The microRNA-7-mediated reduction in EPAC-1 contributes to vascular endothelial permeability and eNOS uncoupling in murine experimental retinopathy

Aims To investigate the consequences of oxidative stress and hypoxia on EPAC-1 expression during retinopathy. Methods Oxygen-induced retinopathy was induced in mice and EPAC-1 expression investigated by immunofluorescence. In silico analyses were used to identify a link between EPAC-1 expression and microRNA-7-5p in endothelial cells and confirmed by western blot analyses on cells expressing microRNA-7-5p. In vitro, endothelial cells were either incubated at 2% oxygen or transfected with microRNA-7-5p, and the effects of these treatments on EPAC-1 expression, endothelial hyperpermeability and NO production were assessed. In the Ins2Akita mouse model, levels of EPAC-1 expression as well as microRNA-7-5p were assessed by qPCR. Endothelial nitric oxide synthase was assessed by immunoblotting in the Ins2Akita model. Results Hypoxia induces the expression of microRNA-7-5p that translationally inhibits the expression of EPAC-1 in endothelial cells, resulting in hyperpermeability and the loss of eNOS activity. Activation of EPAC-1 by the cAMP analogue 8-pCPT-2′-O-Me-cAMP reduced the sensitivity of EPAC-1 to oxidative stress and restored the endothelial permeability to baseline levels. Additionally, 8-pCPT-2′-O-Me-cAMP rescued eNOS activity and NO production. In mouse models of retinopathy, i.e., oxygen-induced retinopathy and the spontaneous diabetic heterozygous Ins2Akita mice, EPAC-1 levels are decreased which is associated with an increase in microRNA-7-5p expression and reduced eNOS activity. Conclusion/Interpretation In retinopathy, EPAC-1 expression is decreased in a microRNA-7-mediated manner, contributing to endothelial dysfunction. Pharmacological activation of remnant EPAC-1 rescues endothelial function. Collectively, these data indicate that EPAC-1 resembles an efficacious and druggable target molecule for the amelioration of (diabetic) retinopathy. Electronic supplementary material The online version of this article (doi:10.1007/s00592-017-0985-y) contains supplementary material, which is available to authorized users.


Introduction
The vascular endothelium exhibits multiple structural and functional abnormalities in response to hypoxia that may contribute to the pathogenesis of several vascular diseases, including (diabetic) retinopathy [1]. Hypoxia is associated with an increment in oxidative stress [2] and the disruption of endothelial adhesion molecules [3,4], resulting in increased endothelial permeability [5] and impairment of vasodilation [6,7].
Although the downstream consequences of EPAC activation on endothelial function receive increasing attention [20], factors regulating the expression of EPAC during pathology remain elusive. In retinopathy hypoxia might contribute to EPAC dysregulation. We recently uncovered that EPAC-1 is targeted by microRNA-7 in human pulmonary smooth muscle cells, and microRNA-7 expression is associated with increased oxidative stress levels [21]. Therefore, we hypothesized that microRNA-7 might induce EPAC-1 deregulation during retinal hypoxia or in diabetic conditions.
Here, we describe that EPAC-1 expression is inhibited by hypoxia in vivo in the oxygen-induced retinopathy mouse model [22,23] and in endothelial cell cultures exposed to hypoxia. Furthermore, we show that the reduction in EPAC-1 expression is associated with the hypoxia-induced expression of microRNA-7, resulting in translational repression. Activation of the remnant EPAC-1 in endothelial cells counteracts hypoxia-induced endothelial hyperpermeability and reverses the NO/ROS imbalance through eNOS activation. Moreover, in the Ins2 Akita mouse model for diabetic retinopathy (DR), EPAC-1 expression is vastly reduced, which coincides with a marked increase in microRNA-7 expression. These data indicate that EPAC-1 is a pivotal regulator of endothelial function in (diabetic) microangiopathies involving endothelial dysfunction associated with hypoxia, and might serve as promising therapeutic targets to ameliorate these conditions.

Animals, materials and methods
Animals and ethical approval C57BL/6J mice and spontaneous diabetic heterozygous Ins2Akita ?/mice (Jackson Laboratory, Charles River, Sulzfeld, Germany) were used throughout the study. Agematched non-diabetic homozygous Ins2Akita -/mice served as control. All experimental procedures were performed according to the guidelines of the statement for animal experimentation issued by the Association for Research in Vision and Opthalmology and were approved by the local board for animal care (Medical Faculty Mannheim, Germany).

Oxygen-induced retinopathy mouse model
Oxygen-induced retinopathy was induced in C57BL/6J mice as described previously [22,23]. In short, newborn mice (n = 6) at postnatal day (p) 7 were exposed to hyperoxia (75% oxygen) in an incubation chamber (Stuart Scientific, Redhill, UK) with their nursing mothers for 5 days and then returned to ambient air, creating a relative hypoxic environment. Control mice (n = 6) were kept at ambient air and used as a control group. Mice were killed at p12 (i.e., 6 h of relative hypoxia) and p13 (i.e., 24 h of relative hypoxia), and the retinas were isolated as described previously [23].

Retinopathy in Ins2Akita mice
After 6 months of diabetes, mice (n = 6/group) were killed and the retinas were isolated as described previously [23].
Parallel cultures were maintained under normoxic (21% oxygen tension) and hypoxic conditions (2% oxygen tension) for 48 h prior to experiments. To establish hypoxia, cell culture medium was deoxygenated by bubbling gaseous N 2 through the medium at room temperature for 30 min. Cells were maintained in an hypoxic cell culture incubator at 37°C containing 2% O 2 , 93% N 2 , and 5% CO 2 .

Immunofluorescence
Endothelial cells were cultured in eight-well Lab-Tek Ò chamber slides (Nunc, IL, USA) until 80% confluence was reached. Cells were incubated under normoxic or hypoxic conditions for 24 h. Samples were fixed in 2% paraformaldehyde in PBS for 20 min and permeabilized with 0.3% Triton X-100 for 10 min. Blocking of unspecific antibody activity was performed using in 2% BSA. Samples were incubated with rabbit polyclonal antibodies to human EPAC-1 (Abcam, #ab21236, 1:100) and eNOS (BD #61098, 1:200) diluted in 2%BSA/PBS overnight at 4°C. Samples were washed and incubated with Alexa Fluor Ò 594-conjugated goat anti-rabbit IgG antibodies (Molecular Probes, Invitrogen, OR, USA) diluted 1:500 in DAPI/PBS for 1 h at room temperature. Samples were mounted in Citifluor AF1 (Agar Scientific, UK), and cells were examined using a Leica TCS SP8 (Leica Microsystems, Germany) laser scanning fluorescence confocal microscope using a 63x/1.40 oil objective.

Nitrite and ROS measurements
Nitrite levels in the cell culture medium, a stabile indicator of NO production, were quantified using the Measure-iT TM High-Sensitivity Nitrite Assay Kit (Molecular Probes, Eugene, OR) according to the manufacturer's protocol. Obtained nitrite concentrations were normalized against the total amount of cellular protein using the DC protein.
Reactive oxygen levels were determined by incubating the cells with 50 lM 2 0 ,7 0 -dichlorofluorescin diacetate (DCFH-DA, Sigma-Aldrich, St. Louis, MO) for 10 min in dark. Cells were dissociated using Accutase TM solution (PAA Laboratories, Austria), pelleted by centrifugation, and suspended in PBS. The generation of intracellular ROS was determined by flow cytometry on the FACSCalibur and WinList version 6.0 software (both BD Biosciences, CA, USA).

MicroRNA in situ hybridization
Double DIG-labeled MicroRNA-7-5p and scrambled control probes, microRNA ISH buffers, and reagents were obtained Exiqon (Vedbaek, Denmark) and used according to the manufacturers' protocol. Hybridization was performed for 16 h at 44°C in a humidified chamber.

Data presentation and statistical analysis
Data is expressed as mean ± SEM. Significant differences between two means were determined by Mann-Whitney two-tailed U test for unpaired data or by one-way analysis of variance (ANOVA) followed by Dunnett's post hoc test, where appropriate. p values \0.05 were considered statistically significant.

Hypoxia decreases EPAC-1 expression
Oxygen-induced retinopathy is associated with a marked decrease in EPAC-1 expression (Fig. 1a). After 6 h and 24 h of relative hypoxia, retinal EPAC-1 gene transcript levels were reduced (2.0-and 1.9-fold, respectively, p \ 0.05, Fig. 1b Conversely, miR-7 expression was increased in the retina of OIR-mice (2.7-and 3.2-fold, p \ 0.01) compared to normoxic control mice (Fig. 1c). The effect of hypoxia on EPAC-1 expression was confirmed in endothelial cell cultures exposed to 2% (hypoxia) or 20% (normoxia) oxygen, where EPAC-1 protein expression decreased (2.4-fold, p \ 0.01, Fig. 1d, e) when cells were exposed to hypoxia for 24 h. These data indicate that the expression of EPAC-1 is oxygen sensitive and its expression is decreased during hypoxic stress.
cAMP signaling counteracts hypoxia-induced endothelial oxidative stress Endothelial NOS protein expression by cells exposed to hypoxia was reduced *40% (p \ 0.05, Fig. 4a, b) compared to normoxic controls, which resulted in decreased NO production (1.4-fold, p \ 0.05, Fig. 4c) and the increased generation of reactive oxygen species (ROS) (1.6-fold, p \ 0.01, Fig. 4d). The addition of miR-7 mimics to endothelial cells did not cause a reduction in eNOS expression level (Fig. 4e). However, the addition of miR-7 mimics to endothelial cells did imitate the hypoxia-induced loss of eNOS activity as indicated by a reduction in eNOS phosphorylation at serine 1177 (p-eNOS/eNOS ratio; 1.7-fold decrease, p \ 0.01; Fig. 4e).
Besides EPAC-1 [21], miR-7 targets a number of additional gene transcripts (Suppl. Table 1). We analyzed the expression of 23 reported miR-7-5p target genes in endothelial cells that express miR-7-5p (Suppl. Table 2) and in retinal isolations from diabetic Ins2Akita mice (Suppl. Table 4) and found no other transcript which was decreased in both conditions.

Discussion
Here, we show that hypoxia induces the expression of miR-7 by endothelial cells in vitro and in vivo, reducing the expression of EPAC-1. The hypoxia-induced reduction in EPAC-1 levels results in endothelial hyperpermeability and a NO/ROS imbalance and is associated with the development of (diabetic) retinopathy. Activation of EPAC-1 by forskolin or the cAMP analogue 8-pCPT reduces the sensitivity of EPAC-1 to oxidative stress and restores the endothelial permeability barrier and rescues NO production by eNOS. These data suggest that EPAC-1 is an appropriate drug target for the treatment of endothelial dysfunction during (diabetic) retinopathy.
(Diabetic) Retinopathy is characterized by hypoxia-induced vascular dysfunction, resulting in the degradation of the blood-retinal barrier (BRB) and concomitant macular edema formation. Herein, hypoxia induces the loss of endothelial cell-cell junctions and oxidative stress [24,25]. Although cAMP signaling is commonly known to regulate  endothelial permeability and NO production [26,27], little is known on the role of cAMP signaling in retinopathy. Yet, agonist to the b-adrenergic system [28,29] and taurine [30] prevent retinal endothelial hyperpermeability in part through the activation of cAMP signaling. Considering these antecedents, we investigated cAMP signaling in the oxygen-induced retinopathy model [22] and found a marked decrease in the expression of the cAMP signaling intermediate EPAC-1. Concurrently, exposing endothelial cells to a hypoxia challenge in vitro similarly reduced EPAC-1 expression levels, suggesting that the loss of EPAC-1 might contribute to the hypoxiainduced retinopathy. Indeed, adenosine reduces inflammatory retinopathy by activating EPAC1-1 [31].
We had previously found an association between EPAC-1 and miR-7 expression levels in airway smooth muscle cells [21]. MicroRNAs are endogenous translational repressors of gene expression. Hence, we investigated if hypoxia could affect EPAC-1 expression through miR-7. Indeed, hypoxia induced the expression of miR-7 in endothelial cells and transfection of endothelial cells with miR-7 mimics reduced EPAC-1 expression by *50%.
The integrity of the BRB is highly dependent on endothelial adherence junctions that consist of VE-cadherin and associated catenins. Hypoxia reduces the presence of VE-cadherin in the endothelial cell-cell junctions [32], resulting in endothelial hyperpermeability. This change in permeability might be derived from the hypoxiainduced reduction in EPAC-1 expression, as EPAC-1 is pivotal in maintaining VE-cadherin at the cell-cell junction through the activation of Rac [10,26]. Corroboratively, inhibition of EPAC-1 activity with the small molecule ESI-09 or miR-7-5p mimics increases the endothelial permeability. Interestingly, hypoxia-mediated endothelial hyperpermeability was associated with a decreased in VEcadherin expression, whereas hyperpermeability induced by miR-7-5p mimics was not. As both models are characterized by a similar reduction in EPAC-1, these data suggest that hypoxia-driven hyperpermeability involves at least one other cascade that results in the degradation of e Association between the EPAC-1 and miRNA-7-5p expression levels in non-diabetic and diabetic Ins2Akita mice. f Levels of phosph-eNOS (Ser1177) and eNOS in non-diabetic and diabetic Ins2Akita mice VE-cadherin. Activation of the remnant EPAC-1 by forskolin or the cAMP analogue 8-pCPT antagonizes the hypoxia or miR-7-induced endothelial hyperpermeability. These data are corroborated by earlier findings of Aslam et al. [32], who describe the restoration of VE-cadherin at the cell-cell junction by 8-pCPT under hypoxic conditions.
As our data implies that miR-7-5p underlies hyperpermeability in oxygen-induced retinopathy and eNOS uncoupling in diabetic retinopathy, inhibition of miR-7-5p, or EPAC-1 activation seem a promising approach to restore retinopathy. Unfortunately, the current methodology to assess permeability in vivo, i.e., Evans Blue dye leakage, or FITC-Dextran leakage, is too insensitive to provide reproducible results on the relative permeability in Ins2Akita. The development of novel technologies, such as scanning laser ophthalmoscopy or optical coherence tomography [33] may provide better resolution in the near future. Alternatively, the streptozotocin-induced rat model for retinopathy, which produced a high retinal hyperpermeability, could be used to investigate the efficacy of EPAC activating drugs.
Additionally, it would be of interest to investigate if EPAC-1 activation can rescue the VEGFa-induced endothelial hyperpermeability, as this is not only associated to (diabetic) retinopathy, but also with endothelial hyperpermeability in tumors.
Besides endothelial hyperpermeability, in retinopathy, hypoxia contributes to endothelial oxidative stress orchestrated by eNOS [34]. Here, we found a marked reduction in eNOS expression and phosphorylation during hypoxia, which coincided with decreased NO levels and elevated ROS production. Interestingly, the reduction in eNOS mRNA stability and eNOS activity are dependent on the activity of Rho-kinase [35]. It is conceivable that the hypoxia-induced suppression of EPAC-1 expression and activity would concomitantly increase Rho-kinase activity, which would in turn cause the observed reduction in eNOS expression and activation. Besides, the observed reduction in eNOS phosphorylation at Ser1177 might result in eNOS uncoupling and the concomitant generation of ROS. Indeed, in the present study we observe decreased eNOS activity under hypoxic conditions, reflected by reduced nitrite formation, and increased ROS production. Moreover, endothelial cells that were transfected with miR-7-5p mimics showed decreased eNOS activity and eNOS activity is decreased in the retinae of diabetic Ins2Akita mice. Activation of EPAC-1 in endothelial cells that received a hypoxic challenge using forkolin or 8-pCPT completely rescued this phenotype and restored NO production and inhibited the formation of ROS.
A potential limitation of our study is the use of umbilical vein endothelial cells, which are macrovascular cells. Although we cannot fully exclude that retinal microvasculature endothelial cells would behave different with respect to miR-7-5p or EPAC-1 stimulation, in preliminary experiments we have found no difference between the umbilical vein endothelial cells and dermal microvascular endothelial cells (data not shown).
In summary (Fig. 6), here we report that the hypoxiainduced reduction in EPAC-1 expression and activity contributes to the generation of retinopathy through the disruption of the endothelial barrier in the OIR model, or by eNOS uncoupling (disturbing the NO/ROS balance) in the Ins2Akita model. The reduction in EPAC-1 expression is in part due to the hypoxia-induced expression of miR-7. Pharmacological activation of EPAC-1 by forskolin or 8-pCPT antagonizes the hypoxia-induced endothelial dysfunction. Therefore, EPAC-1 resembles an efficacious and druggable target molecule for the amelioration of (diabetic) retinopathy. Fig. 6 Hypoxia-mediated repression of EPAC-1 by microRNA-7 in retinopathy. Schematic representation of the study outcomes. Hypoxia during retinopathy increases the expression of microRNA-7, which in turn reduces the protein availability of EPAC-1. The loss of EPAC-1 in endothelial cells causes endothelial junctional instability and concurrently hyperpermeability, as well as the loss of eNOS expression and eNOS activity, resulting in oxidative stress. Combined, hyperpermeability and oxidative stress might further reduce the oxygen transport creating a feed-forward mechanism that aggravates retinopathy. Stimulators of cAMP signaling and EPAC-1 efficiently activate the remnant EPAC-1 protein, which antagonizes the hypoxiainduced damage. Therefore, EPAC-1 is an appropriate drugable target for the treatment of endothelial dysfunction during (diabetic) retinopathy Diabetic Microvascular Complications (GRK1874/DIAMICOM). GK is supported by an Innovational Research Incentive (VENI #916.11.022) from the Netherlands Organization for Health Research and Development (ZonMW).
Author contributions VGM, JF, LJ, and GK designed and performed experiments and were involved in the acquisition of data, analysis and interpretation of data, and the statistical analysis of the data. VG, MCT, HPH, MS, and GK were involved in drafting the manuscript, and critical revision of the manuscript for important intellectual content. GK had overall supervision on the research. All authors approved the final version to be published.

Compliance with ethical standards
Conflict of interest The authors declare no conflicts of interest.
Ethical standard All animal experimental procedures were performed according to the guidelines of the statement for animal experimentation issued by the Association for Research in Vision and Opthalmology and were approved by the local board for animal care (Medical Faculty Mannheim, Germany).
Human and animal rights disclosure This article does not contain any studies with human subjects performed by the any of the authors.

Informed consent None.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://crea tivecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.