Hub genes and molecular mechanisms related to hypospadias based on transcriptome sequencing data and bioinformatics analysis

Abstract

Background

Hypospadias, a prevalent congenital anomaly affecting patients’ quality of life, remains incompletely understood in terms of its pathogenesis. This study aimed to identify key genes and molecular mechanisms through transcriptome sequencing and bioinformatics analysis.

Methods

Transcriptome sequencing was conducted on foreskin samples from 15 patients with hypospadias and 15 controls. Differential expression analysis and WGCNA were used to identify differentially expressed genes (DEGs) and key modules. Genes common to both approaches were further analyzed through PPI network filtering, followed by feature selection using LASSO and SVM-RFE algorithms. Hub genes were identified based on an AUC > 0.8. Functional, immune, and regulatory analyses were performed to investigate their biological roles.

Results

Three hub genes—CSF3R, SELL, and FCN1—demonstrated significant diagnostic potential (AUC > 0.8). Functional enrichment analysis revealed their association with cytokine receptor interaction and chemokine signaling pathways. Eleven immune cell types, including activated dendritic cells, were found to be elevated in the patient cohort. All hub genes were upregulated in the disease group.

Conclusion

CSF3R, SELL, and FCN1 represent potential diagnostic biomarkers for hypospadias and provide valuable insights into its molecular mechanisms and potential therapeutic approaches.

Introduction

Hypospadias is among the most prevalent congenital penile malformations, with an incidence ranging from 0.2 to 4.1 per 1000 live births [1, 2]. This condition is characterized by three primary anatomical abnormalities: mislocation of the urethral meatus, penile curvature, and abnormal foreskin distribution. The ectopic urethral opening, resulting from incomplete fusion of the penile folds during embryogenesis, is typically located on the ventral aspect of the penis, between the glans and the perineum [3]. The etiology of hypospadias is multifactorial, encompassing genetic predisposition, insufficient prenatal hormonal influence, maternal placental factors, and environmental exposures [4, 5]. Among them, genetic factors play a significant role in the occurrence of hypospadias. Existing studies have clearly identified several genetic markers associated with an increased risk of male infant hypospadias. For example, an increase in the number of MAMLD1 gene variations may lead to hypospadias [6]. Genetic variations in TRIM67 and DAB2IP genes can participate in the occurrence and development of hypospadias by enhancing p38 MAPK phosphorylation [7]. Chen have clearly demonstrated that mutations in the ODNAH gene increase the risk of hypospadias by affecting steroid biosynthesis and testosterone synthesis [8].

Surgical interventions for hypospadias are varied, but studies indicate that postoperative lower urinary tract symptoms (LUTS) occur at twice the rate in affected individuals compared to controls. Approximately 39% of patients experience voiding dysfunction, with hesitancy and urinary splashing being most common [9]. While mild hypospadias repair yields only minimal differences in maximum flow rate (Qmax) compared to controls, those with severe hypospadias exhibit significantly reduced Qmax post-surgery, as well as potential sexual dysfunction, including impaired erection and ejaculation [10]. Obstructive urinary flow patterns are frequently observed following tubularized incised plate urethroplasty (TIP) [11]. Long-term complications, such as meatal stenosis, fistulas, and urethral strictures, may persist for years after the initial repair, highlighting the need for ongoing follow-up [12]. The optimal timing for hypospadias surgery is typically between 6 and 18 months, depending on the severity of the condition, penile development, and the surgical approach required [13]. Previous studies have reported that after TIP surgery in children at this age, most of them maintain good long-term urinary function; in terms of sexual satisfaction, erectile function, and frequency of sexual intercourse, there are no significant differences compared to the healthy control group. However, the sexual climax function of men after hypospadias surgery is slightly lower than that of the control group, and this phenomenon is more obvious in patients with proximal hypospadias [14, 15]. This conclusion also confirms that the good long-term prognosis of TIP urethral reconstruction largely depends on timely surgical intervention within this time window, thereby highlighting the importance of early diagnosis.

However, early diagnosis of hypospadias remained fraught with challenges at that time. Prenatal ultrasound has a reported positive predictive value of 72% for detecting hypospadias [1]. Its accuracy is significantly influenced by the gestational age stage and maternal risk factors. Studies have shown that the mid-pregnancy period (20–28 weeks) is the best window for prenatal ultrasound screening for hypospadias. The differentiation of the fetal external genitalia begins at 8 to 11 weeks, and the fetal gender cannot be determined through ultrasound before 12 weeks of pregnancy. Moreover, the fusion process of the fetal urethral plate is delayed before 20 weeks, which may misjudge normal development as hypospadias; and in the late pregnancy period, due to factors such as fetal position restriction and reduced amniotic fluid, the screening sensitivity significantly decreases [16]. Various factors at the maternal level are also related to this disease, including early pregnancy obesity, pre-pregnancy infection with hepatitis B, previous pregnancy with birth defects history, fetal growth restriction, pre-pregnancy use of multiple vitamins, and rarely cooking and eating at home, etc. All these situations may increase the risk of children developing hypospadias or cryptorchidism [17]. Consequently, over 30% of hypospadias cases remain undiagnosed until birth. This not only contributes to the increasing prevalence of the condition but also may delay optimal surgical timing, placing a substantial strain on healthcare systems. Given the rising incidence and the demand for early diagnosis to ensure favorable outcomes, identifying key genetic markers for hypospadias diagnosis remains crucial [18].

Transcriptome sequencing, or RNA sequencing (RNA-Seq), utilizes high-throughput sequencing technology to quantify RNA molecules, providing insights into gene expression levels and regulatory mechanisms. This technique aids in understanding gene function, disease mechanisms, and drug action pathways [19]. It also offered a critical technical underpinning for studies focusing on the early diagnosis of hypospadias.

To summarize, this study on hypospadias, a congenital malformation, adopts a similar network analysis and multi-algorithm integration approach to identify core genes with both diagnostic relevance and mechanistic significance. This aligns with established paradigms for biomarker discovery in complex diseases. Diagnostic, functional, and immunological analyses, along with the construction of regulatory networks, were conducted to explore the biological functions and regulatory mechanisms of these hub genes in hypospadias, offering valuable insights for its screening and diagnosis. The specific analysis workflow is detailed in Fig. 1.

Fig. 1.

Specific analysis workflow

Materials and methods

Sample collection and RNA extraction

The collection of materials was approved by the Scientific Research and Clinical Trial Ethics Committee of the First Affiliated Hospital of Zhengzhou University under reference number 2022-KY-0269-003. Thirty foreskin samples were collected for analysis, comprising 15 from patients with hypospadias (case group) and 15 from normal controls (control group). RNA was extracted from the tissue samples using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. The RNA concentration and purity were assessed using a NanoDrop ND-1000 spectrophotometer (NanoDrop). RNA integrity was evaluated using a Bioanalyzer 2100 and agarose gel electrophoresis to confirm the absence of degradation. A minimum of 1 µg of total RNA was required for subsequent experiments.

cDNA library construction and sequencing

The first cDNA strand was synthesized using Invitrogen SuperScript™ II reverse transcriptase, followed by complementary cDNA strand synthesis using Escherichia coli DNA polymerase I and RNase H. The cDNA ends were repaired with a dUTP solution, and A-tails were added. After purifying the cDNA using magnetic bead selection, the second strand was digested with UDG enzyme to generate a cDNA library with a size range of 300 bp ± 50 bp. This library underwent paired-end PE150 High-throughput sequencing on the Illumina HiSeq platform.

Data preprocessing

Subsequent data analysis was performed using the R programming environment. Filtering was carried out using Trimmomatic, followed by quality assessment of the filtered sequences with FastQC to ensure high-quality clean reads [20]. The high-quality reads were aligned and annotated to the human reference genome Homo_sapiens.GRCh37 using Hisat2 software [21], and quantitative count values were derived using featureCounts. These counts were then converted into TPM values according to a previously established method [22].

Difference analysis

As no batch processing steps, such as RNA extraction or library construction, were involved, and no batch effects were observed during the sequencing process, the influence of batch correction on differentially expressed gene (DEG) analysis was minimal. Consequently, no batch correction methods, such as ComBat, were applied. Differential expression analysis was performed directly on the raw sequencing data. DEGs were identified using the DESeq2 package (version 1.38.3) with a threshold of |log₂ fold change (FC)| > 1 and adjusted P-value < 0.05 [23].

Weighted gene co-expression network analysis (WGCNA)

The WGCNA package (version 1.71) [24] was used to identify modules strongly associated with the traits, utilizing case and control groups as trait data. Initial clustering of the self-sequencing data was performed to identify and remove any outlier samples, ensuring the accuracy of subsequent analyses. To maintain the scale-free nature of gene interactions, a soft threshold (β) was selected to achieve a scale-free R² approaching 0.85 and a mean connectivity close to 0. Gene adjacency was then calculated to assess gene similarity, which was used to compute the dissimilarity coefficient between genes and construct a hierarchical clustering tree. Using the mixed dynamic tree cutting algorithm, modules with fewer than 200 genes were merged to consolidate similar modules. Pearson correlation analysis was performed between the resulting modules and the trait data, identifying the module most correlated with the trait as the key module, which contained the key module genes.

Acquisition of candidate genes

DEGs were intersected with key module genes, and the intersection was visualized using the “ggvenn” package (version 0.1.9) [25], defining the overlapping genes as key DEGs. To explore the relationships among these key DEGs, a protein-protein interaction (PPI) network was constructed using the STRING database (http://string.embl.de/) with a confidence threshold of 0.4. The top 20 genes based on centrality metrics in the PPI network were identified using four centrality algorithms (MCC, DMNC, MNC, Degree) via the CytoHubba plugin in Cytoscape software [26]. The overlap of the top 20 genes identified by all four algorithms was visualized using the “UpSetR” package (version 1.4.0) [27], and the intersecting genes were defined as candidate genes. Chromosomal locations of these candidate genes were analyzed using the “RCircos” package (version 1.2.2) [28].

Enrichment analysis

To investigate the functions of the candidate genes, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted separately leveraging the clusterProfiler package (version 4.7.1.003) [29], identifying shared functions and relevant pathways among the genes (P-value < 0.05).

Machine learning

For further feature selection, the least absolute shrinkage and selection operator (LASSO) and support vector machine-recursive feature elimination (SVM-RFE) methods were applied using the glmnet package (version 4.1-4) [30] and e1071 package (version 1.7–13) [31], respectively. Given the distinct principles and logic of the two algorithms, the “eulerr” package [27] was used to identify the overlap of genes identified by both methods for further analysis. The intersecting genes underwent receiver operating characteristic (ROC) curve analysis using the pROC package (version 1.18.0) [32] to assess their diagnostic performance, with genes exhibiting an area under the curve (AUC) value greater than 0.8 defined as hub genes. AUC values are commonly used to evaluate predictive models, with models considered “good” or “excellent” if their AUC exceeds 0.7, 0.8, or 0.9 [33, 34].

Construction of nomogram

A diagram model incorporating core genes was developed using the rms package (version 6.5-0) [35]. The model’s predictive performance was evaluated through the calibration curve, the Hosmer-Lemeshow test (HL), and the ROC curve.

Interactions and functions analysis

To investigate gene interactions and shared functions among hub genes, network analysis was performed using the GeneMANIA database (http://genemania.org/). Functional similarity analysis of hub genes was conducted with the “GOSemSim” package (version 2.24.0) [36], enabling the exploration of gene likeness. Hub genes were ranked based on their correlation coefficients with all other genes, utilizing their expression levels. Gene Set Enrichment Analysis (GSEA) was then performed using the background gene set c2.cp.kegg.v7.0.symbols.gmt, implemented through the clusterProfiler package (version 4.7.1.003) [37], with the criteria |NES| > 1 and P.adjust < 0.05.

Immune analysis

The ssGSEA algorithm from the GSVA package (version 1.46.0) [38] was employed to calculate immune cell scores for each sample, assessing the scores of 28 immune cell types across all samples. Wilcoxon rank-sum tests were conducted between the control and case groups to identify immune cells with significant differences. Spearman correlation analysis was then performed between hub genes and differentially expressed immune cells, with a threshold of |correlation coefficient (cor)| > 0.3 and p < 0.05.

Construction of network

Transcription factors (TFs) for hub genes were predicted using the ChIP-X Enrichment Analysis 3 (ChEA3) database (https://amp.pharm.mssm.edu/ChEA3), focusing on TFs supported by ChIP-seq data from the ENCODE database (https://www.encodeproject.org/). To further explore the regulatory mechanisms of hub genes, microRNAs (miRNAs) were sourced from the PITA (http://genie.weizmann.ac.il/pubs/mir07/mir07_data.html) and TargetScan (https://www.targetscan.org/vert_80/) databases. Intersecting miRNAs were selected for subsequent analysis. Key long non-coding RNAs (lncRNAs) upstream of these target miRNAs were retrieved from the starBase database (https://starbase.sysu.edu.cn/), with target lncRNAs filtered based on the condition clipExpNum > 4. To predict potential targeted drugs for hypospadias and identify new therapeutic targets, the DGIdb database (https://dgidb.genome.wustl.edu/) was used to forecast medications for hub genes. Finally, mRNA-TF interactions and hub gene-drug interactions were visualized using Cytoscape software.

Statistical analysis

Data processing and analysis were conducted in R software, and statistical significance between the two cohorts was assessed using the Wilcoxon rank-sum test. A P-value < 0.05 was considered statistically significant.

Results

A total of 303 key module genes were obtained

After preprocessing the self-sequencing data, differential expression analysis identified 157 DEGs between the case and control cohorts. Of these, 143 genes were upregulated, while 14 were downregulated (Fig. 2a and b). Clustering analysis of the samples revealed an outlier (ZZ2A10), which was subsequently excluded from further analysis (Fig. 2c). With a β = 8, the R² value approached 0.85, indicating that the network topology was nearing a scale-free distribution. Simultaneously, the mean connectivity approached 0, suggesting that most genes had low connectivity, with only a few acting as core nodes, consistent with the sparse nature of biological networks. A smaller β value would result in excessive network connectivity (high mean connectivity) and unclear module division, while a larger β value would reduce connectivity and potentially obscure important co-expression relationships. Thus, β = 8 was selected as the optimal threshold, as it maintained a scale-free distribution while preserving essential co-expression signals (Fig. 2d). After module clustering and merging, seven modules were identified (Fig. 2e). Correlation analysis showed that the green module exhibited the strongest association with the case group (cor = 0.54, p = 0.003) (Fig. 2f), making it the key module, containing a total of 303 genes.

Fig. 2.

Identification of DEGs and key modules. a Volcano plot of differential genes: The orange dots represent upregulated differentially expressed genes (DEGs), the green dots represent downregulated DEGs, and the gray dots represent genes with no significant statistical difference. Each dot corresponds to an individual gene. b Heatmap of differential gene expression: The horizontal axis represents the samples, and the vertical axis represents the genes. The top of the heatmap shows that green represents control samples, and orange represents case samples. Highly expressed genes are indicated in red, while lowly expressed genes are shown in blue. c Sample clustering before excluding outliers: The branches represent the samples, and the ordinate shows the hierarchical clustering height. Red indicates case samples, while white indicates control samples. d Sample clustering after excluding outliers: The branches represent the samples, and the ordinate shows the hierarchical clustering height. Red represents case samples, and white represents control samples. e Soft threshold screening: The horizontal axis represents the power value of the weight parameter (β), and the vertical axis of the left plot shows the scale-free fit index (signed R2). A higher R2 indicates a network approaching a scale-free distribution. The vertical axis of the right plot represents the mean adjacency value of all genes within the corresponding module. f Hierarchical clustering tree of modules: The upper half of the figure shows the module clustering tree, while the lower half displays the individual modules, with each color representing a different module. g Module-trait relationships heatmap: The colors represent the direction (positive or negative) and strength of the correlation. Red indicates a positive correlation, blue indicates a negative correlation, and darker colors indicate stronger correlations

Identification of candidate genes in PPI network analysis

DEGs were intersected with key module genes, resulting in 64 significant DEGs (Fig. 3a). A PPI network was constructed for these key DEGs to further investigate their interactions. Within the network, core genes such as C5AR1, FCGR3A, and FCGR2A exhibited extensive interactions with other key DEGs (Fig. 3b). The top 20 genes, ranked by centrality across four algorithms (MCC, DMNC, MNC, Degree), were identified, leading to the selection of 15 candidate genes (Fig. 3c and g).

Fig. 3.

Screening of key DEGs and protein interaction networks. a Key DEGs identified via Venn diagram. b PPI network of key DEGs. c-f Gene screening using four centrality algorithms. In the figures, color intensity corresponds to centrality score, with darker colors indicating higher scores. g Upset diagram for four centrality algorithms in the PPI networks. The lower left bar graph (Set Size) represents the number of genes included in each centrality algorithm. Intersection points correspond to genes common across algorithms, with the upper bar graph indicating the number of overlapping genes

Functional analysis of candidate genes

Chromosomal distribution of the candidate genes was examined. These genes were predominantly located on autosomes; for instance, CXCR4 and CXCR1 were both located on chromosome 2 (Fig. 4a). To explore the functional roles and pathways of the candidate genes, GO and KEGG analyses were conducted. GO analysis revealed significant enrichment in processes such as leukocyte migration, secretory granule membrane, and immune receptor activity (Fig. 4b). KEGG analysis indicated involvement in pathways such as neutrophil extracellular trap formation, tuberculosis, and complement and coagulation cascades (Fig. 4c).

Fig. 4.

Functional and chromosomal analysis of candidate genes. a Chromosomal localization of candidate genes. b Lollipop plot of GO enrichment analysis. The ordinate (Count) shows the number of genes enriched in each GO term, while the abscissa represents the GO term names. c KEGG enrichment analysis. The block size represents the number of genes in each pathway, with color intensity indicating the P-value, where darker colors reflect smaller P-values

Identification of hub genes for disease diagnosis

Three intersecting genes, namely CSF3R, SELL, and FCN1, were identified by intersecting the characteristic genes selected by the LASSO algorithm (CSF3R, SELL, CSF3, FCN1) and the SVM-RFE algorithm (CXCR1, CD300A, SELL, CSF3R, FPR1, VNN2, TREM1, FCN1, CR1, ITGAX) (Fig. 5a and d). A diagnostic analysis was then conducted to evaluate the diagnostic performance of these three genes. The AUC values for these genes exceeded 0.8, indicating strong diagnostic efficacy for the disease (Fig. 5e-f). Consequently, these three genes were designated as hub genes.

Fig. 5.

Machine learning-based hub gene identification. a Lasso regression analysis. The abscissa represents log(Lambda), and the ordinate represents cross-validation error. b Lasso regression analysis. Deviance indicates the proportion of residuals explained by the model, illustrating the relationship between the number of hub genes and the proportion of residuals explained (dev). The ordinate represents gene coefficients. c SVM-RFE screening of candidate genes. The abscissa represents the number of genes. d Intersection of gene sets obtained by Lasso and SVM-RFE, yielding hub genes. e Confidence interval of AUC values for the hub genes. f ROC curve of hub genes. The abscissa represents the false positive rate, the ordinate represents the true positive rate, and the area under the curve (AUC) is shown in the figure

Evaluation of the nomogram model for disease prediction

Using the identified hub genes, a nomogram model was constructed (Fig. 6a). The slope of the calibration curve for the nomogram model was close to 1, indicating high accuracy and reliability (Fig. 6b). Furthermore, the HL test yielded a p-value of 0.658, suggesting a good fit for the model. The model’s AUC value was 0.964, demonstrating strong predictive precision (Fig. 6c).

Fig. 6.

Nomogram model for hypospadias prediction. a Nomogram model. b Calibration curve of the nomogram model. The abscissa represents the predicted probability, the ordinate represents the actual probability, with a slope closer to 1 indicating higher predictive accuracy. c ROC curve of the nomogram model. The abscissa represents the false positive rate, the ordinate represents the true positive rate, and the area under the curve (AUC) is shown in the figure

Functional analysis of hub genes

Significant positive correlations were observed among the three hub genes, with a correlation coefficient of 0.85 (P < 0.001) between CSF3R and SELL (Fig. 7a). The GeneMANIA network revealed that these hub genes and their interacting partners were associated with processes such as complement activation and humoral immune response (Fig. 7b). Functional similarity analysis indicated an overall functional similarity of approximately 0.5 among the three hub genes, with a notably high similarity between them (Fig. 7c). GSEA further explored the biological pathways associated with these hub genes, showing that all three were linked to cytokine-cytokine receptor interaction and chemokine signaling pathways (Fig. 7d and f).

Fig. 7.

Functional interactions and pathways of hub genes. a Heatmap of hub gene correlations. The intensity of the orange color indicates the strength of the positive correlation. Significance levels: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. b GeneMANIA analysis of hub genes. c Functional similarity among hub genes. d-f GSEA enrichment analysis of CSF3R, SELL, and FCN1. The graph is divided into two sections: the upper part shows the Enrichment Score (ES), with the horizontal axis representing sorted genes and the vertical axis indicating Running ES. A peak in the plot indicates the Enrichment Score for the pathway gene set, with genes preceding the peak considered core genes within the set. The lower section highlights the genes involved in this gene set

Differential immune cell abundances and gene correlations

The box plot shows the scores for 28 immune cell types across all samples (Fig. 8a). Notably, 11 immune cell types, including activated dendritic cells, central memory CD8 T cells, eosinophils, immature B cells, immature dendritic cells, macrophages, mast cells, MDSCs, neutrophils, plasmacytoid dendritic cells, and Type 1 T helper cells, showed higher abundances in samples from the case cohort (Fig. 8b). Correlation analysis demonstrated a significant positive correlation between CSF3R, SELL, and FCN1 with neutrophils (|cor| > 0.3, p < 0.05) (Fig. 8c).

Fig. 8.

Immune cell profiling and correlation with hub genes. a Stacked bar plot of immune cell scores. b Boxplots of immune cell scores in the control and case groups. The abscissa represents the 28 immune cell types, and the ordinate represents the immune cell scores. Significance levels: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. c Heatmap of correlations between hub genes and differential immune cells. The abscissa represents the immune cells, and the ordinate represents the biomarkers. The intensity of the orange color indicates the strength of the positive correlation. Significance levels: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001

Regulatory mechanisms and therapeutic insights

To elucidate the regulatory mechanisms involving the hub genes, TFs predicted to regulate these genes were identified using relevant databases. An mRNA-TF network consisting of 11 nodes and 8 relationships was constructed. The network revealed that SELL was concurrently regulated by RELA, CTCF, MAFK, and IKZF1 (Fig. 9a). Additionally, a molecular regulatory network was developed, incorporating three key genes, 15 miRNAs, and 87 lncRNAs. Among the regulatory interactions, SELL was linked to hsa-miR-4770-TSIX, among others (Fig. 9b). Furthermore, the three hub genes were utilized to predict potential target drugs, offering insight into therapeutic options for the disease. In the hub gene-drug network, BALUGRASTIM and FILGRASTIM were identified as target drugs for CSF3R, while ASELIZUMAB was identified as a target for SELL (Fig. 9c). Finally, significant expression differences were noted for all three hub genes, all of which were upregulated in the case cohort (Fig. 9d).

Fig. 9.

Regulatory networks and therapeutic insights. a mRNA-TF network. Red triangles represent mRNAs, and orange rectangles represent TFs. b Molecular regulatory network of key genes. c Hub gene-therapeutic drug network. No drugs showed significant association with FCN1. Thus, no gene-drug network was constructed. Red triangles represent hub genes, and blue hexagons represent therapeutic drugs. d Box plot of hub gene expression

Discussion

Hypospadias is a common congenital malformation characterized by an abnormal urethral opening or incomplete urethral canal development, significantly impacting the quality of life for affected individuals [39]. However, the exact pathogenesis of hypospadias remains unclear. It is widely believed to be a polygenic disorder influenced by factors such as maternal diet during pregnancy, environmental pollutants, and genetic mutations [40]. This study identified three hub genes—CSF3R, SELL, and FCN1—through self-sequencing and constructed a diagnostic model along with a biological function analysis based on these genes. This provides a deeper understanding and a new direction for the diagnosis, treatment, and prevention of hypospadias.

CSF3R, a type 1 cytokine receptor, binds to granulocyte colony-stimulating factor (G-CSF), a cytokine essential for granulocyte proliferation and differentiation [41]. Activating mutations in the gene encoding CSF3R lead to altered downstream kinase signaling via SRC family-TNK2 or JAK kinases, resulting in differential responses to kinase inhibitors [42]. Lian Cai first demonstrated that CSF3 and CSF3R are expressed in porcine reproductive organs, cells, and embryos, suggesting their potential role in mammalian reproduction by maintaining embryonic pluripotency, promoting anti-apoptotic activity, and supporting proliferation [43]. Mutations in CSF3R have been linked to severe congenital neutropenia and other genetic disorders [44], however, no studies have yet investigated its involvement in hypospadias. In summary, we speculate that the CSF3R gene may participate in the embryonic morphogenesis of the urethral tissue by regulating granulocyte function, mediating downstream kinase signaling pathways, enhancing the chemotactic and infiltrative capacities of neutrophilsand modulating the proliferation, and differentiation of embryonic cells. Subsequent studies will further verify the pathological role of the CSF3R gene in hypospadias through in vitro and in vivo experiments.

L-selectin, an adhesion molecule encoded by the SELL gene, plays a key role in leukocyte adhesion to the endothelium and the inflammatory response [45, 46]. The SELL gene is a known risk factor for dysplasia [47], and genetic variations in L-selectin are associated with sickle cell disease (SCD), an autosomal recessive disorder [48]. This suggests that hypospadias may be associated with inflammatory processes during embryonic development. Previous studies have confirmed that L-selectin mediates the interaction with the uterus, and this adhesion mechanism is crucial for establishing human pregnancy [49]. Additionally, the specific ligand of L-selectin is expressed and localized on the surface of endometrial epithelial cells in the uterine cavity during the embryo implantation period. This ligand not only assists the blastocyst in completing the early attachment process, but also has a high expression level that is positively correlated with endometrial receptivity, thereby promoting embryo implantation and development [50, 51]. These results suggest that the SELL gene may regulate the embryonic inflammatory response and cell adhesion process through dual regulation, affecting the morphogenesis of urethral tissues, abnormal expression or function of this gene may be one of the potential molecular triggers for hypospadias.

The FCN1 gene encodes Ficolin-1 [52], a protein involved in the innate immune system that functions as a recognition molecule in the complement system [53]. Kexin Zhang observed significantly higher expression of immune-related proteins in patients with hypospadias [54], while Xixi Chen identified FCN1 as a novel and promising diagnostic biomarker for pediatric inflammatory bowel diseases [55]. Additionally, polymorphisms in the FCN2 gene have been linked to an increased risk of systemic lupus erythematosus (SLE), an autoimmune disorder with a familial genetic predisposition [56]. These findings suggest that FCN1 may influence the development of hypospadias through its impact on inflammatory processes.

Functional analysis revealed that all three were significantly enriched in cytokine-cytokine receptor interactions and chemokine signaling pathways. This was confirmed by Gene MANIA network analysis, which indicated their close association with complement activation, humoral immune responses, and other processes. This suggests that they form a functional synergy network in immune signal transduction. Previous studies have shown that the complement activation process may produce inflammatory mediators, which can promote the occurrence of hypospadias during embryonic development [57]. The chemokine signaling pathway and complement activation have cross-regulatory relationships. After activation, the former promotes the synthesis of key molecules of the complement system (such as C3, C5), further amplifying local inflammatory damage [58, 59]. Additionally, if the mother has an infection or exposure to environmental toxins, it will activate fetal or maternal immune cells to secrete excessive pro-inflammatory cytokines. For example, dibutyl phthalate may affect the development of the reproductive nodules by down-regulating the Wnt/β-catenin pathway in male rats of the fetus, and ultimately may induce hypospadias [60]. The normal embryonic development microenvironment requires a dynamic balance between anti-inflammatory and pro-inflammatory factors [61]. Disruption of the cytokine receptor pathway can lead to insufficient secretion of anti-inflammatory factors, unable to effectively counteract pro-inflammatory responses, causing the hypospadias region to remain in a chronic inflammatory state for a long time, disrupting normal physiological processes of cell adhesion and migration, and potentially leading to penile curvature and urethral deformities [62, 63]. Therefore, these three hub genes may regulate these pathways to interfere with the normal development of urethral tissues during the embryonic stage, ultimately leading to the occurrence of hypospadias. This speculation provides a more targeted mechanism explanation for the pathological role of core genes and points out a clear direction for subsequent verification of their causal relationship with hypospadias.

GATA TFs regulate androgen production by activating genes involved in steroidogenesis within gonadal cells [64], suggesting that this family of TFs may contribute to the pathogenesis of hypospadias through modulation of the androgen synthesis pathway. GATA3, a key driver of cell fate decisions and tissue morphogenesis, is essential for embryonic survival, as its deletion can be lethal during early development [65]. This indicates that normal GATA3 expression is crucial for both embryonic survival and tissue differentiation. Even partial dysfunction of GATA3 may disrupt cell fate determination in urethral tissues during embryonic development, potentially affecting the differentiation of urethral interstitial cells into smooth muscle cells or interfering with epithelial cell migration and proliferation, ultimately leading to abnormal urethral development. CTCF plays a significant role in early embryonic development, and its dysfunction can result in developmental defects such as hypospadias [66]. CTCF may indirectly influence urethral development by maintaining chromatin structure and the stability of the gene expression regulatory network in the embryonic genome. As a chromatin insulator, CTCF regulates the spatial and temporal specificity of gene expression by mediating interactions between gene loci. Abnormal CTCF function can lead to uncontrolled expression of genes critical for urethral development, such as the core genes CSF3R and SELL identified in this study, or disrupt the expression of key regulatory factors like the androgen receptor (AR), potentially contributing to hypospadias.

P65 (RelA), a pivotal member of the nuclear factor-κB (NF-κB) family, is closely involved in mesoderm differentiation and plays a pivotal role in normal embryonic development [67]. This suggests that P65 may influence urethral development by regulating mesoderm differentiation. The urethral tissue originates from the mesoderm, and abnormal mesoderm differentiation leads to developmental defects in the urethral stroma, such as impaired spongy tissue formation. Additionally, the P65-mediated NF-κB pathway is central to the inflammatory response; its abnormal activation may induce excessive local inflammation in the urethra, disrupting the developmental microenvironment. The combined effect of these factors increases the likelihood of urethral atresia. Therefore, TFs such as the GATA family (including GATA3), CTCF, and P65 are all implicated in the core pathogenic mechanisms of hypospadias—such as androgen regulation, embryonic development, cell differentiation, and the inflammatory microenvironment—through various mechanisms. When integrated with the core genes CSF3R, SELL, and FCN1 identified in this study, it is plausible that these TFs may directly or indirectly regulate the expression of these core genes, thereby forming a regulatory network of “TFs → core genes → pathological pathways”.

Immune infiltration analysis in this study revealed a significant positive correlation between CSF3R, SELL, and FCN1 and neutrophils, suggesting that these core genes may contribute to the pathological process of hypospadias by modulating neutrophil infiltration or function. Another study demonstrated that neutrophil levels were significantly associated with predicting complications after hypospadias repair [68]. Hyperbaric oxygen therapy (HBOT) has been shown to promote angiogenesis by reducing inflammation, stimulating endothelial proliferation, and enhancing the activity of fibroblasts, lymphocytes, and macrophages, all of which contribute to urological wound healing [69]. These findings suggest that inflammatory responses, immune defense, and tissue repair play pivotal roles in the prognosis of hypospadias.

The diagnostic analysis results further confirmed the core significance of the aforementioned functions and mechanisms. When CSF3R, SELL, and FCN1 were used as diagnostic markers alone, their AUC values all exceeded 0.8. The combined nomogram model constructed from them had an AUC of up to 0.964. Its excellent ability to distinguish diseases is essentially due to their key functional positions in the pathogenesis of hypospadias, in other words, by regulating the core immune inflammatory pathways and related regulatory networks to affect the disease progression.

Moreover, within the hub gene-drug network, BALUGRASTIM and FILGRASTIM were identified as drugs targeting CSF3R, while alemtuzumab was found to target SELL. Barugestim and Nogestim, recombinant human G-CSF drugs, are primarily used to increase neutrophil counts to prevent infections after chemotherapy [70, 71]. However, their mechanism of action involves regulating granulocyte activity through binding to CSF3R. This suggests that if future studies confirm that abnormal CSF3R activation in hypospadias is a key driver of “excessive inflammation,” there may be potential to explore therapeutic strategies to mitigate neutrophil infiltration and alleviate inflammatory disorders. Such approaches could involve regulating the activity of these drugs (e.g., inhibiting CSF3R overactivation rather than enhancing its function). Atezolizumab, a PD-L1 inhibitor primarily used in cancer immunotherapy [72, 73], targets SELL and can influence immune cell adhesion, further highlighting the therapeutic potential of modulating immune interactions in hypospadias.

This study was based on self-generated sequencing data and utilized an integrated strategy of “WGCNA + machine learning + network construction” to identify the core hub genes (CSF3R, SELL, and FCN1) related to hypospadias. Through various bioinformatics analyses, the pathogenic chain “core genes → activation of inflammatory pathways → infiltration of neutrophils → imbalance of embryonic development microenvironment” was revealed, which not only explained the pathogenesis of hypospadias but also confirmed the crucial interfering role of inflammatory disorders in the tissue morphogenesis during the embryonic stage. This provided a reference paradigm for the study of inflammatory mechanisms of other embryonic developmental-related congenital malformations such as congenital urethral atresia and penile curvature. At the same time, the multi-level regulatory network clarified the synergistic regulatory relationship between 15 miRNAs, 87 lncRNAs and transcription factors, filling the research gap in the multi-gene interaction mechanism of hypospadias and providing a reference analytical paradigm for deciphering the “gene-pathway-phenotype” association in multi-gene genetic diseases, promoting the transformation of this research from “exploration of single genes” to “analysis of network regulation”. Although this study has made the aforementioned progress in molecular network analysis and understanding of inflammatory mechanisms, it still has certain limitations. Additionally, the efficient combined diagnostic model of hub genes provides new ideas for the early non-invasive diagnosis of hypospadias.

First, the results may be constrained by data quality and sample size, which should be further validated and expanded. Secondly, although bioinformatics methods can conduct comprehensive analysis of large-scale data, relying solely on transcriptome data and bioinformatics analysis still cannot fully confirm the direct regulatory relationship between the core genes (CSF3R, SELL, FCN1) and neutrophil infiltration. The accuracy of the related mechanisms still requires further verification through in vitro cell experiments and in vivo animal models. Third, due to the current limitations in public data resources on hypospadias transcriptome studies, especially in databases like GEO, there are few independent cohort data sets specific to this disease. Consequently, external independent cohort validation has not yet been conducted. Future research will expand and deepen in three aspects: By combining multiple centers to increase the sample size and establishing an independent external validation dataset, the stability and universality of CSF3R, SELL, and FCN1 as diagnostic biomarkers for hypospadias will be verified. Through core gene overexpression or silencing, neutrophil migration and adhesion experiments, and cell co-culture, the key signaling pathways and downstream molecules regulating neutrophil infiltration will be analyzed. Constructing an animal model of hypospadias, and conducting in vivo intervention on the expression of core genes, the effects of these genes on neutrophil infiltration, the urinary tract development microenvironment, and phenotypes will be verified, clarifying the causal relationship and potential therapeutic targets.

This will enable a deeper exploration of the specific roles of these core genes in the disease’s occurrence and progression, offering new insights and strategies for treating hypospadias.

Conclusion

The three hub genes (CSF3R, SELL, and FCN1) were found to regulate neutrophil infiltration and function, EMT, and the inflammatory microenvironment through pathways such as cytokine-cytokine receptor interactions, chemokine signaling, and the NF-κB pathway, contributing to the pathogenic progression of hypospadias. These findings provide critical genetic and pathway targets for elucidating the molecular mechanisms behind hypospadias and establish a theoretical foundation for the development of diagnostic biomarkers and targeted interventions.

Authors’ contributions

Y HY: Conceptualization, Data curation, Validation, Visualization, Writing–original draft, Writing–review & editing. F YZ: Data curation, Validation, Visualization, Writing–review & editing. L LP: Validation, Writing–review & editing. Z SL: Visualization, Writing–review & editing. D KF: Conceptualization, Supervision, Writing–review & editing. Y JM: Conceptualization, Project administration, Supervision, Writing–review & editing. This work currently described has not been published, is not being considered for publication elsewhere, and its publication was approved by all authors.

Funding

This research received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

The datasets analysed in this study are available in the National Center for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov/), including [PRJNA1201896].

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the provisions of the Declaration of Helsinki. Ethical approval has been obtained from Scientific Research and Clinical Trial Ethics Committee of the First Affiliated Hospital of Zhengzhou University, with the approval number: 2022-KY-0269-003 and approval date: 2022-11. All patients provided written informed consent for the use of their clinical samples in transcriptome data sequencing experiments, ensuring that the research process complies with ethical standards and fully respects the rights, interests and wishes of the patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.