LncRNA TUG1 attenuates ischaemia‐reperfusion‐induced apoptosis of renal tubular epithelial cells by sponging miR‐144‐3p via targeting Nrf2

Abstract Renal ischaemia/reperfusion (I/R) injury may induce kidney damage and dysfunction, in which oxidative stress and apoptosis play important roles. Long noncoding RNAs (lncRNAs) and microRNAs (miRNAs) are reported to be closely related to renal I/R, but the specific molecular mechanism is still unclear. The purpose of this research was to explore the regulatory effect of lncRNA TUG1 on oxidative stress and apoptosis in renal I/R injury. This research revealed that in renal I/R injury and hypoxia/reperfusion (H/R) injury in vitro, the expression level of lncRNA TUG1 was upregulated, and oxidative stress levels and apoptosis levels were negatively correlated with the expression level of lncRNA TUG1. Using bioinformatics databases such as TargetScan and microRNA.org, microRNA‐144‐3p (miR‐144‐3p) was predicted to be involved in the association between lncRNA TUG1 and Nrf2. This study confirmed that the level of miR‐144‐3p was significantly reduced following renal I/R injury and H/R injury in vitro, and miR‐144‐3p was determined to target Nrf2 and inhibit its expression. In addition, lncRNA TUG1 can reduce the inhibitory effect of miR‐144‐3p on Nrf2 by sponging miR‐144‐3p. In summary, our research shows that lncRNA TUG1 regulates oxidative stress and apoptosis during renal I/R injury through the miR‐144‐3p/Nrf2 axis, which may be a new treatment target for renal I/R injury.

the past ten years, there has been accumulating evidence showing that miRNAs play important roles in the development of various diseases, including renal IRI-induced AKI. 9,10 Here, by RNA sequencing, we found that the expression of microRNA-144-3p was significantly changed in IRI-induced kidneys. In addition, microRNA-144-3p has been reported to be closely related to oxidative damage in many studies. Therefore, we first studied the role of microRNA-144-3p in renal IRI.
However, long noncoding RNAs (lncRNAs) are a newly discovered group of RNA molecules that contain more than 200 nucleotides and represent important regulatory roles. 11 LncRNAs control gene expression on multiple levels, including epigenetic, transcriptional, posttranscriptional and miRNA-mediated mRNA transcription. 11 According to recent studies, lncRNAs also play important roles in renal IRI. [12][13][14][15] LncRNA taurine upregulated gene 1 (TUG1), which plays important regulatory roles in models of oxidative damage induced by ischaemia, 16,17 was also reported to target miR-144-3p to play biological roles. 18,19 However, it is still unknown whether TUG1 may regulate apoptosis by targeting miR-144 in renal IRI. To test this hypothesis, follow-up experiments were conducted.

| Animal experiments and I/R model
Male C57BL/6 mice weighing 20-25 g (10-12 weeks of age) obtained from Hubei Center for Disease Control and Prevention were selected for animal experiments. All mice were kept in standard cages with free access to food and water. The room temperature was 18-25°C, and the humidity was controlled at 45-55%. All mice were bred normally for 1 week to adapt to environmental conditions and confirmed to be healthy according to the clinical examinations.
All experiments were performed in accordance with the guidelines of the Ethics and Research Committee of Wuhan University Medical School (Wuhan, China).
A mouse model of renal IRI was established according to experimental requirements. In short, all mice were fasted overnight before surgery and anesthetized by an intraperitoneal injection of 1% pentobarbital sodium (0.3 ml/100 g). A midline abdominal incision was made to the mice in the IRI group, and the renal pedicle was dissected. Then, the bilateral renal pedicle was clamped with non-traumatic clamps for 30 min. Subsequently, the non-traumatic clamps were removed, and the kidney perfusion was restored for 48 h. Finally, the mice were euthanized by injecting excessive pentobarbital sodium, and their the kidney tissues were carefully collected for subsequent experiments. Mice in the sham operation group only received a median abdominal incision. Additionally, for transfection groups, miR-144-3p agomir (1 nmol/g/day), miR-144-3p antagomir (5 nmol/g/day), ASO-TUG1 (5 nmol/g/day) and their negative controls (equal dosage) (RiboBio, Guangzhou, China) were injected from the tail vein for 3 consecutive days before the surgery. Blood samples were collected from individual mice to measure the concentration of serum creatinine.

| Real-time quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted from TCMK cells and the cortex of left kidneys with TRIzol reagent following the manufacturer's instructions.
For the detection of lncRNA, total RNA was reverse-transcribed into cDNA by the reverse transcription kit (RR047A, Takara, Japan).
Furthermore, for the detection of miRNA, the tailed method was applied, and the NCode TM miRNA First-Strand cDNA Synthesis Kit (MIRC10, Invitrogen) was employed to polyadenylate the isolated RNA. The samples were loaded following which the samples were subjected to reverse transcription quantitative polymerase chain reaction (RT-qPCR) with the 7900 HT RT-PCR System (Applied Biosystems, USA). Three replicate wells were set per sample to obtain variable data. Additionally, GAPDH or U6 were used as an internal reference. The following primer sequences were used in the study: miR-144-3p, forward 5'-GCTG GGATATCATCATATACTG-3' and reverse 5'-CGGACTAGTACATCATCTATACTG-3'; TUG1,

| TUNEL detection
In situ apoptosis detection kit (Promega) was used to detect the level of apoptosis in different tissues. Observed under a light microscope, the nuclei of apoptotic cells were stained brown, while the nuclei of negative cells were stained blue. The apoptosis rate was calculated based on the percentage of positive cell nuclei, which were stained brown in the visual field. The cell count under the microscope was independently completed by two professional pathologists in a blind manner.

| Haematoxylin and eosin (HE) staining
The obtained kidney tissues were fixed in 4% formalin, dehydrated and embedded in paraffin. 5μm-thick paraffin sections were cut out for subsequent staining. The sections were deparaffinized and stained with haematoxylin and eosin. Kidney damage was divided into the following grades according to HE staining: 0 (absence of necrosis), 1 (mild necrosis), 3 (moderate necrosis) and 5 (severe necrosis).

| Transmission electron microscopy
Transmission electron microscopy (TEM) was conducted to observe possible mitochondrial morphological changes induced by IRI in vivo.
Briefly, sesame-sized kidney cortex tissues from each group were fixed in 2.5% paraformaldehyde at 4°C for 24 h and then treated with 2% osmium tetroxide at 4°C for 2 h. After multiple rinse cycles with double-distilled water, the samples were dehydrated in a graded series of acetone and then embedded in Spurr's resin for ultrathin sectioning. Ultrathin sections were cut to 50-nm thickness on an ultramicrotome using diamond knives. Subsequently, the sections were stained with 2% uranyl acetate and citrate at 25°C for 1 h.
Finally, the sections were visualized under a transmission electron microscope.

| Plasma creatinine detection
Mouse blood samples for testing were obtained through cardiac puncture and accumulated in a tube. Plasma creatinine was obtained after centrifugation at 3000 rpm at 4°C for 10 min; then, the sarcosine oxidase method with a commercial creatinine assay kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) was used to detect the samples. Creatinine levels were expressed as mg/dl.

| Western blot analysis
Proteins from kidney tissue samples or TCMK cells were obtained

| Statistical analysis
All experimental data are presented as the means ± standard deviation (SD). Each group of experiments was repeated at least three times or each group contained at least six mice. The statistical software package SPSS version 19.0 was used for statistical analysis.
Differences between two groups were compared using two-side unpaired t test, while the comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. p < 0.05 was considered significant.

| MiR-144-3p was downregulated in I/R-injured kidney tissue in vivo and H/R-challenged TCMK cells in vitro
To identify which microRNAs are differentially expressed during kidney ischaemia/reperfusion injury, we performed RNAsequencing analysis of I/R-injured kidney tissue compared with control kidney tissue from C57BL/6 mice. A total of 80 differentially expressed miRNAs with at least twofold changes and their p-values were all less than 0.05. As shown in Figure 1A, the heat map indicated differentially expressed miRNAs and the relative expression levels identified by microarray assay were displayed.
Then, the level of miR-144-3p was further validated by real-time PCR assay, and the results showed that miR-144-3p levels were significantly lower in I/R-injured kidney tissues than in normal tissues ( Figure 1B). To further determine the function of miR-144-3p in the I/R-injured kidney model, we further investigated miR-144-3p expression by RT-PCR assay in a cellular hypoxia reperfusion model using TCMK cells. As shown in Figure 3C, the cells were deprived of oxygen (1% O2)

| Nrf2 was a direct target of miR-144-3p
TargetScan (http://www.targe tscan.org/), microRNA.org (http:// www.micro rna.org/) and miRDB (http://www.mirdb.org) were used to predict the target genes of miR-144-3p. The results in Figure 2A revealed that miR-144-3p can regulate multiple target genes and that Nrf2 was one of them. TargetScan software was first applied to identified the potential binding site of the mouse Nrf2 gene for mmu-miR-144-3p ( Figure 2B). Next, the luciferase construct containing the F I G U R E 1 MiR-144-3p was downregulated in ischemia reperfusion-injured kidney tissue in vivo and hypoxia reperfusion-challenged TCMK cells in vitro. A, IIIumina's Solexa sequencing technology was used to analyze the differentially expressed miRNAs between I/Rinjured and control kidney samples in mice. B, The level of miR-144-3p was measured by RT-qPCR in normal and I/R-injured kidney tissue in vivo (n = 3). C, The level of miR-144-3p in TCMK cells under different hypoxia times (all the groups reperfused for the same time, 6 h) was detected by RT-qPCR (n = 3). Differences between two groups were compared using unpaired t test. *p < 0.05, **p < 0.01 Wt or mut binding sequence of Nrf2 was co-transfected into TCMK cells with the miR-144-3p mimic or miR-NC. The Luciferase report assay was subsequently carried out, and the results showed that the miR-144-3p mimic suppressed luciferase expression containing the WT-Nrf2 sequence but not the Mut-Nrf2 sequence ( Figure 2C). The above results all showed that Nrf2 was a direct target of miR-144-3p in TCMK cells.

| Influence of miR-144-3p on H/R-induced renal tubular epithelial cell apoptosis in vitro
MiR-144-3p, which plays an important role in many diseases, can induce cell apoptosis by regulating oxidative damage. Our experiments confirmed that Nrf2, a key gene that regulates oxidative damage, is one of the direct targets of miR-144-3p. Therefore, it F I G U R E 2 Nrf2 was a direct target of miR-144-3p. A, A schematic diagram of searching target genes of miR-144-3p in three databases. B, A schematic diagram showing the potential binding site between miR-144-3p and Nrf2. C, After luciferase construct and miR-144-3p mimic or miR-NC were co-transfected into TCMK cells, the relative luciferase activity was measured in Nrf2-Wt group and Nrf2-mut group (n = 3). All data are expressed as the mean ± SD; data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. **p < 0.01   . All data are expressed as the mean ± SD; data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. *p < 0.05, **p < 0.01

| TUG1 was upregulated in kidney IRI in vivo and in vitro, and miR-144-3p was a direct target of TUG1
It is known the lncRNA TUG1 can alleviate oxidative stress level and inhibit apoptosis in many diseases. In addition, miR-144-3p is reported to be a direct target of TUG1. To explore whether TUG1 also plays an important role in renal tubular epithelial cells during IRI, the expression of TUG1 in vivo and in vitro were measured by real-time PCR assay. In vivo, TUG1 in I/R-injured kidney tissues was significantly upregulated compared to normal tissues ( Figure 7A). To confirm whether miR-144-3p was a direct target of TUG1, TargetScan

F I G U R E 5
MiR-144-3p regulates I/R-induced kidney damage probably though adjusting Nrf2 expression, oxidative stress and mitochondrial damage in vivo. A, Representative images and respective quantification of Nrf2 expression via immunofluorescence, TUNEL staining, NGAL and cleaved-caspase3 expression via immunohistochemical staining, and mitochondrial structure by TEM in mouse kidneys with different treatments. The black arrows show swollen or vacuolar mitochondria. Data were obtained from 3 images per mouse, with 6 mice per group. B, Levels of serum creatinine in mice with miR-144-3p agomir or miR-144-3p antagomir treatments (n = 6). C, D, SOD and MDA levels in kidney tissue with miR-144-3p agomir or miR-144-3p antagomir treatments (n = 6). All data are expressed as the mean ± SD; data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. *p < 0.05, **p < 0.01 was used to identify the potential binding site of TUG1 for mmu-miR-144-3p ( Figure 7B). After the luciferase construct containing the Wt or mut binding sequence of TUG1 was co-transfected into TCMK cells with the miR-144-3p mimics or miR-NC, the luciferase reporter assay was performed and results showed that miR-144-3p mimics significantly reduced the luciferase activity in TUG1-Wt group ( Figure 7C). Luciferase report assay revealed that miR-144-3p was a direct target of TUG1. Moreover, the result of RNA pulldown assay indicated that miR-144-3p directly interacted with TUG1-Wt but not TUG1-mut, indicating that the interaction between TUG1 and miR-144-3p is sequence-specific ( Figure 7D). Finally, the results of RIP assay showed that TUG1 was detected in Ago2 immunoprecipitates in control group. In contrast, the expression of TUG1 was significantly reduced in Ago2 complexes purified from cell samples transfected with miR-144-3p inhibitor, indicating that TUG1 is likely in the miR-144-3p RISC complex ( Figure 7E). In summary, the above results indicated that TUG1 was upregulated in kidney during renal IRI and that miR-144-3p was directly regulated by TUG1 in TCMK cells.

| TUG1 plays an important role in H/R-induced cell apoptosis possibly through regulating the Nrf2-HO-1 pathway, oxidative stress, mitochondrial damage and endoplasmic reticulum stress via targeting miR-144-3p
To evaluate the specific effect of TUG1 in TCMK cells after H/R treatment, we overexpressed TUG1 via transfection of recombinant adenoviruses and knocked down TUG1 expression via transfection of TUG1 smart silencer, which consisted of three small interfering RNA (siRNA) sequences and three antisense oligonucleotide (ASO) sequences. As a direct target of TUG1, the level of miR-144-3p expression was then detected. Figure 8A shows that TUG1 knockdown increased miR-144-3p expression and that TUG1 overexpression decreased miR-144-3p expression.
Immediately thereafter, we assessed the activation of the Nrf2-HO-1 pathway, degree of oxidative stress, mitochondrial damage and endoplasmic reticulum stress. As shown in Figure 8B,C, the

| The TUG1-miR-144-3p-Nrf2 axis regulates H/R-induced cell apoptosis by adjusting oxidative stress and endoplasmic reticulum stress in vitro
According to our data, the opposite regulation and functional roles of TUG1 and miR-144-3p in I/R-or H/R-induced TCMK cells verified that TUG1, by targeting miR-144-3p, suppressed oxidative stress, mitochondrial damage and endoplasmic reticulum stress-triggered cell death by adjusting the Nrf2-HO-1 pathway. To explore the specific role of miR-144-3p or Nrf2 in the TUG1-miR-144-3p-Nrf2 F I G U R E 7 TUG1 was upregulated in kidney tissues during renal IRI, and miR-144-3p was a direct target of TUG1. A, The TUG1 level was measured in normal and I/R-injured kidney tissues by RT-qPCR (n = 6). B, A schematic diagram showing the potential binding site between TUG1 and miR-144-3p. C, After luciferase construct and miR-144-3p mimic or miR-NC were co-transfected into TCMK cells, the relative luciferase activity was measured in TUG1-Wt group and TUG1-mut group (n = 3). D, RNA pulldown analysis using TUG1-Wt showed that TUG1 directly interacted with miR-144-3p in TCMK cells, and TUG1-Mut failed to pull down miR-144-3p (n = 3). E, After RIP, RT-qPCR was used to detect the total amount of TUG1 bound to Ago2 or IgG (n = 3). All data are expressed as the mean ± SD. Differences between two groups were compared using unpaired t test. Data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. *p < 0.05, **p < 0.01, NS, no significant difference axis, further experiments were performed. Based on Western blot analysis and SOD and MDA detection, we found that TUG1 overexpression reduced cell apoptosis by adjusting the Nrf2-HO-1 pathway, oxidative stress and endoplasmic reticulum stress, which could be reversed by transfected with si-Nrf2 or miR-144-3p mimics. However, TUG1 knockdown exacerbated cell apoptosis, which could be also reversed by transfected with the miR-144-3p inhibitor.
However, this protective treatment could be reversed again once Nrf2 was knocked down at the same time ( Figure 10).

| DISCUSS ION
Renal IR is a common phenomenon in clinical operations such as organ procurement, vascular surgery and renal transplantation. Furthermore, In addition, renal IRI is one of the main clinical causes of ARF and is accompanied by high mortality. 20,21 Therefore, it is clear that it is very necessary to understand the specific molecular mechanism of renal IRI and explore its treatment methods.
In recent years, accumulating researches have revealed that miR-NAs are closely related to the pathogenesis and progression of renal IRI. [22][23][24] In order to clarify the changes of microRNAs systematically and comprehensively during renal IRI, RNA-sequencing analysis was firstly performed. Among all these obviously changed miRNAs, miR-144-3p was preliminarily selected for further studies because it was never studied in renal IRI. Subsequent RT-qPCR detection also confirmed . All data are expressed as the mean ± SD; data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. *p < 0.05, **p < 0.01 closely related to cell apoptosis. Therefore, it is reasonable to speculate that miR-144-3p is a key regulatory factor in renal IRI.
MicroRNAs were widely confirmed to perform functions by inhibiting protein translation via binding to their target mRNA. Nrf2, as an extensively studied gene, was demonstrated to play key roles in oxidative damage related disease including renal IRI. [26][27][28] Besides, several previously studies had already confirmed that Nrf2 was a potential target of miR-144-3p. 29,30 However, the potential relationship F I G U R E 9 TUG1 knockdown aggravated I/R-injured kidney damage by adjusting Nrf2 expression, oxidative stress and mitochondrial damage in vivo. A, Expression of TUG1 in mouse kidneys detected by RT-qPCR (n = 6). B, Levels of serum creatinine in mice treated with TUG1-ASO (n = 6). C, D, SOD and MDA Levels in kidney tissue with TUG1-ASO treatments (n = 6). E, Representative images and respective quantification of Nrf2 expression via immunofluorescence, TUNEL staining, NGAL expression via immunohistochemical staining, HE staining and mitochondrial structure by TEM in TUG1-ASO-treated mouse kidneys. The black arrows show swollen or vacuolar mitochondria. Data were obtained from three images per mouse, with six mice per group. All data are expressed as the mean ± SD; data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. *p < 0.05, **p < 0.01, NS, no significant difference between Nrf2 and miR-144-3p was never studied in a renal IRI reactive cysteine residues and inactivated. 35 As a result, Nrf2 is released from Keap1, and its intracellular expression level has also been improved. Subsequently, Nrf2 enters the nucleus and binds to the sMaf protein to form a heterodimer, 36 which is combined with the antioxidant response element (ARE 5′-TGACXXXGC-3′) in the promoter sequence of the target genes of Nrf2 to promote the expression of its target genes, 37 and HO-1 is one of them. 38  LncRNAs are a set of RNA transcripts of more than 200 nucleotides which can influence posttranscriptional regulation through molecular sponge effects to silence miRNA expression, thus reducing miRNA-mediated suppression of its target genes. [40][41][42] Accumulating studies have revealed the important regulatory roles of lncRNAs in renal IRI including TUG1. 43 obviously increased the expressions of miR-144-3p and accelerated F I G U R E 1 0 The TUG1-miR-144-3p-Nrf2 axis regulates H/R-induced cell apoptosis by adjusting oxidative stress and endoplasmic reticulum stress in vitro. A, B, TUNEL stain and quantitative analysis for H/R-injured TCMK cells with different transfections (n = 3). C-J, The expression levels of Nrf2, H0-1, CHOP, GRP78, Bax, Bcl-2 and cleaved-caspase3 in H/R-injured TCMK cells with different transfections were measured by Western blot analysis. Representative protein bands are shown in F, and the quantitative analysis of protein expression is shown in C-E, G-J (n = 3). K, L, The levels of SOD and MDA in different transfection groups (n = 3). All data are expressed as the mean ± SD; data comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) with Tukey's post hoc test. *p < 0.05, **p < 0.01

F I G U R E 11
Schematic illustrations explains the signal transduction pathway. In short, the TUG1-miR-144-3p-Nrf2 axis regulates ischemia reperfusion-induced cell apoptosis by adjusting oxidative stress, mitochondrial damage and endoplasmic reticulum stress cell apoptosis through inhibiting the Nrf2/HO-1 pathway and the physiological function of mitochondria or endoplasmic reticulum.
However, this proapoptotic effect was rescued when miR-144-3p inhibitor was co-transfected, but reversed again when si-Nrf2 was simultaneously transfected. Our results demonstrated that TUG1 activates the Nrf2/HO-1 pathway by sponging miR-144-3p to play a protective role in renal IRI.
It is important to note that we are not the first to investigate the role of TUG1 in renal IRI, previous two studies had recently studied the effect of TUG1 in regulating renal IR-induced apoptosis or inflammation. 14,48 Interestingly, the results and the conclusions seem inconsistent which could be possibly induced by the differences of animal species, different ischaemia or reperfusion time or differentially used reagents and so on. Besides, it is widely believed that one lncRNA may have multiple interacting miRNAs, and interacting on different miRNAs may conduct completely opposite effects. Therefore, studies on different targets (miRNAs) of TUG1 may lead to different conclusions. On the contrary, our study is generally convincing and many reports support our findings and the final conclusion. TUG1 was previously reported to play protective roles in regulating mitochondrial bioenergetics in diabetic nephropathy, 49,50 and mitochondria were commonly regarded as a key organelle in mediating ischaemia/reperfusion injury-induced cell apoptosis, 51 which indicated that TUG1 may play protective roles through regulating mitochondrial bioenergetics in renal IRI. In addition, TUG1 was also confirmed to have connections with Nrf2 and protect tissues from oxidative damage. 52,53 Meanwhile, the connections between TUG1 and miR-144-3p or miR-144-3p and Nrf2 have been fully discussed above. At last, only male mice were included in our study and the lack of inclusion of female mice might be a limitation of our research. 54 In conclusion, our present study firstly demonstrates that TUG1 plays a protective role in renal IRI by sponging miR-144-3p, activating the Nrf2/HO-1 pathway, improving oxidative stress, mitochondrial damage and endoplasmic reticulum stress, which eventually attenuates apoptosis of renal tubular epithelial cells ( Figure 11). This study provides new insights into the molecular mechanism of renal IRI, and the TUG1/miR-144-3p/Nrf2/HO-1 pathway axis may be a promising new therapeutic target for renal IRI.

CO N FLI C T S O F I NTE R E S T
We declare that we have no commercial or associative interests that conflict with the work submitted.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data supporting the conclusions of this study can be obtained from the corresponding author.