A spontaneously immortalized human MGC line (MIO-M1), kindly provided by G. Astrid Limb (UCL Institute of Ophthalmology and Moorfields Eye Hospital, London, United Kingdom), was used. Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Buenos Aires, Argentina) containing 4500 mg/L glucose, sodium pyruvate) with 2 mM L-glutamine (GlutaMAX; Invitrogen), 10% vol/vol fetal bovine serum (FBS), and 50 U/mL penicillin/streptomycin (Invitrogen). Bovine aortic endothelial cells (BAECs) were grown in DMEM supplemented with 20% vol/vol FBS, and penicillin (100 U/mL)/streptomycin (100 U/mL) (Invitrogen). BAECs’ media contained 0, 03 mg/ml of endothelial cell growth supplement from bovine pituitary (Sigma). Cell cultures were kept at 37°C in a 5% CO₂ humidified environment.

For gas hypoxia, cells were grown at 50–60% confluence in normal conditions and then transferred to a gas culture chamber (StemCell Technologies, Vancouver, Canada) supplied with 1% O₂, 94% N₂, and 5% CO₂. Control cells were kept in normoxia (21% O₂). For chemical hypoxia, cells were grown at 50–60% confluence in 10% FBS and then a solution of CoCl₂ (Cicarelli, Santa Fe, Argentina) diluted in PBS was added to the cell medium. Cell experiments were conducted at two different time points (4 and 24 h). To assess the autophagy flux, Chloroquine (CQ) diluted in PBS was added to result in a final concentration of 10 μM (24 h of exposure) or 50 μM (4 h of exposure). Some experiments included incubations with inductors (50 μM Rapamycin, 100 μM Resveratrol) of the autophagic flux, in presence or absence of CQ. All experiments were carried out in 10% FBS, as fasting constitutes an additional stimulus for autophagy.

For reoxygenation, hypoxic medium was replaced for standard medium and cells were incubated for 1 h at 37°C with 5% CO₂.

To determine cell viability, the colorimetric MTT assay was performed. Briefly, 2.000 cells per well were seed in a 96-well plate and incubated under different experimental conditions (normoxia, hypoxia, CoCl₂, with or without Rapamycin or Resveratrol) for 24 h. Then, 10 μl of the yellow tetrazolium salt 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) (5 mg/ml) was added to culture medium. Cells were incubated with the MTT reagent for additional 3 h at 37°C. After that, cell medium was carefully removed and 100 μl of DMSO was added in order to solubilize the crystal violet. Finally, the optical density values were measured at a wavelength of 570 nm using a SpectraMax M5 plate reader (Molecular Devices, United States).

Following treatment, cells were lysed with a solution of Triton X-100 1% in PBS, 1 mM PMSF and protease inhibitor cocktail (Sigma Aldrich, St. Louis, MO). Lysates were cleared of insoluble material by centrifugation at 15,000 g for 10 min at 4°C. Protein concentration of MIO-M1 cell extracts were measured by a Bicinchoninic Acid Protein Assay Kit (Pierce BCA, Thermo Scientific, United States) according to the manufacturer’s protocol and 20 μg of proteins were electrophoresed in 15% SDS-PAGE. After electrophoresis, proteins were transferred to nitrocellulose membranes (Amersham Hybond ECL; GE Healthcare Bio-Sciences AB, Uppsala, Sweden). To prevent nonspecific binding, membranes were blocked with 5% milk in TBS containing 0.1% Tween-20 (TBST) during 1 h at room temperature (RT). Then, blots were incubated with primary antibodies diluted in 1% BSA in TBST overnight at 4°C. The following antibodies were used: rabbit polyclonal anti-LC3 (1/1000; L7543, Sigma Aldrich), mouse monoclonal anti p62 (1/1000; ab56416, Abcam), rabbit polyclonal anti-GFAP (1/1000; Dako, Carpinteria, CA), mouse monoclonal anti-Vimentin (1/1000; M7020, Dako) and mouse monoclonal anti-β-actin (1/2000; ab8226, Abcam). Blots were incubated with IRDye 800 CW donkey anti-rabbit Ig, or IRDye 800 CW donkey anti-mouse IgG antibodies (1/15000 in TBS with 5% BSA) for 1 h, protected from light. After washing with TBST, membranes were visualized and quantified using the Odyssey Infrared Imaging System (LI-COR, Inc., Lincoln, NE, United States).

Briefly, cells were grown at 30% confluence on coverslips. For staining with the lysosome marker LysoTracker Red (Thermo Fisher Scientific) 1 μl/ml of the reagent was added to the cell medium in the final hour of the incubation time. After stimuli, coverslips were washed twice with precooled PBS and fixed in PBS containing 4% PFA/4% sucrose for 15 min at RT. Free aldehyde groups were blocked with 50 mM NH₄Cl in PBS during 20 min. The cells were then incubated in a blocking buffer (PBS containing 0.05% saponin/0.2% BSA) for 30 min. Antibodies were prepared in blocking buffer. Cells were then incubated overnight at 4°C with the following primary antibodies: rabbit polyclonal anti-LC3 (1/100; L7543, Sigma Aldrich), mouse monoclonal anti p62 (1/100; ab56416, Abcam). Then, cells were washed three times with PBS plus 1% BSA, and exposed to secondary antibodies diluted 1:800 for 45 min at RT. After a thorough rinse with PBS plus 1% BSA, cell nuclei were also counterstained with Hoechst 33258 (1:1000; Molecular Probes) for 7 min. Cells were washed with PBS and mounted on glass slides with Mowiol 4–88 (Calbiochem; Merck KGaA, Darmstadt, Germany). Fluorescent images were obtained with a confocal laser-scanning biological microscope (Olympus FluoView FV1200; Olympus Corp., New York, NY, United States). Finally, images were processed with ImageJ FIJI software (National Institutes of Health, Bethesda, MD, United States).

Total RNA was extracted from cultured cells using Trizol Reagent (Invitrogen) (Ridano et al., 2012). Briefly, 1 μg of total RNA was reverse-transcribed in a total volume of 20 μl using random primers (Invitrogen, Buenos Aires, Argentina) and 50 U of M-MLV reverse transcriptase (Promega Corp.). Then, cDNA was mixed with 1x SYBR Green PCR Master Mix (Applied Biosystems) and forward and reverse primers: Beclin-1 forward: CCATGCAGGTGAGCTTCGT/Beclin-1 reverse: GAA​TCT​GCG​AGA​GAC​ACC​ATC; ATG5 forward: AAA​GAT​GTG​CTT​CGA​GAT​GTG​T/ATG5 reverse: CAC​TTT​GTC​AGT​TAC​CAA​CGT​CA. VEGF-A forward:CCGCAGACGTGTAAATGTTCCT/VEGF-A reverse: CGG​CTT​GTC​ACA​TCT​GCA​AGT​A; PEDF forward: GCTGAGTTACGAAGGCGAAGT/PEDF reverse: TTG​ATG​GGT​TTG​CCT​GTG​AT. qPCR were carried out on an Applied Biosystems 7500 Real-Time PCR System with Sequence Detection Software v1.4. The cycling conditions included a hot start at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. Specificity was verified by melting curve analysis. Results were normalized to β-actin (Forward: AAATCTGGCACCACACCTTC/Reverse: GGG​GTG​TGA​AGG​TCT​CAA​A). Relative gene expression was calculated according to the ^2−ΔΔCt method. Each sample was analyzed in triplicate. No amplification was observed in PCRs using as template water (data not shown).

The assay was performed according to Arnaoutova y Kleinman, (Arnaoutova and Kleinman, 2010). Briefly, BAEC (≈1.5 × 10⁴ cells) were placed on a 96 well half area plate previously coated with 30 uL of Matrigel. The plates were incubated for 18 h (37°C, 5% CO₂) in the presence of normoxic or hypoxic MIO-M1 supernatants containing vehicle or autophagy inductors. Sodium suramine 30 uM and Axitinib 10 uM were used as positive control of inhibition. Control samples with and without their respective vehicles were simultaneously run. The images were obtained with an Olympus CKX41 inverted microscope and analyzed with the software ImageJ. The tubular structures were quantified and the percentages of inhibition (I%) were calculated I% = [1−(Total tube length treatment/Total tube length control)] x 100.

Statistical analysis was performed using the GraphPad Prism 7.0 software. A p-value < 0.05 was considered statistically significant. Parametric or nonparametric tests were used according to variance homogeneity evaluated by F or Barlett’s tests. Two-tailed unpaired t or Mann Whitney tests were used in analysis of two groups. One-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparison post-test or Kruskal-Wallis followed by Dunn´s multiple comparison post-test were used to determine statistical significance among more than two different groups. Mean ± standard error (SEM) are shown in graphs analysed with parametric tests and median with interquartile range are shown when data were analysed with nonparametric test.

Previous studies conducted in MGCs exposed to hyperglycemia showed an altered autophagic flux given that autophagosomes could not reach the degradation step (Lopes de Faria et al., 2016). However, in a hypoxic environment rat MGCs displayed an increase in LC3B II protein levels (Fung et al., 2016). Thereby, our first aim was to evaluate changes in the autophagic flux in human MGCs (MIO-M1), under hypoxia. Two experimental hypoxic models were used: the gas culture chamber (cells exposed to 1% O₂) and a chemical model (cells exposed to CoCl₂). Our experimental design included controls with and without Chloroquine (CQ) in order to assess the accumulation of the autophagosomes.

Although MGCs are very perceptive to environmental changes, they are considered the most resistant retinal cells to external injuries (Coughlin et al., 2017). In order to examine MIO-M1 survival under both hypoxic models, we carried out MTT assays at 24 h (Figure 1). MIO-M1 viability was reduced by 15% when exposed to 1% O₂ compared to normoxia (p = 0.001) (Figure 1A). For chemical hypoxia, we tested increasing concentrations of CoCl₂. Viability of MIO-M1 cells exposed to CoCl₂ up to 250 μM did not change significantly respect to the control (p = 0.1358) (Figure 1B). Of note, no significant differences were observed in the MIO-M1 viability for shorter exposure periods (4 and 8 h, data not shown). Then, we studied changes in the expression of autophagic proteins in MIO-M1 cells exposed to CoCl₂ for 4 h by Western blot (Figure 1C). LC3B II (a structural protein of the autophagosomes) and p62 (an adaptor protein that is degraded with the autophagosome content) were evaluated. The quantitative analyses of LC3B II and p62 bands evidenced a significant increase in the autophagic protein levels since a concentration of 250 μM respect to vehicle (p = 0,027 for LC3B II and p = 0,031 for p62). Therefore, we selected 250 μM CoCl₂ as a concentration for in vitro studies as it represents the minimum concentration that achieves two relevant requisites: it is not detrimental for cell viability and it is able to induce the autophagic flux.

Autophagy is a catabolic process rapidly activated in response to environmental stimuli that enables the cell adaptation to the new living conditions (Mizushima and Komatsu, 2011). Hence, we decided to evaluate if hypoxia modified the autophagic flux in human MGCs exposed to short periods. For this purpose, the MIO-M1 cells were exposed to hypoxia or normoxia during 4 h, in presence of CQ or vehicle in the above indicated conditions. Simultaneously, a comparative study was performed by incubating MIO-M1 cells with CoCl₂ 250 μM, including controls with and without CQ (Figure 2). By Western blot assay we observed no significant changes in LC3B II and p62 protein expression levels in gas hypoxia or CoCl₂ respect to normoxia in absence of CQ. However, in samples treated with CQ a significant increase in LC3B II protein levels was observed under both hypoxic treatments (p = 0.0447 for 1% O₂ and p = 0.0021 for CoCl₂) respect to normoxia.

Under 1% O₂, MIO-M1 cells incubated with CQ showed a significant increase in p62 protein levels (p = 0.0006) (Figures 2A,B) respect to normoxia-CQ. Although we observed a non-significant increase in the expression of p62 in presence of CoCl₂ 250 μM (p = 0.3959), incubation with CQ showed a significant accumulation of LC3B II in MIO-M1 cells respect to vehicle, suggesting that the autophagic vesicles follow the degradation pathway properly.

Correlatively, immunofluorescence assays with classic markers of autophagy showed an increased number of LC3B puncta per cell in MIO-M1 exposed to gas hypoxia (p < 0.001) and CoCl₂ (p = 0.005), respect to normoxia (Figures 2C,D). In line with these results, a higher colocalization of LC3B and p62 was determined in both hypoxic conditions respect to normoxia (p = 0.018 for 1% O₂ and p = 0.0069 for CoCl₂) (Figure 2E).

Cell response to an injurious stimulus can vary depending on its duration. Therefore, we decided to evaluate MGCs behaviour under prolonged exposure to hypoxia (24 h) by analysing changes in their autophagic flux. Experimental conditions were similar to those in shorter hypoxic stimuli. By Western blot assays, similar protein expression levels of LC3B II and p62 were observed in MIO-M1 cells under gas hypoxia and normoxia when incubated with CQ (p > 0.9999 for LC3BII and p = 0.2821 for p62) (Supplementary Figures S1A,B). Then, we quantified mRNA levels of Beclin-1 and ATG5. Results showed a not statistically significant decrease in Beclin-1 and ATG-5 transcript levels in MIO-M1 cells under 1% O₂ (p = 0.1248 for Beclin-1 and p = 0.0753 for ATG5) (Supplementary Figure S1C). In contrast, an increase in autophagy protein levels was observed in cells incubated with CoCl₂ for 24 h in presence of CQ (p = 0.0011 for LC3BII and p < 0.0001 for p62) (Supplementary Figures S1A,B). In spite of this result, mRNA levels of Beclin-1 and ATG5 were not modified (p = 0.7368 for Beclin-1 and p = 0.3638 for ATG5), suggesting a slight activation of the autophagic flux (Supplementary Figure S1C). Confocal images of MIO-M1 cells exposed to hypoxia (1% O₂) and normoxia revealed similar number of LC3B puncta per cell (p = 0.3081) and colocalization of LC3B and p62 (p = 0.2975) (Supplementary Figures S1D–F). On the other hand, CoCl₂ induced an increase in the number of vesicles decorated with LC3B (p = 0.0373) and a slight rise in the colocalization of LC3B and p62 respect to control (p = 0.0024) (Supplementary Figures S2A–C). These results demonstrate a higher number of autophagosomes when MIO-M1 cells were exposed to CoCl₂ for 24 h respect to control. The late stage of the autophagy pathway involves the degradation of the autophagosome content and its inner membrane by fusion with lysosomes. In order to evidence the fusion of autophagosomes with lysosomes, immunofluorescence assays were performed with LC3B and LysoTracker, an acidic compartment and late endosomes marker. The colocalization analysis evidenced no difference in the overlap of the proteins when MIO-M1 cells were incubated with vehicle or CoCl₂ during 24 h (p = 0.9738) (Supplementary Figures S2D,E). Moreover, we also evaluated the lysosomal membrane integrity by immunostaining with Galectin-1 (Gal-1). In our experiments, Gal-1 staining showed a homogeneous cytoplasmic distribution in MIO-M1 cells in both hypoxic models, suggesting that lysosomal structure was preserved after 24 h of exposure (Supplementary Figure S2F).

Clinical studies provided evidence about the variation of oxygen levels in neuronal tissues in patients with PDR (Daulatzai, 2017). Hence, we next evaluated the autophagic flux in MIO-M1 cells exposed to 4 h hypoxia followed by 1 h of reoxygenation. Quantitative analysis of Western blot assays revealed a significant increase in LC3B II (p < 0.001) in reoxygenated MIO-M1 cells compared to the hypoxic ones, in conditions with CQ (Figures 3A,B). In line with this results an increase in vesicles decorated by LC3B were observed in reoxygenated MIO-M1 cells in presence of CQ (Figure 3C), suggesting an increase in the number of autophagosomes when returned to normoxia after the hypoxic period. To determine the role of autophagy in reoxygenation, we assessed MIO-M1 survival by MTT assay (Figure 3D). The cell viability of the reoxygenated MIO-M1 cells was significantly lower than in hypoxic condition (p = 0.0009). In addition, inductors of the autophagic flux were unable to restore MIO-M1 survival (p = 0.305 for Reoxygenation + Rapamycin; p = 0.9999 for Reoxygenation + Resveratrol). Notably, the blockade of the flux with CQ induced cytoplasmatic retraction and nuclear condensation in MIO-M1 cells under reoxygenation (Figure 3C, white arrow).

In response to an injury or stressor, MGCs trigger protective mechanisms in order to preserve retinal structure and functionality, a process called reactive gliosis (Subirada et al., 2018). We have previously demonstrated in a mouse model of Oxygen-induced retinopathy (OIR) that MGCs become reactive after the hyperoxic phase and modified the expression of stress and detoxifying proteins from postnatal days 12 to 26 in OIR mice (Ridano et al., 2017). Then, our next aim was to analyse changes in the expression levels of GFAP after hypoxic exposure. This set of experiments was performed with gas hypoxia, at the time where a main response to hypoxia was detected (4 h).

After 4 h of hypoxia, MIO-M1 cells showed an increase in protein levels of GFAP (p < 0.0001) in accordance with the activation of glial cells (Figures 4A,B), although no differences in Vimentin expression was detected (p = 0.1012) by Western blot assays (Figures 4C,D).

We have previously demonstrated in the OIR mouse model, a decrease in the GFAP protein levels after a single intraocular injection of Rapamycin at postnatal day 26 OIR (Subirada et al., 2019). Therefore, we aimed to determine if inductors of the autophagic flux prevented the increase of GFAP expression under hypoxic conditions. For this purpose, MIO-M1 cells were incubated in normoxia or hypoxia (1% O₂) and in presence of Rapamycin or Resveratrol for 4 h. Western blot assays showed a substantial decrease in GFAP protein expression levels by Rapamycin in MIO-M1 cells exposed to 1% O₂ (p = 0.0438). A trend to diminish GFAP protein levels was observed in hypoxia/Resveratrol, however, this decrease was not statistically significant (p = 0.6558) (Figures 4E,F). We confirmed the induction of the autophagic flux in this experiment by analysis of LC3B II and p62 protein levels (Figures 4E,G). Lastly, we evaluated if autophagy inductors could prevent cell death after 24 h of hypoxia, the only experimental time point where decreased viability was observed (Figure 1A). The MTT assay showed that MIO-M1 cells incubated with either of the inductors exhibited similar cell survival respect to hypoxia vehicle (p = 0.0834 for hypoxia + Rapamycin; p = 0.1337 for hypoxia + Resveratrol) (Figure 4H).

It is known that MGCs maintain the vascular structure through a balance between synthesis and secretion of pro- and anti-angiogenic factors. Thus, our next goal was to evaluate MGCs ability to induce tubulogenesis of endothelial cells (ECs) under hypoxia and how the inductors of the autophagy modified this response. For this purpose, we incubated MIO-M1 cells under normoxia or hypoxia (1% O₂) with an inductor of the autophagic flux or vehicle, and obtained their supernatants. Then, tubulogenesis assays were performed on Matrigel by incubating ECs with the supernatants of MIO-M1 cells during 24 h (Figure 5A). Suramin and Axitinib were used as negative controls of tubulogenesis. Quantitative analysis of the tubular network formed by ECs showed that hypoxic MIO-M1 supernatants increased the formation of vascular tubules respect to normoxic supernatants. We observed an increase in the mesh number (polygonal vascular structures) and the total segment length of vessel tubules in ECs incubated with hypoxic supernatants (Figures 5A,B). Then we compared ECs tubulogenesis incubated with supernatants containing vehicle respect to those containing an autophagy inductor. MIO-M1 supernatants exposed to hypoxia and in presence of Rapamycin decreased the mesh number, increased the average mesh area and decreased the total segment length respect to hypoxic + vehicle supernatant. Those changes were not observed between vehicle and Rapamycin in normoxic supernatants. On the other hand, Resveratrol modified all the quantified parameters, showing an evident decrease in the tubule formation respect to vehicle. Resveratrol had a similar effect when ECs were incubated with both normoxic and hypoxic MIO-M1 supernatants. In line with these results, both treatments in hypoxic supernatants showed a significant increase in the percentage of tubulogenesis inhibition respect to hypoxic-vehicle supernatant (Figure 5C). Finally, we examined if autophagy inductors had a direct effect over ECs. For that purpose, we incubated ECs with Rapamycin and Resveratrol for 24 h. We observed no variations in the formation of polygonal structures when ECs were incubated with either autophagy inductor respect to vehicle (Supplementary Figure S3).

It has been demonstrated that under hypoxic conditions MGCs are one of the cells responsible for the increase of VEGF retinal levels, the main pro-angiogenic factor involved in angiogenesis (Wang et al., 2010). To restrict the vascular growth, MGCs secrete PEDF, which counterbalance pro-angiogenic factors (Yang et al., 2012). In order to determine if trophic factors levels were modified by the autophagy inductors, we further quantified the transcript levels of VEGF and PEDF in hypoxic MIO-M1 cells (Figure 5D). Quantitative RT-PCR analysis showed an increase in VEGF mRNA levels in hypoxia respect to normoxia. Moreover, Rapamycin induced a significant decrease of VEGF mRNA respect to hypoxia vehicle (p = 0.0334). VEGF transcript was not modified in MIO-M1 incubated in hypoxia + Resveratrol. By contrast, PEDF mRNA was only increased in hypoxic MIO-M1 cells treated with Resveratrol (p = 0.0003). In summary, the results indicate that Rapamycin and Resveratrol prevents tubule formation by modulating different proteins involved in angiogenesis.