XAF1 is identi ed as a novel hub gene and associated with the clinical characteristics of lupus nephritis

Xingruo Zeng Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Tian Xie Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Jiaxing Sun Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Yufei Lei Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Lin Jia Department of Nephrology, The Central Hospital of Wuhan, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, Hubei 430014, China Zimeng Wei Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Muhammad Jamal Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Hong Qin Department of Medical Insurance Management, Zhongnan Hospital of Wuhan University, Wuhan, China qiuping zhang (  qpzhang@whu.edu.cn ) Department of Immunology, School of Basic Medical Science, Wuhan University, Wuhan, Hubei 430071, China Hubei Provincial Key Laboratory of Developmentally Originated Disease, Wuhan University, Wuhan, Hubei 430071, China


Results
The 14 genes (13 upregulated and 1 downregulated) ultimately screened out as candidate hub genes of the pathogenesis of LN. Moreover, Correlation analysis between the unexplored hub genes and clinical features of LN suggested that XAF1 may involve in the progression of LN. Finally, our data demonstrated that the expression level of XAF1 was upregulated in LN compared with IgA nephropathy (IgAN) and related to the WHO Lupus Nephritis Class and the quantitative 24 h proteinuria of LN patients.

Conclusions
The current study proposed XAF1 as a novel hub gene in LN which may perform as a brand-new biomarker or therapeutic target of LN in the future.

Background
Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by the loss of selftolerance and formation of nuclear auto-antibody and immune complexes resulting in in ammation of multiple organs. Over the past 10-20 years, with the advent of earlier diagnosis and recognition of disease, as well as the introduction of newer less toxic therapeutic measures, the mortality in SLE patients have certainly decreased [1]. However, increased mortality remains part of the natural history of lupus [2][3][4] and SLE patients still have two to ve times the risk of death compared with the general population [1]. Lupus nephritis (LN) is one of most serious complication of SLE [5,6] which affects up to 60% of SLE patients [7,8] and is a major predictor of poor prognosis in patients with SLE. The development of LN involves multiple pathogenic pathways including aberrant apoptosis, autoantibody production, immune complex deposition and complement activation [9]. However, the local tissue effects are independent of hematopoietic cell in uence, which are major contributors to end-organ damage in LN [10]. Thus, the pathogenesis involved in the local tissue damage should be of great concern and the therapies that limit tissue damage by targeting renal parenchymal cells may also prove useful in the treatment of LN. In spite of great progresses have been made in understanding pathogenesis of LN through the use of genetic variant identi cation, mouse models, gene expression studies, and epigenetic analyses, the pathogenesis of LN remains unclear and also has been hampered by disease heterogeneity.
Recently, with the development of bioinformatics technology, the gene expression pro ling analysis of the whole transcriptome has increasingly been used to explore the pathogenesis-associated genes, classify different types of disease and predict clinical outcome [11]. A series of bioinformatics analysis methods, including differentially expressed genes (DEGs) investigation, function and pathway enrichment analyses, as well as protein-protein interaction (PPI) network analyses, were performed based on gene expression pro les [11][12][13][14]. LN is an in ammatory condition of the kidneys that encompasses various patterns of renal disease including glomerular and tubulointerstitial pathology. In view of the crucial role for local tissue effects in the pathogenesis of LN, exploring the underlying mechanisms, and nding e cient therapeutic strategy for retarding renal damage are quite necessary.
The aim of this study is to identify common marker genes across glomerulus and tubulointerstitial in LN.
We analyzed the gene expression in glomerulus and tubulointerstitial of LN using Gene Expression Omnibus (GEO) database, and eventually identi ed 14 hallmark genes which may be related to the pathogenesis of LN. Correlation analysis between the unexplored hub genes and clinical features of LN suggested that XAF1 may be a novel hub gene involved in the progression of LN. Moreover, the immunohistochemistry results of LN patients' kidney biopsies demonstrated that the expression level of XAF1 is upregulated in LN group and related to the WHO Lupus Nephritis Class of LN and 24 h urine protein quantitation. The current study proposed XAF1 as a novel candidate gene in LN and may perform as a brand-new biomarker or therapeutic target of LN in the future.

Methods
The microarray dataset The microarray dataset GSE32591 was analyzed with the GPL14663: Affymetrix GeneChip Human Genome HG-U133A Custom CDF. A total of 47 samples were used in this dataset, including 15 healthy living donors and 32 LN patients. Demographic, clinical, and histologic characteristics of these patients were no statistical difference in any parameters between the LN cohorts used in arrays and RT-PCR [15]. All specimens were kidney biopsy.

Identi cation of DEGs
The gene expression pro les GSE32591 were acquired from GEO database [15]. The array data of the dataset consists of 32 glomeruli and 32 tubulointerstitium from LN patients and 15 glomeruli and 15 tubulointerstitium from control living donors. DEG was obtained from GEO database by a way of GEO2R analysis (http://www.ncbi. nlm.nih.gov/geo/geo2r/). The adj. P < 0.05 and |log2FC|>1 was set as DEGs cutoff criterion.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of DEGs
To predict the biological processes or pathways that DEGs may be involved in, GO and KEGG enrichment analyses of DEGs were performed by R package "clusterPro ler" [16]. GO can be divided into three parts: molecular function (MF), biological process (BP) and cellular component (CC). R package "ggplot2" was used to visualize GO terms and KEGG pathways.

PPI network construction and module analysis
The PPI network was predicted using Search Tool for the Retrieval of Interacting Genes (STRING; https://string-db.org/) (version 11.0) online database [17]. Analyzing the functional interactions between proteins may provide insights into the mechanisms of generation or development of diseases. In the present study, PPI network of DEGs was constructed using STRING database, and an interaction with a combined score > 0.9 was considered statistically signi cant. Cytoscape (version 3.6.0) is an open source bioinformatics software platform for visualizing molecular interaction networks [18]. The plug-in Molecular Complex Detection (MCODE) (version 1.5.1) of Cytoscape is an APP for clustering a given network based on topology to nd densely connected regions [19]. The PPI networks were drawn using Cytoscape and the most signi cant module in the PPI networks was identi ed using MCODE. The criteria for selection were as follows: MCODE scores > 5, degree cut-off = 2, node score cut-off = 0.2, k-score = 2 and Max depth = 100.

Nephroseq analysis
Nephroseq analysis (https://www.nephroseq.org/resource/login.html#) is a platform to the academic for integrative data mining of genotype/phenotype data. Nephroseq combines a wealth of publicly available renal gene expression pro les, which is gathered and managed by an experienced term of data scientists, bioinformaticians, and nephrologist [20]. We performed analysis on human kidney biopsy samples of comparison of all genes in LN vs other diseases, and examined the function of unexplored hub genes.

Human kidney sample preparation
Eighteen LN and seventeen IgA nephropathy (IgAN) kidney tissues were collected from the Central Hospital of Wuhan. Kidney tissues were immediately snap-frozen in liquid nitrogen, and stored at -80 °C for further usage. The present study was approved by the Ethics Committee of the Central Hospital of Wuhan. Written informed consents were obtained from all patients prior to enrollment in the study and anonymity was guaranteed.

Immunohistochemistry staining
Kidney cryo-sections at 3-µm thickness were xed for 15 min in 4% paraformaldehyde, followed by permeabilization with 0.2% Triton X-100 in 1 × phosphate-buffered saline for 5 min at room temperature. After blocking with 2% donkey serum for 60 min, the slides were immunostained with anti-XAF1 (cat: ab17204, Abcam Biotechnology, USA). Slides were viewed with an Olympus Epi-uorescence microscope equipped with a digital camera.
Statistical analysis SPSS version (v. 21.0) and GraphPad Prism (v. 8.0) were used for statistical analysis and generating gures. Unpaired t-test and one-way ANOVA followed by Dunnett's test was used to compare the expression of XAF1 in different groups. Pearson correlation analysis was used to analyze the correlation between XAF1 expression and clinical characteristics. P < 0.05 was considered statistically signi cant.

Identi cation of DEGs
The gene expression pro le of GSE32591 was used to screen out signi cant differently expressed genes (DEGs) in the glomeruli and the tubulointerstitial respectively. Using adj. P < 0.05 and |log2FC| > 1 as cutoff criterion, 351 DEGs (250 upregulated and 101 downregulated) were identi ed in the glomeruli

Functional and pathway enrichment analysis of DEGs
To analyze the biological classi cation of DEGs, functional and pathway enrichment analyses were performed by using R package "clusterPro ler". In glomeruli, GO analysis results showed that changes in BP of DEGs were signi cantly enriched in response to virus and defense response to virus, negative regulation of viral life cycle, type I interferon signaling pathway, cellular response to type I interferon, negative regulation of viral process, response to type I interferon (Fig. 2a). Changes in CC of DEGs were mainly enriched in the secretory granule membrane, cytoplasmic vesicle lumen, vesicle lumen and membrane microdomain (Fig. 2a). Changes in MF were mainly enriched in inorganic acid binding, glycosaminoglycan binding and cytokine binding (Fig. 2a). As for the pathways in glomerulus, the results of KEGG enrichment analysis showed that DEGs were enriched in In uenza A, Tuberculosis, Staphylococcus aureus infection, Epstein-Barr virus infection, Phagosome (Fig. 2b).
In tubulointerstitial, GO analysis results showed that DEGs were enriched in various BPs, the top 5 terms were type I interferon signaling pathway, cellular response to type I interferon, response to type I interferon, response to virus, defense response to virus, response to interferon-gamma, negative regulation of viral process, regulation of multi-organism process (Fig. 2c). Changes in CC of DEGs were mainly enriched in MHC protein complex, blood microparticle, extracellular matrix (Fig. 2c). Changes in MF were mainly enriched in organic acid binding, glycosaminoglycan binding and cytokine binding (Fig. 2c). KEGG pathway analysis revealed that the DEGs were mainly enriched in Epstein-Barr virus infection (Fig. 2d).

PPI network construction and module analysis
The PPI network of DEGs in glomeruli was constructed with 195 nodes and 789 edges (Fig. S1a) and the most signi cant module was obtained using Cytoscape (Fig. 3a), while in tubulointerstitial, the PPI network of DEGs was constructed with 84 nodes and 481 edges (Fig. S1b) and the most signi cant module showed in Fig. 3b. There are 22 and 23 genes in the most signi cant modules of glomeruli (module 1) and tubulointerstitial (module 2), respectively.

Hub gene selection and analysis
A total of 24 genes in glomeruli (hub gene 1) and 23 genes in tubulointerstitial (hub gene 2) with degrees ≥ 20 were identi ed. In our study, we selected the overlap genes of "module 1", "module 2", "hub gene 1" and "hub gene 2" as shown in the Venn diagram. The results showed that 14 DEGs, including 13 upregulated genes (IFITM1, IFIT1, IFI6, IFITM3, ISG15, MX2, XAF1, IFIT3, IFIT2, RSAD2, OAS1, IFI27, MX1) and 1 down-regulated genes (EGR1), were the common genes between these four groups, and were identi ed as hub genes for further investigation (Fig. 4a). The gene symbols, full names, implications, and expression changes of these 14 hub genes are shown in Table 1. We further performed GO analysis of the 14 hub genes, with the criteria of adj. P < 0.0001 and q value < 0.0001. Following these criteria, 18 GO terms were signi cantly enriched (Fig. 4b). All the 14 genes were enriched in the type I interferon related terms (type I interferon signaling pathway, cellular response to type I interferon, response to type I interferon) ( Table 2).

Analysis of the DEGs based on Nephroseq
Based on the results from GEO databases and a series of analysis, we found that the important 14 genes may be involved in the pathogenesis of lupus nephritis. To further verify the induction of these 13 upregulated genes in LN, we analyzed the Nephroseq database based on Comparison of All Genes in Ju Chronic Kindey Disease Glomeruli (Ju CKD Glom) and tubulointerstitial (Ju CKD TubInt) study, lupus nephritis vs other diseases. The dataset showed that the expression of these genes was also increased in the kidney glomeruli and tubulointerstitial of LN patients (Fig. 5a-b). Subsequently, the unexplored hub genes IFITM3 and XAF1 in SLE/LN were further analyzed. The correlation between unexplored hub genes and clinical manifestation was performed on Nephroseq online platform. We found that the expression of XAF1 increased in lupus mouse model with proteinuria compared with no proteinuria (Fig. 5c).

The expression of XAF1 in kidneys of lupus nephritis
The  Table S1. In comparison with the IgAN group, the expression of XAF1 was upregulated in LN tissue by immunohisto-chemistry ( Fig. 6a-b). In LN group, the expression of XAF1 is related to the WHO Lupus Nephritis Class, which was increased in Class V + IV LN and Class V LN compared to Class IV LN, respectively (Fig. 6c). Moreover, the expression of XAF1 shows a positive correlation with the quantitative 24 h proteinuria (P < 0.05) (Fig. 6d).

Discussion
LN is the most common severe complication of SLE [5] and contributes signi cantly to mortality in this disease [21,22]. Despite currently available aggressive treatments, up to 50% of patients progress to endstage renal disease within 5 years of diagnosis [21,22]. As previously noted, most research and therapeutic target in clinical practice focus almost exclusively on glomerular pathology. More and more researches support the importance of tubulointerstitial in ammation in determining prognosis and patient outcomes [23][24][25]. Thus, kidney involvement in LN can affect either glomerular or tubulointerstitial compartments as well as combinations thereof. Here we used the bioinformatics analysis to identify the hub genes in glomerular and tubulointerstitial of LN. The hub genes could be used to elucidate the pathogenesis of this disease, and might be important biomarkers and/or therapeutic targets for LN.
In our study, microarray dataset was used to identify the DEGs in both glomerular and tubulointerstitial of LN, and total of 351 DEGs (250 upregulated and 101 downregulated) and 129 DEGs (104 upregulated and 25 downregulated) were identi ed in glomerular and tubulointerstitial, respectively. Next, we predicted the DEGs functions based on GO and KEGG pathway enrichment analysis. Based on the PPI network, 14 DEGs, including 13 up-regulated and 1 down-regulated genes were recognized as hub genes.
Unexpectedly, GO analysis of the 14 hub genes showed that all these 14 genes were enriched in the type I IFN related terms. It is well documented that the type I IFN signature is a feature of LN. Increased level of IFN in serum of patients with SLE was already described 40 years ago and were later identi ed as type I IFN [26]. IFN is important in both the in ammatory process and development of damage in LN. Kidney biopsies of patients with SLE showed increased expression of IFN-inducible genes [27][28][29][30] and plasmacytoid dendritic cells accumulate in glomeruli of patients with active disease [31].
Type I IFN, as a central mediator in the pathogenesis of LN, may activate innate and adaptive immunity and intrarenal pathogenic mechanisms. Both direct and indirect effects of IFNs result from induction of a subset of genes, called IFN stimulated genes. The 13 up-regulated genes including IFITM1, IFIT1, IFI6,   IFITM3, ISG15, MX2, XAF1, IFIT3, IFIT2, RSAD2, OAS1, IFI27, MX1, were almost IFN-inducible genes. The demonstration of a broad IFN-I-induced gene transcript signature in SLE PBMCs emerged from several laboratories [32,33]. Recent data from epigenetic analyses of hypomethylated genome sites support activation of many genes related to type I IFN signaling in SLE patients. IFIT1 is the rst gene described as a candidate gene for SLE, and may function by activating Rho proteins through interaction with Rho/Rac guanine nucleotide exchange factor [34]. Wang J, et al. have found that IFIT3 is one of the genes that contributes to the overactive cGAS/STING signaling pathway in human SLE monocytes [35]. IFITM1 were found to be up-regulated in platelets from SLE patients compared with healthy volunteers [36]. The ISG15 mRNA level was higher in whole blood cell counts of SLE patients when compared with the disease control and healthy control groups and ISG15 expression correlated with lymphocytopenia in active SLE patients [37]. The epigenome-wide DNA methylation study in lupus showed signi cant hypomethylation of differentially methylated sites was associated with several interferon-related genes, including MX1, IFI44L, IFIT1, RSAD2 and IRF7 in PBMCs [38]. However, the role of IFITM3 and XAF1 in SLE/LN has not been reported. In our study, XAF1 was found to be upregulated in both glomerular and tubulointerstitial of LN based on dataset GSE32591. Additionally, the clinical manifestation detection showed the XAF1 expression could be associated with proteinuria in the lupus mouse model. Therefore, we speculated that XAF1 participants in the progression of LN, and may be a novel biomarker and therapeutic target for LN.
XAF1, a novel IFN stimulated gene, was identi ed in gene array studies in IFN-sensitive melanoma cells (WM9) [39]. XAF1 was discovered in a yeast two hybrid studies as a XIAP (X-linked inhibitor of apoptosis protein) -interacting protein [40] and seemed to function as a negative regulator of members of the IAP (inhibitor of apoptosis protein) family. Overexpression of XAF1 resulted in neutralization of XIAP's ability to inhibit cell death [40]. It is well known that, XAF1 as a proapoptotic tumor suppressor is always inactivated in multiple human cancers. XAF1 was identi ed ubiquitously in all normal adult and fetal tissues but was present in very low levels in a variety of cancer cell lines [41][42][43][44][45]. Both IFN-α2 and IFN-β were found to induce XAF1 transcription. Type I IFN may therefore inhibit XIAP function by the induction of XAF1, and then negatively regulate the inhibitor of apoptosis. XAF1 was upregulated in whole peripheral blood from the Sjögren's syndrome patients compared with controls [46].  [48]. Therefore, we speculated that the expression of XAF1 might be induced by the subepithelial immune complex deposits in lupus kidney tissue and associated with podocyte injure. Meanwhile, the expression of XAF1 was positive correlation with quantitative 24 h proteinuria, which indicated that XAF1 may be implicated in the kidney ltration barrier and tubular reabsorption dysfunction.

Conclusions
In summary, the bioinformatics analysis indicated the up-regulation of XAF1 in kidney tissue may be involved in the pathogenesis of LN. Based on this nding, our further detection of renal tissue indicated that XAF1 was upregulated in the kidneys of LN compared with IgAN, and the expression of XAF1 was associated with pathological type and the proteinuria. Our study may highlight the novel biomarker and therapeutic targets for LN. However, regarding the limited patient number included in this study, the results are preliminary and more studies are still needed to further decipher the role of XAF1 involved in the pathogenesis of LN. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by the National Natural Science Foundation of China (Nos. 81770180) and Hubei Provincial Natural Science Fund for Creative Research Groups (2018CFA018).
Authors' contributions QPZ designed and managed the whole research; XRZ, TX performed the experiments, analyzed the data and wrote the main manuscript text; XJS, YFL, ZMW, Jamal revised the manuscript; LJ provided professional advices about the research. All authors read and approved the nal manuscript.