The miR-15a-5p-XIST-CUL3 regulatory axis is important for sepsis-induced acute kidney injury

Abstract Background: Acute kidney injury (AKI) refers to a sudden loss of renal function. This study was performed to identify the key RNAs acting in the mechanism of sepsis-induced AKI. Methods: Microarray dataset GSE94717 (including six sepsis-induced AKI samples and three control samples) was downloaded from Gene Expression Omnibus database. Differentially expressed miRNAs (DE-miRNAs) were identified. The miRNA targets were predicted and enrichment analysis was performed. Protein-protein interaction (PPI) and competing endogenous RNA (ceRNA) regulatory networks were constructed. Mouse podocytes were treated with lipopolysaccharide (LPS), following by cell viability and PCR analysis. Cellular apoptosis and the ceRNA network were validated. Results: Thirty-one common DE-miRNAs (two up-regulated and 29 down-regulated) by AKI versus control and male AKI versus control were identified. We found the targets of miR-15a-5p, miR-15b-5p, and miR-16-5p were involved in mTOR signaling pathway, and those of miR-29b-3p and miR-16-5p were enriched in PI3K-Akt signaling pathway. RNAs including miR-15b-5p, miR-15a-5p, miR-107, XIST, miR-16-5p, and cullin 3 gene (CUL3) were included in the ceRNA regulatory network. The downregulation of miR-15a-5p and miR-15b-5p and the upregulation of lncRNA XIST and CUL3 gene were validated using qPCR. The miR-15a-5p-XIST-CUL3 regulatory axis was identified and was validated. We confirmed that LPS inhibited the growth of mouse podocytes and seven of the ten miRNAs, but upregulated XIST and CUL3. Transfection analysis showed XIST siRNA enhanced LPS-induced MPC5 cell apoptosis and miR-15a-5p inhibitor reserved it, so did as CUL3 overexpression for miR-15a-5p mimics. Conclusion: The miR-15a-5p-XIST-CUL3 regulatory axis was related to the pathogenesis of sepsis-induced AKI. Highlights Totally, 31 miRNAs were dysregulated between disease and control groups. MiR-15a-5p, miR-15b-5p, and miR-16-5p were involved in mTOR signaling pathway. MiR-16-5p and miR-29b-3p were implicated in PI3K-Akt signaling pathway. The miR-15a-5p-XIST-CUL3 axis was critical for sepsis-induced AKI.


Introduction
Acute kidney injury (AKI) is a sudden loss of renal function that happens in less than seven days [1]. It can be induced by various processes, and the most common causes are dehydration, sepsis, and nephrotoxic drugs [2,3]. AKI is usually diagnosed based on increased blood and urea nitrogen, or insufficient urine produced by the kidneys [4]. The complications of AKI include high potassium content, metabolic acidosis, variation of fluid balance, and uremia [5]. Sepsis-induced AKI occurs in 40% patients with sepsis, namely septic AKI [6]. Recent studies demonstrated that patients with AKI may have increased renal blood flow, tubular injury but not necrosis and apoptosis [6,7]. Therefore, the mechanisms of AKI are complex and should be deeply explored to improve the therapies of the disease.
In the current study, the miRNA expression profile of sepsis-induced AKI was downloaded and analyzed. Through differential expression analysis, miRNA-target prediction, enrichment analysis, and network analysis, the crucial RNAs and ceRNA regulatory relationships were identified in sepsis-induced AKI. This study might further reveal the pathogenesis of the disease and provide theoretical support for its clinical treatment.

Data preprocessing and differential expression analysis
The matrix data of the microarray dataset was acquired and subjected to background correction and normalization using the R package limma (version3.10.3, http:// www.bioconductor.org/packages/2.9/bioc/html/limma. html) [15]. Subsequently, the expression matrix was divided into sepsis-induced AKI group and control group, and the significance of p values of expression difference was calculated based on the unpaired t-test in limma package [15]. The differentially expressed miRNAs (DE-miRNAs) were identified with the thresholds of jlog fold change (FC)j!1 and p values <.01.
Bioinformatics enrichment was performed for the genes included in the miRNA-gene pairs. Gene Ontology (GO), including cellular component (CC), biological process (BP), and molecular function (MF) categories [18] and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway [19] enrichment was conducted combined with DAVID online tool [20] (version 6.8; https://david-d.ncifcrf.gov/). Meanwhile, the R package clusterprofiler [21] (https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html) was utilized to perform KEGG enrichment analysis for the miRNAs in the miRNA-target regulatory network with the number of target genes ranked in the top 10. The significant thresholds for selecting the results of enrichment analysis were set as gene count ! 2 and p values < .05.

Protein-protein interaction (PPI) network analysis for the target genes
The interactions among the genetic productions of the targets were identified in STRING database [22] (version 10.0; http://string-db.org/; combined score > 0.4). PPI network was visualized using the Cytoscape software [17]. To obtain the key target genes, the network topology property index Degree Centrality was used to analyze the scores of network nodes. The higher the node score, the more important the location of the node was in the network. The significant network modules were screened using the MCODE plug-in [23] in Cytoscape software, with the threshold of score > 5.

CeRNA regulatory network analysis
The miRNA-lncRNA pairs involving the DE-miRNAs were screened in starBase database [24] (version 2.0, http:// starbase.sysu.edu.cn/), with the thresholds of low stringency ! 1 and number of cancer types ! 1. The lncRNA and mRNA regulated by the same miRNAs were screened from the miRNA-mRNA pairs and miRNA-lncRNA pairs, namely the miRNA-lncRNA-mRNA or ceRNA pairs. ceRNA regulatory network was visualized using Cytoscape software [17].

Patient collection
A total of five patients (male ¼ 4 and female ¼ 1, aged 45.6 ± 6.9 years old) were collected from Department of ICU, the first people's hospital of Xiaoshan District, Hangzhou, during February 2019 to July 2019. Five sexand age-matched healthy controls (47.1 ± 8.2 years old), without known diseases, were collected from our hospital. The fasting peripheral blood samples were collected from all patients and healthy controls. Blood samples were prepared and RNA was isolated and stored at À20 C before analysis. The human experiments were approved by the Ethics Committee of the First People's Hospital of Xiaoshan District, Hangzhou. Written informed consents were obtained from all participants before blood sampling.

Cell proliferation assay
Cell viability was tested using the Cell Counting Kit 8 (CCK8) assay kit (Beyotime, Shanghai, China). Cells were harvested at 0, 12, 24, and 48 h post LPS induction and then incubated in CCK8 solution for 2 h. Cell viability was detected using a microplate reader (Bio-Rad Labs, Sunnyvale, CA) and the optical density at 450 nm was detected. Each experiment was performed in triplicate.

Cell transfection
MPC5 (1 Â 10 5 cells/well) were seeded into 24-well plates, and then transfected with siRNAs targeting XIST, miR15b-5p mimics, inhibitors, and the scramble sequences (NC; GenePharma, Shanghai, China) for 6 h. For the overexpression of CUL3 gene, cells were transfected with CUL3-overexpressing (OE-CUL3) plasmids constructed by cloning the full length of human CUL3 gene coding region into pcDNA3.1 vectors (Genechem). Empty pcDNA3.1 was used as control for OE-CUL3 transfection. Cell transfections were conducted using the Lipofectamine 2000 (Invitrogen) according to the manufactures' instruction. Then MPC5 cells were treated with 100 ng/mL LPS (Sigma-Aldrich) for 48 h. Cell proliferation was detected after LPS treatment.

Cellular apoptosis
LPS-induced MPC5 cell apoptosis was detected using Annexin V/PI double staining (BD Biosciences, San Jose, CA, USA). Transfected cells (5 Â 10 5 cells/ml) were placed into 6-well plates and then treated with LPS as previously reported. Cells were harvested and then digested into single-cell suspensions, which were then incubated with Annexin V-FITC/PI staining solutions (BD Biosciences) and then detected using a BD FACS Calibur TM flow cytometry (BD Biosciences). Each experiment was performed in triplicate.

Real-time PCR analysis
Cellular and blood RNAs were extracted using TRIzol reagent (Invitrogen). Reverse transcription of miRNA and mRNA was performed with reverse transcription and a DBI Bestar qPCR RT Kit (DBI Bioscience, Shanghai, China), respectively, following the manufacturer's instructions. Primers were synthetized by Sangon (Shanghai, China; Table 1). Applied Biosystems 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) was employed for the PCR amplification. GAPDH was used as reference gene for mRNA and lncRNA, and U6 for miRNA, respectively. The relative expression level of each RNA was calculated using 2 ÀDDCt methods.

Statistical analyses
All data are expressed as the mean ± standard deviation (SD). Student's t-test was used for differences between groups. A p value <.05 was considered statistically significant.

Differential expression analysis
We firstly identified the DE-miRNAs from male patients (n ¼ 4), and 51 DE-miRNAs, including 10 up-regulated Table 1. The sequences of the used PCR primers.

Gene
Primers and 41 down-regulated DE-miRNAs were identified versus controls (Figure 1(A)). With the mixing of two female samples (whole), 31 DE-miRNAs (two up-regulated and 29 down-regulated) were identified in sepsisinduced AKI group (male ¼ 4, female ¼ 2) compared with control group (male ¼ 1, female ¼ 2; Figure 1(A)), and all the 31 DE-miRNAs were common DE-miRNAs between the two comparisons. The bidirectional clustering heatmap of 31 common DE-miRNAs could clearly distinguish the samples in different groups (Figure 1(B)). We then focused on the function of the 31 common DE-miRNAs.

MiRNA-target prediction and enrichment analysis
A total of 564 miRNA-target pairs (involving 17 downregulated miRNAs) were predicted ( Figure 2). The top 10 miRNAs with higher interaction degrees in the regulatory network are listed in Table 2. Enrichment analysis showed that genes in the miRNA-target pairs were enriched into 111 GO_BP (such as positive regulation of transcription from RNA polymerase II promoter), 33 GO_CC (such as cytoplasm), and 43 GO_MF (such as protein binding), and 19 KEGG pathways (such as focal adhesion). The top 20 terms in each category are presented in Figure 3. Clusterprofiler analysis showed the top 10 miRNAs (down-regulated) were associated with 41 KEGG pathways, and the top 5 pathways enriched for each miRNA are shown in Figure 4. Especially, miR-15a-5p, miR-15b-5p, and miR-16-5p were involved in mTOR signaling pathway, and miR-16-5p and miR-29b-3p were enriched in PI3K-Akt signaling pathway.

PPI network analysis for the target genes
The PPI network for the target genes included 281 nodes (productions) and 776 edges (interactions; Figure 5). Three significant network modules (module A, score ¼ 15, involving 15 nodes and 105 edges; module B, score ¼ 8.2, involving 11 nodes and 41 edges; module C, score ¼ 6, involving 6 nodes and 15 edges) were identified ( Figure 6). The top 10 nodes in the PPI network and the nodes in the three significant network modules are listed in Table 3.
compared with sepsis-induced AKI and sepsis-non AKI [27]. However, some miRNAs were also identified to be downregulated in sepsis-induced AKI versus sepsis-non AKI, like miR-15b-5p, miR-15a-5p and miR-16-5p. We identified several key miRNAs, including miR-15a-5p, miR-15b-5p, and miR-16-5p were involved in mTOR signaling pathway, and miR-16-5p and miR-29b-3p were associated with PI3K-Akt signaling pathway, the similar pathways identified by Ge et al. These two pathways have been reported to be impressive in ischemia-reperfusion (I/R)-induced AKI [28]. Additionally, miR-29b-3p, miR-15b-5p, miR-15a-5p, miR-107, XIST, miR-19a-3p, miR-16-5p, miR-19b-3p, miR-92a-3p, and CUL3 were the top 10 nodes in the ceRNA regulatory network. Therefore, the significant downregulation of these miRNAs might suggest the important roles of them in sepsis-induced AKI via involving mTOR and PI3K-Akt signaling pathways. The overexpression of miR-29a was found to be independent risk factor for the mortality of patients with septic AKI, and thus they can be applied for predicting the 28-day mortality of the disease [29]. Huo et al reported that the expression level of miR-29a were positively correlated with the levels of serum creatinine, cystatin C, and kidney injury molecule-1 (KIM-1) in patients with septic AKI. However, Drummond et al reported that miR-29a-3p was increased in the kidney of mice exposed to e-cigarette air. The expression of its targets including Collagen 1A1 and Fibrillin 1 as well as renal fibrosis was significantly increased [30]. Overexpressed miR-107 Figure 8. The validation experiments. A and B, the qPCR analysis for several differentially expressed miRNAs in five patients with AKI (male ¼ 4 and female ¼ 1) versus healthy controls (sex and age-matched), and that in LPS-induced cellular AKI model. C, the cell viability by CCK8 assay in MPC5. Ã p < .05 vs. blank. ÃÃ p < .01 vs. blank. LPS: lipopolysaccharide. Figure 9. The results of dual-luciferase reporter assay. The interaction between lncRNA and miRNA (A) and between miRNA and target (B) were predicated using the LncBase Predicted v(0).2 and miRTarBase, respectively. The interaction was validated using dual-luciferase reporter system. ÃÃ p < .01 vs. NC. causes TNF-a secretion through regulating dual-specificity phosphatase 7(DUSP7) in circulating endothelial cells, which may lead to tubular cell damage in sepsisinduced AKI [31]. Our present study demonstrated that the DUL3 is a predictive target of miR-107 and was upregulated by LPS treatment in MPC5. CUL3 is part of the ubiquitin proteasomal system and the Cul3 disruption is associated with tubulointerstitial fibrosis [32]. Saritas et al reported that the upregulation of CUL3 in patients with AKI and Cul3 disruption increased the expression of cyclin E and p21 and the promotion of proximal tubule injury in mice [32]. In addition, other studies identified the elevation of miR-92a-3p in chronic kidney injury (CKD) and CKD-associated atherosclerosis [33,34]. These suggested that miR-29b-3p, miR-107, and miR-92a-3p might be correlated with the pathogenesis of sepsisinduced AKI. The specific mechanism mediated by them should be explored. The downregulation of miR-15a-5p has been identified in CKD with hypertension [35,36]. In addition, Ge et al suggested the downregulation of miR-16-5p, miR-15a-5p and miR-15b-5p in sepsis-induced AKI versus sepsis non AKI [27]. This consistency showed the important roles of these miRNAs in AKI. The difference is our present study identified lncRNAs related to these miRNAs. The expression of lncRNAs XIST and NEAT1 are significantly increased in glomerular and tubular epithelial cells, and urinary XIST serves as a potential marker for detecting membranous nephropathy [37]. Elevated XIST, which could be induced by LPS, is correlated with glomerular nephritis [37]. Our in vitro experiment using LPS-treated MPC5 cells showed XIST was elevated by LPS stimulus. We identified XIST regulated the expression of upregulated CUL3 by sponging the top 10 downregulated miRNAs, including miR-15a-5p, miR-16-5p and miR-107. The upregulation of CUL3 has been identified in patients with AKI [32]. Mutations in CUL3 reduced ubiquitination in the kidneys [38,39]. These studies indicated that XIST plays impressive roles in AKI via regulating CUL3 by sponging miRNAs. The miR-15a-5p-XIST-CUL3 regulatory axis in the ceRNA regulatory network was validated using the in vitro cellular experiment. We also demonstrated that miR-15a-5p inhibition reserved MPC5 cell apoptosis that was enhanced by XIST siRNA, so did as the overexpression of CUL3 for miR-15a-5p mimics-induced apoptosis in MPC5 cells. These data suggested that miR-15a-5p-XIST-CUL3 played crucial roles in sepsis-induced AKI, and the management targeting this axis might be of great value for controlling AKI development.

Disclosure statement
No potential conflict of interest was reported by the authors. Figure 10. The apoptosis of MPC5 cells. A and B, the validation of XIST/miR-15a-5p axis and miR-15a-5p/CUL3 axis on cell apoptosis, respectively. ÃÃ p < .01 vs. Blank. # and ## notes p < .05 and .01 vs. LPS, respectively. LPS: lipopolysaccharide.