Overexpression of circRNA circUCK2 Attenuates Cell Apoptosis in Cerebral Ischemia-Reperfusion Injury via miR-125b-5p/GDF11 Signaling

Abstract

Circular RNAs (circRNAs) are expressed at high levels in the brain and are involved in various central nervous system diseases. However, the potential role of circRNAs in ischemic stroke-associated neuronal injury remains largely unknown. Herein, we uncovered the function and underlying mechanism of the circRNA UCK2 (circUCK2) in ischemia stroke. The oxygen-glucose deprivation model in HT-22 cells was used to mimic ischemia stroke in vitro. Neuronal viability and apoptosis were determined by Cell Counting Kit-8 (CCK-8) assays and TUNEL (terminal deoxynucleotidyltransferase-mediated deoxyuridine triphosphate nick end labeling) staining, respectively. Middle cerebral artery occlusion was conducted to evaluate the function of circUCK2 in mice. The levels of circUCK2 were significantly decreased in brain tissues from a mouse model of focal cerebral ischemia and reperfusion. Upregulated circUCK2 levels significantly decreased infarct volumes, attenuated neuronal injury, and improved neurological deficits. circUCK2 reduced oxygen glucose deprivation (OGD)-induced cell apoptosis by regulating transforming growth factor β (TGF-β)/mothers against decapentaplegic homolog 3 (Smad3) signaling. Furthermore, circUCK2 functioned as an endogenous miR-125b-5p sponge to inhibit miR-125b-5p activity, resulting in an increase in growth differentiation factor 11 (GDF11) expression and a subsequent amelioration of neuronal injury. Consequently, these findings showed that the circUCK2/miR-125b-5p/GDF11 axis is an essential signaling pathway during ischemia stroke. Thus, the circRNA circUCK2 may serve as a potential target for novel treatment in patients with ischemic stroke.

Graphical Abstract

Chen and colleagues found that overexpressed circRNA UCK2 (circUCK2) attenuated ischemic neuronal injury by regulating TGF-β/Smad3 signaling. Furthermore, circUCK2 functioned as an endogenous miR-125b-5p sponge to inhibit miR-125b-5p activity, resulting in an increase in GDF11 expression and a subsequent amelioration of neuronal injury.

Introduction

Ischemic stroke is a common neurological disorder that is a leading cause of death and long-term disability worldwide.1, 2, 3 Acute ischemic stroke, triggered by intracranial artery occlusion or extracranial cervical artery occlusion, accounts for approximately 85% of total strokes.^4,5 During the hours after an ischemic stroke, neurons become permanently damaged and undergo cell death.⁶ Therefore, therapeutic options that rescue damaged neurons are important; however, to date, only a few therapeutic agents have been reported to relieve neurological deficits after stroke. Clinical treatments, such as tissue plasminogen activator-mediated thrombolysis, are often restricted by a narrow therapeutic time window and insufficient long-term effects.7, 8, 9, 10 Novel therapeutic approaches that decrease neuronal injury are urgently needed.

Circular RNAs (circRNAs), which are newly identified endogenous non-coding RNAs (ncRNAs), are characterized by back-splicing resulting from covalently closed continuous loops.^11,12 circRNAs play a role in several diseases, including stroke, cancer development, and heart disease.^11,13 For instance, circHIPK3 regulates cell proliferation and migration by sponging miR-124 and regulating Aquaporin 3 (AQP3) expression in hepatocellular carcinoma.^14,15 circRNA_001569 promotes cell proliferation through absorbing miR-145 in gastric cancer.¹⁶ Many circRNAs are abundantly expressed in the brain and exert their functions by serving as endogenous competing RNA of mRNA via interacting with miRNAs in ischemia stroke.^17,18 The circRNA TLK1 aggravates neuronal injury and neurological deficits after ischemic stroke via the miR-335-3p/TIPARP axis. As shown in our previous studies, circular RNA SHOC2 (circSHOC2) ameliorates ischemic damage by regulating the autophagy of neurons.¹⁹ In the present study, the circRNA circUCK2 was upregulated during the acute period after focal ischemia. circUCK2 is expressed in neurons, but its function remains unclear. Previous studies have shown that circUCK2 was significantly downregulated in patients with ischemic stroke. However, the function of circUCK2 after ischemic stroke is poorly understood. Herein, we investigated the role of circUCK2 in neuronal injury after middle cerebral artery occlusion (MCAO) in adult mice.

MicroRNAs (miRNAs) are endogenous ncRNAs that post-transcriptionally inhibit gene expression.20, 21, 22 Aberrant expression of a miRNA was associated with ischemia stroke.23, 24, 25 miR-125b-5p expression levels were upregulated in the brain tissue and blood plasma of mice 24 hours after ischemic stroke.²⁶ Therefore, we speculated that miR-125b-5p may regulate the neuroprotection in ischemia stroke. We found that upregulated circUCK2 acted as a sponge for miR-125b-5p, thereby increasing the expression of growth differentiation factor 11 (GDF11), which reduces neuronal injury, and thus contributes to neuroprotection.

Results

circUCK2 Ameliorated Neuronal Injury In Vitro

A previous circRNAs microarray study in an established MCAO stroke model revealed that circUCK2 is one of the circRNAs downregulated during ischemic stroke.²⁷ Considering that circUCK2 plays a critical role in stroke, further experiments were needed to dissect the detailed mechanisms. In this study, we first detected the expression of circUCK2 in a murine ischemic stroke model (MCAO) using quantitative reverse transcription PCR (qRT-PCR), and the results showed that circUCK2 expression was decreased in the brain tissue of mice subjected to the MCAO model and in oxygen glucose deprivation (OGD)-induced HT-22 neurons (Figures 1A and 1B). We then transferred circUCK2 mimics into neurons before OGD treatment, and the apoptotic rate of the neurons was evaluated by flow cytometry. As shown in Figure 1C, neuronal apoptosis was significantly increased in OGD-subjected HT-22 cells, and these changes were reversed after the transfection of circUCK2 (Figure 1D). Furthermore, neuronal apoptosis was also evaluated by TUNEL (terminal deoxynucleotidyltransferase [TdT]-mediated deoxyuridine triphosphate [dUTP] nick end labeling) staining. After the OGD challenge, the apoptotic rate of the cells significantly increased as the number of TUNEL-positive cells increased, which was abolished by circUCK2 overexpression. Moreover, the expression of proteins related to apoptosis was detected by western blot to further clarify the effect of circUCK2 on neuronal apoptosis. Consistent with the flow cytometry and TUNEL results, overexpression of circUCK2 in HT-22 cells upregulated the levels of Bax and cleaved caspase-3 but downregulated the levels of Bcl-2 after MCAO injury (Figures 1E and 1F). However, the expression of caspase-3, Bax, and Bcl-2 under normal conditions was similar in the circUCK2 and circRNA-NC groups (data not shown).

Figure 1.

circUCK2 Ameliorated Neuronal Injury In Vitro

(A) The relative expression of circUCK2 was significantly decreased in mouse brain treated with MCAO as compared to that in matched normal brain tissues. ∗p < 0.05. (B) The relative expression of circUCK2 was significantly decreased in HT-22 cells treated with OGD as compared to that in matched untreated HT-22 cells. ∗p < 0.05. (C) Cell apoptosis analysis after the transfection of circUCK2 mimics. ∗p < 0.05. (D) TUNEL staining shows the apoptotic rates of neurons treated with circUCK2 after OGD. (E) The expression levels of Bax, Bcl-2, and caspase-3 proteins treated with circUCK2 mimics after OGD. (F) Quantification of Bcl-2, Bax, and caspase-3 expression. ∗p < 0.05.

Overexpression of circUCK2 Ameliorated Neurological Deficits after Focal Ischemia In Vivo

We next evaluated the role of circUCK2 in neuronal deficits after stroke in vivo. We established four groups of mice as follows: (1) sham operation group, (2) MCAO model group, (3) MCAO + circRNA-NC group, and (4) MCAO + circUCK2 group. After 24 h, we assessed the neurological deficits in each group by calculating the neurological function score. Neurological deficit scoring, an adhesive removal test, and a cylinder test were conducted before and 3 and 7 days after MCAO. The neurological deficit scores in the MCAO + circUCK2 group were significantly lower than those in the MCAO group (Figures 2A–2C). Consistent with these behavioral improvements, circUCK2 lentivirus treatment also ameliorated the increase in Bax and decreased the MCAO-dependent induction in caspase-3 (Figure 2D). Cerebral infarction volume is an important indicator in acute ischemia stroke therapies. Following circUCK2 treatment, the cerebral infarction volume was profoundly decreased compared with the MCAO group (Figures S1A and S1B). All of these results indicated that the neuronal damage in a MACO mouse model was alleviated by circUCK2 treatment.

Figure 2.

Overexpression of circUCK2 Ameliorated Neurological Deficits after Focal Ischemia In Vivo

(A) Neurological deficits were measured by calculating the modified neurological severity scores at 3 and 7 days after MCAO. ∗p < 0.05. (B) Cylinder test of forelimb symmetry after MCAO. ∗p < 0.05. (C) Adhesive removal test of forelimb function at 3 and 7 days after MCAO. ∗p < 0.05. (D) Western blot analyses of Bax, Bcl-2, and caspase-3 proteins levels after treatment with circUCK2 mimics after MCAO. ∗p < 0.05.

circUCK2 Served as a Molecular Sponge for miR-125b-5p and Negatively Regulated miR-125b-5p Expression

Accumulating evidence has indicated that the circRNAs that are highly expressed in mouse tissues and in human cell lines act as competing endogenous RNA (ceRNA) sponges, which interact with miRNAs and influence the expression of their target proteins. We performed bioinformatics analysis to identify the potential target miRNA of circUCK2. The miRNA target prediction software from starBase (http://starbase.sysu.edu.cn/) found that the 3′ (untranslated region) UTR of circUCK2 contains regions matching the seed sequences of miR-125b-5p, miR-7061-3p, and miR-770-5p. Among these three miRNAs, only the levels of miR-125b-5p were significantly increased within the ischemic cortex at 6 h after MCAO (Figure 3A). In order to determine whether circUCK2 targets and regulates miR-125b-5p expression, a luciferase reporter assay was performed. The expression of miR-125b-5p after the transfection of miR-125b-5p mimic in HT-22 cells was examined to verify transfection. As shown in Figure 3B, miR-125b-5p overexpression inhibited the luciferase activity of circUCK2-wild-type (WT) but had no obvious inhibitory effects on circUCK2-mutant (Mut). Additionally, we also found that circUCK2 and miR-125b-5p co-localized in the cytoplasm via qRT-PCR analysis (Figure 3C). We next sought to determine whether circUCK2 physically bound to miR-125b-5p. Therefore, a biotin-coupled miR-125b-5p mimic was then used in order to determine whether miR-125b-5p pulled down circUCK2. We observed an enrichment of circUCK2 in the miR-125b-5p-captured fraction and not in the fractions captured by Mut miR-125b-5p mimics, which had disrupted base pairing between circUCK2 and miR-125b-5p (Figure 3D). These findings were confirmed in an inverse affinity isolation assay for miR-125b-5p using a biotin-labeled specific circUCK2 probe (Figure 3E). Taken together, these data demonstrated that circUCK2 served as a molecular sponge for miR-125b-5p and negatively regulated miR-125b-5p expression.

Figure 3.

circUCK2 Served as a Molecular Sponge for miR-125b-5p and Negatively Regulated miR-125b-5p Expression

(A) Expression of miR-125b-5p, miR-7061-3p, and miR-779-5p in neurons treated with 6-h OGD. ∗p < 0.05. (B) Relative luciferase activity of wild-type (WT) and 3ʹ UTR mutant (Mut) constructs of circUCK2 co-transfected with miR-125b-5p mimics and miRNA negative control. ∗p < 0.05. (C) RT-PCR analysis of circUCK2 in the cytoplasmic and nuclear compartments. (D) A dual-luciferase reporter assay was conducted to verify the relationship between miR-125b-5p and circUCK2. ∗p < 0.05. (E) The levels of total and pulled down miR-125b-5p and U6 by circNCX1 probe or control probe were analyzed by qRT-PCR. The relative pellet/input ratios were calculated. Random, scrambled control probes. ∗p < 0.05.

miR-125b-5p Increased the Level of Apoptosis in OGD-Treated Neurons

We further explored the effect of miR-125b-5p on OGD-induced cell death of neurons. Since circUCK2 exerts an inhibitory effect on miR-125b-5p expression, we further explored the effect of miR-125b-5p on the OGD-induced cell death of neurons. Transfection of a miR-125b-5p mimic significantly increased miR-125b-5p expression both under normal and OGD conditions, while a miR-125b-5p inhibitor significantly decreased miR-125b-5p expression under both conditions (Figure 4A). Cell Counting Kit-8 (CCK-8) assays showed that miR-125b-5p overexpression decreased the cell viability of neurons under both normal conditions and OGD conditions, while a miR-125b-5p inhibitor had the opposite effect (Figures 4B and 4C). This was consistent with the findings that miR-125b-5p overexpression also increased caspase-3 activity, while a miR-125b-5p inhibitor suppressed caspase-3 activity (Figures 4D and 4E). To confirm whether miR-125b-5p could influence apoptosis in HT-22 cells 6 h after OGD, a TUNEL assay was performed. TUNEL-positive cells were rarely seen in the normoxic group, whereas the number of TUNEL-positive cells increased in the 6 h OGD group. Transfection of miR-125b-5p was significantly enhanced, while a miR-125b-5p inhibitor attenuated the number of TUNEL-positive cells (Figures 4F and 4G). The miR-125b-5p mimics decreased glucose uptake in HT-22 cells under OGD conditions (Figure 4H). Conversely, inhibiting the expression of miRNA-125b-5p can accelerate glucose uptake.

Figure 4.

miR-125b-5p Increased Apoptosis in OGD-Treated Neurons

(A) The relative expression of miR-125b-5p in HT-22 cells treated with miR-125b-5p mimics and a miR-125b-5p inhibitor. ∗p < 0.05. (B) A CCK-8 assay was used to detect the viability of HT-22 cells with miR-125b-5p mimics. ∗p < 0.05. (C) A CCK-8 assay was used to detect the viability of HT-22 cells with a miR-125b-5p inhibitor. ∗p < 0.05. (D) Caspase-3 activity was detected by a caspase-3 detection kit in HT-22 cells with miR-125b-5p mimics. ∗p < 0.05. (E) Caspase-3 activity was detected by a caspase-3 detection kit in HT-22 cells with a miR-125b-5p inhibitor. ∗p < 0.05. (F) Number of TUNEL-positive neurons treated with circUCK2 mimics after OGD. ∗p < 0.05. (G) Number of TUNEL-positive neurons treated with circUCK2 inhibitor after OGD. ∗p < 0.05. (H) The percent of glucose uptake by HT-22 cells was assayed after transfection with circUCK2 mimics or a circUCK2 inhibitor. ∗p < 0.05.

GDF11 Was a Target of miR-125b-5p

In order to explore the potential targets of miR-125b-5p during ischemic stroke injury, we predicated miR-125b-5p-targeted genes using TargetScan (http://www.targetscan.org/) online tools. First, we observed that among these genes, GDF11 expression changed significantly after the overexpression of miR-125b-5p (Figure 5A). Moreover, GDF11 was downregulated in a MCAO model mouse (Figure 5B). The predicted binding site or respective mutated sequences were then cloned into the pGL3-enhancer luciferase reporter. The luciferase reporter activity was measured and plotted after normalizing against Renilla luciferase activity for 24 h. Co-transfection of an miR-125b-5p-overexpressing vector and a pmiR-GLO plasmid with the WT GDF11 3′ UTR resulted in the downregulation of luciferase activity, and this effect was reversed in HT-22 cells transfected with a mutated GDF11 3′ UTR. However, miR-125b-5p failed to influence the luciferase density of the mutated GDF11-3′ UTR plasmid (Figure 5C). In the following experiments, qRT-PCR, western blot, and an ELISA assay confirmed that overexpression of miR-125b-5p significantly downregulated GDF11 expression, while interfering miR-125b-5p upregulated GDF11 levels (Figures 5D–5F).

Figure 5.

GDF11 Was a Target of miR-125b-5p

(A) The predicted binding site of miR-125b-5p on the GDF11 3ʹ UTR; the blue part is the mutated base sequence. (B) The bar graph represents western blot analyses of GDF11 levels in the ischemic cortex after the circUCK2 mimics injection at 24 h after MCAO. ∗p < 0.05. (C) Relative luciferase activity of 3ʹ UTR WT and 3ʹ UTR Mut constructs of the GDF11 mRNA co-transfected with the miR-125b-5p mimic (miR-125b-5p) and the miRNA blank control (miR-NC).∗p < 0.05. (D) qRT-PCR was used to detect Gdf11 mRNA expression after overexpression of circUCK2 in miR-125b-5p mimics cell line. ∗p < 0.05. (E) Western blot was used to detect the protein expression of GDF11. (F) ELISA was used to detect the protein expression of GDF11. ∗p < 0.05.

circUCK2 Reduced Neuronal Apoptosis via Targeting miR-125b-5p/GDF11 through the TGF-β/Smad3 Signaling Pathway

Previous studies have shown that GDF11 can regulate cell growth via activating tropomyosin receptor kinase B (TrkB), thereby activating the transforming growth factor β (TGF-β)/mothers against decapentaplegic homolog 3 (Smad3) signaling pathway. Presently, in order to explore the downstream molecular mechanism of the circUCK2/miR-125b-5p/GDF11 axis, we first examined the protein expression of GDF11 via western blot. Transfection with the circUCK2 small interfering RNA (siRNA) (si-circUCK2) significantly inhibited the OGD-induced increase in GDF11 levels in HT-22 cells (Figure 6A). Transfection with si-circUCK2 attenuated the increase in GDF11 expression induced by the miR-125b-5p inhibitor (Figure 6B). The OGD + si-circUCK2 + GDF11 group abolished the apoptosis effect induced by circUCK2 knockdown and demonstrated similar neuron damage to the OGD + circUCK2 group (Figure 6C). To further investigate the effect of GDF11 on neuron viability, HT-22 cells were treated with 25 ng/mL GDF11 for a 72-h period after OGD, followed by CCK-8 assays. The results showed that overexpression GDF11 significantly rescued the OGD-induced decrease in neuronal survival in HT-22 cells, suggesting that GDF11 ameliorates OGD-induced apoptosis in neurons (Figure 6D). GDF11 is a member of the TGF-β superfamily. It has recently been reported that activation of TGF-β has a protective effect in ischemic stroke-induced neuronal apoptosis. To elucidate the underlying mechanisms that drive the inhibitory effects of GDF11 on neuroprotection, we then checked whether the TGF-β/Smad3 signaling pathway was involved. A qRT-PCR assay showed that Smad3 was influenced by circUCK2 or GDF11 overexpression in a similar manner. Smad3 expression was enhanced in the OGD + circUCK2 group, and the increase was reversed in the OGD + circUCK2 + GDF11 inhibitor group (Figure 6E; Figure S2). Our results suggested that circUCK2 modulates OGD-induced neuronal apoptosis via targeting miR-125b-5p/GDF11-mediated TGF-β/Smad3 signaling.

Figure 6.

circUCK2 Reduced Neuron Apoptosis via Targeting miR-125b-5p/GDF11 through TGF-β/Smad3 Signaling

(A) Western blot analysis detected the expression of GDF11 protein after circUCK2 inhibitor (circUCK2 siRNA [si-circUCK2]) was added to the HT-22 cell line after OGD treatment. ∗p < 0.05. (B) Western blot was used to detect the expression variable of GDF11 protein and the expression of relative protein after miR-125b-5p inhibitor (siRNA-125b-5p) was added to the HT-22 cell line after OGD. ∗p < 0.05. (C) Transfection of HT-22 with the miR-125b-5p inhibitor significantly reduced the si-circUCK2-induced increase in cell viability, as determined by a CCK-8 assay. ∗p < 0.05. (D) GDF11 attenuated the OGD-induced decrease in cell viability, as determined by a CCK-8 assay. ∗p < 0.05. (E) Western blot analyses showed the relative expression levels of Smad2 in HT-22 cells after the transfection of circUCK2 mimics after OGD. ∗p < 0.05.

The Effect of the circUCK2/miR-125b-5p/GDF11 Axis on Ischemic Stroke-Induced Neuronal Injury In Vivo

To further validate the effect of circUCK2 on ischemic stroke-induced neuronal injury, we injected adenovirus carrying circUCK2 overexpression vectors into the hippocampus of mice (Figure 7A). The expression of miR-125b-5p in the brain cells of overexpressing-circUCK2 mice was significantly downregulated, and the mRNA expression level of GDF11 was remarkably upregulated versus control (Figures 7B and 7C). To demonstrate the role of the TGF signaling pathway in circUCK2-mediated protective effects during OGD-induced neuronal injury, we treated mice with the GDF11 inhibitor follistatin (FST). A TUNEL assay revealed that the apoptotic index in the brain cells of mice overexpressing circUCK2 was significantly decreased, while FST increased the number of TUNEL-positive cells (Figures 7D and 7E). Additionally, GDF11 expression was significantly increased in the brains of mice overexpressing circUCK2, and FST significantly repressed the expression of GDF11 (Figures 7F and 7G). This indicated that circUCK2 activates the downstream TGF signaling pathway via modulation of miR-125b-5p/GDF11 expression, thereby improving the cell survival rate and ameliorating ischemic stroke-induced neuronal injury.

Figure 7.

The Effect of the circUCK2/miR-125b-5p/GDF11 Axis on Ischemic Stroke-Induced Neuronal Injury In Vivo

(A) The Expression of circUCK2 in an overexpressing-circUCK2 mouse was detected by qRT-PCR. ∗p < 0.05. (B) The expression of miR-125b-5p in an overexpressing circUCK2 mouse was detected by qRT-PCR. ∗p < 0.05. (C) The expression of GDF11 mRNA in an overexpressing circUCK2 mouse was detected by qRT-PCR. ∗p < 0.05. (D) Images represent TUNEL staining of neurons treated with circUCK2 and a GDF11 inhibitor (FST) after MCAO. (E) The bar graph represents TUNEL staining of neurons treated with circUCK2 and a GDF11 inhibitor (FST) after MCAO. ∗p < 0.05. (F) Images represent western blot analyses of GDF11 levels in the brain tissue injected with circUCK2 mimics and a GDF11 inhibitor (FST, 0.5 mg/kg) after MCAO. (G) Bar graph represents western blot analyses of GDF11 levels in the brain tissue injected with the circUCK2 mimics and GDF11 inhibitor (FST, 0.5 mg/kg) after MCAO. ∗p < 0.05.

Discussion

Due to the high mortality and disability rate, a stroke is a serious burden to families and society.^24,28 Current stroke research is centered on the development of acute neuroprotective agents.^29,30 However, the failure of these agents in clinical trials suggests that future stroke research must expand its horizons to include both acute and long-term neuroprotection. In recent years, the high specificity and stability of circRNAs make them suitable candidates as biomarkers. circRNAs are abundant in the mammalian brain, and their roles in many nervous system diseases, such as Alzheimer’s disease, multiple system atrophy, neuropathic pain, major depressive disorder, drug abuse, and ischemic stroke, have been explored.31, 32, 33 Furthermore, certain circRNAs have been proposed as transcriptional and translational regulators in pathological aspects of several cerebrovascular diseases.^34,35 Previous studies have shown that the circRNA DLGAP4 ameliorates ischemic stroke.³⁶ However, the role of the circRNA circUCK2 has not been evaluated in ischemic stroke. A recent study showed that circUCK2 was significantly downregulated in patients with ischemic stroke.²⁷ The overexpression of circUCK2 demonstrated substantial protective effects on neurons. As expected, our results elucidated the role of the circUCK2/miR-125b-5p/SIRT1 axis in stroke-associated neuronal injury and neurological deficits. Our findings regarding the expression and function of circUCK2 indicated that it represents a promising biomarker for reducing the damage caused by ischemic insults.

Increasing evidence has suggested that circRNAs are abundant in miRNA-binding sites, which play a role in the adsorption of miRNA in cells and block or reduce the inhibitory effect of miRNA on genes, thereby promoting the expression of target genes.37, 38, 39 For instance, circSLC8A1 was reported to bind miR-130b/miR-494, which regulated bladder cancer progression.⁴⁰ Another study showed that circRNA DLGAP4 inhibited miR-143 activity in order to decrease the size of infarct areas and attenuate neuronal deficits and blood-brain barrier (BBB) damage in MCAO mice.⁴¹ In the present study, we used the starBase 2.0 databases to predict the miRNA binding of the verified circRNAs. Our prediction analysis suggested that circUCK2 contained multiple binding sites of miR-125b-5p. A luciferase reporter assay and RNA pull-down confirmed the interaction between circUCK2 and miR-125b-5p. Our study showed that circUCK2 expression might function as a miRNA sponge for certain miRNAs, such as miRNA-125b-5p, thereby regulating the downstream TNF signaling pathway to reduce neuron apoptosis. Next, we investigated the role of miR-125b-5p on neurons. miR-125b-5p was first identified to be associated with neuronal death, as the expression of miR-125b-5p increased the apoptosis of HT-22 cells. Furthermore, we found that circUCK2 acts as an miR-125b-5p sponge to decrease the sizes of infarct areas and attenuate neuronal deficits in the MCAO model mice.

In the present study, our bioinformatics prediction highlighted a potential interaction site between miR-125b-5p and GDF11. GDF11 is closely associated with neuroprotection. The gene encoding GDF11 has been mapped to a region of chromosome 12q13.2, which encodes a novel protein of 407 aa with a signal sequence for secretion, an RXXR proteolytic processing site, and a carboxyl terminal region containing a highly conserved pattern of cysteine residues.42, 43, 44, 45 Improving GDF11 expression has been shown to reduce ischemic stroke injury in mice.^46,47 Our results showed a negative regulation between miR-125b-5p and GDF11, indicating a potential neurotoxic role of miR-125b-5p. To the best of our knowledge, our study is the first to demonstrate that GDF11, as a target of miR-125b-5p, functions in neurons induced by OGD, lending credence to our speculation of the function of the circUCK2/miR-125b-5p/GDF11 axis during stroke.

GDF11 is a member of the activin subfamily, which is also classified as a member of the TGF-β superfamily and is capable of activating the Smad3 signaling pathway via binding to the TGF-β type I and II receptors.⁴⁸ A previous study effectively demonstrated the neuroprotective potential of GDF11 in stroke.^49,50 GDF11 functions by phosphorylating and activating Smad3, similar to TGF-β. Activation of TGF-β signaling has diverse functions, including an increase in angiogenesis and neurogenesis.^51,52 Neuronal apoptosis is another indicator of ischemia injury. Smad3 is closely involved with apoptosis.⁵¹ A previous study demonstrated that overexpression of Smad3 not only plays an anti-apoptotic role after cerebral ischemia and reperfusion but also has the potential to inhibit inflammation and reduce acute brain injury.⁵² Our results indicated that circUCK2 activates the downstream TGF-β/Smad3 signaling pathway via modulating miR-125b-5p/GDF11 expression, thereby improving the cell survival rate and ameliorating OGD-induced neuronal injury.

Taken together, our results demonstrated for the first time that circUCK2 acts as an endogenous miR-125b-5p sponge to inhibit miR-125b-5p activity, resulting in the increased expression of GDF11. Restoration of circUCK2 ameliorated neuronal deficits and decreased infarct areas. The results of our study have identified a novel biomarker and potential therapeutic target for ischemic stroke.

Materials and Methods

Cell Culture

The mouse brain neuronal cell line HT-22 was obtained from ATCC (Manassas, VA, USA) and cultured in DMEM supplemented with 10% fetal bovine serum (FBS) (Invitrogen, Carlsbad, CA, USA), 100 U/mL penicillin (Sigma-Aldrich, St. Louis, MO, USA), and 100 U/mL streptomycin (Invitrogen, Carlsbad, CA, USA). The cells were maintained at 37°C in humidified air containing at 5% CO₂. The medium was changed every 3 days.

Cell Model of OGD

HT-22 cells were cultured with glucose-free deoxygenated DMEM (Thermo Fisher Scientific, Frederick, MD, USA) supplemented 100 mg/mL streptomycin and 100 U/mL penicillin in an incubator with premixed gas (95% N₂ and 5% CO₂) for 4 h. The cells were then given normal DMEM containing 10% FBS and placed in a CO₂ incubator (95% air and 5% CO₂) for 24 h. The cells in the control group were cultured with normal DMEM and 10% FBS for the same amount of time. Cells maintained in normal media and normal conditions were used as controls.

Cell Viability Assay

The CCK-8 assay (DoJindo, Japan) is widely used to assess cell viability. Neuronal injury was measured 6 h after OGD using the CCK-8 assay according to the manufacturer’s protocol. The cells were then placed into six-well plates containing 2 mL of complete medium and cultured (3.5 × 10⁵ cells/well). A microplate reader (Bio-Rad, Hercules, CA, USA) was used to measure optical density (OD) at OD₄₅₀. Images were collected at 0 and 24 h and observed under a phase-contrast optical microscope (Olympus, Japan). ImageJ software (version 1.8.0, http://wsr.imagej.net/distros/osx/ij153-osx-java8.zip) was used to analyze the images in this assay.

MCAO

MCAO was performed according to a previous report.^53,54 Adult male C57BL/6J mice (6–8 weeks old and weighing 16–22 g) were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). Briefly, anesthesia for mice was induced with 3% isoflurane mixed with 30% oxygen and 70% nitrous oxide in an anesthetic chamber and maintained with 1.5% isoflurane via a facemask. A silicone rubber-coated 6-0 nylon filament (Doccol, Sharon, MA, USA) was inserted into the external carotid artery and advanced for 9–10 mm to the mouse’s carotid artery bifurcation along the internal carotid artery and to the origin of the MCA. After 1 h of occlusion, the filament was removed to restore the blood flow to the MCA region. The control group was subjected to the same surgery but was not subjected to an occluded MCAO. Mice were housed in a room with a controlled temperature (22°C ± 3°C) and humidity (60% ± 5%) under a 12-h light/12-h dark cycle. All animal experimental designs were approved by the Animal Ethics Committee of Southern Medical University, and all experiments conformed to relevant regulatory standards.

qRT-PCR Analysis

According to the manufacturer’s protocol, total RNA was isolated from murine plasma and HT-22 cells using TRIzol (Invitrogen, Carlsbad, CA, USA). The RNA was then reverse transcribed with a stem-loop RT primer (RiboBio, Guangzhou, China) using HiScript Q Select RT SuperMix for the qPCR kit (Vazyme, Piscataway, NJ, USA) and quantified using AceQ qPCR SYBR Green master mix (Vazyme, Piscataway, NJ, USA). The levels of miR-125b-5p mRNA were determined via qRT-PCR and were normalized to the level of Gapdh. For mature miRNAs, the RNA was reverse transcribed with a stem-loop RT primer (RiboBio, USA) using HiScript Q Select RT SuperMix for the qPCR kit (Vazyme, Piscataway, NJ, USA). The cDNA was then set up in a qPCR reaction using AceQ qPCR SYBR Green master mix (Vazyme, Piscataway, NJ, USA). Each procedure was performed in triplicate. The primers used to amplify the circRNAs, and mRNA transcripts, were synthesized by Invitrogen (Carlsbad, CA, USA). The sequences of the primers are listed in Table S1.

Neurological Deficits and Behavioral Tests

Neurological function was determined using the modified neurological severity score test and performed by a researcher who was blinded to the experimental groups. The score was graded on a scale of 0–14 points (the normal score was 0 points). The higher the score, the more severe the injury. Behavioral tests were performed before MCAO and 3 and 7 days after MCAO by an independent investigator who was blinded to the experimental groups. For the adhesive removal somatosensory test, two small pieces of adhesive-backed paper dots (of equal size, 25 mm²) were used as bilateral tactile stimuli. The time it took for the mice to remove each stimulus from the forelimb was recorded in three mice per day. The cylinder test was performed according to the standard protocol.55, 56, 57, 58 A cylinder made of thick glass was used to assess forepaw asymmetry. The mouse was placed in the glass cylinder for 10 min, and the video was recorded from the side. All forepaw contacts with the wall of the cylinder were counted by two individuals blinded to the experimental groups.

Flow Cytometry Analysis of Cell Apoptosis

Flow cytometry was used to analyze the effects of circUCK2 overexpression on cell apoptosis with an annexin V-fluorescein isothiocyanate (FITC) apoptosis detection kit (eBioscience, San Diego, CA, USA). HT-22 cells grown in six-well plates were collected 48 h post-transfection and centrifuged at 400 × g for 5 min. After washing the cells with PBS three times, an equal number of the transfected cells were plated in 12-well plates. Cells were grouped into three groups as follows: (1) control group, (2) OGD group, and (3) OGD + circUCK2 mimic group. OGD was performed in all of the groups except for the control group. Cells were subsequently incubated with 5 μL of annexin V-FITC and 10 μL of propidium iodide (Sigma-Aldrich, St. Louis, MO, USA) staining solution for 15 min at room temperature (20°C–25°C) in the dark. The percentage of apoptotic HT-22 cells was determined using a Beckman Coulter Navios EX flow cytometer (Beckman Coulter, Shanghai, China). Apoptotic cells were analyzed using a BD FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA, USA) and BD CellQuest software (Becton Dickinson, San Jose, CA, USA).

ELISA Assay

In order to examine the levels of TGF-β1 in the supernatant of HT-22 cells, corresponding ELISA assay kits (Mlbio, Shanghai, China) were used according to the manufacturer’s instructions. The OD at 450 nm was measured using a microplate reader (Bio-Rad, Hercules, CA, USA) and then a standard curve was calculated. Each procedure was performed in triplicate.

TUNEL Staining

Cell apoptosis was also measured by using a one-step TUNEL apoptosis assay kit (Beyotime, Nanjing, China). The OGD model was described as above.59, 60, 61 Hanks’ balanced salt solution (HBSS, Gibco, CA, USA) was replaced with normal medium for 6 h to reoxygenate the cells. Neurons in six-well plates were washed three times in PBS and fixed with 4% paraformaldehyde for 30 min at room temperature. Fixed neurons were rinsed three times with PBS and incubated in 3% H₂O₂ in methanol in the dark for 20 min. After washing three times with PBS, neurons were incubated in a mixture of TdT and dUTP (1:9) from the TUNEL kit (Roche, Mannheim, Germany) at 37°C for 60 min, followed by incubation with converter-POD at 37°C for 30 min. Next, neurons were incubated with DAPI for nuclear staining. TUNEL-positive cells were observed under a light microscope (Olympus, Tokyo, Japan). Each treatment group had three wells, and each assay was repeated three times independently.

Dual-Luciferase Reporter Assay

HT-22 cells were seeded in 24-well plates. The cells were co-transfected with pmirGLO-circUCK2-WT, pmirGLO-circUCK2-Mut, or pmirGLO-GDF11-3′ UTR WT or pmirGLO-GDF11-3′ UTR-Mut reporter plasmids and mock (negative control) miR-125b-5p and anti-miR-125b-5p. Then, 48 h post-transfection, a Dual-Luciferase reporter assay system (Promega, Madison, WI, USA) was used according to the manufacturer’s instructions. Dual-luciferase activity was normalized to firefly luciferase activity and reported as a percentage of the control. Three wells were used for each treatment group, and each assay was repeated three times independently.

Pull-Down Assay

A total of 2 × 10⁶ HT-22 cells were seeded 1 day before transfection. On the following day, the cells were transfected with 3′ end-biotinylated miR-125b-5p or control RNA (GenePharma, Shanghai, China) at a final concentration of 50 nm for 36 h. Then, the cells were harvested and lysed in radioimmunoprecipitation assay (RIPA) lysis buffer (Abcam, Cambridge, MA, USA). The supernatant was incubated with 500 pM anti-sense oligonucleotides supplemented with RNase inhibitor overnight at 4°C. Subsequently, 10 μL of streptavidin agarose beads (Thermo Fisher Scientific, Frederick, MD, USA) was added to the cells, and the mixture was homogenized for 2 h at 4°C. The streptavidin agarose beads were then incubated with the supernatant for 2 h. The complex was centrifuged at a speed of 3,000 rpm for 10 min. The lysates were precleared, and 50-μL aliquots of the samples were prepared for input. The RNA was extracted using TRIzol according to the manufacturer’s protocol and subjected to RT-PCR analysis. The bound RNA was purified using TRIzol to measure circUCK2 levels.

Cell Transfection

circUCK2 mimics, si-circUCK2, miR-125b-5p mimics, miR-125b-5p inhibitor, or the corresponding negative controls were obtained from GenePharma (Shanghai, China). Transfection was performed using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) based on the manufacturer’s recommendations.

Western Blot Analysis

As described in our previous studies, proteins were extracted with RIPA lysis buffer (P0013B, Beyotime, Nanjing, China), separated on SDS polyacrylamide gels (10% and 12%) and electrophoretically transferred onto polyvinylidene fluoride (PVDF) membranes (Bio-Rad, Hercules, CA, USA). Bicinchoninic acid (BCA) protein analysis reagent (Thermo Fisher Scientific, Frederick, MD, USA) was used to measure the protein concentration.62, 63, 64 Membranes were probed with antibodies recognizing caspase-3 (Cell Signaling Technology, Beverly, MA, USA), GDF11 (ab234647, Abcam, Cambridge, MA, USA), Smad3 (ab227223, Abcam, Cambridge, MA, USA), and GADPH (Cell Signaling Technology, Beverly, MA, USA) overnight at 4°C. Membranes were then incubated with a horseradish peroxidase (HRP)-conjugated goat anti-mouse immunoglobulin G (IgG) secondary antibody (7076P2, Cell Signaling Technology, Beverly, MA, USA) and an HRP-conjugated goat anti-rabbit IgG secondary antibody (7074P2, Cell Signaling Technology, Beverly, MA, USA). Immunoreactivity was detected using an enhanced chemiluminescence (ECL) detection system (Thermo Fisher Scientific, Frederick, MD, USA).

Statistical Analysis

The values were expressed as the mean ± standard deviation (SD). The statistical differences among groups were analyzed using a one-way ANOVA using SPSS 19.0 software (IBM, Chicago, IL, USA), version 19.0 and GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA). p <0.05 was considered to be statistically significant.

Data Availability

The datasets used or analyzed in this study may be obtained from the authors upon reasonable request.

Author Contributions

W.C. contributed to the figures and tables and to the manuscript. H.W., and J.F. were in charge of experiments and data.

Conflicts of Interest

The authors declare no competing interests.