Cross-Tissue Integrative Transcriptomic and Multimodal Analysis Suggests Shared Immune Signatures Linking Lupus Nephritis and Cutaneous Lupus Erythematosus

Meilu Li^1*Changze Song²Zilong Wang²Huisheng Yuan²Zhiqiang Zhang²Yu Sun²Fu Zhang²Songjuan Wang²Hongcheng Sun³Hanshu Zhao³Linyu Zhu²Di Wang²Yuzhen Li⁴

• ¹Department of Scientific Research Center, Seventh Affiliated Hospital, Sun Yat-Sen University, Shenzhen, China
• ²The Seventh Affiliated Hospital Sun Yat-sen University, Shenzhen, China
• ³First Affiliated Hospital of Harbin Medical University, Harbin, China
• ⁴Second Affiliated Hospital of Harbin Medical University, Harbin, China

Background: Lupus nephritis (LN) and cutaneous lupus erythematosus (CLE) are major organ manifestations of systemic lupus erythematosus (SLE), imposing significant health and economic burdens due to their chronic course. This study aims to explore putative shared molecular signatures and hypothesis-generating therapeutic targets by examining the expression profiles of genes associated with LN and CLE. Methods: We analyzed gene expression profiles from LN and CLE using bulk transcriptome analysis, single-cell RNA sequencing, and machine learning approaches. Differentially expressed genes (DEGs) were identified, and weighted gene co-expression network analysis (WGCNA) was employed to reveal gene modules associated with clinical traits. Functional enrichment analyses were performed to characterize implicated pathways. Machine learning algorithms, including LASSO, SVM-RFE, and random forest, were applied to screen for putative biomarkers. Single-cell datasets were used to determine the cellular distribution of candidate genes, and validation was conducted in the lupus mouse model C57BL/6-FasLpr. Results: A total of 361 DEGs in LN and 711 DEGs in CLE were identified, with 99 overlapping genes. We combined overlapping genes from WGCNA and DEGs to conducted enrichment analysis, highlighted disease-associated mechanism enriched in immune-related pathways, particularly type I interferon signaling. Machine learning analysis identified six hub genes, PDE4B, ISG20, IFI27, PARP12, IFI44 and GATA3, most of which demonstrated diagnostic value with AUC values >0.7. Single-cell RNA sequencing confirmed their expression in T cells, B cells, and NK cells, implicating them in immune dysregulation. In vivo validation in Lpr mice revealed elevated serum IFN-β levels and decreased ratio of CD4/CD8, consistent with human transcriptomic findings. Conclusion: This integrative analysis establishes a shared molecular signature between LN and CLE. The identified hub genes represent promising hypothesis-generating molecular signatures and hypothesis-generating therapeutic targets, with potential to improve risk stratification, guide early intervention, and support precision medicine approaches for lupus comorbidities.