Systemic lupus erythematosus (SLE) is one of the autoimmune diseases predominantly affecting young women and involving multiple organs (1). It is characterized by a broad spectrum of clinical symptoms from mild fatigue and rash to severe, life-threatening organ damage, such as hematologic abnormalities, renal injury, and neuropsychiatric involvement (2). The pathogenesis of SLE focuses on abnormal immune system activating and attacking healthy cells and tissues throughout the body (3). SLE is prevalent among Asians, Hispanics, and Americans. The incidence of SLE in China is approximately 30 per 100,000 (4).Due to the huge impact on patients’ quality of life, uncovering the underlying molecular characteristics and mechanism of SLE is very important to discover reliable biomarkers for diagnosis and effective therapy.

At present, transcriptomic and microarray analyses have been widely used to explore promising biomarkers to improve the diagnosis and treatment of disease at the genome level (5–7). Numerous bioinformatic studies have demonstrated different abnormal expression levels of genes associated with the development of SLE. Lu et al. reported that the methylation level of IFI44L promoter could distinguish patients with SLE from healthy people (8). IFIT1 expression was also proven to be associated with podocytes damage and be capable of suppressing some proteins essential to glomerular filtration in MRL/lpr mice (9). Abnormal elevation in IFIT3 was demonstrated to be associated with overactive cyclic GMP-AMP synthase/stimulator of interferon (IFN) genes signaling in SLE monocytes (10). Moreover, other bioinformatic studies have reported abnormal high levels of IFI27, RPL26L1, FBXW11, FOXO1, and SMAD7 in patients with SLE (7, 11). Therefore, exploring the accurate molecular targets included in the occurrence and progression of SLE is imperative. Through the combination of microarray and bioinformatics analyses, exploring potential key genes and pathway networks that are closely related to the development of SLE is possible.

In this study, microarray datasets GSE49454 and GSE20864 for patients with SLE and healthy controls (HCs) were first downloaded from the Gene Expression Omnibus (GEO) database. A total of 178 patients with SLE and 65 HCs were included. The data were integrated and reanalyzed. A total of 136 common differentially expressed genes (DEGs) were identified, including 122 upregulated genes and 14 downregulated genes. DEGs were clustered with Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. Furthermore, a protein–protein interaction (PPI) network was constructed using the online tool STRING, and Cytoscape was used to identify cluster modules and hub genes related to SLE. Subsequently, the selected hub genes were validated using the GEO dataset GSE65391. The diagnostic accuracy of the identified hub genes for SLE was evaluated with the area under the curve of the receiver operating characteristic curve (ROC-AUC). The odds ratios (OR) and 95% confidence intervals (CI) of the relationship between clinical manifestations and hub genes were estimated with multivariable logistic regression. The diagnostic accuracy of the identified hub genes in diagnosing different clinical manifestations was also demonstrated by ROC-AUC. The relationships between the hub gene expression and the SLE disease activity index (SLEDAI) were analyzed through linear correlation. Repeated measures ANOVA was applied to compare the expression levels of the hub genes in patients with SLE at three follow-ups.

SLE is one of the autoimmunity diseases with largely unknown molecular mechanisms (23). As the huge impact on patients’ quality of life, an urgent need for an enhanced understanding of the detailed mechanism to diagnose SLE and assess disease activity exists (24). In the present study, a series of bioinformatic analysis identified 10 hub genes (IRF7, IFI35, IFIT3, ISG15, OAS1, MX2, OASL, IFI6, IFITM1, and IFIH1) between SLE and HC samples based on the gene expression profiles obtained from the GSE49454 and GSE20864 datasets. The results were validated using the GSE65391 dataset. According to the results of ROC-AUC, the 10 hub genes had certain to good diagnostic performances. In addition, the combination model of different hub genes was good in defining different groups among patients with SLE. The expression levels of seven hub genes (except IFI6, OAS1 and IFIT3) were positively associated with disease activity. One-way repeated measures ANOVA revealed that the expression levels of four hub genes (IRF7, IFI35, IFIT3, and ISG15) decreased compared with the baseline expression levels (not remarkably, 0.005<P < 0.05). Furthermore, the biological functions of these common DEGs were investigated. Our GO analysis revealed that these DEGs were significantly associated with IFN response, in line with previous studies (25–27).

IRF7, a member of the IFN regulatory transcription factor family, is a key player in the regulation of many facets of innate immune responses (28). It is reported to be able to induce transcription of IFN-α and IFN-α-induced gene downstream of endosomal toll-like receptors (29). Available data have recently indicated that targeting TLR7-IRF7 signaling may be a promising treatment choice in SLE. One study reported a steady reduction in IRF7 mRNA gradually occurred after autologous hematopoietic stem cell transplant in patients with SLE, coinciding with the relatively low levels of IFN-regulated gene expression. Miyamoto et al. found that the nuclear translocation of IRF7 and IFN-α production in plasmacytoid dendritic cells could be suppressed by BAY11, one of the NF-kB inhibitors, in MRL-lpr mice (30, 31). Our study has detected that IRF7 had good diagnostic performance and was positively associated with disease activity. The expression level of IRF7 was negatively associated with the incidence of alopecia. Meanwhile, IRF7 was significantly decreased during patient follow-up. Therefore, IRF7 could function as a valuable marker for the diagnosis, assessment of disease activity, and treatment response of patients with SLE.

IFI35, IFIT3, IFITM1, and OASL, which are all proteins induced by IFN, have antimicrobial and innate defense abilities. IFI35 is associated with cellular proliferation, and IFIT3 is overexpressed in PBMCs of patients with SLE (32). Lihua Zhang et al. have reported that IFI35 enhanced the proliferation of mesangial cells in patients with lupus nephritis (33). A meta-analysis demonstrated that the gene expression of IFI35 was increased and hypomethylated in SLE and suggested to be a therapeutic target for SLE (34). In 2018, IFIT3 was found to be one of the genes that contribute to the overactive cGAS-STING signaling pathway in SLE monocytes, indicating that IFIT3 may serve as a therapeutic target to block the production of IFN and other proinflammatory cytokines generated by the cGAS-STING signaling pathway (35). The IFIH1 gene, better known as melanoma differentiation-associated gene 5, is located at 2q24.3, and it encodes a cytoplasmic dsRNA helicase belonging to the RLR family (36). A former study has demonstrated an association of IFIH1 rs1990760 polymorphism with SLE susceptibility in a Chinese population (37). OASL belongs to the 2-5-oligoadenylate synthetase (OAS) family (38). Similar to the present finding, a previous study reported upregulated OASL in patients with SLE. However, the exact relationship between OASL and disease activity has not been fully unveiled. ISG15, also related to enhanced type I IFN response, promotes the synthesis of IFNγ by T and B cells. One study recently showed that neutrophil extracellular traps from patients with SLE were characterized by the expression of ISG15 (39). Therefore, ISG15 in lupus pathogenesis may act as a durable proinflammatory stimulation. In the present study, IFI35, IFIT3, IFIH1, OASL, and ISG15, demonstrated good diagnostic ability, and they were associated with disease activity in patients with SLE. Moreover, IFI35, IFIT3, and ISG15 were discovered to be substantially decreased during patient follow-up. All these findings suggested that the IFN signaling pathway acted as a key factor in the pathological process of SLE.

Limited studies focused on the role of IFITM1 in SLE. Notably, IFITM1 plays an essential role in IFN-γ-induced cell proliferation of B-lymphocytes (40). IFITM1 is overexpressed in a large number of solid human tumors (40). The present study showed that IFITM1 has the ability to predict disease activity and identify patients with alopecia. Therefore, further studies are needed to discover its function in SLE pathogenesis.

Our study also demonstrated the relationship between clinical manifestations and different hub genes. Seven clinical manifestations were associated with different hub genes. Furthermore, the diagnostic efficacy of these identified hub genes in recognizing different clinical manifestations were tested with ROC-AUC. The combination model of IFIT3, ISG15, MX2, and IFIH1 was good in diagnosing mucosal ulcers among patients with SLE. The combination model of MX2 and IFIH1 could determine high-risk patients with leukopenia, and the combination model of IFIT3 and IFIH1 had a definite diagnostic accuracy in detecting fever among patients with SLE. These results might help us hierarchically manage patients.

The following limitations should be pay attention to in our study. On the one hand, we only focused on the hub genes, whereas the other DEGs, which may also have a diagnostic and therapeutic effect, were ignored. On the other hand, significance cutoff should be 0.005 after correction of multiple hypothesis testing for 10 genes. Therefore, none of the hub genes had significantly decreased during follow-up. Therefore, more precisely designed clinical studies are necessary to verify these potential genes.

In conclusion, on the basis of integrated bioinformatic analyses, differences in the biological functions in SLE compared with HC samples were identified. In particular, seven valuable genes, namely, IRF7, IFI35, ISG15, MX2, OASL, IFITM1, and IFIH1, were identified as potential biomarkers for the diagnosis and assessment of the disease activity of patients with SLE. Among them, IRF7, IFI35, IFIT3, and ISG15 decreased during the follow-up of patients with SLE (not significantly, 0.005<P < 0.05). In addition, the relationship between the expression levels of different hub genes and the clinical manifestations were investigated, and different combination models were effective in distinguishing patients with different features. Thus, the clinical values of the identified hub genes in SLE were demonstrated, and they might serve as a reference for future experimental research and clinical transformation.

In conclusion, on the basis of integrated bioinformatic analyses, differences in the biological functions in SLE compared with HC samples were identified. In particular, nine valuable genes, namely, IRF7, IFI35, IFIT3, ISG15, OAS1, MX2, OASL, IFITM1, and IFIH1, were identified as potential biomarkers for the diagnosis and assessment of the disease activity of patients with SLE. Among them, IRF7, IFI35, IFIT3, and ISG15 decreased during the follow-up of patients with SLE (not significantly, 0.005<P < 0.05). In addition, the relationship between the expression levels of different hub genes and the clinical manifestations were investigated, and different combination models were effective in distinguishing patients with different features. Thus, the clinical values of the identified hub genes in SLE were demonstrated, and they might serve as a reference for future experimental research and clinical transformation.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

LX designed the study. LX and FZ wrote the manuscript. SL revised the manuscript. All authors read and approved the final manuscript.

This research was supported by Hainan Provincial Natural Science Foundation of China (820QN386). This project is supported by Hainan Province Clinical Medical Center.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.