Jiejie Zhu,1,* Yishan Du,2,* Luyao Gao,3 Jiajia Wang,3 Qiao Mei1 

1Department of Gastroenterology, The First Affiliated Hospital of Anhui Medical University, Hefei City, Anhui Province, People’s Republic of China; 2Geriatric Department, The First Affiliated Hospital of Ningbo University, Ningbo City, Zhejiang Province, People’s Republic of China; 3Department of Pharmacology, School of Basic Medical Sciences, Anhui Medical University, Hefei City, Anhui Province, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Qiao Mei, Email meiqiao@hotmail.com Jiajia Wang, Email wjj@ahmu.edu.cn

Background: The course of Crohn’s disease (CD) is prolonged and many of them may develop kidney stone disease (KSD) with the need for surgical treatment. Therefore, finding biomarkers that can predict CD with KD become increasingly important.
Methods: We obtained three CD and one KSD dataset from GEO database. DEGs and module genes were identified utilizing Limma and WGCNA. We constructed a protein-protein interaction (PPI) network and employed machine learning algorithms to pinpoint potential hub genes (HGs) for diagnosing CD with KSD. We developed a nomogram and receiver operating characteristic (ROC) curve. Additionally, human intestinal cell and proximal tubular epithelial cell models were established to explore the HG levels. Next, we used Cytoscape to build the regulatory networks. Finally, single-cell analysis was performed to investigate specific cell types displaying these biomarkers in CD.
Results: We identified 36 common genes associated with CD and KSD. PYY, FOXA2, REG3A, REG1A, REG1B were identified as HGs utilizing the machine learning algorithm. The nomogram and all five potential HGs exhibited strong diagnostic capabilities. Cell experiments also verified that these genes were markedly expressed in cell models of CD and KSD. Meanwhile, we pinpointed four microRNAs and three transcriptional regulators intimately linked to five crucial genes. Finally, single-cell analysis indicated FOXA2, REG3A, REG1A and REG1B exhibited elevated expression in goblet cells, whereas PYY demonstrated high expression levels in coloncytes.
Conclusion: We determined five biomarkers, including PYY, FOXA2, REG3A, REG1A, REG1B. Our results offer useful perspectives for identifying CD with KSD.

Keywords: Crohn’s disease, kidney stone disease, hub genes, bioinformatics analysis, machine learning, single-cell RNA-seq