microRNA-182 Mediates Sirt1-Induced Diabetic Corneal Nerve Regeneration

Patients with diabetes exhibit various types of diabetic keratopathy, resulting in irreversible visual impairment (1,2). The cornea receives a dense sensory innervation from the trigeminal ganglion (TG) (3), and corneal nerve fibers are sensitive to low-intensity mechanical, chemical, and thermal stimulation (4). The corneal nerve supports corneal epithelial nerve function (5) and plays an important role in corneal epithelial wound healing through the interactions with TG sensory neurons (6). Decreased corneal sensitivity and neuronal abnormalities in patients with diabetes were thought to be the main cause of diabetic keratopathy (7).

Silent mating type information regulation 2 homolog 1 (Sirt1) is an emerging focus in neuroprotection in both the central nervous system (8,9) and peripheral nervous system (10–12). In our previous study, overexpression of Sirt1 promoted epithelial wound healing in diabetic corneas (13,14). However, the underlying mechanism by which Sirt1 induces diabetic corneal nerve regeneration remains elusive.

microRNAs (miRNAs) are small, noncoding RNA molecules found in multicellular eukaryotes and play essential regulatory roles in translational repression (15). Several wound healing–related miRNAs, such as miR-146a, were identified in the corneas of people with diabetes (14,16,17). Abnormal miR-146a upregulation may be an important mechanism of delayed wound healing in the diabetic cornea (17,18).

Sirt1 is a direct target of miR-204-5p in regulating the epithelial cell cycle in the corneas of persons with diabetes, suggesting that Sirt1 is involved in the miRNA–mRNA regulatory network (14), whereas the downstream miRNA targets of Sirt1 in corneal nerve regeneration in diabetes remain unclear.

miR-182 is a sensory organ–specific miRNA (19) and was implicated in diabetic retinopathy, which is damage to the retina (the transparent, light-sensitive structure at the back of the eye) as a result of diabetes. miR-182 was also found to be associated with oxidative stress (20,21) and has developed with selective axonopathy in sensory neurons (22). Here we used the adenoviral system to activate Sirt1 in TG tissue through subconjunctival injection. Using miRNA screening, we revealed that the expression of miR-182 was upregulated by Sirt1 as a direct downstream target. Using in silico prediction and a luciferase reporter assay, we found NADPH oxidase 4 (NOX4), which belongs to the NOX family—a major source of reactive oxygen species (ROS) in peripheral sensory neurons (23)—may be one miR-182 target gene in corneal nerve regeneration in diabetes.

The animal experiments were approved by the Institutional Animal Care and Use Committee, Shandong Eye Institute (Qingdao, Shandong, China). The procedures using and the handling of the animals during this study conformed to the guidelines of the Association for Research and Vision and Ophthalmology statement for the use of animals in ophthalmic and vision research. In this study we used BKS.Cg-m^+/+Leprdb/J (db/db) mice as the model of type 2 diabetes. db/db and nondiabetic control db/+ mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Glucose concentrations in blood from the tail vein were determined using a commercial glucometer (Bayer Diabetes Care, Elkhart, IN). A blood glucose concentration >15 mmol/L was considered to be indicative of diabetes. Experiments on mice were performed more than 24 weeks after the onset of diabetes based on a previous report (24). The blood glucose and body weight values of db/db and nondiabetic control db/+ mice are shown in Table 1.

The in vivo corneal injury model was established as previously described (25). The entire corneal epithelium, including the limbal region (marked using a 3-mm trephine), was scraped with an AlgerBrush II corneal rust ring remover (Alger Equipment Co., Lago Vista, TX), and the right eye was wounded in each animal. Fluoro-Gold (2% in saline; Fluorochrome, Denver, CO) was used as a retrograde tracer to identify neurons in the TG by detecting the expression of miR-182 after subconjunctival injection with miR-182 agomir (RiboBio, Guangzhou, China). The mice were injected with an agomir negative-treated control (NTC), miR-182 agomir, adenovirus (Ad)–expressing Sirt1 (Hanbio, Shanghai, China), or Ad-expressing Sirt1 small interfering RNA (siRNA). Ad-expressing green fluorescent protein (GFP) served as a control. Agomir NTC (2.5 nmol/L) or miR-182 agomir with 2 µL Fluoro-Gold was injected into the subconjunctival site of the right eye on the same day as the corneal epithelium was injured. This experiment was performed at least twice. All of the mice were assessed by using 3 μL 0.25% fluorescein sodium and were photographed at 0, 24, 48, and 72 h under a dissecting microscope (Canon, Tokyo, Japan) and a tungsten light source with a cobalt blue filter (Welch Allyn, Inc., Skaneateles Falls, NY). The photographs were analyzed to quantify the area of the epithelial wound using Adobe Photoshop software (Adobe Systems, Mountain View, CA).

Ophthalmic branches of the trigeminal nerves were collected from db/db mice and nondiabetic db/+ mice, and primary TG neuronal cells were cultured as described previously (17). Culturing neurons from diabetic animals were treated with 5 mmol/L d-glucose (considered normal glucose [NG]) and 25 mmol/L d-glucose (considered high glucose [HG]). The osmotic pressure of the NG medium was adjusted to the same level of the HG medium by adding 20 mmol/L l-glucose. In addition, TG cells were treated with 1 μmol/L SRT1720 (Selleck Chemicals, Houston, TX) or 10 μmol/L resveratrol (RSV) (Sigma-Aldrich, St. Louis, MO) in HG medium. Cells cultured in NG medium were treated with these chemicals for use as a control.

To detect Sirt1 mRNA expression, total RNA was isolated from murine TG using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and transcribed using the SuperScript Reverse Transcriptase (Invitrogen, Carlsbad, CA). Primer sequences are listed in Supplementary Table 1. Data were normalized using the murine ribosomal protein L5 housekeeping gene and analyzed by the 2^−ΔCt values of the normalized data.

To analyze miRNA expression profiling, 24 TG samples were harvested from 12 wild-type C57BL/6 mice aged 6–8 weeks after 3 days of subconjunctival injection with Ad-expressing Sirt1 or GFP (serving as a control). These samples were pooled into six groups and subjected to the miRNA microarray assay (Ad-Sirt1 infection groups 1–3 and Ad-GFP infection groups 1–3, each of which comprised four TGs from two mice). The array was analyzed—including labeling, hybridization, scanning, normalization, and data analysis—by CloudSeq Biotech Inc., Shanghai, China, using a miRCURY LNA microRNA Array Kit version 16.0 (Exiqon, Vedbaek, Denmark). The specific primers for detecting miR-182 expression were designed and synthesized by RiboBio (Guangzhou, China) using the SYBR Green Quantitative PCR Protocol. U6 was used as an endogenous control to normalize the expression level of miR-182. All reactions were performed in triplicate. The TG infected with Ad-GFP was used as a calibrator, and the data were presented as the fold change relative to the calibrator.

Total RNA was isolated and cDNA was synthesized as described previously (13). The real-time PCR array and data analyses were performed using a RT² Profiler PCR Array (Mouse Oxidative Stress and Antioxidant Defense PCR Array, PAMM-065A; SABiosciences, Frederick, MD).

A 2,000-bp DNA fragment containing the putative promoter region of mouse miR-182 was amplified by PCR using mice genomic DNA as a template. This was cloned into the pGL3 Luciferase Reporter Vector (Promega, Madison, WI). The primers used for the PCR were as follows:

• mmu-miR-182-p-F: GAATTGGTACCCCTTGGTGGAGGCTTTGCTGAGACC

• mmu-miR-182-p-R: GGAATACGCGTCAGCAGCCAGACCAGTAAGCCTATG

miR-182-Luc reporter plasmid (200 ng) containing the firefly luciferase gene and pRL-TK (50 ng) (Promega) were cotransfected with Sirt1 or the control vector (800 ng) into human HEK 293T cells. Luciferase activities were measured with the Dual-Luciferase Reporter Assay System (Promega) and Centro LB 960 detection system (Berthold, Germany).

Mouse NOX4 3′ untranslated region (UTR) containing the putative target site for miR-182 was amplified from genomic DNA by PCR amplification and inserted into the pmiR-REPORT (RiboBio). The mutant reporter plasmid at the miR-182 complementarity site (TGCCAAA to ACGGTTT) was generated by the QuikChange II Site-Directed Mutagenesis Kit (Stratagene California, La Jolla, CA). SH-SY5Y neuroblastoma cells were transiently transfected with wild-type or mutant reporter plasmid and miRNA agomir/mimic using Lipofectamine 2000 (Invitrogen).

Corneas from db/db and control db/+ mice were fixed in 4% paraformaldehyde and stained with an Anti-β III Tubulin antibody (ab7751; Abcam, Cambridge, MA). Representative images were taken with a Nikon Eclipse Ni-U fluorescence microscope (Nikon, Tokyo, Japan). Four pie-shaped quadrants of corneal specimens were examined: superior (10.30–1.30 o’clock), inferior (4.30–7.30 o’clock), nasal (right eye) and temporal (left eye) (1.30–4.30 o’clock), and temporal (right eye) and nasal (left eye) (7.30–10.30 o’clock). The corneal nerves entering each quadrant were counted using ImageJ software (National Institutes of Health, Bethesda, MD).

Cornea and TG samples were frozen in Tissue-Tek optimum cutting temperature compound (Sakura Finetechnical Co., Tokyo, Japan). Frozen sections were fixed in 4% paraformaldehyde. The samples were stained with primary antibodies (Supplementary Table 2) and subsequently incubated with fluorescein-conjugated secondary antibodies. All stained samples were observed under a microscope as described previously (14).

Samples were homogenized in 100 μL radioimmunoprecipitation assay buffer supplemented with a proteinase inhibitor cocktail. The homogenates, which contained 10–20 μg protein, were then separated in 15% SDS–polyacrylamide gel and transferred to polyvinyl difluoride membranes. The blots were probed with primary antibodies as described previously (14).

The presence of miR-182 in trigeminal sensory neurons was tested using SmartFlares (Merck Millipore, Billerica, MA). SmartFlares (1.5 µL) was injected below the conjunctiva in the right eye after 24 h of miR-182 agomir or agomir NTC treatment. After another 24 h, the TG samples were collected to detect miR-182 expression. We also double stained Neurofilament 200 (NF200; Sigma-Aldrich, Shanghai, China) and miR-182 with SmartFlare RNA detection probes according to the immunocytochemistry protocol for murine primary neuronal cells.

The results are presented as the means ± SEMs. The statistical analyses were performed using an unpaired t test or one-way ANOVA by comparing the groups using the Student-Newman-Keuls test. The least significant difference procedure was performed with GraphPad Prism software 5.0 (GraphPad Software, Inc., San Diego, CA). A P value <0.05 was considered statistically significant.

We first examined the blood glucose concentrations and body weights of db/db and nondiabetic control db/+ mice (Table 1). As expected, db/db mice had significantly higher concentrations of blood glucose and higher body weights than the control mice (n = 20 per group). The expression of Sirt1 was significantly downregulated at both the mRNA and protein levels in the diabetic TG neurons compared with the control neurons (n = 10 per group) (Fig. 1A-i and B-i).

The mRNA and protein expression levels of Sirt1 were also reduced in TG cells with HG treatment compared with the group receiving NG treatment (n = 10 per group) (Fig. 1A-ii and B-ii). Coimmunostaining results showed that Sirt1 was partially colabeled with NF200-expressing neuronal cell bodies of TG sensory neurons (Fig. 1C). There were significantly fewer Sirt1-positive cells in the diabetic mice, which confirmed the downregulation of Sirt1 in db/db mice.

In view of the downregulation of Sirt1 in TG sensory neurons of db/db mice, we hypothesized that Sirt1 might be essential for sensory neuron protection. Therefore we used the adenoviral system to overexpress Sirt1 in TG tissue through subconjunctival injection, and Sirt1-overexpressing TG tissues were collected to analyze the miRNA expression profile. Differentially expressed miRNAs matching these stringent criteria are listed in Supplementary Table 3. As shown in Fig. 2A and Supplementary Table 3, several miRNAs (e.g., miR-1a-3p, miR-5129-5p, miR-470-5p, miR-182) were deregulated in Sirt1-activated TG tissues (n = 3 per group). Among these differentially expressed genes, miR-182 was significantly upregulated (2.27-fold; P = 0.00018) in the TG tissue infected with Ad-Sirt1 compared with that infected with Ad-GFP.

Further, we studied the role of miR-182 in TG sensory neurons because 1) miR-182 had the minimum P value; 2) miR-182 is a sensory organ–specific miRNA (19); and 3) miR-182 has been implicated in diabetic retinopathy (26), which occurs in parallel with nerve fiber alterations of the subbasal nerve plexus of diabetic corneas (27).

As we expected, the mRNA expression of miR-182 was decreased in the diabetic TG samples compared with that in control mice (n = 10 per group) (Fig. 2B). We also found miR-182 was downregulated in the primary HG-treated TG cells compared with NG-treated TG cells (n = 4 per group) (Fig. 2C). We treated the primary cultured TG cells with two different Sirt1 activators, SRT1720 and RSV, which resulted in efficient upregulation of Sirt1 and miR-182 expression (n = 4 per group) (Supplementary Fig. 1 and Fig. 2D). In addition, the expression of miR-182 was increased in TG cells infected with an Ad-expressing Sirt1 and decreased in TG cells infected with an Ad-expressing Sirt siRNA (n = 4 per group) (Fig. 2F). We revealed that the relative luciferase activity of the miR-182 promoter was enhanced by cotransfection with Sirt1 in a dose-dependent manner (n = 4 per group) (Fig. 2G). We also performed chromatin immunoprecipitation assay to evaluate Sirt1 binding activity on the miR-182 promoter in chromatin prepared from the TG samples. Computational methods were used to predict the binding regions. The results in Fig. 2H demonstrate that Sirt1 binds to this predicted promoter region. Taken together, these results indicate that miR-182 is upregulated by Sirt1 in TG sensory neurons.

Primary cultured TG sensory neurons were transfected with miR-182 agomir (200 μmol/L) to achieve robust and predictable miR-182 overexpression in TG neurons derived from both control and db/db mice (Fig. 3A). After 3 days of transfection with miR-182 agomir, diabetic TG neurons exhibited more and longer neuritis than those transfected with miRNA agomir NTC in sensory neurons from db/db and control mice (Fig. 3B). Statistical analysis showed that the total neurite length increased by 2.98 ± 0.029–fold in miR-182 agomir–transfected TG sensory neurons (n = 4 per group) (Fig. 3C). These data demonstrate that miR-182 promotes the growth of neurites from diabetic trigeminal sensory neurons in vitro.

After 3 days of subconjunctival injections, miR-182 was expressed at 5.12 ± 0.20–fold higher levels in TG sensory neurons versus neurons in the NTC-treated mice (n = 10 per group) (Fig. 4A). Representative images of TG colabeled with miR-182 and Fluoro-Gold in diabetic db/db mice are presented in Fig. 4B; these show that more Fluoro-Gold-labeled cell bodies were observed in the medial region of the ipsilateral TG compared with that in NTC-treated mice. Moreover, miR-182 was observed to be costained with NF200-positive cells (Fig. 4C). We also got the same results in control db/+ mice (data not shown). These results suggest that miR-182 is overexpressed in a large proportion of trigeminal sensory neurons.

Quantitative PCR analysis showed that miR-182 was expressed at a 71.24 ± 0.14–fold higher level in the corneas of the db/db mice with a subconjunctival injection of miR-182 agomir versus the NTC-treated mice (n = 10 per group) (Fig. 5A). Punctate fluorescence staining showed a significant difference with regard to the corneal epithelial healing rate within 48 h of the corneal epithelium scrape between control and miR-182 agomir–injected db/db mice (Fig. 5B-i). The size of the corneal epithelium defect in the miR-182 agomir–injected diabetic mice (9.73 ± 0.43%; n = 10) was remarkably improved compared with that of the diabetic mice (54.78 ± 0.49%; n = 10) and the NTC-treated diabetic mice (52.72 ± 0.45%; n = 10), which reached a level comparable to that of the control mice (18.59 ± 0.36%; n = 10) (Fig. 5B-ii). Representative images of innervation of the subbasal layer are presented in Fig. 5C-i. At day 28 after abrasion, the diabetic mice that received miR-182 agomir treatment displayed a significant increase in corneal nerve density in both the central and peripheral zones compared with the agomir NTC-treated diabetic mice (n = 6 per group) (Fig. 5C-ii and iii). Corneal nerve regeneration is associated with equal functional recovery. At day 28 after corneal epithelial debridement, the corneal sensation in the diabetic mice receiving the miR-182 agomir injection increased significantly compared with that in the NTC-treated mice and returned almost to the level of the control db/+ mice (n = 10 per group) (Fig. 5C-iv).

To study the molecular mechanism of miR-182 in diabetic TG sensory neurons and diabetic keratopathy, we used in silico prediction and a luciferase reporter assay to search for potential miR-182 target genes. Using in silico prediction, we followed three different miRNA target prediction algorithms: PicTar, miRanda, and TargetScan. Three genes were implicated in oxidative stress, including membrane protein, palmitoylated 1 (MPP1); protein phosphatase 1, regulatory (inhibitor) subunit 13B (PPP1R13B); and NOX4. To investigate whether miR-182 targets these 3 genes, we performed a luciferase reporter assay in SH-SY5Y neuroblastoma cells. In view of the upregulation of NOX4 in TG neurons from diabetic mice, we found the possibility of a direct link only between miR-182 and NOX4. A significant decrease of the relative luciferase activity was observed when pmiR-RB-REPORT-NOX4–3′ UTR was cotransfected with an miR-182 agomir (n = 4 per group) (Fig. 6A). Notably, the mutation of the perfectly miR-182 complementary site in the 3′ UTR of NOX4 (TGCCAAA → ACGGTTT) abolished the suppressive effect of miR-182 through the disruption of the interaction between miR-182 and NOX4.

The NOX4 protein level was decreased in TG cells by miR-182 agomir treatment, but it increased with miR-182 antagomir treatment (n = 4 per group) (Fig. 6B-i and ii). However, miR-182 generated no effect on the expression of NOX2, another NAPDH isoform, with same treatment (Fig. 6B-i and iii). We then detected the effect of miR-182 on the regulation of the NOX4 or NOX2 protein in TG cells under HG conditions in vitro. In contrast to the downregulation of Sirt1 and miR-182 expression, NOX4 protein levels were increased upon HG treatment (Fig. 6C). In addition, the NOX4 expression levels were significantly downregulated after 48 h of miR-182 agomir treatment, whereas NOX2 expression remained at comparable levels (n = 3 per group) (Fig. 6C). Next, diabetic adult sensory neurons from db/db or db/+ mice were treated with miR-182 agomir and cultured in the HG medium. Representative results showed that NOX4 was colocalized with NF200 and that the expression of NOX4 was significantly downregulated in the diabetic TG cells after miR-182 agomir treatment (Fig. 6D).

Mouse Oxidative Stress and Antioxidant Defense RT² PCR arrays were performed to investigate the oxidant-antioxidant imbalance that was involved in the diabetic TG neurons. Of the 84 genes assayed in the scatter plot (Fig. 7A), 24 transcripts were downregulated (Supplementary Table 4) and 5 transcripts were upregulated (Supplementary Table 5). Among these genes, NOX4 was expressed at a 10.11-fold higher level in diabetic TG neurons compared with the control neurons. The expression levels of NOX4 mRNA (n = 10 per group) and protein (n = 3 per group) were significantly upregulated in the diabetic TG neurons compared with the controls (Fig. 7B and C). Representative results show the NOX4-positive cells were significantly increased in the neuronal cells of diabetic mice, with stronger fluorescence upon staining, and NOX4 was colocalized with NF200-positive TG neurons (Fig. 7D), suggesting that NOX4 expression was markedly upregulated in diabetic TG neurons.

An Ad-NOX4, Ad-expressing siRNA-NOX4, or Ad-GFP viral preparation was injected below the conjunctiva on the same day the corneal epithelium was injured after miR-182 agomir or NTC treatment for 24 h. After injury (48 h), the NOX4 knockdown in corneal epithelia with miR-182 agomir efficiently promoted the wound healing process (Fig. 8A), and the wound area in diabetic mice that were administered Ad-expressing siRNA-NOX4 plus miR-182 was most significantly decreased relative to other groups (Fig. 8B). Conversely, NOX4 overexpression remarkably antagonized the promotion effect of miR-182 agomir on corneal wound healing. At day 28 after abrasion, nerve staining analysis revealed that knockdown of NOX4 in corneal epithelia with miR-182 agomir promoted corneal nerve regeneration in diabetic mice (Fig. 8C). The diabetic mice that received Ad-expressing siRNA-NOX4 plus miR-182 agomir treatment displayed a significant increase in corneal nerve density compared with the NTC-treated diabetic mice (n = 6 per group) (Fig. 8D). Importantly, the miR-182–enhanced corneal nerve growth was obviously hindered in diabetic mice treated with Ad-NOX4 with miR-182 (Fig. 8D). After subconjunctival injection, NOX4 was localized in the corneal epithelium (Fig. 8E). Compared with the corneal epithelia of the group infected with Ad-expressing siRNA-NOX4, the corneal epithelia of the Ad-NOX4-infected group receiving miR-182 agomir treatment were characterized by high NOX4 mRNA expression (Fig. 8F). In addition, several factors that have been used as markers for corneal wound healing—phospho-AKT, epidermal growth factor receptor, and Ki67 (28)—were also detected by immunofluorescence. The expression levels of phospho-AKT, epidermal growth factor receptor, and Ki67 were significantly increased in diabetic mice with Ad-expressing siRNA-NOX4 infection and miR-182 treatment (Fig. 8E), further demonstrating that downregulation of NOX4 by either miR-182 targeting or NOX4 siRNA results in an increase in corneal nerve growth.

In this article we report on miR-182 as a Sirt1 downstream effector to protect against peripheral nerve damage in TG neurons and keratopathy in an experimental mouse model of diabetes. We used the adenoviral system to overexpress Sirt1 in TG tissues and identified miR-182 as an upregulated miRNA. Moreover, the promoter activity of miR-182 is enhanced by Sirt1. These results indicate that miR-182 maybe a downstream effector of Sirt1.

Several miRNAs could be used in a potential therapeutic approach for peripheral nerve regeneration, including miR-21 and miR-29b. miR-21 was able to promote axon growth in adult dorsal root ganglion neurons by targeting the Sprouty2 protein (29). miR-29b was reported to exhibit neuronal protective effects in sciatic nerve regeneration (30). In this study miR-182 promoted corneal sensory nerve growth in a chronic diabetic neuropathy mice model. The colocalization of miR-182 and Fluoro-Gold by immunofluorescence indicates the retrograde transport of miR-182 agomir from the cornea to the TG. The responses to miR-182 overexpression mainly occurred in TG sensory neurons; this is supported by the colocalization of miR-182 and NF200 in TG neurons. Furthermore, miR-182 agomir treatment significantly increased the corneal nerve density in diabetes. These results demonstrate that miR-182 significantly increased corneal nerve regeneration in diabetic mice.

Accumulating evidence implies that Sirt1 has a critical neuroprotective effect on nerve injury (31,32). Consistent with this, Ogawa et al. (33) reported on the expression of Sirt1 in the TG of developing mice. Although several miRNAs are reported to be regulated by Sirt1, such as miR-134 in the brain (34) and miR-138 in mammalian axon regeneration (35), the spectrum of Sirt1-regulated miRNAs in diabetic TG sensory neurons remains to be explored in detail. Previous studies proposed that miR-182 was implicated in multiple functional processes, such as oncogenesis and light/dark transition in the retina (36), T helper cell clonal expansion (37), and lipid homeostasis. miR-182 was recently reported to be tightly associated with metastasis of primary sarcomas (18) and ROS-induced premature senescence (38).

Most investigators believe the theory that abnormal oxidative stress mediates changes in diabetic neuropathy and ROS accumulation–induced impaired antioxidant capabilities in response to hyperglycemia; this has been considered as one of the critical pathological mechanisms in diabetic neuropathy (39,40). Consistent with this notion, inhibition of ROS signaling prevents hyperglycemia-induced complications, including diabetic keratopathy (28,41). As targets for diabetic complications (42), NOX4 expression and NOX4-derived ROS production could be significantly induced by diabetes and HG (43,44). These findings suggest that ROS-related NOX4 is likely to be involved in diabetic neuropathy.

miR-182 actually has many targets, and several of these have been reported previously, such as cortactin and Rac1 in amygdala-dependent memory formation (45). That we investigated only one target gene of miR-182 in this study may seem like an oversimplification. We focused on investigating NOX4 as a miR-182 target gene because 1) NOX4 is a major source of ROS in peripheral sensory neurons; 2) NOX4 expression was significantly different between diabetic TG tissues and control tissues based on the PCR array test; and 3) NOX4 was present in the miRNA predictions using three methods. As we expected, NOX4 is directly targeted by miR-182, and it acts as a key downstream target to mediate functions elicited by the Sirt1-miR-182 axis in corneal nerve regeneration in diabetes; this strongly demonstrates that NOX4 is involved in diabetic keratopathy. To understand further how miR-182 targets NOX4, we performed rescue experiments in a corneal injury model with overexpression or knockdown of NOX4 expression. The promotion effect of miR-182 on corneal wound healing and nerve regeneration was counteracted by NOX4 overexpression, whereas downregulating the expression of NOX4 enhanced the functional effect of miR-182.

Acknowledgments. The authors thank Xiaolong Cai, Hanbio Biotechnology Co., Ltd. (Shanghai, China); Meifang Dai, CloudSeq Biotech Inc. (Shanghai, China); Lei Zhang, Hubei University of Technology (Wuhan, China); and Ji Xu, RiboBio Co., Ltd. (Guangzhou, China), for help with preparing the manuscript. The authors acknowledge the editorial assistance of Scribendi Inc., Chatham, Canada.

Funding. This study was supported by the State Key Basic Research (973) Project of China (2012CB722409), the National Natural Science Foundation of China (30901637, 81370990, 81300742), and the Shandong Province Natural Science Foundation (BS2012YY030, BS2013YY013).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.W. interpreted the data and wrote the manuscript. Y.W., X.Z., and X.W. collected and analyzed the data. Y.W. and L.X. obtained funding for the study. X.Z. and Y.D. contributed the animal model. P.C. performed the cell culture and treatment. All authors read and approved the final manuscript. Y.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.