neuroretina, whereas as happens in mouse retina, endostatin was unobservable in the optic disc. The figure is a mosaic composite obtained from different captures at 63X in the laser scan­ ning confocal microscope. OD: Optic disc; ON: Optic nerve; CR: Central retinal artery and vein; R: Retina. Scale bar: 474.92 μm; 226.15 μm (inset). anti-mouse MMP-9 (Proteintech, Rosemont, IL, USA) at 1:100 dilution. blocking, paraffin sections were incubated with Mouse Seroblock FcR All primary antibody solutions were diluted with donkey serum (Sigma- (Bio-Rad Laboratories, Hercules, CA, USA) at 1:100 dilution during 1 h Aldrich) at 1:10 dilution as a blocking agent to prevent secondary before the start of the immunohistochemistry process in order to prevent antibody unspecific binding. After washing the sections in PBS, they non-specific binding of the Fc fragment from the Endostatin’s IgG to the were incubated 2 h at room temperature with specific fluorochrome- mouse Fc receptors FcRγIII and FcR&gamma:II. For isotype control conjugated secondary antibodies diluted at 1:100: donkey anti-goat assay, we used an antibody matching the host and IgG isotype of anti- Alexa 488 (Life Technology, Carlsbad, Ca, USA); goat anti-rabbit 488, endostatin antibody that targets an antigen absent in the retina. Anti- rabbit anti-goat Alexa 568, and goat anti-rabbit 568 (Invitrogen). endostatin antibody’s host is rabbit, which has only one subclass of Hoechst (Sigma-Aldrich) diluted in PBS (1∶100 dilution) was used for IgG isotype (Weber et al., 2017). Thus, we used a rabbit antibody against nuclear counterstaining. Slides were mounted in Fluoromount (Sigma- collagen I (abcam), which is highly expressed in cornea and absent in the Aldrich) medium for further analysis using a laser scanning confocal retina (Ruberte et al., 2017).To assess retinal morphology and location microscope (TCS SP2 or TCS SP5; Leica Microsystems GmbH, Heidel­ of intravitreal vessels, eye paraffin sections were stained with Hema­ berg, Germany). Immunohistochemistry negative control was done by toxylin/Eosin and were observed with a light microscope (Nikon Eclipse omission of the primary antibody in a sequential tissue section. To verify E800). anti-endostatin antibody specificity, blocking peptide assay, FcR blocking, and isotype control assay were performed on eye paraffined sections. For blocking peptide assay, the neutralization of the primary 2.5. Western blot analysis anti-endostatin antibody was done in a previous incubation step of 1 h before immunohistochemistry by adding 10-times excess of recombi­ For protein extraction, mouse eyes were enucleated and each retina nant mouse endostatin (R&D Systems, MN, Minneapolis), in contrast to was immediately dissected in PBS and homogenized in RIPA lysis buffer the control incubated with the anti-endostatin antibody alone. For FcR (25 mM Tris base, pH 8.2, 150 mM NaCl, 0.5% NP-40, 0.1% SDS, 0.5% sodium deoxycholate), containing protease inhibitor cocktail tablets 6 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 7. Comparison between human and mouse endostatin distribution in neuro­ retina. Immunolabelling with anti- endostatin antibody in human and mouse retinas showed a similar, although with some differences, endostatin distribution in both species, mainly located at bipolar cells and photoreceptors. (A) Human bipolar ter­ minal axons (arrows) showed lower expres­ sion of endostatin compared to mouse retina. (B) Bipolar cell perikaryal were stained with endostatin in both human and mice retina with a similar intensity. (C) In contrast with mice, where only scarce cones (arrowheads) showed endostatin accumula­ tion in their inner segments, human retina showed a higher signal of endostatin at photoreceptors, suggesting that both cone and photoreceptors accumulated endostatin. The same anti-endostatin antibody was used on both human and mice retinal sections. GCL: ganglion cell layer; IPL: inner plexi­ form layer; INL, inner nuclear layer; OPL: outer plexiform layer; ONL: outer nuclear layer; IS: inner segments of photoreceptors. Scale bars: 38.3 μm; 10.64 μm (A, B, C). (Complete Mini, Roche, Basel, Switzerland). After vortex, retinas were 0.05% Twen 20) for 1 h at room temperature and incubated with rabbit centrifuged at 8000 g for 10 min. BCA assay reagent (Sigma-Aldrich) anti-mouse endostatin antibody (Invitrogen) at 1:100 dilution overnight was used for each sample to determine total protein concentration. at 4 ◦ C. After washing in TBST, the membrane was incubated with For immunoblotting, 50 mg of protein of each sample were indi­ secondary antibody horseradish peroxidase-conjugated goat anti-rabbit vidually resuspended in 50% sample loading buffer (140 mM Tris base, (Bionova Scientific, Fremont, CA, USA) at 1∶5000 dilution for 45 min at pH 6.7, 6.8% SDS, 33% glycerol, 0.004% bromophenol) and 5% room temperature. Both primary and secondary antibodies were also β-mercaptoethanol (Sigma-Aldrich) and heated at 95 ◦ C for 5 min. Af­ diluted with 2% BSA in a 0.05% TBST. terwards, 12 μl of each sample were loaded in a 12% pre-cast SDS-PAGE Following antibodies incubation, the reagent Luminata Crescendo gel (Bio-Rad) for electrophoresis followed by protein transfer to a pol­ (Merck Millipore) was used for chemiluminescent detection. The bands yvinylidene difluoride (PVDF) membrane (Merck Millipore). PVDF revealed with Fluor-S Max imager (Bio-Rad) were analyzed with Image J membranes were blocked with 2% BSA in TBST (Tris buffered saline, software (National Institutes of Health, Bethesda, MD, USA) and rabbit 7 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 to visualize retinal blood vessels. 10 μl of 5% sodium fluorescein (Sigma- Aldrich) were injected subcutaneously in each mouse and angiography was performed with HRA2 Imaging Instrument (Heidelberg Engineer­ ing, Heidelberg, Germany). Angiography was performed in every mouse retina in both eyes and all vascular plexi were analized, paying especial attention to the vessels that protruded into the vitreous from the optic disc. 2.7. Vessel density quantification Vascular density of the entire retina and the central area was quan­ tified using Angiotool software (Zudaire et al., 2011) in both groups. The results were represented in % of vessels area. 2.8. Statistical analyses Statistical analyses were performed using R Statistical Software (version 4.1.0; R Foundation for Statistical Computing, Vienna, Austria). Shapiro-Wilk test and F test were performed for the evaluation of data distribution and variance homogeneity, respectively. For normal dis­ tribution and homogenous variances, the parametric method unpaired t- test was used. For data not normally distributed the Wilcoxon test was performed. Data are expressed as mean ± standard error of the mean (SEM). Difference between groups was considered significant at P- value<0.05. 3. Results 3.1. Endostatin is expressed in mouse and human retinal neurons Cross-reactivity of anti-endostatin antibodies with collagen XVIII, the endostatin precursor, is one of the challenges to immunolocalize cleaved active endostatin, since antibodies used to date in retinal studies also recognize uncleaved endostatin and mark almost exclusively blood basement membranes and the internal limiting membrane (Pehrsson et al., 2019). To try to solve this situation we used a polyclonal anti-endostatin antibody which previously had demonstrated its ability to detect endostatin intracellularly (Martín-Granado et al., 2017; Hua­ Fig. 8. Endostatin is decreased in db/db mouse retinas. (A) Western blot man et al., 2019). To confirm antibody specificity for cleaved endo­ analysis revealed that endostatin protein expression was 0.5-fold decreased in statin, several assays were performed in retinal sections. First, a blocking 12 weeks-old diabetic mouse retinas, especially in the ganglion cell layer. peptide assay was used to neutralize the primary anti-endostatin anti­ Statistical analysis showed significant differences in endostatin band intensity body with recombinant mouse endostatin (Fig. 1A). In addition to this, between db/+ and db/db mice (1.275 ± 0.126 arbitrary units in db/+ mice vs. blocking Fc assay (Fig. 1B) that specifically blocks mouse Fc receptors in 0.753 ± 0.055 arbitrary units in db/db mice; p = 0.019; n = 3). α-tubulin was the retina, and isotype control assay (Fig. 1C) that guarantees used as a loading control (B) Immunofluorescence representative images non-specific binding of similar IgG isotypes, were performed. Further­ showing decreased endostatin expression in db/db retinas. Nuclei were coun­ terstained with Hoechst. OD: Optic disc; CR: Central retinal vessel; R: Retina. more, Western blot analyses showed that bands of retinal endostatin and Scale bar: 22.22 μm. recombinant endostatin were comparable and around 20 kDa (Fig. 2), the stablished molecular weight of cleaved endostatin (Marneros et al., 2004; Cristante et al., 2018). Lane 1 shows a different number of bands anti-mouse α-tubulin (Abcam) at 1:500000 dilution was used as a compared to lane 2 (Fig. 2), probably due to the fact that lane 1 corre­ loading control to normalize the levels of the protein of interest. Re­ sponds to native endostatin from the retinal tissue and lane 2 corre­ combinant mouse endostatin was also run in the same gel to compare the sponds to recombinant endostatin. Recombinant endostatin is a purified molecular wheight of retinal endostatin and the isolated peptid. A total peptide that might have not suffered the same post-translational modi­ of 3 repetitions were performed in each Western blot. fications. In addition to this, mouse endostatin peptides can have de­ letions or extensions of few residues at the N or C-terminal (Yamaguchi 2.6. Scanning laser ophthalmoscopy (SLO) et al., 1999) that could explain why native endostatin shows more bands and a slightly different weigh compared to the recombinant endostatin. Mice were anaesthetized using a ketamine (12 mg/kg) and xylazine In mouse retinas, endostatin distribution was similar throughout the (100 mg/kg) cocktail diluted in physiological saline solution and central and peripheral retina, but at the level of the optic disc endostatin injected intraperitoneally. Toe pinch was performed to assure a correct was unobservable (Fig. 3A). Endostatin signal was found inside retinal anesthetic status and respiratory function was monitored during the neurons, without marking neither the internal limiting membrane nor process. Once mice were anaesthetized, eye drops of anesthetic collyr­ the blood vessel basement membranes (Fig. 3B), which confirmed that ium and tropicamide (Alcon, Barcelona, Spain) were administered to the antibody was only detecting cleaved endostatin and not collagen induce mydriasis. To prevent the eye from drying and developing cat­ XVIII. aracts, Viscofresh (Allergan, Dublin, Ireland) was continuously applied Given that the distribution of the endostatin signal was compatible on both eyes and custom-made mouse lenses were placed on the corneas with the morphology of bipolar cells (Fig. 4), dual immunostaining with 8 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 9. Vascular density in central retina was increased in db/db mice. In vivo fundus fluorescein angiography was performed to assess retinal vascularization. (A) No significant differences in vascular density were found when comparing the entire retinal vascular network (34.82 ± 2.32% vessel area in db/+ retinas vs. 33.35 ± 1.56% vessel area in db/db retinas; p = 0.652; n = 5) in 12 weeks-old mice. (B) However, a statistically significant increase in vascular density was found in the central part of the retina of db/db mice (52.47 ± 1.12% vessel area in db/+ retinas vs. 64.87 ± 13.39% vessel area in db/db retinas; p = 0.046; n = 5) in 12 weeks- old mice. anti-PKCα antibody, a specific marker of rod bipolar cells (Haverkamp human retina (Fig. 7C). et al., 2003), was performed. The result showed that endostatin was accumulated in bipolar cell axon terminals but not in the axons them­ 3.2. Endostatin is decreased in db/db mouse retinas selves (Fig. 4A). Dendrites, including their terminals, and the perikaryon of bipolar cells also accumulated endostatin (Fig. 4B). Although endo­ Once established the presence of endostatin in mouse and human statin was mainly located at the internal neuroretina, sporadic photo­ retinal neurons and given that eye concentrations of endostatin in pa­ receptors also showed intracellular endostatin (Fig. 5). To identify what tients with diabetic retinopathy significantly correlates with the severity type of photoreceptor accumulates endostatin, cone arrestin, a specific of the disease (Funatsu et al., 2003), we next investigated the concen­ marker of cone photoreceptors (Zhu et al., 2002), was employed. Dual tration of endostatin in the retina during experimental hyperglycemic immunostaining with anti-cone arrestin antibody showed that endo­ conditions. Western blot analysis of db/db mice, a well stablished model statin was accumulated in the inner segments of cone photoreceptors. of type 2 diabetes (Robinson et al., 2012; Bogdanov et al., 2014), showed The cone synaptic terminals did not accumulate endostatin (Fig. 5). a statistically significant decrease (0.5-fold change) of endostatin in the Interestingly, the distribution pattern of endostatin in human retina neuroretina of diabetic mice (1.275 ± 0.126 arbitrary units in db/+ was similar to the mouse retina. Endostatin was also undetectable in the mice vs. 0.753 ± 0.055 arbitrary units in db/db mice; p = 0.019; n = 3) human optic disc (Fig. 6) and was also accumulated in bipolar cells and (Fig. 8A). In addition, in some db/db mice endostatin decrease was photoreceptors (Fig. 7). However, some minor differences were found, exacerbated in ganglion cell layer (Fig. 8B). namely a lower accumulation of endostatin in the bipolar cell axons terminals (Fig. 7A), although bipolar perikaryal were stained with 3.3. Optic disc intravitreal vascularization in db/db mouse retinas similar intensity (Fig. 7B), and a higher number of photoreceptors with intracellular endostatin (Fig. 7C). The morphology of these photore­ Given that endostatin is an inhibitor of angiogenesis preventing ceptors suggested that both cone and rods accumulated endostatin in the endothelial cell proliferation, migration, and invasion (Kim et al., 2002), 9 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 10. Intravitreal vessel angiography in db/db mouse retinas. After fluorescein injection, three consecutives focal planes of the eye fundus (A, B, and C) were performed to assess intravitreal vessels. 50% of db/db mice presented intravitreal vessels while only one db/+ presented a single tiny intravitreal vessel (n = 8). In the most part of db/db mice several vessels, usually three, invaded the vitreous from the optic disc. we next analyzed whether the decrease of endostatin in the retina of area in db/+ retinas vs. 33.35 ± 1.56% vessel area in db/db retinas; p = diabetic mice led to neovascularization. Retinal blood vessels were 0.652; n = 5) (Fig. 9A). This agrees with most of the data provided by the investigated using in vivo eye fundus fluorescein angiography (Fig. 9) bibliography, which indicate that neovascularization is not a frequent and the AngioTool software (Zudaire et al., 2011) was used to quantify finding in db/db mouse retina (Li et al., 2010; Bogdanov et al., 2014; vascular density. The results showed not significant differences in Kim et al., 2014). To our knowledge, neovascularization in db/db mice vascular density between 12-week-old db/db and db/+ mice, when the retina has only been reported in a single study using long-standing entire retinal vascular network was compared (34.82 ± 2.32% vessel 15-month-old mice (Cheung et al., 2005). However, when the vascular 10 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 3.4. Optic disc intravitreal vessel phenotype in db/db mice Once established that 50% of 12-week-old db/db mice studied had vessels that from the optic disc penetrated the vitreous, we tried to answer the question about the nature of these vessels. Basically, two possibilities existed: (a) they were persistent hyaloid vessels or (b) they were neovessels triggered by the pro-angiogenic environment induced by the decrease of endostatin in the db/db mouse retina. During development, the intraocular structures are nourished by the hyaloid vascular system (Saint-Geneiz and D’Amore, 2004). This vascular system in mouse starts to form at embryonic day 10.5 and be­ comes fully developed by embryonic day 13.5 (Smith et al., 2002). It is widely accepted that the hyaloid vascular system in mice initiates its regression after birth and totally disappears around postnatal days 14–24 (Ito et al., 1999; Fukai et al., 2002; Ohlmann et al., 2004; Ritter et al., 2005). In a previous study from our group, 15% of 25-week-old wild type C57BL/6J mice showed functional persistent hyaloid vessels (McLenachan et al., 2015), which matches with the presence of a small intravitreal vessel found in one db/+ mice (representing the 16% of the db/+ analyzed). However, 50% of db/db retinas presented intravitreal vessels, which largely exceeded the normal percentage of expected persistent hyaloid vessels in a non-diabetic mouse. In this regard, mice lacking collagen XVIII/endostatin showed ocular abnormalities including a delay in normal postnatal regression of hyaloid vessels (Fukai et al., 2002). Similarly, hyaloid vessels also persisted in collagen XVIII human mutations (Knobloch syndrome) (Sertié et al., 2000). Although these arguments could suggest that intravitreal vessels found in db/db mouse retinas were persistent hyaloid vessels, delayed regression of hyaloid vessels in collagen XVIIII/endostatin null mice leads to abnormal disorganized vascularization of the entire retina (Fukai et al., 2002), situation that we never found in db/db mouse ret­ inas (Fig. 8). Furthermore, the morphological aspect of intravitreal vessels in db/db mice was similar to human papillovitreal neovessels during diabetic retinopathy (L’Esperance, 1998). Thus, to discriminate whether intravitreal vessels found in db/db mice were persistent hyaloid vessels or neovessels, we analyzed them by immunofluorescence in 8-week-old db/db mice. Intravitreal db/db vessels showed an incom­ plete lining of GFAP positive signal (Fig. 12), what would indicate that db/db intravitreal vessels were partially surrounded by neuroglial cells. Fig. 11. Intravitreal vessel histology in db/db mouse retinas. Paraffin sections Similarly, it is well stated that preretinal neovessels expressed GFAP in of the retina sectioned along the eye axis were counterstained with Hematox­ their walls during proliferative diabetic human retinopathy (Smith et al., ylin/Eosin. A significant increase of intravitreal vessels (arrows) in db/db mice 1999). In contrast, published data about the presence of neuroglia sur­ was observed (1.143 ± 0.261 number of intravitreal vessels in db/+ mice (n = rounding persistent hyaloid vessels is contradictory. In Nuc1 sponta­ 7) vs. 2.200 ± 0.374 number of intravitreal vessel in db/db mice (n = 5); p = neous rat and in humans with persistent fetal vasculature (PFV) it has 0.037). OD: Optic disc; R: Retina. Scale bar: 76.45 μm (A); 12.23 μm (B). been described that astrocytes sheath the persistent hyaloid artery (Zhang et al., 2005). This abnormal association of astrocytes with the density was compared in the central part of the retina, a significant slight hyaloid artery may impede the normal macrophage-mediated remod­ vascular density increase was observed in diabetic mice (52.47 ± 1.12% elling and regression of the hyaloid system (Zhang et al., 2011). How­ vessel area in db/+ retinas vs. 64.87 ± 13.39% vessel area in db/db ever, in collagen XVIII/endostatin null mice persistent hyaloid vessels retinas; p = 0.046; n = 5) (Fig. 9B). are devoid of neuroglial coverage (Hurskainen et al., 2005). To confirm whether this increased vascular density in the central Intravitreal db/db vessels showed a thickening of their basement part of the retina was compatible with optic disc vessels protruding into membrane due to increased collagen IV deposition (Fig. 13A), which is the vitreous, we acquired three different focal planes of the eye fundus the main component of the retinal vessel basement membrane (Inoue, using SLO (Fig. 10 A, B, C). Using this approach, optic disc vessels 1989), compared to the intraretinal vessels (Fig, 13B). Basement mem­ protruding the vitreous can be easily identified (Fig. 10). The results brane thickening and increased collagen IV deposition are hallmarks of obtained showed that 50% of the db/db mice (n = 8) presented vessels diabetic retinopathy (Roy et al., 2010). However, this is not a differen­ that originated from the optic disc and invaded the vitreous. Further­ tial clue used to distinguish between persistent hyaloid vessels and more, in comparison with db/+ mice, where only one animal presented intravitreal neovessels, because degenerating hyaloid vessels (Kishimoto a single small vessel contacting the vitreous, in most db/db mice several et al., 2018) and persistent hyaloid vessels (Hurskainen et al., 2005) vessels, usually three, invaded the vitreous (Fig. 10). This result corre­ present also increased collagen IV deposition in their thicker basement lated with histological analysis in paraffin eyes sections studied at the membranes. level of the optic disc, where the number of recognizable intravitreal Degradation of extracellular matrix components is thought to be blood vessels identified by Hematoxylin/Eosin staining was significantly essential for the development of neovessels in proliferative diabetic increased in db/db mice (1.143 ± 0.261 number of intravitreal vessels in retinopathy. Several species of MMPs are capable of degrading various db/+ mice (n = 7) vs. 2.200 ± 0.374 number of intravitreal vessel in db/ extracellular matrix molecules and have been detected in the fibrovas­ db mice (n = 5); p = 0.037) (Fig. 11). cular tissues in eyes with diabetic retinopathy (Roy et al., 2010). 11 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 12. GFAP expression surrounding intravitreal vessels in db/db retinas. Double immunofluorescence using collagen IV and GFAP showed that intravitreal vessels in 8 weeks db/db mice were covered partially by neuroglia. Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer. Scale bar: 23.82 μm; 12.54 μm (inset). Specifically, MMP-2 and -9 have been immunohistochemically detected in the wall of intravitreal human neovessels (Noda et al., 2003). In the same way, intravitreal db/db vessels showed MMP-2 (Fig. 14) and MMP-9 (Fig. 15) co-localizing with collagen IV, a substrate of these MMPs (Somerville et al., 2003). Although finding both basement membrane thickening and degradation by MMPs in intravitreal db/db vessels may seem counterintuitive, it may be explained as a compensa­ tory response to increased synthesis of basement membrane components during diabetic retinopathy (Roy et al., 2010). 4. Discussion Endostatin, the carboxyl-terminal fragment of type XVIII collagen, is a natural inhibitor of angiogenesis and its effect is being tested and proven effective in several clinical eye studies. RetinoStat, an equine infectious anemia virus-based lentiviral gene therapy vector that ex­ presses endostatin and angiostatin, has been demonstrated to have ac­ tivity against choroidal neovascularization and it was well tolerated and Fig. 13. Thickening of intravitreal vessel basement membrane in db/db mice. capable of persistent expression after subretinal delivery (Kachi et al., Immunodetection of collagen IV in 8 weeks-old db/db mice retinas showed a thickening in the basement membrane of intravitreal vessels (A) compared to 2009). Furthermore, Tat PTD-Endostatin-RGD, a novel fusion protein, the intraretinal vessels (B). Nuclei were counterstained with Hoechst. GCL: inhibited abnormal angiogenesis in retina via eye drops (Li et al., 2016). ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer. Scale However, despite its therapeutic use, comprehensive research about the bar: 35.74 μm; 8.69 μm (A); 7,35 μm (B). role of endostatin during diabetic retinopathy is scarce (Noma et al., 12 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 14. MMP-2 expression in the basement membrane of intravitreal vessels of db/db mice. Immunofluorescence analysis showed that collagen IV and MMP-2 co- localized in the basement membrane of 8-week-old db/db mice intravitreal vessels. Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer. Scale bar: 18.49 μm; 9.24 μm (inset). 2002; Funatsu et al., 2003). Exosomes are nanometer-sized vesicles that cells can release in a This study revealed for the first time the presence of intracellular controlled manner to mediate a plethora of cellular activities, including endostatin in mouse and human retinal neurons. The observations made neovascularization (Klingeborn et al., 2017). Very little research on the showed that endostatin was accumulated in the dendrites and axons of role of exosomes in diabetic proliferative retinopathy has been done to bipolar cells and in photoreceptors, mainly in their inner segments. The date, although it is known that exosomes containing endostatin suppress question about whether endostatin has intracellular or extracellular choroidal neovascularization (Hajrasouliha et al., 2013). Investigating location is an important variable when considering likelihood of sub­ the role of endostatin containing exosomes from retinal neurons may strate/enzyme interaction. To date, endostatin in neuroretina has been identify antiangiogenic based mechanisms, which could help to develop only found extracellularly in basement membranes (Takahashi et al., novel therapeutic targets to control aberrant retinal neovascularization. 2003; Bhutto et al., 2004; Ohlmann et al., 2004; Ramesh et al., 2004; Diabetic db/db mice showed a reduction of endostatin in their ret­ May et al., 2006; Määttä et al., 2007), where collagen XVIII may be inas. Furthermore, in some db/db mice endostatin was especially cleaved by MMPs, elastases, and other extracellular proteases (Ferreras decreased in the ganglion cell layer. However, this result is not et al., 2000). However, similarly as what occurs in our study, healthy conclusive and further studies will be necessary to clarify to what extend neurons containing intracellular endostatin have been found in the endostatin decrease is different among retinal layers in db/db mice. human cortex (Deininger et al., 2002). An explanation for this intra­ Endostatin reduction was associated with the appearance of vessels that cytoplasmic localization of endostatin could be cathepsin B, a lysosomal originated from the optic disc and penetrated to the adjacent vitreous in protease, which can generate endostatin from collagen XVIII (Im et al., the 50% of db/db mice. Whether these intravitreal vessels found in db/ 2005). Cathepsin B has been found to colocalize intracellularly with db mice were persistent hyaloid vessels at 84 postnatal day (12-week- endostatin in neurons. Although intracellular localization of endostatin old), when in fact they should have disappeared completely at 14–24 has been correlated with cell apoptosis (Hou et al., 2010), other studies postnatal days (Ito et al., 1999; Fukai et al., 2002; Ohlmann et al., 2004; in isolated retinal astrocytes suggested that intracellular endostatin is Ritter et al., 2005), or neovessels induced by the pro-angiogenic envi­ secreted extracellularly by exosomes (Hajrasouliha et al., 2013). ronment developed by the decrease of endostatin, is a question that our 13 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Fig. 15. MMP-9 expression in the basement membrane of intravitreal vessels of db/db mice. Immunofluorescence analysis showed that collagen IV and MMP-9 co- localized in the basement membrane of 8-week-old db/db mice intravitreal vessels. Nuclei were counterstained with Hoechst. GCL: ganglion cell layer; INL: inner nuclear layer; ONL: outer nuclear layer. Scale bar: 17.82 μm 8.91 μm (inset). study has not been fully able to answer. 5. Conclusions The intravitreal vessel phenotype in db/db mice revealed features such as GFAP positive sheath of neuroglia and a thicker basement For the first time, this study revealed the presence of endostatin in membrane, which are hallmarks of neovessels in diabetic retinopathy bipolar cells and photoreceptors, in both mouse and human retinas. (Smith et al., 1999; Roy et al., 2010). However, they also appear in Endostatin expression was inappreciable at the level of the optic disc, persistent hyaloid vessels (Hurskainen et al., 2005; Zhang et al., 2005). which could explain why this location is susceptible to neo­ In contrast, the presence of MMP-2 and MMP-9 at the basement mem­ vascularization in diabetic retinopathy. Furthermore, endostatin was brane in db/db intravitreal vessels, which is characteristic of human decreased in diabetic mouse retina. Diabetic db/db mice also presented diabetic neovessels (Noda et al., 2003), has not been described, to our significantly increased number of intravitreal vessels protruding from knowledge, in persistent hyaloid vessels. Since endostatin inhibits the optic disc. These intravitreal vessels showed a GFAP positive MMP-2 (Kim et al., 2000), its decrease in db/db mice retinas could neuroglia sheath, basement membrane thickening, and expressed met­ explain the presence of metalloproteinase in the wall of intravitreal alloproteinases in the vascular wall, suggesting that they could be neo­ vessels. All together, these findings could suggest that neo­ vessels similar to the papillovitreal neovessels observed during human vascularization may occur at the optic disc in hyperglycemic db/db diabetic retinopathy. However, we canot ruled out the possibility that mice. However, it cannot be ruled out that these intravitreal vessels may hese vessels were persistent hyaloid vessels triggered by the decrease of also represent non-degenerated hyaloid vessels. retinal endostatin in diabetic mice. Finally, the human optic disc represents a preeminent site for the appearance of neovessels during proliferative diabetic retinopathy. Two Funding situations occur during diabetic neovascularization: appearance of new blood vessels on the optic disc (NVD), which frequently lead to visual This work was supported by the Instituto de Salud Carlos III, Min­ loss, and appearance of new blood vessels elsewhere on the retina isterio de Ciencia e Innovación, Spain (grant number PI16/00719); the (NVE), which do not pose such a serious threat to vision (Valsania et al., Fundação para a Ciência e a Tecnologia, Ministerio da Educação e 1993). The almost null presence of anti-angiogenic endostatin in the Ciência, Portugal (grant numbers SFRH/BD/95330/2013 and SFRH/ human optic disc, could explain why the optic disc is prone to neo­ BPD/102573/2014); and Fondo Europeo de Desarrollo Regional vascularization during human diabetic retinopathy. (FEDER). 14 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 Declaration of competing interest codelivery of endostatin and angiostatin driven by retinal pigmented epithelium- specific VMD2 promoter inhibits choroidal neovascularization. Hum. Gene Ther. 20 (1), 31–39. https://doi.org/10.1089/hum.2008.046. A. Bonet, None; A. Valença, None; L. Mendes-Jorge, None; A. Case­ Kim, J., Kim, C.S., Lee, I.S., Lee, Y.M., Sohn, E., Jo, K., Kim, J.H., Kim, J.S., 2014. Extract llas, None; A. Rodríguez-Baeza, None; V. Nacher, None; D. Ramos, None; of Litsea japonica ameliorates blood-retinal barrier breakdown in db/db mice. J. Pampalona, None; R. Simó, None; Ruberte, None. Endocrine 46 (3), 462–469. https://doi.org/10.1007/s12020-013-0085-x. Kim, Y.M., Jang, J.W., Lee, O.H., Yeon, J., Choi, E.Y., Kim, K.W., Lee, S.T., Kwon, Y.G., 2000. Endostatin inhibits endothelial and tumor cellular invasion by blocking the Acknowledgements activation and catalytic activity of matrix metalloproteinase 2. Cancer Res. 60 (19), 5410–5413. Kim, Y.M., Hwang, S., Kim, Y.M., Pyun, B.J., Kim, T.Y., Lee, S.T., Gho, Y.S., Kwon, Y.G., The authors thank Lorena Noya, Verónica Melgarejo, and Ángel 2002. Endostatin blocks vascular endothelial growth factor-mediated signaling via Vázquez for their technical assistance. direct interaction with KDR/Flk-1. J. Biol. Chem. 277 (31), 27872–27879. https:// doi.org/10.1074/jbc.M202771200. Kishimoto, A., Kimura, S., Nio-Kobayashi, J., Takahashi-Iwanaga, H., Park, A.M., References Iwanaga, T., 2018. Histochemical characteristics of regressing vessels in the hyaloid vascular system of neonatal mice: novel implication for vascular atrophy. Exp. Eye Res. 172, 1–9. https://doi.org/10.1016/j.exer.2018.03.024. Bhutto, I.A., Kim, S.Y., McLeod, D.S., Merges, C., Fukai, N., Olsen, B.R., Lutty, G.A., 2004. Klingeborn, M., Dismuke, W.M., Bowes-Rickman, C., Stamer, W.D., 2017. Roles of Localization of collagen XVIII and the endostatin portion of collagen XVIII in aged exosomes in the normal and diseased eye. Prog. Retin. Eye Res. 59, 158–177. human control eyes and eyes with age-related macular degeneration. Invest https://doi.org/10.1016/j.preteyeres.2017.04.004. Ophtalmol Vis Sci 45 (5), 1544–1552. https://doi.org/10.1167/iovs.03-0862. L’Esperance, F.A., 1998. The Natural History and Classification of Diabetic Retinopathy. Bogdanov, P., Corraliza, L., Villena, J.A., Carvalho, A.R., Garcia-Arumí, J., Ramos, D., Diabetic Renal-Retinal Syndrome. Springer Netherlands, pp. 117–150. Ruberte, J., Simó, R., Hernández, C., 2014. The db/db mouse: a useful model for the Li, J., Wang, J.J., Yu, Q., Chen, K., Mahadev, K., Zhang, S.X., 2010. Inhibition of reactive study of diabetic retinal neurodegeneration. PLoS One 9 (5), e97302. https://doi. oxygen species by lovastatin downregulates vascular endothelial growth factor org/10.1371/journal.pone.0097302. expression and ameliorates blood-retinal barrier breakdown in db/db mice: role of Boulton, M.E., Mc Leod, D., Garner, A., 1988. Vasoproliferative retinopathies: clinical, NADPH oxidase 4. Diabetes 59 (6), 1528–1538. https://doi.org/10.2337/db09- morphogenetic and modulatory aspects. Eye 2, 124–139. https://doi.org/10.1038/ 1057. eye.1988.139. Li, Y., Li, L., Li, Z., Sheng, J., Zhang, X., Feng, D., Zhang, X., Yin, F., Wang, A., Wang, F., Cheung, A.K.H., Fung, M.K.L., Lo, A.C.Y., Lam, T.T.L., So, K.F., Chung, S.S.M., Chung, S. 2016. Tat PTD-Endostatin-RGD: a novel protein with anti-angiogenesis effect in K., 2005. Aldose reductase deficiency prevents diabetes-induced blood-retinal retina via eye drops. Biochim. Biophys. Acta 1860 (10), 2137–2147. https://doi.org/ barrier breakdown, apoptosis, and glial reactivation in the retina of db/db mice. 10.1016/j.bbagen.2016.05.031. Diabetes 54 (11), 3119–3125. https://doi.org/10.2337/diabetes.54.11.3119. Määttä, M., Heljasvaara, R., Pihlajaniemi, T., Uusitalo, M., 2007. Collagen XVIII/ Cristante, E., Liyanage, S.E., Sampson, R.D., Kalargyrou, A., De Rossi, G., Rizzi, M., endostatin shows a ubiquitous distribution in human ocular tissues and endostatin- Hoke, J., Ribeiro, J., Maswood, R.N., Duran, Y., Matsuki, T., Aghaizu, N.D., containing fragments accumulate in ocular fluid samples. Graefes Arch. Clin. Exp. Luhmann, U.G., Smith, A.J., Ali, R.R., Bainbridge, J.W.B., 2018. Late Ophthalmol. 245 (1), 74–81. https://doi.org/10.1007/s00417-006-0281-y. neuroprogenitors contribute to normal retinal vascular development in a Hif2a- May, C.A., Ohlmann, A.V., Hammes, H., Spandau, U.H.M., 2006. Proteins with an dependent manner. Development 145 (8), 157511. https://doi.org/10.1242/ endostatin-like domain in a mouse model of oxygen-induced retinopathy. Exp. Eye dev.157511. Res. 82 (2), 341–348. https://doi.org/10.1016/j.exer.2005.07.005. Deininger, M.H., Fimmen, B.A., Thal, D.R., Schluesener, H.J., Meyermann, R., 2002. Marneros, A.G., Keene, D.R., Hansen, U., Fukai, N., Moulton, K., Goletz, P.L., Aberrant neuronal and paracellular deposition of endostatin in brains of patients Moiseyev, G., Pawlyk, B.S., Halfter, W., Dong, S., Shibata, M., Li, T., Crouch, R.K., with Alzheimer’s disease. J. Neurosci. 22 (24), 10621–10626. https://doi.org/ Bruckner, P., Olsen, B.R., 2004. Collagen XVIII/endostatin is essential for vision and 10.1523/JNEUROSCI.22-24-10621.2002. retinal pigment epithelial function. EMBO J 23 (1), 89–99. https://doi.org/10.1038/ Ferreras, M., Felbor, U., Lenhard, T., Olsen, B.R., Delaissé, J.M., 2000. Generation and sj.emboj.7600014. degradation of human endostatin proteins by various proteinases. FEBS Lett. 486 (3), Martín-Granado, V., Ortiz-Rivero, S., Carmona, R., Guitiérrez-Herrero, S., Barrera, M., 247–251. https://doi.org/10.1016/s0014-5793(00)02249-3. San-Segundo, L., Sequera, C., Perdiguero, P., Lozano, F., Martín-Herrero, F., Funatsu, H., Yamashita, H., Noma, H., Mochizuki, H., Mimura, T., Ikeda, T., Hori, S., González-Porras JR., Muñoz-Chápuli, R., Porras, A., Guerrero, C., 2017. C3G 2003. Outcome of vitreous surgery and the balance between vascular endothelial promotes a selective release of angiogenic factors from activated mouse platelets to growth factor and endostatin. Invest. Ophthalmol. Vis. Sci. 44 (3), 1042–1047. regulate angiogenesis and tumor metastasis. Oncotarget 8 (67), 110994–111011. https://doi.org/10.1167/iovs.02-0374. https://doi.org/10.18632/oncotarget.22339. Fukai, N., Eklund, L., Marneros, A.G., Oh, S.P., Keene, D.R., Tamarkin, L., Niemelä, M., McLenachan, S., Magno, A.L., Ramos, D., Catita, J., McMenamin, P.G., Chen, F.K., Ilves, M., Pihlajaniemi, T., Olsen, B.R., 2002. Lack of collagen XVIII/endostatin Rakoczy, E.P., Ruberte, J., 2015. Angiography reveals novel features of the retinal results in eye abnormalities. EMBO J. 21 (7), 1535–1544. https://doi.org/10.1093/ vasculature in healthy and diabetic mice. Exp. Eye Res. 138, 6–21. https://doi.org/ emboj/21.7.1535. 10.1016/j.exer.2015.06.023. Hajrasouliha, A.R., Jiang, G., Lu, Q., Lu, H., Kaplan, H.J., Zhang, H.G., Shao, H., 2013. Miosge, N., Simniok, T., Sprysch, P., Herken, R., 2003. The collagen type XVIII Exosomes from retinal astrocytes contain antiangiogenic components that inhibit endostatin domain is co-localized with perlecan in basement membranes in vivo. laser-induced choroidal neovascularization. J. Biol. Chem. 288 (39), 28058–28067. J. Histochem. Cytochem. 51 (3), 285–296. https://doi.org/10.1177/ https://doi.org/10.1074/jbc.M113.470765. 002215540305100303. Hanai, J., Dhanabal, M., Karumanchi, S.A., Albanese, C., Waterman, M., Chan, B., Nawaz, I.M., Rezzola, S., Cancarini, A., Russo, A., Costagliola, C., Semeraro, F., Ramchandran, R., Pestell, R., Sukhatme, V.P., 2002. Endostatin causes G1 arrest of Presta, M., 2019. Human vitreous in proliferative diabetic retinopathy: endothelial cells through inhibition of cyclin D1. J. Biol. Chem. 277 (19), characterization and translational implications. Prog. Retin. Eye Res. 72, 100756. 16464–16469. https://doi.org/10.1074/jbc.M112274200. https://doi.org/10.1016/j.preteyeres.2019.03.002. Haverkamp, S., Ghosh, K.K., Hirano, A.A., Wässle, H., 2003. Immunocytochemical Noda, K., Ishida, S., Inoue, M., Obata, K.I., Oguchi, Y., Okada, Y., Ikeda, E., 2003. description of five bipolar cell types of the mouse retina. J. Comp. Neurol. 455 (4), Production and activation of matrix metalloproteinase-2 in proliferative diabetic 463–476. https://doi.org/10.1002/cne.10491. retinopathy. Investig. Ophthalmol. Vis. Sci. 44 (5), 2163–2170. https://doi.org/ Hou, Q., Ling, L., Wang, F., Xing, S., Pei, Z., Zeng, J., 2010. Endostatin expression in 10.1167/iovs.02-0662. neurons during the early stage of cerebral ischemia is associated with neuronal Noma, H., Funatsu, H., Yamashita, H., Kitano, S., Mishima, H.K., Hori, S., 2002. apoptotic cell death in adult hypertensive rat model of stroke. Brain Res. 1311, Regulation of angiogenesis in diabetic retinopathy: possible balance between 182–188. https://doi.org/10.1016/j.brainres.2009.11.033. vascular endothelial growth factor and endostatin. Arch. Ophtalmol. 120 (8), Huaman, J., Naidoo, M., Zang, X., Ogunwobi, O.O., 2019. Fibronectin regulation of 1075–1080. https://doi.org/10.1001/archopht.120.8.1075. integrin B1 and SLUG in circulating tumor cells. Cells 8 (6), 618. https://doi.org/ Ohlmann, A.V., Adamek, E., Ohlmann, Andreas, Lütjen-Drecoll, E., 2004. Norrie gene 10.3390/cells8060618. product is necessary for regression of hyaloid vessels. Invest. Ophthalmol. Vis. Sci. Hurskainen, M., Eklund, L., Hägg, P., Fruttiger, M., Sormunen, R., Ilves, M., 45 (7), 2384–2390. https://doi.org/10.1167/iovs.03-1214. Pihlajaniemi, T., 2005. Abnormal maturation of the retinal vasculature in type XVIII O’Reilly, M.S., Boehm, T., Shing, Y., Fukai, N., Vasios, G., Lane, W.S., Flynn, E., collagen/endostatin deficient mice and changes in retinal glial cells due to lack of Birkhead, J.R., Olsen, B.R., Folkman, J., 1997. Endostatin: an endogenous inhibitor collagen types XV and XVIII. FASE J. 19 (11), 1564–1566. https://doi.org/10.1096/ of angiogenesis and tumor growth. Cell 88 (2), 277–285. https://doi.org/10.1016/ fj.04-3101. s0092-8674(00)81848-6. Im, E., Venkatakrishnan, A., Kazlauskas, A., 2005. Cathepsin B regulates the intrinsic Pehrsson, M., Bager, C.L., Karsdal, M.A., 2019. Chapter 18-Type XVIII Collagen. angiogenic threshold of endothelial cells. Mol. Biol. Cell 16 (8), 3488–3500. https:// Biochemistry of Collagens, Laminins and Elastin: Structure, Function and doi.org/10.1091/mbc.e04-11-1029. Biomarkers, second ed. Elsevier, Herlev, Denmark, pp. 149–162. Inoue, S., 1989. Ultrastructure of basement membranes. Int. Rev. Cytol. 117, 57–98. Ramesh, S., Bonshek, R.E., Bishop, P.N., 2004. Immunolocalisation of opticin in the https://doi.org/10.1016/S0074-7696(08)61334-0. human eye. Br. J. Ophthalmol. 88 (5), 697–702. https://doi.org/10.1136/ Ito, M., Yoshioka, M., 1999. Regression of the hyaloid vessels and pupillary membrane of bjo.2003.031989. the mouse. Anat. Embryol. 200 (4), 403–411. https://doi.org/10.1007/ Ritter, M.R., Aguilar, E., Banin, E., Scheppke, L., Uusitalo-Jarvinen, H., Friedlander, M., s004290050289. 2005. Three-dimensional in vivo imaging of the mouse intraocular vasculature Kachi, S., Binley, K., Yokoi, K., Umeda, N., Akiyama, H., Muramatu, D., Iqball, S., Kan, S., Naylor, S., Campochiaro, P.A., 2009. Equine infectious anemia viral vector-mediated 15 A. Bonet et al. Experimental Eye Research 212 (2021) 108801 during development and disease. Investig. Ophthalmol. Vis. Sci. 46 (9), 3021–3026. Valsania, P., Warram, J.H., Rand, L.I., Krolewski, A.S., 1993. Different determinants of https://doi.org/10.1167/iovs.05-0153. neovascularization on the optic disc and on the retina in patients with severe Robinson, R., Barathi, V.A., Chaurasia, S.S., Wong, T.Y., Kern, T.S., 2012. Update on nonproliferative diabetic retinopathy. Arch. Ophthalmol. 111 (2), 202–206. https:// animal models of diabetic retinopathy: from molecular approaches to mice and doi.org/10.1001/archopht.1993.01090020056023. higher mammals. Dis Model Mech 5 (4), 444–456. https://doi.org/10.1242/ Waszczykowska, A., Podgórski, M., Waszczykowski, M., Gerlicz-Kowalczuk, Z., dmm.009597. Jurowski, P., 2020. Matrix metalloproteinases MMP-2 and MMP-9, their inhibitors Roy, S., Ha, J., Trudeau, K., Beglova, E., 2010. Vascular basement membrane thickening TIMP-1 and TIMP-2, vascular endothelial growth factor and sVEGFR-2 as predictive in diabetic retinopathy. Curr. Eye Res. 35 (12), 1045–1056. https://doi.org/ markers of ischemic retinopathy in patients with systemic sclerosis—Case series 10.3109/02713683.2010.514659. report. Int. J. Mol. Sci. 21 (22), 8703. https://doi.org/10.3390/ijms21228703. Ruberte, J., Carretero, A., Navarro, M., 2017. Morphological Mouse Phenotyping. Eye Weber, J., Peng, H., Rader, C., 2017. From rabbit antibody repertoires to rabbit and Related Structures. Panamericana, Madrid, pp. 475–520. monoclonal antibodies. Exp. Mol. Med. 49, e305. https://doi.org/10.1038/ Saint-Geniez, M., D’Amore, P.A., 2004. Development and pathology of the hyaloid, emm.2017.23. choroidal and retinal vasculature. Int. J. Dev. Biol. 48 (8–9), 1045–1058. https://doi. Wickström, A.S., Alitalo, K., Keski-Oja, J., 2002. Endostatin associates with integrin org/10.1387/ijdb.041895ms. alpha5beta1 and caveolin-1, and activates Src via a tyrosyl phosphatase-dependent Sertié, A.L., Sossi, V., Camargo, A.A., Zatz, M., Brahe, C., Passos-Buenos, M.R., 2000. pathway in human endothelial cells. Cancer Res. 62 (19), 5580–5589. Collagen XVIII, containing an endogenous inhibitor of angiogenesis and tumor Yamaguchi, N., Anand-Apte, B., Lee, M., Sasaki, T., Fukai, N., Shapirolvo, R., Lowik, Q., growth, plays a critical role in the maintenance of retinal structure and in neural Timpl, R., Olsen, B., 1999. Endostatin inhibits VEGF-induced endothelial cell tube closure (Knobloch syndrome). Hum. Mol. Genet. 9 (13), 2051–2058. https:// migration and tumor growth independently of zinc binding. EMBO J. 18, doi.org/10.1093/hmg/9.13.2051. 4414–4423. https://doi.org/10.1093/emboj/18.16.4414. Simó, R., Carrasco, E., Garcia-Ramirez, M., Hernández, C., 2006. Angiogenic and Zhang, C., Gehlbach, P., Gongora, C., Cano, M., Fariss, R., Hose, S., Nath, A., Green, W.R., antiangiogenic factors in proliferative diabetic retinopathy. Curr. Diabetes Rev. 2, Goldberg, M.F., Zigler, J.S., Sinha, D., 2005. A potential role for β- and γ-crystallins 71–98. https://doi.org/10.2174/157339906775473671. in the vascular remodeling of the eye. Dev. Dynam. 234 (1), 36–47. https://doi.org/ Smith, G., McLeod, D., Foreman, D., Boulton, M., 1999. Immunolocalisation of the VEGF 10.1002/dvdy.20494. receptors FLT-1, KDR, and FLT-4 in diabetic retinopathy. Br. J. Ophthalmol. 83 (4), Zhang, C., Asnaghi, L., Gongora, C., Patek, B., Hose, S., Ma, B., Fard, M.A., Brako, L., 486–494. https://doi.org/10.1136/bjo.83.4.486. Singh, K., Goldberg, M.F., Handa, J.T., Lo, W.K., Eberhart, C.G., Zigler Jr., J.S., Smith, R.S., John, S.W.M., Nishina, P.M., Sundberg, J.P., 2002. Systematic Evaluation of Sinha, D., 2011. A developmental defect in astrocytes inhibits programmed the Mouse Eye: Anatomy, Pathology, and Biomethods, first ed. CRC Press. regression of the hyaloid vasculature in the mammalian eye. Eur. J. Cell Biol. 90 (5), Somerville, R.P.T., Oblander, S.A., Apte, S.S., 2003. Matrix metalloproteinases: old dogs 440–448. https://doi.org/10.1016/j.ejcb.2011.01.003. with new tricks. Genome Biol. 4 (6), 216. https://doi.org/10.1186/gb-2003-4-6-216. Zhu, X., Li, A., Brown, B., Weiss, E.R., Osawa, S., Craft, C.M., 2002. Mouse cone arrestin Takahashi, K., Saishin, Yoshitsugu, Saishin, Yumiko, Lima-Silva, R., Oshima, Y., expression pattern: light induced translocation in cone photoreceptors. Mol. Vis. 8, Oshima, S., Melia, M., Paszkiet, B., Zerby, D., Kadan, M.J., Liau, G., Kaleko, M., 462–471. Connelly, S., Luo, T., Campochiaro, P.A., 2003. Intraocular expression of endostatin Zudaire, E., Gambardella, L., Kurcz, C., Vermeren, S., 2011. A computational tool for reduces VEGF-induced retinal vascular permeability, neovascularization, and retinal quantitative analysis of vascular networks. PLoS One 6 (11), e27385. https://doi. detachment. Faseb. J. 17 (8), 896–898. https://doi.org/10.1096/fj.02-0824fje. org/10.1371/journal.pone.0027385. 16