Stabilization of HIF-1α in Human Retinal Endothelial Cells Modulates Expression of miRNAs and Proangiogenic Growth Factors

Retinal hypoxia is one of the causative factors of diabetic retinopathy and is also one of the triggers of VEGF release. We hypothesized that specific dysregulated miRNAs in diabetic retinopathy could be linked to hypoxia-induced damage in human retinal endothelial cells (HRECs). We investigated in HRECs the effects of chemical (CoCl2) hypoxia on the expression of HIF-1α, VEGF, PlGF, and of a focused set of miRNAs. We found that miR-20a-5p, miR-20b-5p, miR-27a-3p, miR-27b-3p, miR-206-3p, miR-381-3p correlated also with expression of TGFβ signaling pathway genes in HRECs, challenged with chemical hypoxic stimuli. In conclusion, our data suggest that retinal angiogenesis would be promoted, at least under HIF-1α activation, by upregulation of PlGF and other factors such as miRNAs, VEGFA, and TGFβ1.


INTRODUCTION
Diabetic retinopathy (DR), a complication of diabetes, is a microvascular disease with a strong inflammatory imprinting. Vascular endothelial growth factor (VEGF) is a key player in retinal neovascularization, and intraocular injections of anti-VEGF agents are currently the established therapies for diabetic macular edema, along with steroids (Bandello et al., 2012). Although not fully elucidated, alterations in retinal hemodynamics and reduced blood flow may be detrimental for DR, along with uncontrolled hyperglycemia (Schmetterer and Wolzt, 1999;Schmidl et al., 2015). Furthermore, during DR progression, local or global changes in retinal oxygenation may cause the development of hypoxic areas (Arden and Sivaprasad, 2012) and oxidative stress (Bucolo et al., 2006). Similar to the etiopathogenesis of retinopathy of prematurity (ROP), induction of hypoxiainducible factor-1 a (HIF-1a) may be responsible for the production of vascular endothelial growth factor (VEGFA), which is the main cause of retinal neovascularization (Aiello et al., 1994;Arjamaa and Nikinmaa, 2006;Abu El-Asrar et al., 2012). Furthermore, HIF-1a and VEGFA crosstalk in ocular neovascularization has been widely investigated (Ozaki et al., 1999;Rodrigues et al., 2016). In particular, the HIF-1a inhibition strategy has also been explored for treatment of retinal neovascularization (Iwase et al., 2013;D'Amico et al., 2015;D'Amico et al., 2017;Zeng et al., 2017).
Therefore, we hereby hypothesized that these eight miRNAs could also be involved in activation of HIF-1/angiogenic axis in retinal endothelial cells. With this aim, we stabilized, by cobalt chloride treatment, HIF-1a protein in human retinal endothelial cells (HRECs), in order to analyze the activation of HIF-1/ VEGFA-PlGF axis, along with expression of a focused set of miRNAs, previously found to be dysregulated in an in vivo model of DR . A bioinformatic approach guided the identification and in vitro validation of alternative target genes of miRNAs, dysregulated after inhibition of HIF-1a degradation. We analyzed the expression of genes of the TGFb (Transforming growth factor beta) signaling pathway, which is an emerging target in DR Stafiej et al., 2018) and was found to be one of top pathways modulated by HypoxamiRs target genes (Gupta et al., 2018).

Induction of Chemical Hypoxia In Vitro
Cobalt chloride (CoCl 2 ) is commonly used to stabilize HIF-1a, because it inhibits the HIF-1a degradation, as shown in several in vitro settings, including primary human retinal endothelial cells cultures as previously described (Gao et al., 2008;Hu et al., 2012;Li et al., 2017;He et al., 2019). Preliminary studies were carried out and HRECs cultures were treated with various concentrations of CoCl 2 (100-200 mM), in order to assess cell tolerability for 24 h with MTT test (Supplementary Data). The concentration used for all experiments was 200 mM, accordingly to previous CoCl 2 concentrations tested on retinal ganglion cells (Balaiya et al., 2012;Li et al., 2017). Cells were seeded in Petri dishes (passage number 4, cell density 4×10 5 ); after reaching confluence (approximately 80%), cells were treated with CoCl 2 for 30 min, 2 and 8 h to induce HIF-1a accumulation/ nuclear translocation.

Western Blot
HRECs were cultured in 60 mm Petri dishes (cell density 4×10 5 ). Proteins from cell lysates were extracted with RIPA Buffer, including protease and phosphatase inhibitors cocktail (Sigma-Aldrich, St. Louis, MO, USA). Total protein content, in each cell lysate sample, was determined by the BCA Assay Kit (Pierce ™ BCA Protein Assay Kit, Invitrogen, Life Technologies, Carlsbad, CA, USA). Extracted proteins (40 µg) were loaded on 4%-12% tris-glycine gel. After electrophoresis proteins were transferred into a nitrocellulose membrane (Invitrogen, Life Technologies, Carlsbad, CA, USA). Immunoblot was preceded by addition of Odyssey Blocking Buffer (LI-COR Lincoln, NE, USA) to membranes. Therefore, membranes were incubated overnight (4°C) with appropriate primary HIF-1a (1:200 dilution) and anti-GAPDH (1:500 dilution) antibodies. GAPDH was selected as control for protein expression, accordingly to previous reports (Botlagunta et al., 2011;Ao et al., 2015;Evrard et al., 2016;Gao et al., 2019). After overnight incubation, the membranes were then incubated with secondary fluorescent antibodies (1: 10,000 dilution) for 1 h at room temperature. Immunoblot was detected through Odyssey imaging system (LI-COR, Lincoln, NE, USA). Densitometry analyses of blots were performed at nonsaturating exposures and analyzed using the ImageJ software (NIH, Bethesda, MD, USA; available at http://rsb.info.nih.gov/ij/index. html). Values were normalized to GAPDH, which was also used as loading control (see supplemental information for whole gel membranes immunoblots).

Extraction of Total RNA and cDNA Synthesis
Extraction of the total RNA was performed with TRIzol Reagent (Invitrogen, Life Technologies, Carlsbad, CA, USA). The A260/ A280 ratio of the optical density of RNA samples (measured with Multimode Reader Flash di Varioskan ™ ) was 1.95-2.01. This RNA purity was confirmed with the electrophoresis in nondenaturing 1% agarose gel (in TAE), that showed an adequate RNA purity, concentration, and integrity. cDNA was synthesized from 2 µg RNA with a reverse transcription kit (SuperScript ™ II Reverse Transcriptase, Invitrogen, ThermoFisher Scientific, Carlsbad, CA, USA).

MicroRNA or TGFb Signaling Pathway Genes Expression Determination Analysis
CFX Manager ™ Software (Bio-Rad, Hercules, CA, USA) was used to calculate Cycle threshold (Ct) values. Data analysis was carried out with the 2 -DD Ct method. Particularly, DCt value for each miRNA or gene profiled was calculated as DCt = Ct miRNA -Ct Ce_miR-39-5p or as DCt = Ct gene -Ct GAPDH . Then, DDCt was calculated as DCt time x -DCt time 0 , where time x is the analyzed time point and time 0 is the expression of the target miRNA normalized to Ce-miR-39-5p or of the target gene normalized to GAPDH (Livak and Schmittgen, 2001). Where data are reported as fold-regulation, this was the inverse negative of fold change (2^-DD Ct) for fold change values lower than one (downregulation). In case of upregulation, the fold-regulation was equal to fold change (2^-DD Ct) for fold change values greater than 1.

Statistical Analysis
All results were reported as mean ± SD from four independent in-vitro experiments, where each group was triplicated in plates as technical replicate. The results were analyzed using one-way ANOVA, followed by Tukey-Kramer post-hoc multiple comparisons test. Differences between groups were considered significant for p-value < 0.05. Graphs design and statistical analysis were carried out with GraphPad Prism 5 software (GraphPad Inc., San Diego, CA, USA).
After 8 h, similarly to VEGFA expression, we found a shift in expression pattern of miRNAs, compared levels detected 2 h after CoCl 2 treatment ( Figure 3). Experimental validated miRNA : VEGFA mRNA interactions were found for miR-20a-5p and miR-20b-5p , and in hepatocellular carcinoma for miR-381-3p (Tsai et al., 2017;Wang et al., 2018). Therefore, VEGFA expression levels could be related to the expression pattern of miRNAs, 2 to 8 h after stabilization of HIF-1a, because VEGFA is a target of miR-20a, miR-20b, miR-381, and indirectly of miR-27b (Veliceasa et al., 2015). On the contrary, PlGF is not a validated or predicted target of any miRNAs dysregulated in HRECs treated with CoCl 2 . Particularly, the role of PlGF in regulation retinal angiogenesis, under hypoxic stimuli, is still unknown. On the other hand, several reports support the detrimental role of PlGF in the pathogenesis and progression of DR (Carmeliet et al., 2001;Huang et al., 2015), likely through HIF-1a, or indirectly by miRNAs and the PI3K/AKT signaling pathways (Figure 4) (Zhou et al., 2016;Jin et al., 2018). Therefore, our hypothesis is based on retinal angiogenesis regulated by miRNAs under hypoxic stimuli, and miRNAs can be considered alternative and/or ancillary components to VEGFA and PlGF pathways. Indeed, we analyzed other putative miRNAs targets (gene and pathways) and identified, through a bioinformatic approach, the TGFb signaling pathway as the top-scored pathway dysregulated by identified miRNAs (Figure 4). Then, we found that miRNAs, dysregulated after CoCl 2 treatment, (miR-20a-5p, miR-20b-5p, miR-27a-3p, miR-27b-3p, miR-206-3p, miR-381-3p) influenced mRNA levels of TGFb1, TGFbR1, TGFbR2 and SMAD2, according to experimental validated miRNA:mRNA interactions ( Figures 5 and 6, Table 1). TGFb1, TGFbR2 and SMAD2, were upregulated 2 and 8 h after HIF-1a stabilization. Interestingly, the expression of TGFbR1 receptor, which is target of most of analyzed miRNAs (Table 1), correlated with expression pattern shift of miRNAs at 2 h and 8 h after CoCl 2 treatment. Several reports support a detrimental role of TGFbR1 in DR, particularly, TGFbR1 immunoreactivity was found to be increased   in retinal capillaries of diabetic rats (Gerhardinger et al., 2009;van Geest et al., 2010).
The HIF-1/TGF-b1 axis, and related stimulation of angiogenesis, has been investigated in different experimental settings (Han et al., 2013;Mingyuan et al., 2018), including endothelial cells (Iruela-Arispe and Sage, 1993;Peshavariya et al., 2014). On the contrary, few reports demonstrated a putative link between HIF-1a/miRNAs/ TGFb signaling pathway and angiogenesis (Xing et al., 2014). Furthermore, only one study analyzed the role miRNAs in regulation of hypoxia-TGFb-angiogenesis pathway in a model of corneal neovascularization (Zhang Y. et al., 2019). According to our findings, miR-27 was reported to be involved in regulation of HIF-1/TGFb axis, at least in an in vitro model of cardiac ischemia . However, there are still no evidences about a putative link in retinal disease between hypoxia, miRNAs, VEGFA, and TGFb pathway.
High throughput miRNA expression analysis on retinal endothelial cells, challenged with chemical hypoxic stimuli, could reveal the involvement of other miRNAs, along with the focused set analyzed in this study. However, those high throughput analyses are expensive and need quantitative qPCR validation (de Ronde et al., 2018). Despite the small set of analyzed miRNAs, our study suggested that ocular neovascularization, during hypoxia, would be promoted by the upregulation of PlGF and other factors induced by HIF-1a/ miRNAs, i.e. VEGFA, and genes of the TGFb1 signaling pathway (Figure 7). Therefore, these data warranting further in vivo studies to explore the use of pharmacological/molecular approach such as antagomiRs and agomir.
Indeed, the present findings highlighted that proangiogenic factors are worthy to be further explored as potential targets for pharmacological modulation of local retinal hypoxic events, which are generally transient but detrimental in retinal degenerations.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

AUTHOR CONTRIBUTIONS
CB, MD'M, and SR made substantial contributions to conception, design, and interpretation of data. FL, MT, and CP carried out experiments. FL, MT, CP, FP, and CG carried out formal analysis of data. FL, MT, CP, and CB wrote initial draft of the manuscript. CB, MD'M, SR, FD, and MG reviewed the manuscript critically for important intellectual content and gave final approval of the version to be submitted.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphar.2020. 01063/full#supplementary-material The microRNAs (miRNAs), targeting each gene, were predicted with application of Tarbase, or whenever written with microT-CDS algorithm.
FIGURE 7 | Proposed model of angiogenic shift in retinal endothelial cells exposed to chemical hypoxia. CoCl 2 -induced hypoxia leads to the stabilization of HIF-1a, with the subsequent translocation into the nucleus and transcription of hypoxia-related genes.