Circ_0004354 might compete with circ_0040039 to induce NPCs death and inflammatory response by targeting miR-345-3p-FAF1/TP73 axis in intervertebral disc degeneration

The abnormal function of nucleus pulposus cells (NPCs) plays a crucial role in the pathogenesis of intervertebral disc degeneration (IVDD). Recent studies have demonstrated that circular RNAs (circRNAs) are involved in the pathological process of IVDD by regulating NPCs' function. Nevertheless, the investigation on circRNA-circRNA interaction has not yet been reported. Here, we identified the top upregulated circ_0040039 and circ_0004354 in IVDD, derived from the syntrophin beta 2 gene but had different degrees of biological functions. Accumulating studies have reported PANoptosis is composed of apoptosis, pyroptosis, and necroptosis. Based on this, we think there should be a new pro-inflammatory cell death PAoptosis in the form of apoptosis and pyroptosis. Circ_0004354 might compete with circ_0040039 to induce the development of IVDD by modulating miR-345-3p-FAF1/TP73 axis-mediated PAoptosis, inflammatory response, growth inhibition, and ECM degradation of NPCs. Thus, these findings offer a novel insight into the circRNAs-mediated posttranscriptional regulatory network in IVDD, contributing to further clarification of the pathological mechanism of IVDD to develop a promising therapeutic target for IVDD diseases.


Introduction
In 2019, the analysis of the global burden of diseases identified low back pain (LBP) as the predominant cause leading to dyskinesia in patients [1]. Intervertebral disc degeneration (IVDD) is one of the most critical contributors to trigger LBP [1][2][3], which is known to cause serious socialeconomic problems [1,4]. The intervertebral disc (IVD) is a complex structure containing the central nucleus pulposus (NP) [5]. IVDD often begins with the degeneration of NP cells (NPCs). NPCs play a crucial role in supporting IVD structure and biological functions, as well as in maintaining IVD homeostasis by synthesizing extracellular matrix (ECM) components, especially aggrecan (ACAN) and collagen II alpha 1 (COL2A1, COL2) [6,7]. Inflammatory cell death, encompassing NPCs apoptosis and pyroptosis, also influences the biological functions of IVD [8,9]. The elevated secretion of proinflammatory cytokines, especially tumor necrosis factor (TNF)-α and interleukin (IL)-1β, is another vital trait of the degeneration of the NPCs [2,3,10]. During IVDD, the developing inflammation presents a waterfall-like cascade reaction, promoting hyperalgesia and nerve ingrowth, enhancing NPCs death and ECM breakdown [2,3]. Thus, we hypothesized that the proinflammatory factors in the IVD microenvironment caused by various factors would continue to increase, and eventually form an inflammatory cascade, which is the key mechanism leading to LBP and aggravation of IDD [2,3,10]. Therefore, it was critical to conduct an in-depth investigation of the mechanism of inflammatory cell death in IVDD to find a novel method to eliminate inflammation and NPCs death.
Apoptosis is divided into the extrinsic and intrinsic apoptosis pathways, which is executed by the cleavage caspase3 (c-CASP3) [11][12][13]. Gasdermin E (GSDME) is the key executive protein of pyroptosis, which can be driven by CASP3 [13][14][15]. The occurrence of cells pyroptosis is accompanied by the secretion of IL-1β [13][14][15]. Since cells can undergo extensive crosstalk under pathological conditions, cell death usually does not occur independently but in a mixed form [11,12]. Karki et al. [11] found that there was a mixed cell death PANoptosis composed of apoptosis, pyroptosis, and necroptosis. To facilitate the investigation of the relationship between cells death and inflammation, we think there should be a new pro-inflammatory cell death PAoptosis in the form of apoptosis and pyroptosis. However, the relationship between apoptosis and pyroptosis and the mixed death PAoptosis has not yet been reported in NPCs.
Circular RNAs (circRNAs), without a 5'-3' polarity and polyadenylation tail, are known to be highly resistant to RNase R [16]. Exonic circRNA is the most common type of circRNAs, which comes from the cyclization of at least one exon from a single gene [16,17]. Exonic circRNAs is usually located in the cytoplasm, which mediate NPCs' functional changes by acting as competitive endogenous RNA (ceRNA) [7,[17][18][19][20][21]. microRNAs (miRNAs) can recognize and bind to the target gene's 3'-UTR by the seed sequence to degrade mRNA or inhibit translation by binding to argonaute 2 (AGO2) protein to form RNA-induced silencing complex (RISC) [22]. The dysregulation of the expressions of circRNAs and miRNAs act as a hallmark characteristic of IVDD. Increasing evidence has revealed that circRNAs and miRNAs play a crucial role in mediating the occurrence and progression of IDD through participating in the modulation of NPCs apoptosis, proliferation, inflammatory response, and ECM metabolism [7,[17][18][19][20][21]. However, these studies have focused on the regulation of miRNA by one cir-cRNA; the impact of multiple circRNAs originating from the same parental gene has been neglected by the researchers. Furthermore, the role of circRNA-mediated NPCs pyroptosis is unclear. Thus, the goal of this study was to study the role and biological significance of multiple circRNAs in the IVDD progress.

Materials and methods
2.1. Ethics statement and NP tissues selection. This study was supervised and approved by the Tianjin Hospital Ethics Committee. All donors signed the informed consent form before surgery. We obtained 30 human degenerative NP tissues from patients with IVDD who were diagnosed with lumbar disc herniation, lumbar spinal stenosis, and lumbar spondylolisthesis and then conducted surgery on account of the severe LBP, neuralgia, or acute complications. The normal tissues were obtained from 16 patients with scoliosis, fresh thoracolumbar fracture, and spinal cord injury who were undergoing surgery due to spinal deformity, instability, or neurological deficits. Patients with rheumatoid arthritis, immune diseases, seropositive and negative spondyloarthropathy, thyroid diseases, tumors, and tuberculosis were excluded from the study. Supplementary Table 1 presents detailed information for each patient. The severity of IVDD was evaluated according to Pfirrmann's classification method [24]. Patients with Pfirrmann grade I/II were assigned to the normal group, whereas those with Pfirrmann grade III/IV constituted mild degeneration and grade V constituted the severe degeneration group. Supplementary Figure 1 shows the partial data on the patients' magnetic resonance imaging (MRI) information.
2.2. Analysis of circRNA, miRNA, and mRNA microarray datasets. CircRNA (GSE67566), miRNA (GSE63492/ GSE116726), and mRNA (GSE56081) microarray datasets were downloaded from the GEO database [23] and analyzed by the limma package in R [25]. The screening criteria for differentially expressed circRNAs (DECs) was: -log 10 FDR>2 and |log 2 fold-change (FC)| > log 2 5. The volcano plot of differentially expressed miRNAs (DEMs) was made according to the analysis of GSE116726 with the standard of FDR<0.05 and |log 2 (FC)| >1. The upregulated mRNAs in GSE56081 were selected with the standard of P-value <0.01 and log 2 (FC) >1. Table 1 shows the detailed information for each dataset.

2
Oxidative Medicine and Cellular Longevity The screened DEMs and DEGs were used to construct circRNAs-miRNAs-mRNAs network using Cytoscape software version 3.7.1 [32]. Gene Ontology (Go) analysis was conducted through the Database for Annotation, Visualization, and Integrated Discovery (DAVID) tools (https:// david-d.ncifcrf.gov/) [33] and visualized by Sangerbox tool (http://sangerbox.com/tool), a free online platform for data analysis. Furthermore, the clueGO plugin in Cytoscape software was applied to display TP73-mediated biological processes. P-value <0.05 was considered to be statistically significant. Additionally, a circRNA online database (https://circinteractome.nia.nih.gov/bin/circsearch Test) [34] was used to analyze the potential AGO2 protein binding sites of key circRNAs.
2.4. Acquirement, culture, and treatment of human NPCs. The precise method has been described in our previously study [20,35]. We obtained the primary human NPCs from ScienCell Research Laboratories (Sciencell, Cat. #4800, USA). NPCs were cultured in NPC Medium (Sciencell, Cat. #4801, USA) containing 5 mL NPCs growth supplement, 5 mL penicillin/streptomycin solution, and 10 mL fetal bovine serum. They were incubated at 37°C in a humidified environment with 5% CO 2 . The medium was changed every 2 days, and the NPCs were passaged once a week. Then the well-grown NPCs were used in the following experiments. As described by previous studies, 10 ng/mL TNF-α and 10 ng/mL IL-1β (Peprotech) were used to treat NPCs 35 or chondrocytes [36] for 24 h before experiments, so we used the same TNF-α/IL-1β concentration and treatment time to treat the NPCs after transfection of plasmids or miR-345-3p to simulate the micro-environment of IVDD in vitro. 2.6. Quantitative real-time RT-PCR. The TRlzol Reagent (Life Technologies, Thermo Fisher Scientific, USA) was used to extract total RNAs from NPCs or NP tissues. To detect RNA concentration and purity, 1 μL RNA sample was added to micro spectrophotometer (Nano-300, Allsheng, Hangzhou, China) and recorded the A 260 / 280 ratio and RNA concentration. Then 1% agarose gel electrophoresis was performed using electrophoresis (Junyi, Beijing, China) to further identify RNA purity and integrity. Subsequently, we used 1 μg total RNAs and 1 μL Geneseed® Enzyme Mix (Geneseed, Guangzhou, China) to reverse into 20 μL complementary DNA (cDNA) using Geneseed® II First Strand cDNA Synthesis Kit (Geneseed, Guangzhou, China). Subsequently, 10 μL Geneseed® qPCR SYBR® Green Master Mix (Geneseed, Guangzhou, China), 0.5 μL Forward (F) primer, and 0.5 μL Reverse (R) primer were prepared to perform qRT-PCR on ABI 7500 system (Applied Biosystems, USA). Supplementary Table 2 shows all primers. CircRNAs and mRNAs expression was normalized to GAPDH, whereas miR-345-3p expression was normalized to U6. The relative expression levels of circRNAs, miRNA, and mRNAs were detected and analyzed using the 2 -△△ Ct method.

Vectors construction and
2.7. RNase R treatment and Actinomycin D (ACT-D) assay. The total RNAs (10 μg) were extracted from NPCs and then incubated at 37°C for 30 min in the presence or absence of RNase R (Epicentre Technologies, Madison, WI, USA). Next, 2 mg/L ACT-D (Sigma-Aldrich, St. Louis, MO, USA) was used to treat NPCs at different time points to repress RNA synthesis. The relative expression of circ_0040039, circ_0004354, and SNTB2 mRNA were assessed through qRT-PCR.
2.8. Western blot analysis. RIPA lysis buffer containing phenylmethanesulfonylfluoride (Beyotime, Shanghai, China) was employed to extract proteins from NPCs. We then employed the Micro Bicinchoninic Acid Protein Assay kit (Beyotime, Shanghai, China) to quantify protein concentration. Next, the proteins were isolated through SDS-PAGE and then were transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Germany) at 350 mA for 90 min. Subsequently, the PVDF membranes were closed by 5% non-fat milk for 60 min and incubated overnight with primary antibody at 4°C and washed 10 min ×3 times using phosphate-buffered saline with Tween-20. Then the   (f) Venn diagram displayed the common target miRNAs of circ_0040039 and circ_0004354, as predicted by different algorithms. (g) The circ_0040039/circ_0004354-miRNAs network was constructed based on the above prediction results. Circ_0040039/circ_0004354, IVDDrelated common miRNAs, and miR-345-3p were indicated by a red ellipse, purple squares, and purple triangles, respectively. (h) 75 overlapping targets mRNAs of miR-345-3p were predicted by intersecting four different databases, and GSE56081 upregulated mRNAs. (i) Cytoscape software was utilized to construct the circ_0040039/circ_0004354 -miR-345-3p-mRNA interaction network, of which FAF1 and TP73 were indicated by a red ellipse. Bubble diagram (j) and Go chord diagram (k) revealed the predominant biological process of miR-345-3p is involved in.

(e)
Normal (16) Severe (13) Mild (   Oxidative Medicine and Cellular Longevity membranes incubation with a secondary antibody for 60 min at room temperature and followed by washing again. Finally, the PVDF membranes were put into a chemiluminescent substrate and developed for 2 minutes. Then the chemiluminescence system (Bio-Rad, CA, USA) was employed to detect the signals. Supplementary Table 3 shows the list of antibodies used in this study.

Cell Counting Kit-8 (CCK-8) and Flow cytometry (FCM).
The well-growth NPCs were inoculated into six-hole cell culture plates at a density of 5 × 10 5 cells per well. Then, 200 μL diluted RNAs-lipofectamine 8000 (Beyotime, China) complex were added to the cell wells that had been replaced with an 800 μL of serum-free medium. The NPCs were then cultured for 0, 1, 2, and 3 days at 37°C incubators. For the Normal (16) Severe (13) Mild (17
2.10. Enzyme-linked immunosorbent assay (ELISA). Human IL-1β ELISA kits (Elabscience, E-EL-H0149c) were used to detect the concentration of IL-1β in NPCs under different treatment conditions. IL-1β antibody was added to the ELISA well, and standards and samples were added to the microplates. Each standard and sample were measured through ELISA at a wavelength of 450 nm. IL-1β concentration was evaluated according to the absorbance value.
For RIP, a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Geneseed, Cat.No. P0101, China) was used to conduct RIP experiments. More specifically, 1 × 10 7 NPCs were collected and resuspended in 1000 μL buffer A containing 1% volume protease inhibitor and RNase inhibitors. The cell lysis supernatant (450 μL) was incubated with 5 μg AGO2 or control IgG antibody and protein A + G magnetic beads at 4°C overnight with rotation. Subsequently, the magnetic beads complex was cleaned six times in 1 mL buffer B. The isolated immunoprecipitated RNA was quantified by qRT-PCR to assess the relative expression of circ_ 0040039, circ_0004354, and miR-345-3p. Supplementary Table 3 shows the anti-AGO2 and IgG antibodies used in this study.
For miRNA pulldown assay, biotinylated miR-345-3p mimic or mimic NC (Guangzhou, China) were transfected into NPCs using Lipofectamine 8000 (Beyotime, China). Subsequently, M-280 Streptavidin Magnetic Beads (Invitrogen, USA) were incubated with the NPCs lysates at 4°C overnight. The biotin-coupled RNA complex bound to the beads was pulled down, and the relative expression of circ_ 4 ⁎⁎ ⁎⁎ ⁎⁎ ⁎⁎⁎         Prism software 6 version. The statistical significance between the two groups was compared by an unpaired Student's t-test, whereas the differences among more than two groups were assessed by one-way analysis of variance followed by Turkey's multiple comparisons test. We carried out at least three independent experiments. Results are presented as mean ± standard deviation. P-value <0.05 was considered to be statistically significant.

12
Oxidative Medicine and Cellular Longevity
3.5. The functional difference between circ_0040039 and circ_ 0004354 in NPCs in response to TNF-α treatments. Circ_ 0040039 is derived from the exons 2 to 5 of host gene Sntb2 (chr16: 69279504-69318147), and circ_0004354 is derived from the exon 5 of Sntb2 (chr16: 69317950-69318147) ( Figure 5(a)). The length of the mature sequences from circ_0040039 and circ_0004354 were 765 bp and 197 bp, respectively. According to CircInteractome online database, circ_0040039 has three, and circ_0004354 has two binding sites with AGO2, of which each has one region overlapped with the seed sequences of miR-345-3p ( Figure 5(a)). Next, we extracted circ_0040039, circ_0004354, and linear Sntb2 mRNA from NPCs and treated them with or without RNase R to verify that the circ_0040039 and circ_0004354 were indeed circular. As expected, the expression of circ_ 0040039 and circ_0004354 was insignificantly modified in the presence of RNase R, whereas Sntb2 expression was significantly decreased ( Figure 5(b)). Also, circ_0040039 and circ_0004354 were more stable than Sntb2 mRNA under the treatment of actinomycin D (ACT-d) because ACT-d could repress RNA synthesis ( Figure 5(c)). Previously, our study suggested that circ_0040039 and circ_0004354 had different biological characteristics [35]. Next, we constructed the overexpression vector of circ_ 0040039 and circ_0004354 to investigate their different biological functions and observed that their expression was increased approximately eight-fold in the over-expressed NPCs, respectively ( Supplementary Figures 3(e)-3(f)). TNF-α is known to induce apoptosis [2,3,10,11] and pyroptosis [11,14]. Subsequently, we aimed to confirm whether TNF-α could induce circRNA-mediated NPCs PAoptosis. We used different concentrations of TNF-α to treat NPCs. When the TNF-α concentration was 10 ng/mL, circ_0004354 significantly promoted, whereas circ_0040039 slightly promoted NPCs death. With an increase in the concentration of TNF-α, the degree of NPC death increased rapidly in circ_0040039 but slowly in circ_0004354 overexpressed NPCs (Figure 5(e)). Additionally, the ability of circ_0040039 to inhibit NPCs growth was lesser than that of circ_0004354 ( Figure 5(d)). Under the stimulation of TNF-α, the overexpression of circ_0040039 displayed the robust cleavage of apoptotic CASP3 and BAX and the slight cleavage of pyroptotic GSDME, while the over-expression of circ_0004354 showed the opposite results (Figures 5(f) and 5(g)). The expression of IL-1β was also elevated, but the effect of circ_0004354 in promoting IL-1β elevation was superior to circ_0040039 (Figures 5(f) and 5(i)). We also observed that circ_0004354 more significantly repressed ECM expression than circ_0040039 (Figures 5(f) and 5(h)). Moreover, the mRNA and protein expression of miR-345-3p target genes, FAF1 and TP73, were upregulated to varying degrees by circ_0040039 and circ_0004354 ( Figure 5(j)). These phenotypes could be reversed in the presence of miR-345-3p mimic. Thus, these data revealed that circ_0040039 and circ_0004354 promoted NPCs PAoptosis and inflammatory response probably by co-adsorbing miR-345-3p; circ_0040039 had a stronger ability to promote apoptosis, while circ_0004354 was more inclined to promote pyroptosis.
3.6. Circ_0004354 competes with circ_0040039 to adsorb miR-345-3p in NPCs at 10 ng/mL TNF-α concentration. Next, FISH, RIP, and RNA pulldown assays were conducted at a TNF-α concentration of 10 ng/mL to further elucidate the mechanism of functional difference between circ_ 0040039 and circ_0004354 in NPCs. This concentration was selected since circ_0040039 and circ_0004354 expressions were relatively low in normal NPCs and relatively high in 10 ng/mL TNF-α-induced NPCs. First, the FISH assay confirmed that circ_0040039 and circ_0004354 were colocalized and both preferentially cytoplasmic (Figure 6(a)), implying that there may be an interaction between them. Second, the RIP results revealed that circ_0040039, circ_ 0004354, and miR-345-3p were remarkably enriched by the anti-AGO2 antibody rather than the anti-IgG antibody, and the expression abundance of circ_0004354 was higher than circ_0040039, suggesting that they all existed in RISC ( Figure 6(b)). Third, a biotinylated miR-345-3p pulldown assay showed that the circ_0004354 had a more remarkable enrichment compared with circ_0040039 in NPCs ( Figure 6(c)). Next, a circ_0040039 or circ_0004354 fragment with WT or MUT complementary binding sites for miR-345-3p were established and inserted into psiCHECK2 luciferase reporter vectors, respectively, to further verify their interaction (Figures 6(d) and 6(f)). The results of the assay found that miR-345-3p mimic remarkably decreased and miR-345-3p inhibitor markedly elevated the luciferase activity of circ_0040039 or circ_0004354 WT and MUT1-2 reporters, whereas the luciferase activity of MUT3 reporters (sites 1 and 2 mutate together) was not noticeably altered, revealing that both circ_0040039 and circ_0004354 could directly bind to miR-345-3p via the two complementary target sites (Figures 6(e) and 6(g)). These results indicated that circ_0004354 might have a stronger binding ability to miR-345-3p at 10 ng/mL TNF-α concentration. The schematic sketch of mechanisms by which circ_0004354 competes with circ_0040039 to induce IVDD was showed in Figure 7

Discussion
Previous studies have shown that the DECs play a critical role in the pathological process of IVDD by regulating NPCs' biological functions [7,[17][18][19][20][21]. However, the biological functions and mechanisms of multiple circRNAs in IVDD have not yet been identified. The present study identified circ_0040039 and circ_0004354, both of which originated from the circularization of SNTB2 gene exons, triggering TNF-α-induced NPCs PAoptosis through the competitive adsorption of miR-345-3p. Our conclusions were based on the following observations: 1) In the presence of TNF-α, FAF1-activated CASP3-GSDME, resulting in extrinsic NPCs PAoptosis without an increase in IL-1β expression; while TP73 activated CASP3-GSDME pathway, resulting in intrinsic NPCs PAoptosis with an increase in IL-1β expression and a decrease in ECM components expression; 2) MiR-345-3p repressed NPCs PAoptosis, IL-1β expression and promoted the expression of ECM components by targeting FAF1 and TP73; 3) Both circ_0040039 and circ_0004354 promoted TNF-α-induced NPCs PAoptosis, of which circ_0040039 had a stronger ability to promote NPCs apoptosis, while circ_0004354 was more inclined to promote NPCs pyroptosis; 4) Circ_0004354 might compete with circ_0040039 to adsorb miR-345-3p in NPCs, and there was more circ_0004354 binding to miR-345-3p at TNF-α concentration of 10 ng/mL. Thus, this work unraveled a whole new mechanism of how circ_0040039 and circ_ 0004354 induced mixed inflammatory cell death, PAoptosis, by the competitive adsorption of miR-345-3p in NPCs.
Until now, several studies have reported that circRNAs function as ceRNA to modulate the progression of IVDD in a miRNAs-dependent manner, predominately encompassing ECM-related, apoptosis-related, and inflammationrelated pathways [7,[17][18][19][20][21] Cheng et al. [18] corroborated that circ-VMA21 could alleviate pro-inflammatory cytokines-induced NPCs death and ECM decomposition by repressing the miR-200c-XIAP pathway both in vitro and in vivo. Wang et al. [21] showed that circ-4099 could promote ECM synthesis and repress the excretion of TNFα and IL-1β, but the molecular mechanism involved in the dysregulation of IL-1β expression is still unclear. Our previous study demonstrated that circ_0040039 could promote NPCs apoptosis and repress NPCs growth [35]. Shen et al. [36] validated that circSERPINE2 could repress chondrocytes apoptosis and ECM synthesis by targeting miR-1271 and ETS-related gene. Our current study verified that circ_ 0004354 might compete with circ_0040039 to trigger TNFα-induced NPCs PAoptosis, inflammatory response, growth inhibition, and ECM degradation by targeting the miR-345-3p-FAF1/TP73 axis. Nevertheless, we could not rule out the possibility that the increase in IL-1β levels in NPCs also might have enhanced NPC's PAoptosis and ECM degradation. Additionally, further studies are required to determine whether TP73 directly combined with ACAN and COL2 to inhibit their expression.
The cell death pathways have long been believed to function in parallel with little or no overlap. Recent studies have shown that apoptosis and pyroptosis are closely linked and could cross-regulate each other. Hou et al. [46] found that the macrophage-derived TNF-α-activated caspase8, in the presence of GSDMC activated by hypoxia, switching apoptosis to pyroptosis. GSDME was confirmed to switch CASP3-mediated apoptosis to pyroptosis [14,15]. Karki et al. [11] demonstrated the synergism between TNF-α and IFN-γ triggering a mixed cell death PANoptosis composed of apoptosis, pyroptosis, and necroptosis. Here, we found a new circRNAs-mediated inflammatory cell death, PAoptosis, which is composed of apoptosis and pyroptosis in NPCs.
Intriguingly, the degree of circ_0040039 promoting NPCs PAoptosis was positively correlated with the concentration of TNF-α, whereas the degree of circ_0004354 promoting NPCs PAoptosis gradually slowed down with an increase in TNF-α concentration. The possible reasons were as follows: First, the release of IL-1β increased with an increase in NPCs PAoptosis, IL-1β not only negatively feedback inhibited the expression of circ_0004354, which slowed down the rate of PAoptosis and release of IL-1β, but also induced circ_0040039 expression, which accelerated the rate of PAoptosis and the release of IL-1β, forming a feedback regulation mechanism. Furthermore, RIP and RNA pulldown assays showed that there was more circ_0004354 enrichment, suggesting circ_0004354 might have a stronger binding ability to miR-345-3p at 10 ng/mL TNF-α concentration. However, further studies are required to determine the binding ability of circ_0040039/circ_0004354 and miR-345-3p and the effects on the phenotype of NPCs under high concentration of TNF-α or high concentration of TNFα + high concentration of IL-1β.
Therefore, we propose the following IVDD pathological mechanism model (Figure 7): Sntb2 gene was cut into multiple circRNAs, such as circ_0040039 and circ_0004354 under certain pathological conditions. In the early stages of IVDD, the concentration of micro-environmental proinflammatory cytokines (such as TNF-α) was lower, so circ_0004354 might have a stronger binding ability to bind with miR-345-3p, which powerfully promotes TP73 and slightly promotes FAF1 expression, thereby predominantly inducing NPCs pyroptosis and releasing IL-1β. With the increased secretion of pro-inflammatory cytokines in the micro-environment, IL-1β negatively feedback inhibited the expression of circ_0004354, which slowed down the rate of PAoptosis and the secretion of pro-inflammatory cytokines and strived to balance the micro-environment homeostasis again. Simultaneously, the pro-inflammatory cytokines (such as TNF-α and IL-1β) might induce the expression of circ_0040039, which antagonized the binding of circ_ 0004354 to miR-345-3p and enhanced the ability of circ_ 0040039 to bind to miR-345-3p, inducing NPCs apoptosis and releasing IL-1β. Therefore, we hypothesized that under different inflammatory factor concentration gradients, circ_ 0040039/circ_0004354 and miR-345-3p had different binding abilities, forming a feedback regulation mechanism, continuously releasing inflammatory factors, and finally forming an inflammatory cascade, which initiated or aggravated IVDD.
Several circRNAs were reported to have similar functions with their linear counterparts [47,48]. For example, a zinc finger with KRAB and SCAN domains 1 (ZKSCAN1) gene and its corresponding circRNA (circZKSCAN1) both repressed cell growth by mediating distinct signaling pathways [48]. In this study, we identified the role of SNTB2 related circRNAs (circ_0040039 and circ_0004354) on NPCs, while SNTB2 functions remained unclear. Notably, we found that miR-345-3p also repressed SNTB2 mRNA expression (data not shown), but its role in NPCs still unknown. The relationship between SNTB2 and SNTB2derived circRNAs needs further investigation. 18 Oxidative Medicine and Cellular Longevity However, there were also several limitations to this study. First, the microarray data were obtained from GEO, and the clinical sample size was relatively small since the normal NP tissues were difficult to obtain in clinical practice. Second, the animal experiments will be performed in the future. Third, we have not silenced circRNAs to validate their functions, since their expression was relatively low in normal NPCs. Finally, further studies are required to examine the levels of IL-1β and PAoptosis after cotransfection of circ_0040039 and circ_0004354 to NPCs and rat models.

Conclusion
Here, we demonstrated that circ_0004354 might compete with circ_0040039 to adsorb miR-345-3p to positively regulate FAF1/TP73 and their downstream target genes in the induction of low concentration of TNF-α, thereby promoting NPCs PAoptosis, inflammatory response, growth inhibition, and ECM degradation. These findings offer a novel insight into the circRNAs-mediated the posttranscriptional regulatory network in IVDD, which would contribute to the understanding of the pathological mechanism of IVDD to develop an invaluable therapeutic approach to IVDD diseases.

Data Availability
All data relevant to the study are included in the article or uploaded as supplementary information.