This study included 11 treatment-naive patients diagnosed with IgAN via renal biopsy and 11 age- and sex-matched healthy controls. Exclusion criteria encompassed malignancies, severe liver dysfunction, and other rheumatic diseases. Inclusion criteria required: 1) no prior drug therapy or renal tissue puncture before urine collection; 2) IgAN diagnosis based solely on post-collection renal biopsy. All biopsies were evaluated by pathologists blinded to patient outcomes; 3) signed informed consent.

Participants provided three urine samples: first morning, second morning, and a random sample. Urine samples were promptly transferred to 50 mL centrifuge tubes and centrifuged at 490 × g for 10 min. The supernatant was discarded, and the tubes were rinsed with Dulbecco’s phosphate-buffered saline (DPBS) to recover preliminary urine cells. These cells were then centrifuged at 2000 rpm for 5 min, with the supernatant removed and the process repeated twice. The resultant urine cell pellets were stored at −80°C for subsequent analysis. All procedures were carried out on ice to preserve sample integrity.

Urine cells were lysed by adding 30 μL of lysis buffer, followed by centrifugation at 10,000 r/min. The lysate was incubated at room temperature for 6 min. Two microliters of the supernatant were extracted and promptly added to the OneStep reaction system (AccuraCode^® HTP OneStep RNAseq Kit, Singleron, Nanjing, China) while maintained on ice. The mixture was homogenized by pipetting and subjected to reverse transcription in a PCR machine. After reverse transcription, the products were purified and quality-checked. Subsequent steps included fragmentation, adapter ligation, purification of ligated products, PCR enrichment, and library sorting to complete library construction. Libraries that passed quality control were then subjected to bulk RNA sequencing.

This study incorporated 10 independent datasets. Two RNA-seq datasets for IgA nephropathy (n = 41, IgAN = 31, control = 10) were obtained from the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) with accession numbers GSE141295 and GSE210098. Additionally, four microarray datasets for IgAN (n = 170, IgAN = 100, control = 70) were downloaded from GEO with accession numbers GSE37463, GSE93798, GSE99340, and GSE104948. Finally, four scRNA-seq datasets were obtained from GEO: GSE131685 (n = 3, control = 3), GSE171314 (n = 5, IgAN = 4, control = 1), GSE127136 (n = 29, IgAN = 18, control = 11), and GSE140989 (n = 24, control = 24). Specific details regarding each dataset can be found in Supplementary Table 1.

In the analysis of bulk RNA-seq data, the R package sva (version 3.52.0) was used to remove batch effects between samples. Differential expression analysis between normal and IgAN samples in urine and kidney tissue was conducted using the R package edgeR (version 3.42.4) with thresholds of |logFC| > 1 and p-value <0.05. The differentially expressed genes (DEGs) from the three urine sample types were then subjected to an intersection analysis, focusing on genes upregulated only in the second morning urine. These genes were further intersected with those upregulated in IgAN kidney tissue, resulting in a set of genes that are jointly upregulated in both the second morning urine and kidney tissue of IgAN patients. These genes are considered to accurately reflect the kidney’s status. Venn diagrams were generated using the R package VennDiagram (version 1.7.3), and ROC curves were plotted using the R package pROC (version 1.18.5).

To build a robust predictive model with high accuracy, machine learning analysis was performed using the R packages glmnet, caret, and xgboost. Initially, Lasso and Ridge regression were conducted using glmnet to select features, and data preprocessing and model training were performed using the caret package. Cross-validation was employed to optimize model parameters and prevent overfitting. Subsequently, gradient boosting decision trees (GBDT) were trained using the xgboost package to enhance the model’s predictive performance and robustness. An ensemble model was then constructed by combining predictions from multiple models, further improving accuracy and reliability. These steps ensured the reliability and efficiency of the constructed model.

In analyzing microarray data, the R package sva (version 3.52.0) was used to remove batch effects, ensuring data accuracy. The R package pROC (version 1.18.5) was then utilized to plot ROC curves, evaluating the classification performance and predictive efficacy of the models.

Data were downloaded from the Nephroseq v5 renal disease database (http://v5.nephroseq.org/) and subjected to correlation analysis. Scatter plots were generated to illustrate relationships between variables, while box plots were used to display data distribution and differences between groups. Pearson’s correlation coefficient was used to assess the relationship between clinical data and gene expression levels. Differences in gene expression levels between groups were evaluated using Wilcoxon tests. A p-value < 0.05 was considered statistically significant.

ScRNA-seq data were analyzed using the Seurat package (version 5.1.0). Quality control was performed by filtering cells with mitochondrial gene content below 30%, and highly variable genes expressed in at least 3 cells within the 200–2,500 expression range were selected for further analysis. Batch effects were removed using the Harmony package. Cell clusters were constructed using the “FindClusters” and “FindNeighbors” functions and visualized using the t-SNE method. Cells were annotated based on marker genes of different cell types. Gene expression distributions in single-cell RNA-seq data were visualized using the Nebulosa package. Additionally, cell development trajectories and fate decisions in scRNA-seq data were analyzed using the monocle package.

Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed using the R package clusterProfiler (version 4.10.0). GO analysis identified significantly enriched biological processes, molecular functions, and cellular components, while KEGG pathway analysis revealed significant enrichment of target genes in metabolic and signaling pathways. A p-value < 0.05 was considered statistically significant, ensuring the reliability of the enrichment results.

Pseudotime analysis of scRNA sequencing data from Seurat objects was performed using Monocle 2 (version 2.30.1). RNA count matrices were extracted, and a CellDataSet object was created. After normalization and dispersion estimation, highly variable genes were selected. Dimensionality reduction was conducted using the DDRTree method, and pseudotime trajectories were constructed to reveal developmental pathways.

IgAN and normal kidney tissues were fixed in 4% paraformaldehyde for 48 h and subsequently embedded in paraffin. The paraffin-embedded tissues were sectioned into 3-μm-thick slices, followed by deparaffinization and rehydration. Antigen retrieval was performed by heating the sections in EDTA solution, first at medium heat for 8 min until boiling, then at medium-low heat for an additional 7 min. To block endogenous peroxidase activity, the sections were treated with 3% hydrogen peroxide and incubated in the dark at room temperature for 25 min. Non-specific binding was minimized by coating the sections with 3% BSA, followed by a 30-min blocking step at room temperature. After removing the blocking solution, primary antibodies—anti-TYROBP (Zenbio, R383147, 1:100) and anti-HCK (Proteintech, 11600-1-AP, 1:50)—were applied and incubated overnight at 4°C. The following day, sections were incubated with HRP-conjugated secondary antibodies for 50 min at room temperature. Visualization was achieved using diaminobenzidine (DAB) staining. Integrated optical density (IOD) analysis of IHC staining was performed using ImageJ software to quantify the expression levels of TYROBP and HCK in IgAN and normal kidney tissues.

We initially utilized the Drug Signatures Database (DSigDB) to predict potential small molecules that interact with our selected genes. Following this, we employed AutoDock for molecular docking to explore these interactions. The molecular structures of the selected genes were obtained from the PDB database (https://rcsb.org/), while the biomolecular structures of the targets were downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/).

Adhering to standard docking protocols, we performed automated docking simulations between the biomolecules and the small molecules. The strength of the interactions was assessed based on the lowest binding energy. Visualization of the docking results was conducted using CB-Dock2.

All statistical analyses were performed using R (version 4.3.1). A p-value <0.05 was considered statistically significant. Detailed descriptions of additional statistical tools, methods, and thresholds are provided within the methods section.

The overall research approach is outlined in Figure 1. In this study, we collected morning urine, second morning urine, and random urine samples from 11 treatment-naïve IgAN patients and 11 healthy volunteers. Detailed baseline characteristics of the patients are provided in Supplementary Table 2. Following reverse transcription and library preparation, bulk RNA-seq was performed on urine-derived cells, and subsequent data analyses were conducted (Figure 2A). After processing, the final analysis included 5 First morning urine samples, 11 second morning urine samples, and 4 random urine samples from the IgAN group, as well as 11 samples of each type from the healthy control group.

To investigate the differential gene expression in the urine of IgAN patients, we performed differential expression analysis on morning urine, second morning urine, and random urine between the IgAN and healthy control groups (Figures 2B–D). The results revealed 257 upregulated genes in morning urine, 1809 in second morning urine, and 1898 in random urine (Supplementary File 1). Intersection analysis identified 1,269 genes that were specifically upregulated only in second morning urine (Figure 2E). These specific genes (Supplementary File 2) are likely more reflective of the pathological changes in IgAN and may hold potential as biomarkers.

To further explore the role of second morning urine in reflecting renal status, we analyzed IgAN-related datasets (GSE210098 and GSE141295) and identified 549 upregulated genes in IgAN renal tissues (Figure 3A). The complete list of differentially expressed genes (DEGs) can be found in Supplementary File 3. Intersection analysis with the upregulated genes in second morning urine revealed 43 genes that were upregulated in both the second morning urine and renal tissue of IgAN patients (Figure 3B), including genes such as CD86, MMP9, and COL4A1 (Supplementary Table 3). These genes may directly mirror the pathological state of the kidney in IgAN.

KEGG enrichment analyses of these 43 commonly upregulated genes indicated that IgAN is not only associated with the intestinal immune network involving IgA production but also with chemokine signaling pathways, cytokine-cytokine receptor interactions, and the Nucleal factor kappaB (NF-κB) signaling pathway (Figure 3C). GO analysis further revealed that these genes are involved in critical biological processes such as negative regulation of the immune system, leukocyte migration, and myeloid leukocyte differentiation (Figure 3D), as well as being closely associated with cellular structures like the collagen-containing extracellular matrix, the external side of the plasma membrane, and membrane external components (Figure 3E). Functionally, ligand-receptor activity and organic anion transmembrane transporter activity were significantly enriched, highlighting their relevance to IgAN (Figure 3F).

We utilized a comprehensive dataset encompassing 43 genes consistently upregulated in second morning urine samples and renal tissue samples to conduct machine learning-based feature selection and develop a robust diagnostic model. We began our analysis with Lasso regression (Figure 4A), a technique that facilitated regularization and helped isolate a subset of the most significant genes. This was followed by the application of random forest (Figure 4B) and XGBoost algorithms (Figure 4C), both of which are renowned for their ability to manage high-dimensional data and reveal intricate feature interactions.

Through these advanced analytical methods, we honed in on two pivotal genes: TYROBP and HCK (Figure 4D). These genes were subsequently employed to construct a diagnostic model, aimed at improving diagnostic precision and reliability.

To investigate the expression of TYROBP and HCK at single-cell resolution, we analyzed multiple scRNA-seq datasets from the Gene Expression Omnibus (GSE131685, GSE171314, GSE140989, GSE127136). After batch effect correction and principal component analysis (PCA), 70,299 cells were clustered into 20 distinct clusters and categorized into 15 different cell types (Figures 5A, D). Marker genes were sourced from the literature (Liao et al., 2020; Tang et al., 2021; Chen et al., 2024; Menon et al., 2020) and the CellMarker 2.0 database (Hu et al., 2022). Bubble plots illustrate the expression of these marker genes across 20 clusters and 15 cell types (Figures 5B, C).

Violin plots reveal that TYROBP is highly expressed in both monocyte-macrophage and NK-T cell populations, whereas HCK is predominantly expressed in the monocyte-macrophage cluster (Figure 5E). Density plots further corroborate the high expression of TYROBP and HCK specifically in monocyte-macrophages (Figure 5F).

To further investigate the expression patterns of TYROBP and HCK in monocyte-macrophages, we analyzed 3,676 monocyte-macrophage cells and performed subpopulation analysis. t-SNE clustering divided the cells into four subpopulations, with subpopulation 3 predominantly comprising peripheral blood monocytes (Figure 6A). Violin plots reveal that TYROBP is highly expressed across all subpopulations, while HCK is predominantly expressed in subpopulations 0, 1, and 3 (Figure 6B).

Pseudotime analysis indicated that subpopulation 3 primarily consisting of peripheral blood monocytes represent the developmental starting point. Cells were classified into five states along a pseudotime trajectory, transitioning from state 1 through state 5, and gradually shifting towards state 4, reflecting dynamic changes in monocyte-macrophage states (Figure 6C). TYROBP maintains high expression throughout the developmental trajectory and shows an increasing trend over time (Figure 6D). This suggests that TYROBP may be a critical driver in the progression of IgAN. Interestingly, HCK exhibits an expression pattern inversely related to macrophage activation, indicating a possible association with the activity state of monocyte-macrophages (Figure 6D).

These findings underscore the involvement of TYROBP and HCK in the monocyte-macrophage lineage and their potential roles in IgAN pathogenesis and progression. Scatter plots further revealed positive correlations between TYROBP and monocyte-macrophage marker genes, including CD68, CD86, and CD163, within the monocyte-macrophage subpopulations. A similar correlation pattern was observed for HCK. Both TYROBP and HCK were also positively correlated with NFKBIA, indicating a potential link between monocyte-macrophages and the NF-κB signaling pathway in the context of IgAN (Supplementary Figure 1).

IHC staining confirmed the marked upregulation of TYROBP in renal tissues from IgAN patients compared to normal controls (Figure 7A). TYROBP exhibited significantly increased staining intensity in IgAN samples, characterized by a higher density and number of positively stained cells. Similarly, HCK staining was markedly enhanced in IgAN renal tissues relative to normal samples (Figure 7B). These findings highlight the pivotal roles of TYROBP and HCK in IgAN pathophysiology and their potential as diagnostic biomarkers. Quantitative analysis of five randomly selected fields from the IHC images further validated these observations. Integrated optical density (IOD) analysis revealed that the IOD of TYROBP-positive areas in IgAN tissues was significantly greater than in normal tissues, with a statistically significant difference (Figure 7C). A comparable pattern was observed for HCK (Figure 7D).

To ensure staining specificity, IHC negative control experiments were conducted using secondary antibodies without primary antibodies, applied to both IgAN and normal renal tissues. As shown in Supplementary Figure 2, no discernible staining was observed under these conditions, confirming the specificity of the TYROBP and HCK staining patterns observed in Figures 7A, B. Additional validation from the Human Protein Atlas database revealed low baseline expression of TYROBP and HCK in normal kidney tissues (Supplementary Figure 3). This stark contrast emphasizes that their elevated expression is a distinct characteristic of IgAN, rather than a general feature of renal pathology.

Collectively, these findings highlight the potential of TYROBP and HCK as diagnostic markers for IgAN. Their distinct upregulation in IgAN renal tissues, compared to normal tissues, underscores their relevance to the disease’s underlying pathophysiology. Further research is warranted to elucidate their mechanistic roles in IgAN progression, which may provide novel insights into disease biology and inform the development of targeted therapeutic strategies.

We assessed the diagnostic performance of the core genes TYROBP and HCK using receiver operating characteristic (ROC) curves, with second-morning urine data serving as the test set (11 IgAN patients and 11 healthy controls) and GEO datasets (GSE37463, GSE93798, GSE99340, GSE104948) as external validation sets (100 IgAN patients and 70 healthy controls). TYROBP exhibited strong diagnostic capability, achieving an area under the curve (AUC) of 0.909 in the test set and 0.906 in the validation set (Figures 8A, D). Similarly, HCK demonstrated solid performance, with an AUC of 0.843 in the test set and 0.845 in the validation set (Figures 8B, E). When combined, TYROBP and HCK improved diagnostic accuracy, reaching an AUC of 0.942 in the test set and 0.917 in the validation set (Figures 8C, F).

These results highlight the robust potential of TYROBP and HCK as non-invasive biomarkers for diagnosing IgAN using second-morning urine samples. The strong performance across both test and validation sets suggests that these biomarkers could offer a reliable and effective tool for early diagnosis and clinical management of IgAN.

To elucidate the clinical significance of TYROBP and HCK in IgAN, we analyzed their correlation with key clinical parameters using data from the Nephroseq v5 database. Scatter plot analyses revealed a strong inverse correlation between TYROBP expression and glomerular filtration rate (GFR) in IgAN patients (R = −0.68, p = 4e-04) (Figure 9A). Similarly, HCK expression showed a significant negative correlation with GFR (R = −0.68, p = 4e-04) (Figure 9A). In parallel, TYROBP levels were positively correlated with serum creatinine (R = 0.6, p = 0.0015) (Figure 9B), and HCK demonstrated a similar positive association with serum creatinine (R = 0.57, p = 0.003) (Figure 9B).

These findings underscore the potential of TYROBP and HCK as biomarkers for monitoring IgAN progression, given their associations with both decreased renal function and increased serum creatinine levels. Further expression analysis confirmed that both TYROBP and HCK levels were significantly elevated in IgAN patients compared to healthy controls (Figure 9C), reinforcing their potential as key indicators in the pathophysiology and clinical progression of the disease.

In this study, we also employed molecular docking techniques to examine the interactions between key clinical drugs used to slow the progression of IgAN and the proteins TYROBP and HCK. The drugs analyzed included RAS inhibitors (RASi) (Yang et al., 2018; Sugiura et al., 2021), selective sodium-glucose co-transporter 2 inhibitors (SLTG2i) (Dong et al., 2023), Sparsentan (Barratt et al., 2023a; Selvaskandan et al., 2024), and Budesonide (Barratt et al., 2023b; Selvaskandan et al., 2024; Wang J. et al., 2024). Sacubitril valsartan sodium hydrate is a RAS inhibitor that helps lower blood pressure and improve heart function. Dapagliflozin, a selective SGLT2 inhibitor, works by preventing glucose reabsorption in the kidneys, offering renal protection. Sparsentan is a dual-action drug that blocks both endothelin and angiotensin II receptors, targeting renal health. Budesonide is a corticosteroid that reduces inflammation in kidney diseases like IgAN.

Molecular docking analyses revealed that all tested drugs exhibited strong binding affinities for both TYROBP and HCK, with the lowest recorded binding free energies for each compound. Sacubitril valsartan sodium hydrate, for instance, showed binding free energies of −9.1 kcal/mol for TYROBP (Figure 10A) and −8.7 kcal/mol for HCK (Figure 10B). Similarly, Dapagliflozin displayed binding free energies of −10.4 kcal/mol for TYROBP (Figure 10C) and −8.2 kcal/mol for HCK (Figure 10D), while Sparsentan showed binding free energies of −9.5 kcal/mol for TYROBP (Figure 10E) and −8.6 kcal/mol for HCK (Figure 10F). Budesonide demonstrated binding free energies of −7.9 kcal/mol for TYROBP (Figure 10G) and −8.3 kcal/mol for HCK (Figure 10H). In addition, utilizing the DSigDB, we identified Bathocuproine disulfonate as a novel compound that exhibited promising binding to both TYROBP and HCK, with binding free energies of −7.1 kcal/mol for TYROBP (Figure 10I) and −10.7 kcal/mol for HCK (Figure 10J).

These results indicate that all clinically relevant drugs analyzed—Sacubitril valsartan sodium hydrate, Dapagliflozin, Sparsentan, and Budesonide—demonstrate strong binding affinity to the diagnostic proteins TYROBP and HCK, as evidenced by the lowest free energies. Additionally, Bathocuproine disulfonate emerges as a potential novel modulator of IgAN progression, highlighting its promise as a therapeutic agent. The robust interactions between these drugs and key diagnostic proteins underscore the critical role of TYROBP and HCK in IgAN pathophysiology, suggesting that targeting these proteins could enhance therapeutic efficacy. The identification of Bathocuproine disulfonate as a potential new therapeutic candidate presents exciting opportunities for the development of more effective treatment strategies, with the potential to improve patient outcomes in IgAN management.