LncRNA MIAT enhances cerebral ischaemia/reperfusion injury in rat model via interacting with EGLN2 and reduces its ubiquitin‐mediated degradation

Abstract Long non‐coding RNA (lncRNA) MIAT (myocardial infarction associated transcript) has been characterized as a functional lncRNA modulating cerebral ischaemic/reperfusion (I/R) injury. However, the underlying mechanisms remain poorly understood. This study explored the functional partners of MIAT in primary rat neurons and their regulation on I/R injury. Sprague‐Dawley rats were used to construct middle cerebral artery occlusion (MCAO) models. Their cerebral cortical neurons were used for in vitro oxygen‐glucose deprivation/reoxygenation (OGD/R) models. Results showed that MIAT interacted with EGLN2 in rat cortical neurons. MIAT overexpression or knockdown did not alter EGLN2 transcription. In contrast, MIAT overexpression increased EGLN2 stability after I/R injury via reducing its ubiquitin‐mediated degradation. EGLN2 was a substrate of MDM2, a ubiquitin E3 ligase. MDM2 interacted with the N‐terminal of EGLN2 and mediated its K48‐linked poly‐ubiquitination, thereby facilitating its proteasomal degradation. MIAT knockdown enhanced the interaction and reduced EGLN2 stability. MIAT overexpression enhanced infarct volume and induced a higher ratio of neuronal apoptosis. EGLN2 knockdown significantly reversed the injury. MIAT overexpression reduced oxidative pentose phosphate pathway flux and increased oxidized/reduced glutathione ratio, the effects of which were abrogated by EGLN2 knockdown. In conclusion, MIAT might act as a stabilizer of EGLN2 via reducing MDM2 mediated K48 poly‐ubiquitination. MIAT‐EGLN2 axis exacerbates I/R injury via altering redox homeostasis in neurons.

is associated with complex physiological alterations (such as oxidative stress, neuroinflammation and excitotoxicity), leading to induced neuronal apoptosis and neurovascular injury. 2 Therefore, a clear understanding of the molecular mechanisms of ischaemic/reperfusion (I/R) injury would provide new opportunities for palliative interventions.
Long non-coding RNAs (lncRNAs) are a class of mRNA-like transcripts than 200 nucleotides in length but lacking protein-coding activity. They regulate many fundamental biological processes and pathophysiological events at transcriptional, post-transcriptional and translational levels. 3 MIAT (myocardial infarction associated transcript), also known as RNCR2 (retinal non-coding RNA 2), is a lncRNA that is involved in various diseases, such as myocardial infarction, diabetic complications, ischaemic stroke and cancers. 4 For example, it binds to and stabilizes NF-E2-related factor 2 (Nrf2), thereby exaggerating high glucose induced renal tubular epithelial injury. 5 It enhances inflammation and oxidative stress in sepsisinduced cardiac injury by sponging miR-330-5p and activating the downstream TRAF6/NF-κB signalling. 6 It promotes hypoxia/ reoxygenation-induced myocardial cell apoptosis by activating Akt/ GSK-3β signalling. 7 Its upregulation is also observed in brain tissues after ischaemic stroke, 8 which promotes neural cell autophagy and apoptosis via reducing ubiquitin-mediated degradation of regulated in development and DNA damage responses 1 (NEDD1). 8 Therefore, it might be a critical lncRNA mediating I/R injury.
Egl-9 Family Hypoxia Inducible Factor 2 (EGLN2, also known as PHD1) is a member of the prolyl hydroxylase domain proteins (PHDs), including PHD1-3. 9 It is an oxygen sensor and a critical regulator of the response to hypoxia. 9 In normoxia, PHDs mediate hydroxylation targets proteins for proteasomal degradation. 9 One previous study showed that the PHD1 exerts a critical neuroprotective role in ischaemic stroke. 10 PHD1 −/− mice had 71% reduction in infarcted size 24 h after permanent middle cerebral artery occlusion (pMCAO), without vascular changes. 10 Cortical neurons from PHD1 −/− mice are less susceptible to I/R injury due to a shifting of glucose oxidation to oxidative pentose phosphate pathway (oxPPP). 10 Therefore, EGLN2 inhibition has been considered as a potential strategy to diminish neuronal damage under the risk of cerebral ischaemia. 10 In the current study, we investigated the functional partner of MIAT in primary rat neurons and identified a physical interaction between MIAT and EGLN2. MIAT increases EGLN2 stability in cortical neurons after I/R injury and modulates oxidized (GSSG) and reduced (GSH) glutathione ratio.

| Prediction of MIAT binding partners
Proteins that might interact with MIAT were predicted using RNAInter (http://www.rna-socie ty.org/rnain ter/home.html). 11 The protein sequence of rat EGLN2 was obtained from Uniprot (Q6AYU4). The possible binding sites between MIAT (NR_111959.1) and EGLN2 protein sequence were also predicted using the PRIdictor module of RNAInter.

| Primary culture of rat cortical neurons
Primary cortical neurons were obtained from newborn Sprague-Dawley rat brain tissue, according to the methods introduced previously. 12 Briefly, the primary cortical neurons were maintained in a culture medium with 97% Neurobasal Medium, 2% B27, 1% penicillin and streptomycin (complete medium). Six days after isolations, the primary neurons were subjected to oxygen-glucose deprivation/ reoxygenation (OGD/R) treatment. The cells were washed twice and were cultured in glucose-free DMEM. Then, the cells were maintained in a tri-gas incubator containing 94% N 2 , 1%O 2 and 5% CO 2 for 2 h at 37°C. Then, the culture medium was replaced by the complete medium, and the cells were further incubated in a normal incubator for 6, 12 or 24 h.

| Middle cerebral artery occlusion (MCAO) model in rats
Male Sprague-Dawley rats were weighted around 220-250 g were purchased from Chengdu Dashuo Biotechnology Co., Ltd (Chengdu, China). MCAO operation was established as described previously. 14 Briefly, rats were anaesthetized by intraperitoneal (IP) injection of pentobarbital sodium (40 mg/kg). Then, a midline neck incision was performed, and the right common carotid artery (CCA), external carotid artery (ECA) and internal carotid artery (ICA) were surgically exposed. The ECA was permanently ligated. A 4-0 monofilament nylon suture with a 0.26 mm diameter rounded tip was aseptically inserted into the right CCA lumen and gently advanced into the ICA for a point 19-20 mm beyond the bifurcation of the CCA. After 90 min of occlusion, reperfusion was conducted by removing the nylon filament, followed by different time intervals of reperfusion (3, 6, 12, 18 and 24 h). The brain tissues obtained from each group were collected for the following experimental procedures.

| qRT-PCR analysis
RNA extraction, cDNA synthesis and qRT-PCR were conducted according to the protocol introduced in one previous study. 15 Gene expression was quantified by calculating fold changes using the formula 2 −ΔΔCT method. GAPDH expression served as an internal control.
The primers used were provided in Table S1.

| RNA pull-down assay
Briefly, MIAT and its antisense RNA were chemically synthesized and were inserted into the sites between KpnI and SacI in pBlue-

| RNA immunoprecipitation (RIP)-qPCR
RIP assays were conducted using the EZ-Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (Merck Millipore), according to the manufacturer's instruction. Briefly, 72 h after AAV-EGLN2 overexpression, primary rat neurons were harvested, wash and lysed using RIPA buffer. Then, whole-cell lysate was incubated with the RIP immunoprecipitation buffer containing protein A/G magnetic beads coated with EGLN2 antibody. Normal rat IgG was used as the control.
The precipitated RNA fraction was isolated and subjected to qRT-PCR analysis of MIAT expression.

| Western blotting
Western blot was performed as described previously. 15 In brief, total proteins were extracted from cell or tissue samples. Then, protein concentrations were determined using the BCA assay (Pierce, Rockford, IL, USA). Samples containing 30 μg protein were loaded to each lane, subjected to 12% SDS-PAGE and transferred onto nitrocellulose membranes. The membranes were incubated consecutively with primary antibodies followed by appropriate HRP-conjugated secondary antibodies. Protein band signals were developed using BeyoECL Star (Chemiluminescence; Beyotime, Wuhan, China) with ChampGel full automatic gel imaging system (Sage Creation Science, Beijing). The primary antibodies and dilutions were provided in Table S1.

| Co-immunoprecipitation (Co-IP)
Co-IP was conducted using Pierce Co-Immunoprecipitation Kit (Thermo Scientific, Waltham, MA, USA) according to the manufacturer's instruction. In brief, primary anti-Flag tag (10 μg) or anti-Myc tag (10 μg) were immobilized to AminoLink Plus coupling resin. Then, primary rat neurons were collected and lysed using IP lysis buffer. The supernatant of cell lysate was collected, precleaned using control agarose resin and then was added to the spin column containing the antibody-coupled resin.
Then, the spin column was gently wobbled overnight at 4°C. Then, the spin column was centrifuged and washed. The immunoprecipitated proteins were eluted and then were subjected to Western blot analysis.

| HE, Nissl and Terminal dexynucleotidyl transferase (TdT)-mediated dUTP nick end labelling (TUNEL) assay
HE and Nissl staining were performed to assess morphological changes after I/R injury. Brain tissues of each group were collected, fixed with paraformaldehyde, dehydrated, dipping in wax, embedded and sectioned. Then, sections were stained with HE or cresyl violet (Nissl staining). Apoptotic neurons in the brain were visualized by TUNEL staining using an One Step TUNEL Apoptosis Assay Kit (Beyotime, Wuhan, China) following the manufacturer's instructions. Immunofluorescent images were captured under an IX83 fluorescent microscope (Olympus).

| Metabolic flux assays
Oxidative pentose phosphate pathway (oxPPP) flux was measured using the isotopic non-stationary gluconate tracer method, as described previously. 17 Oxidized (GSSG) and reduced (GSH) glutathione ratios were measured using LC-MS, as described previously. 18

| Statistical analysis
All experiments were repeated three times independently.
Quantitative data were reported as means ±standard deviation (SD), based on at least three repeats of three independent tests. Unpaired t test with Welch's correction was used to compare the difference between two groups. One-way ANOVA with Sidak's multiple comparisons test was conducted for multiple-group comparison.
p < 0.05 was considered statistically significant.

| MIAT interacts with EGLN2 in rat cortical neurons
To explore the potential functional interactor of MIAT, we gener- or IgG control ( Figure 1H, 1I). Results indicated that MIAT was significantly enriched in the anti-EGLN2 group ( Figure 1I). Cycloheximide chase analysis showed that MIAT knockdown facilitated EGLN2 degradation ( Figure 2F and 2H), while MIAT overexpression slowed the degradation process ( Figure 2G and 2I).

| MDM2 interacts with EGLN2 and promotes its degradation
To explore the underlying mechanisms of EGLN2 degradation, we predicted E3 ubiquitin ligases interacting with EGLN2 using UbiBrowser 19 (http://ubibr owser.ncpsb.org/). Only the candidates with high confidence interaction score (>0.7) were identified, including murine double minute 2 (MDM2), synovial apoptosis inhibitor 1 (SYVN1) and STIP1 Homology And U-Box Containing Protein 1 (STUB1) ( Figure 3A). MDM2 is an E3 ligase involved in cerebral I/R injury (20), with the highest confidence score. IF staining confirmed the co-localization of EGLN2 and MDM2 in primary rat neurons ( Figure 3B). Then, co-IP assay was conducted in primary neurons with enforced overexpression of flag-MDM2 or myc-EGLN2 ( Figure 3C). Results showed that flag-MDM2 could be immunoprecipitated by anti-myc, while myc-EGLN2 could be immunoprecipitated by anti-flag ( Figure 3C).

| MIAT reduces MDM2 mediated K48-linked poly-ubiquitination of EGLN2
MDM2 overexpression significantly increased poly-ubiquitination of EGLN2, the effect of which was weakened by MIAT overexpression ( Figure 4A). To identify the specific type of ubiquitin linkage catalysed by MDM2 on EGLN2, two Ub mutants with only K48 or K63 residue, but all other lysine residues replaced with arginine were generated. Primary rat neurons were coinfected with myc-EGLN2, flag-MDM2 and wild-type or mutant HA-Ub constructs. Co-IP assay showed that MDM2 enhanced the K48 poly-ubiquitination of EGLN2 ( Figure 4B). MIAT knockdown also increased EGLN2 K48 poly-ubiquitination after OGD/R treatment ( Figure 4C). Bioinformatic prediction in UbiBrowser indicated that MDM2 might interact with the N terminal of EGLN2 ( Figure S1). Therefore, truncated EGLN2 constructs with an N-terminal Myc tag ( Figure 4D). When the EGLN2 constructs were co-expressed with flag-tagged MDM2 in primary rat neurons, only the constructs containing the N terminal (FL and N88) could interact with MDM2 ( Figure 4E). OGD/R treatment increased the binding of N88 with MDM2 ( Figure 4F). MIAT knockdown weakened the binding ( Figure 4F).

| MIAT-EGLN2 axis modulates oxidized (GSSG) and reduced (GSH) glutathione ratio in rat neurons
One recent study showed that in EGLN2 −/− neurons, glycolysis, glucose consumption and glucose oxidation were significantly downregulated. In comparison, the activity of the oxPPP was enhanced. 10 In the oxPPP pathway, NADPH is generated as a reducing equivalent by glutathione reductase to regenerate GSH, which serves as a critical antioxidant in neurons. 20 Thus, we further investigated whether the MIAT-EGLN2 axis modulates carbon metabolism in rat neurons. Seahorse XF96 analyzer was utilized to measure the changes in mitochondrial respiration.
Quantitation of oxidized (GSSG) and reduced (GSH) glutathione 6 h after OGD/R indicated that MIAT overexpression increased the ratio of oxidized glutathione, EGLN2 knockdown drastically reduced the ratio ( Figure 6E). Based on these findings, we infer that I/R injury induces MIAT upregulation in rat neurons ( Figure 6F). MIAT enhances EGLN2 stability via reducing MDM2 mediated ubiquitin-proteasomal degradation ( Figure 6F). EGLN2 can reduce oxPPP flux and enhance reactive oxygen species (ROS) generation, leading to neuronal death ( Figure 6F).  The current study found that in rat cortical neurons, MIAT interacted with EGLN2. Enforced MIAT overexpression or knockdown did not alter EGLN2 mRNA expression. In comparison, MIAT increased its stability after I/R injury via reducing its ubiquitinmediated degradation. MDM2 is an E3 ligase significantly elevated after transient MCAO 28 and serves as a neuroprotective protein by promoting p53 degradation and preventing p53-mediated neuronal apoptotic death. 29 MDM2 knockdown triggers p53 accumulation and increases neuronal susceptibility to OGD/R-induced apoptosis. 30 Similar protect effects of MDM2 were confirmed in primary cultured spinal cord neurons against OGD/R injury. 31  During the phase of reperfusion, oxidative stress exacerbates ROS production. 33 The free radicals are highly reactive to a series of molecular targets, including nucleic acids, proteins and unsaturated lipids in cell membranes, generating oxidation-derived products. 33 Some endogenous mechanisms are utilized to balance ROS production, including glutathione, coenzyme Q and some other enzymes (superoxide dismutase, glutathione reductase, glutathione-S-transferase and glutathione peroxidase). However, when ROS production overwhelms the handing capacity of the antioxidants, cellular damage occurs. 33 EGLN2 acts as an important regulator of neuronal energy metabolism. 10 Glucose oxidation is the major energy source of neurons. EGLN2 knockout does not alter oxygen consumption in neurons but increases oxPPP activity at the expense of glycolysis. 10 This metabolic alteration enables neurons better against ischaemia via a greater capacity to generate NADPH to scavenge oxygen radicals. 10 In this study, we found that MIAT reduced oxPPP flux and increased oxidized/reduced glutathione ratio via stabilizing EGLN2.

F I G U R E 3 MDM2 interacts with EGLN2 and promotes its degradation. (A) Predicted
Bioinformatic analysis in this study also observed that SYVN1 and STUB1 are potential E3 ligase interacting with EGLN2. SYVN1 mediates the ubiquitination and degradation of glutathione peroxidase 5 (GPX5), increasing the generation of ROS and apoptosis of cardiomyocytes. 34 Besides, SYVN1 can enhance I/R induced renal epithelial injury by mediating NRF2 ubiquitylation and degradation. 35 Furthermore, SYVN1 participates in ER-associated degradation of unfolded/misfolded proteins, which is linked to brain ischaemia. 36 Therefore, it is meaningful to explore their potential regulative effect on EGLN2 stability in the future.
In summary, this study identified MIAT as a novel stabilizer of EGLN2, via reducing MDM2 mediated K48 poly-ubiquitination.
MIAT-EGLN2 axis exacerbates I/R injury via altering redox homeostasis in neurons. Future studies are required to explore the therapeutic potential of targeting the MIAT-EGLN2 axis to inhibit neuronal apoptotic cell death after acute ischaemic stroke.

CO N FLI C T O F I NTE R E S T
The authors have on conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.