CircSmox knockdown alleviates PC12 cell apoptosis and inflammation in spinal cord injury by miR‐340‐5p/Smurf1 axis

Abstract Background Spinal cord injury (SCI) is a traumatic central nervous system disorder that leads to irreversible neurological dysfunction. Emerging evidence has shown that differentially expressed circular RNAs (circRNAs) after SCI is closely associated with the pathophysiological process. Herein, the potential function of circRNA spermine oxidase (circSmox) in functional recovery after SCI was investigated. Methods Differentiated PC12 cells stimulated with lipopolysaccharide (LPS) were employed as an in vitro model for neurotoxicity research. Levels of genes and proteins were detected by quantitative real‐time PCR and Western blot analysis. Cell viability and apoptosis were determined by CCK‐8 assay and flow cytometry. Western blot analysis was used to detect the protein level of apoptosis‐related markers. The levels of interleukin (IL)‐1β, IL‐6, IL‐8, and tumor necrosis factor (TNF)‐α. Dual‐luciferase reporter, RIP, and pull‐down assays were used to confirm the target relationship between miR‐340‐5p and circSmox or Smurf1 (SMAD Specific E3 Ubiquitin Protein Ligase 1). Results LPS elevated the levels of circSmox and Smurf1, but decreased the levels of miR‐340‐5p in PC12 cells in a dose‐dependent manner. Functionally, circSmox silencing alleviated LPS‐induced apoptosis and inflammation in PC12 cells in vitro. Mechanistically, circSmox directly sponged miR‐340‐5p, which targeted Smurf1. Rescue experiments showed that miR‐340‐5p inhibition attenuated the neuroprotective effect of circSmox siRNA in PC12 cells. Moreover, miR‐340‐5p suppressed LPS‐triggered neurotoxicity in PC12 cells, which was reversed by Smurf1 overexpression. Conclusion CircSmox enhances LPS‐induced apoptosis and inflammation via miR‐340‐5p/Smurf1 axis, providing an exciting view of the potential involvement of circSmox in SCI pathogenesis.


| INTRODUCTION
Traumatic spinal cord injury (SCI) is one of the most devastating injuries, which can cause irreversible sensory deficit, neurological damage, dysfunction, and necrosis. 1 SCI affects over 2.5 million people worldwide, and approximately 50%-80% of patients with SCI suffer from long-term moderate to severe traumatic pain owing to the lack of effective management. 2,3 SCI is a two-step process where the primary injury is followed by a progressive secondary injury characterized a cascade of biochemical and cellular processes involving the activation of neuroinflammation, proapoptotic signaling, vascular ischemia, cytotoxic debris, and lipid peroxidation. [4][5][6] Thus, further investigation on the molecular mechanism underlying secondary injury are of great significance for developing effective strategy for SCI treatment.
Circular RNAs (circRNAs) are one of noncoding RNAs with a covalently closed continuous loop that lacks 5′−3′ ends, thereby they are highly stable and can resistant to degradation by exonuclease RNase R. 7 Research increasingly reported that circRNAs play a significant potential role in modulating diverse crucial biological processes related to carcinogenesis, metabolism, and inflammation. [8][9][10] Moreover, the dysregulated expression of circRNAs has been observed in various diseases, such as cancer, immune system diseases, cardiovascular disease, and nervous system disease, and broadly participate in the pathogenesis and development of these diseases. [11][12][13][14] Besides that, studies have also showed that differential expression of circRNAs after SCI is tightly associated with the pathophysiological process. 15 CircRNA spermine oxidase (circSmox, ID: rno-Smox_0001) is originated from Smox gene in chr3: 124068796 | 124102267, which was identified to be highly expressed in the rat spinal cord following SCI. 16 However, the function of circSmox in pathogenesis of SCI remains vague.
Herein, the potential function and molecular mechanism of circSmox in SCI pathological process was studied in vitro by treating differentiated PC12 cells, which are widely used as the model of neurons in vitro, 17 with the lipopolysaccharide (LPS), which may provide a novel insight into the pathogenesis of SCI.
PC12 cells were stimulated with increasing doses of LPS (0, 1, 2, 5, or 10 μg/mL, Cat#L2630, Sigma-Aldrich) for 12 h. 5 μg/mL was selected to mimic the model of neuron injury in PC12 cells in vitro, cells treated with same volume of phosphate buffer saline (PBS, Cat#P4474, Sigma-Aldrich) were used as the control, and cells were differentiated before experimental treatments.

| Quantitative real-time PCR (qRT-PCR)
The TRIzol reagent (Cat#15596026, Invitrogen) was employed for the preparation of total RNA. Approximately 3 µg of total RNAs isolated from cultured PC12 cells were incubated with 3 U/μg of RNase R (Cat# R4875, Sigma-Aldrich) or Mock without enzymatic activity for 1 h at 37°C. The PrimeScript RT Reagent Kit (Cat#RR037B, Takara) was used to reverse transcribe the RNA sample into cDNA, and then qRT-PCR analysis was conducted with SYBR™ Green PCR Master Mix (Cat#4309155, Thermo Fisher Scientific) and primers (Table 1). PC12 cells were mixed with 5 μg/mL Actinomycin D (Cat# SBR00013, Sigma-Aldrich) for 0, 6, 12, or 24 h to block transcription, and the half-life of circSmox and linear Smox were measured using qRT-PCR. The relative fold changes were represented by CT value with GAPDH (for circSmox, Smox, and Smurf1) or U6 (rno-miR-340-5p) as an internal reference. 18

| Cell counting kit 8 (CCK-8) assay
PC12 cells were cultured on 96-well plates overnight and subjected assigned transfection. Forty-eight later, cells were exposed to 5 μg/mL LPS for 12 h. Subsequently, 10 μL of CCK-8 solution (Cat#C003, Beyotime) was added and incubated for 3 h. Finally, the optical density was measured at 450 nm to calculate cell viability. 20

| RNA immunoprecipitation (RIP) assay
The lysates of PC12 cells obtained by RIP lysis buffer were incubated with magnetic beads conjugated with AGO2 antibody or a negative IgG antibody at 4°C overnight (Cat#17-704, Millipore). After the addition of proteinase K in the bead/antibody/lysate mixture, the enrichment of miR-340-5p and circSmox with AGO immunoprecipitation was detected by using qRT-PCR.

| Statistical analyses
Each experiment was independently repeated three times. The data in the bar graphs were manifested as mean ± standard deviation. Paired or unpaired t-test, or Mann-Whitney test if not normally distributed was used for the comparison of datasets containing two groups. Statistical difference among multiple groups were conducted by analysis of variance (ANOVA) followed by Tukey's posttest. All statistical analysis was performed with GraphPad Prism 6 software (GraphPad), and p < .05 suggested significant differences (*p < .05, **p < .01. ***p < .001).

| LPS treatment dose-dependently increases circSmox expression in PC12 cells
As shown in Figure 1A, exposure of PC12 cells to different concentrations of LPS (0, 1, 2, 5, or 10 μg/mL) for 12 h led to an increase of circSmox expression in a dose-dependent manner, indicating the potential involvement of circSmox in LPS-induced cytotoxicity. 5 μg/mL LPS was selected for subsequent experiment due to the 50% increase. Thereafter, the circular characteristics of circSmox were analyzed. It was found that circSmox was resistant to the degradation by RNase R relative to the linear Smox in PC12 cells ( Figure 1B). Moreover, the half-life of circSmox exceeded 24 h, while that of linear Smox mRNA was about 6 h in PC12 cells ( Figure 1C), further indicating circSmox is a stable circRNA.

| CircSmox reinforces LPS-induced neurotoxicity in PC12 cells in vitro
Subsequently, to investigate the detailed functions of circSmox in LPS-induced neurotoxicity, we constructed the overexpression and the siRNA vectors of circSmox (OE-circSmox and si-circSmox). The results of qRT-PCR exhibited that circSmox was significantly downregulated or upregulated in PC12 cells transfected with si-circSmox or OE-circSmox compared with the cells transfected with si-NC or pCD-ciR (Figure 2A,B). In addition, as shown in Supporting Information: Figure S1, the viability of PC12 cells stimulated with 5 μg/mL LPS decreased by nearly 50%, thus, 5 μg/mL was selected to mimic the model of neuron injury in PC12 cells in vitro. Then transfected PC12 cells were treated with 5 μg/ mL LPS for 12 h, and the elevation of circSmox caused by LPS was decreased by circSmox knockdown, but increased by circSmox overexpression in PC12 cells ( Figure 2C). Functionally, the results in Figure 2D show that LPS treatment could decrease the viability of PC12 cells, which was rescued by circSmox knockdown, but reinforced by circSmox overexpression. Flow cytometry analysis suggested that circSmox knockdown reversed LPS-induced apoptosis accompanying with the decreases of Bax, c-caspase 3, and c-caspase 9 as well as the increase of Bcl-2, while circSmox overexpression showed opposite effects to enhance LPS-induced apoptosis by elevating the levels of Bax, c-caspase 3, and c-caspase 9 and reducing the level of Bcl-2 in PC 12 cells (Figure 2E-J). Besides that, ELISA analysis suggested that LPS evoked inflammatory response in PC12 cells, evidenced by the increases of TNF-α, IL-1β, IL-6, and IL-8, and when transfected with circSmox overexpression plasmids, this inflammatory response of PC12 cells was significantly increased, while transfection with circSmox siRNA decreased this inflammatory response in PC12 cells ( Figure  2K-N). Taken together, these results suggested that circSmox knockdown had a neuroprotective effect.

| CircSmox acts as a sponge for miR-340-5p
Based on the competing endogenous RNA (ceRNA) hypothesis, 24 the underlying miRNA of circSmox in PC12 cells were probed using bioinformatics analysis based on circAtlas database, and miR-340-5p was predicted to have the putative conserved target site on circSmox ( Figure 3A). After confirming the transfection efficiency of miR-340-5p mimic ( Figure 3B), the dual-luciferase reporter assay was performed. The results showed that miR-340-5p mimic significantly reduced the luciferase activity of the wild-type circSmox reporter vector, but not the mutated one in PC12 cells ( Figure 3C). RIP assay was then conducted, it was found that the levels of circSmox and miR-340-5p were effectively higher in AGO immunoprecipitation than those in the IgG negative control in PC12 cells ( Figure 3D). Moreover, RNA pull-down assay in PC12 cells showed that circSmox was significantly captured by the biotinylated miR-340-5p probe with markedly enhanced fold-changes ( Figure 3E). All these data confirmed the binding between circSmox and miR-340-5p. Thereafter, we found that miR-340-5p expression was dosedependently decreased by LPS treatment in PC12 cells ( Figure 3F). The level of miR-340-5p was downregulated by circSmox overexpression but upregulated by circSmox knockdown in PC12 cells ( Figure 3G). Then the knockdown efficiency of miR-340-5p inhibitor was validated by qRT-PCR analysis ( Figure 3H), and we proved that miR-340-5p inhibitor overtly reduced circSmox knockdown-triggered elevation of miR-340-5p in PC12 cells ( Figure 3I). In all, these results confirmed that circSmox directly sponged miR-340-5p in PC12 cells.

| MiR-340-5p knockdown reverses the neuroprotective effects of circSmox siRNA in PC12 cells
To probe whether circSmox plays its biological role via circSmox/miR-340-5p axis, we performed a series of rescue experiments. The results revealed that inhibition of miR-340-5p counteracted circSmox knockdownevoked viability enhancing role ( Figure 4A) and the inhibitory impacts on apoptosis ( Figure 4B-G) and inflammatory response ( Figure 4H-K) in PC12 cells in the presence of LPS. Collectively, these results demonstrated that circSmox might function as a sponge for miR-340-5p to contribute to LPS-induced neurotoxicity in PC12 cells.

| Smurf1 is a target of miR-340-5p in PC12 cells, and circSmox can regulate Smurf1 by sponging miR-340-5p
To explore the molecular mechanism underlying miR-340-5p, we then predicted the potential targets of miR-340-5p using TargetScan database. The results indicated that miR-340-5p had a putative conserved target site on Smurf1 ( Figure 5A). Thereafter, results of dual-luciferase reporter assay showed that miR-340-5p overexpression markedly reduced the luciferase activity of the wild-type Smurf1 reporter, but failed to affect the mutated one in PC12 cells ( Figure 5B). Smurf1 was found to be increased by LPS treatment in a dose-dependent manner both at mRNA and protein levels ( Figure 5C,D). Besides that, Smurf1 expression was decreased by miR-340-5p overexpression, but increased by miR-340-5p inhibition in PC12 cells ( Figure 5E,F). All these results suggested that miR-340-5p targeted Smurf1 and negatively regulated its expression. Moreover, western blot analysis showed that circSmox knockdown or overexpression led to a decrease or an increase of Smurf1 in PC12 cells under LPS treatment (Supporting Information: Figure S2), besides that, we observed that knockdown of circSmox resulted in a decrease of Smurf1 expression in PC12 cells, which was rescued by the inhibition of miR-340-5p ( Figure 5G,H). In addition, when we validated the transfection efficiency of Smurf1 overexpression plasmids ( Figure 5I,J), it was proved that Smurf1 transfection rescued miR-340-5p mimic-induced decrease of Smurf1 in PC12 cells ( Figure 5K,L). Thus, we identified a circSmox/miR-340-5p/Smurf1 axis in PC12 cells.

| DISCUSSION
Currently, the main clinical treatments for SCI are surgical procedures and high-dose methylprednisolone, which are largely limited to prolong the survival rate of patients, but not recover the injured nerve functions. 25 Thus, in-depth understanding the basic neurobiology of SCI is necessary for the development of available strategies for SCI therapy. At present, a large number of findings suggest that circRNAs play key roles in human diseases and exhibit great potential as biomarkers and therapeutic targets. 26 In SCI, some circRNAs have also been recognized to be involved in the pathogenesis by regulating the secondary damage. For example, circRNA-2960 was demonstrated to be increased in SCI rats, and induced apoptosis and inflammation at the lesion site by targeting miR-124. 27 Circ_0000962 was found to have a neuroprotective effect by activating PI3K/Akt and blocking of NF-κB via miR-302b-3p to suppress inflammatory response in vitro cell model of SCI. 28 Additionally, Sun et al. showed that circTYW1 contributed to functional recovery in rats after SCI and suppressed apoptosis in the lesion site as well as oxygen-glucose deprivation (OGD)-induced PC12 cells by activating ERK1/2 signaling through miR-380/FGF9 axis. 29 Therefore, circRNAs are also have roles in function recovery after SCI. In this study, an increased expression of circSmox was discovered in LPS-induced PC12 cells. Functionally, disrupting circSmox reversed LPS-evoked inflammatory response CircRNAs can serve as a ceRNA to prevent miRNA-mediated degradation of its downstream gene through sponging shared miRNAs. 24,30 Therefore, the underlying miRNA/mRNA network of circSmox was then investigated. This study confirmed that circSmox directly bound to miR-340-5p, which targeted Smurf1. Moreover, circSmox could regulate Smurf1 expression by miR-340-5p, suggesting the circSmox/miR-340-5p/ Smurf1 axis. Gao et al. showed that miR-340-5p overexpression could reduce neuropathic pain and inflammation in by rat models after chronic constriction injury by regulating Rap1A. Moreover, miR-340-5p re-expression induced locomotor function recovery in rat models after SCI and relieved neuroinflammation, oxidative stress, and apoptosis via blocking P38-MAPK signaling. Smurf1 has been shown to be increased in spinal cord of rats after SCI. 31 Disrupting Smurf1 expression ameliorated LPS-evoked neuroinflammation and neuronal necroptosis in PC12 cells. 32 Besides that, Zhao's team showed that Smurf1 attenuated miR-125b-meidated neurological recovery after SCI by promoting KLF2 degradation. 33 All these data indicated the neurotoxic action of Smurf1 in SCI. In the present study, we found that LPS decreased miR-340-5p expression, but elevated Smurf1 expression in PC12 cells. MiR-340-5p also showed a neuroprotective effect by suppressing LPS-induced apoptosis and inflammation in PC12 cells, which were reversed by Smurf1 upregulation. Importantly, miR-340-5p inhibition abolished the neuroprotective action of circSmox siRNA in LPS-stiluated PC12 cells.
In conclusion, this work for the first time confirmed that circSmox contributed to LPS-induced neuroinflammation and apoptosis in PC12 cells via miR-340-5p/ Smurf1 axis (Figure 7), suggesting a novel insight into the mechanisms of secondary injury in SCI and a new target for SCI therapy.