Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2026 May 16;16:22266. doi: 10.1038/s41598-026-53162-7

Exploring dysregulation of cuproptosis-related genes molecular clusters and candidate biomarkers in pterygium

Ji Yang 1,#, Bo-Yu Liang 1,#, Chen-Yan Fang 1,#, Jia-Jie Li 1, Peng Zhang 1, Ping Xiang 2, Li-Wei Zhang 1,✉, Hai Liu 1,✉
PMCID: PMC13373229  PMID: 42143165

Abstract

Pterygium is a common ocular surface disorder, with its prevalence strongly correlated to ultraviolet (UV) exposure in geographic regions. Epidemiological investigations reveal significant demographic variations, with higher incidences observed in areas with intense UV radiation and within specific populations, notably rural individuals. Despite surgical interventions being the standard treatment, recurrent cases underline the necessity for understanding the underlying biological mechanisms contributing to pterygium pathogenesis. Recent advancements in cellular death mechanisms point to cuproptosis, a copper-dependent programmed cell death pathway, as a potential regulatory factor in ocular diseases, including pterygium. This study aims to systematically investigate the immunological significance of cuproptosis-related genes (CuRGs) in pterygium’s pathogenesis using an integrative bioinformatics framework. We performed transcriptomic profiling on pterygium tissues and employed machine learning algorithms to identify pivotal biomarkers for pterygium risk stratification. Comprehensive immune profiling and functional enrichment analyses were conducted to elucidate the interplay between identified CuRGs and the immune microenvironment in pterygium. Our analysis highlighted 19 CuRGs, with eight genes displaying significant dysregulation in pterygium tissues (p < 0.05). We established robust associations between CuRG expression and prominent immune cell infiltrates, notably regulatory T cells and macrophages. Furthermore, three core biomarkers (SERTAD1, JMJD1C, CSRNP1) were identified through machine learning and validated by QPCR, with the support vector machine model demonstrating exceptional predictive performance (AUC = 0.84). Empirical validation corroborated significant downregulation of selected biomarkers in pterygium tissue samples compared to normal conjunctiva. Our findings underscore the vital role of CuRGs in modulating pterygium development through immune and metabolic interactions, establishing their potential as novel therapeutic targets. Nevertheless, our study has limitations, as these findings are hypothesis-generating and require validation in larger patient cohorts.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-53162-7.

Keywords: Pterygium, Cuproptosis-related genes, Machine learning, Prediction model, Immune characteristic

Introduction

Pterygium is characterized by a hyperplastic, centripetally directed growth of altered limbal epithelial cells accompanied by Bowman’s layer dissolution, epithelial- mesenchymal transition, and an activated fibroblastic stroma with inflammation, neovascularization, and matrix remodelling, mediated through the concerted actions of cytokines, growth factors, and matrix metalloproteinases, is a common ocular surface disorder with distinct geographical distribution patterns1,2. Epidemiological studies demonstrate a strong correlation between disease prevalence and ultraviolet (UV) exposure intensity3,4, with incidence rates reaching 10–20% in equatorial regions characterized by high solar radiation (e.g., Southeast Asia, Africa), compared to < 2% in temperate zones such as Northern Europe and North America5. This condition exhibits demographic variations, showing higher prevalence in males (attributed to prolonged outdoor occupational exposure) and rural populations (vs. urban dwellers), likely reflecting cumulative environmental risk factor exposure6. Meanwhile, UV radiation can also affect the skin microenvironment, suppress anti-tumor immune responses, and accelerate the process of malignant transformation. Long-term or excessive exposure to ultraviolet rays is an important risk factor for the development of skin malignancy7. Notably, in Yunnan Province, China, a region with intense UV irradiation and high-altitude terrain—a recent study by Ke8 revealed 22.6% prevalence among adults aged ≥ 40 years, with hyperendemic areas exceeding 35.27%. Emerging evidence implicates environmental pollution and specific lifestyle habits in pterygium pathogenesis9. Particulate matter (PM2.5) has been shown to exacerbate ocular surface inflammation through oxidative stress mechanisms, potentially accelerating fibrovascular proliferation10. At present, owing to the utilization of pterygium excision combined with autologous limbal stem cell transplantation techniques, the recurrence rate of pterygium has seen a notable decline. Research has indicated that the recurrence rate for these surgical procedures stands at roughly 1.7%11. This persistent clinical challenge underscores the critical imperative to elucidate the underlying biological drivers of disease recrudescence and develop targeted therapeutic interventions that address the fundamental pathophysiological mechanisms contributing to pterygium recurrence12,13.

Emerging research has elucidated molecular mechanisms in pterygium pathogenesis, particularly focusing on apoptotic pathways and inflammatory cascades14,15. Oxidative stress has emerged as a key mediator, with its dynamic interplay in immune microenvironment remodeling showing significant influence on disease progression16. While these findings advance our understanding, the precise molecular circuitry governing pterygium initiation and chronicity remains poorly characterized. Recent breakthroughs in cell death mechanisms have identified cuproptosis as a copper-dependent programmed death pathway implicated in malignancies and neurodegeneration17–19. This novel mechanism involves mitochondrial protein aggregation through copper-sulfur cluster formation, particularly affecting lipoylated TCA cycle enzymes. The ocular system demonstrates unique copper homeostasis requirements, where physiological levels are crucial for cytochrome C oxidase function, yet excessive accumulation (as seen in ATP7B mutations) induces β-amyloid aggregation20,21. Notably, despite established connections between copper dysregulation and ocular pathologies like AMD, the potential involvement of dysregulation of cuproptosis-related genes remains unexplored22. The anatomical exposure of pterygium to environmental copper sources, combined with its characteristic inflammatory microenvironment, positions dysregulation of cuproptosis-related genes as a compelling mechanistic candidate worthy of systematic investigation.

This study establishes an integrative bioinformatics framework to decode cuproptosis-related genes (CuRGs) in pterygium pathogenesis. Figure 1 shows our research protocol. Through transcriptomic profiling, functional enrichment mapping, and machine learning-driven biomarker discovery, we systematically characterize CuRGs signatures and their immune microenvironment interactions. This multi-modal strategy identifies clinically relevant molecular networks. Our primary focus centers on developing CuRG-based diagnostic classifiers using multivariate nomogram. The validated prediction system demonstrates significant preliminary clinical relevance for precision management. These findings provide mechanistic insights into copper-associated ocular pathology while offering new therapeutic perspectives for pathway-targeted intervention.

Fig. 1.

Fig. 1

Open in a new tab

Flow chart.

Materials and methods

Data resource

The research protocol received ethical approval from the Institutional Review Board at Yunnan University Affiliated Hospital, with approval number 2022198. All participants provided written informed consent, in compliance with the Declaration of Helsinki. A total of 34 elective pterygium cases (5 males, 29 females) were registered under project PRJNA1147595, and all surgical interventions were performed by a single surgeon under local anesthesia. Our dataset encompasses transcriptome data from 34 pterygia, along with data from 34 autologous normal conjunctival tissues of patients. The normal conjunctival tissues were sourced from the redundant conjunctival tissue at the marginal area of the conjunctival grafts that were trimmed after autologous conjunctival transplantation. According to the pterygium grading method described in the literature, we divided pterygium into three levels. Among them, there were 5 cases of Grade 1 pterygium, 18 cases of Grade 2 pterygium, and 11 cases of Grade 3 pterygium23. All pterygium specimens were primary nasal pterygia, with no recurrent pterygia. All patients were pterygium patients from the plateau region of Yunnan, where the average ultraviolet radiation could reach 260.75 MJ/m2. Genomic sequencing was carried out using the Illumina NovaSeq 6000 platform. The experimental data has been deposited in the NCBI Sequence Read Archive, with SRA accession number PRJNA114759524. Additionally, supplementary dataset GSE51995 are available through the Gene Expression Omnibus repository (https://www.ncbi.nlm.nih.gov/geo/). The GEO submission includes metadata for 4 conjunctivas specimens and 4 pterygiums, which contrast pterygium lesions with matched healthy conjunctival tissues (Supplementary Table 1)25–27.

Discovery and functional characterization of cuproptosis-related genes in pterygium

A total of 19 genes related to cuproptosis (CuRGs) were pinpointed (Supplementary Table 2), based on information from previous literature and the Genecards database28,29. We produced a normalized expression matrix and conducted a differential expression analysis comparing pterygium and conjunctiva samples from the PRJNA1147595 dataset using the ‘LIMMA’ package. The results, considered significant at a p value < 0.05, were depicted using the ‘heatmap’ package30. To illustrate the chromosomal distribution, we generated an RCircos plot. This investigation sought to unravel the biological significance and potential roles of the identified differentially expressed cuproptosis-related genes (DECuRGs). These genes underwent gene ontology (GO) analysis, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis31–33. To pinpoint key CuRGs, we utilized the STRING database (accessible at https://string-db.org/) and Cytoscape software (version 3.10.1)34.

Immune infiltrating and correlation analysis between DECuRGs

To assess immune cell infiltration and characterize the disease-associated immune microenvironment, we employed the CIBERSORT algorithm in conjunction with the LM22 feature matrix35. Immune cell infiltration results were visualized using the ggplot2 package in R. To examine correlations among these infiltrations, Spearman correlation analyses were conducted with the ‘circlize’ and ‘corrplot’ R packages. A p value below 0.05 was deemed statistically significant, suggesting a notable correlation among the immune cell types.

Assessment of phenotypic scores associated with DECuRGs

We collected 34 pterygium samples, which were selected based on the expression profiles of unannotated RNA genes that exhibit differential expression and are linked to the DECuRGs. To assign scores to each sample, we utilized single-sample gene set enrichment analysis (ssGSEA). Subsequently, the samples were divided into high score, low score groups for the cuproptosis-related genes of interest, with the median score serving as the threshold. For the ssGSEA analysis, we leveraged the ‘GSVA’ R package, enabling us to explore the differences in enriched gene sets between the clusters of DECuRGs36.

Weighted gene co-expression network analysis

The WGCNA algorithm was utilized to identify co-expression modules between pterygium and conjunctiva, and to detect cluster-specific module genes in pterygium within the PRJNA1147595 dataset37. WGCNA was used to determine the optimal power value for constructing both a weighted adjacency matrix and a topological overlap matrix (TOM). WGCNA analyses were performed on the top quartile of genes with the highest variance. The WGCNA framework integrated clinical trait-module correlations to identify phenotype-associated expression patterns.

Machine learning model screen the candidate genes

Utilizing the weighted correlation network analysis (WGCNA) results and hub gene expression profiles, four machine learning models were constructed with the “caret” R package38. The models comprised the Generalized Linear Model (GLM), Random Forest (RF), eXtreme Gradient Boosting (XGB), and Support Vector Machine (SVM). The base learners were trained on these subsets, and their predictions were aggregated to generate the integrated model’s final output39. The 34 conjunctiva and 34 pterygium samples were randomly allocated into a validation set (comprising 30%, n = 21) and a training set (comprising 70%, n = 47). The “DALEX” package was utilized to analyze the four aforementioned machine learning models, providing visualizations of residual distribution and feature importance. The ‘caret’ R package automatically optimized the model parameters, which were evaluated using tenfold cross-validation with default settings. The “pROC” package was used to display the area under the ROC curves.

Construction and external validation of a prediction model

A nomogram model was constructed utilizing the “rms” R package, with the aim of forecasting the occurrence of pterygium based on five pivotal predictive genes. A point value was assigned to each of these predictive variables, and the cumulative ‘total points’ were calculated by adding up the points for all five predictors. To assess the model’s predictive prowess, we employed decision curve analysis (DCA) and calibration curves. Additionally, we integrated the GSE51995 dataset to externally validate the model’s predictive capabilities. The outcomes derived from these analyses were then visually represented using the “pROC” R package.

Tissue collection and gene expression analysis

We collected twelve tissue specimens, which included 11 samples of pterygiums and 8 of conjunctivas. During surgical procedures, tissue samples intended for RNA extraction and QPCR analysis were preserved in RNA protectant and stored at −80 °C until required. We extracted RNA from tissue using TRIZOL reagent. To assess the expression levels of candidated genes, we employed a HiScript® III RT SuperMix for QPCR (+ gDNA wiper) kit for reverse transcription and ChamQ™ Universal SYBR® QPCR Master Mix on a CFX96 fluorescence QPCR system (Vazyme, Nanjing, China). The PCR protocol consisted of an initial 30-s pre-denaturation at 95 °C, followed by 40 cycles of 3–10 s of denaturation at 95 °C and 10–30 s of annealing/extension at 60 °C. GAPDH was used as the internal control gene, and the relative expression of the target genes was calculated using the 2−ΔCt method. The primer sequences used are provided in Supplementary Table 3.

Statistics

R software (version 4.3.3) was used to for data visualization. Transformed values derived from RNA-seq and QPCR analyses are expressed as mean ± standard deviation (SD). Differences between groups were evaluated using a t-test. Specifically, for all continuous variables subjected to inter-group comparison using a t-test, we a priori performed tests for normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test). A parametric t-test was employed only when the data met the assumptions of both normal distribution (p > 0.05) and equal variance (p > 0.05). P < 0.05 was considered statistically significant, *represents p < 0.05, ** denotes p < 0.01, and *** indicates p < 0.001.

Results

Cuproptosis-related genes expression and immune infiltration in pterygium

To clarify the role of CuRGs in pterygium progression, we analyzed the expression of 19 CuRGs in normal conjunctiva and pterygium tissues using the PRJNA1147595 dataset. Our analysis identified eight differentially expressed CuRGs, which are depicted in Fig. 2A–C. Specifically, ATP7B, SLC31A1, CDKN2A, and PDHB exhibited upregulation in pterygium, whereas PDHA1, LIPT1, NLRP3, and MTF1 were notably downregulated compared to the conjunctiva. Subsequently, we undertook a correlation analysis of these differentially expressed CuRGs (DECuRGs) to investigate their interactions within pterygium. The resulting relationship network highlighted the tight interplay among these DECuRGs (Fig. 2D). PDHA1 and NLRP3 showed an antagonistic relationship with a correlation of −0.59, while ATP7B and SLC31A1 displayed a synergistic interaction with a correlation of 0.64 (Fig. 2E). Employing the CIBERSORT algorithm, we conducted an immune infiltration analysis to assess variations in the proportions of 22 immune cell types between pterygium and normal conjunctiva. The study found increased levels of regulatory T cells, M0 macrophages, M1 macrophages, and resting dendritic cells in pterygium (Fig. 2F), indicating that local tissue immune system activation might play a role in pterygium pathogenesis. We conducted DECuRGs and immune infiltrating cells correlation analysis to explore the potential role of these genes in immune regulation in pterygium. The results may suggest that M1 macrophages were correlated with seven DECuRGs. NLRP3 and neutrophils (cor = 0.59, p < 0.001), along with ATP7B and neutrophils (cor = 0.65, p < 0.001), potentially indicated the positive correlations. Conversely, NLRP3 and naive B cells displayed the most significant adverse interaction (Fig. 2G). Collectively, these findings imply that these DECuRGs may play immune-modulating roles in pterygium. These findings should be interpreted as exploratory and require further validation in larger, independent cohorts.

Fig. 2.

Fig. 2

Open in a new tab

Transcriptomic profiling of copper metabolism in pterygium. (A) Boxplot analysis of expression profiles for 19 copper-related genes (CuRGs). (B) Hierarchical clustering heatmap showing differential expression patterns of 8 DECuRGs (differentially expressed CuRGs). (C) Genomic distribution mapping of DECuRGs across human chromosomes. (D) Regulatory network diagram with activating (red) and inhibitory (blue) interactions. (E) Correlation matrix visualization using proportional area encoding. (F) CIBERSORT-derived immune cell infiltration proportions across tissue types. (G) Spearman’s correlation heatmap between DECuRGs and immune cells.

Enrichment analysis and PPI network construction

We used the clusterProfiler package to perform enrichment analysis on the 8 differentially expressed copper-related genes (DECuRGs) to explore their potential biological functions and pathways. KEGG analysis found that these DECuRGs were significantly enriched in crucial pathways such as valine, leucine, and isoleucine biosynthesis, the citrate cycle (TCA cycle), butanoate metabolism, pyruvate metabolism, and glycolysis/gluconeogenesis (Fig. 3A, B). The GO analysis revealed that these DECuRGs were enriched in biological processes like acetyl-CoA biosynthesis from pyruvate, copper ion transport, as well as other biological processes (BP), cellular components (CC), and molecular functions (MF) (Fig. 3C, D). We developed a protein–protein interaction (PPI) network to identify the main contributors among these DECuRGs. Six DECuRGs: PDHA1, PDHB, LIPT1, SLC31A1, ATP7B, and MTF1 were identified as hub genes, creating a network with six nodes and edges, and a confidence score exceeding 0.4 (Fig. 3E). This network provides insights into the potential interactions and interdependencies among these genes.

Fig. 3.

Fig. 3

Open in a new tab

Functional annotation of DECuRGs. (A, B) KEGG pathway enrichment shown through (A) bubble plot and (B) column chart. (C, D) GO term analysis categorized by (C) biological processes, (D) cellular components and molecular functions. (E) STRING-derived PPI network highlighting six hub genes (Cytoscape 3.10.2 visualization).

Characterization and immune characteristics of cuproptosis-related clusters in pterygium

Building on our previous analysis, we obtained expression data for eight DECuRGs performed a comparative analysis on 34 pterygium samples. We employed ssGSEA to assess the gene set enrichment in these samples. We divided the samples into low and high score clusters using the median ssGSEA score as a threshold. To identify the molecular signatures differentiating the low and high score clusters, we first examined the expression differences of the eight DECuRGs between these clusters. Our results indicated that the high score cluster was characterized by notably elevated expression levels of ATP7B, SLC31A1, CDKN2A, PDHB, and MTF1, as depicted in Fig. 4A, B presents an expression heatmap illustrating the differential expression of these eight DECuRGs between the low and high score groups. Furthermore, our differential analysis of immune cells revealed a marked alteration in the immune infiltration microenvironment between the two clusters. Specifically, the high score cluster exhibited a significantly higher proportion of resting dendritic cells, as illustrated in Fig. 4C, D.

Fig. 4.

Fig. 4

Open in a new tab

Dysregulation of cuproptosis-related genes associated molecular signatures. (A, B) Cluster-specific expression patterns of DECuRGs by (A) boxplot and (B) heatmap. (C) Differential immune infiltration landscape between clusters. (D) CIBERSORT quantification of immune cell subtype variations.

WGCNA analysis of module gene enrichment results

Gene expression values from disease and control samples were used to create co-expression modules via the WGCNA algorithm. We focused on the top 25% of variable genes for subsequent analysis. The samples were analyzed using cluster analysis with the “flashClust” package. The selected power value was 10, achieving an independence degree of 0.9 and a relatively high mean connectivity degree (Fig. 5A, B). Using the dynamic cutting algorithm, we identified ten unique gene co-expression modules and depicted the corresponding heatmap of the topological overlap matrix (TOM) (Fig. 5C, D). Notably, the tan, salmon, and turquoise modules exhibited the strongest correlation with pterygium, encompassing 647 protein-coding genes. We utilized the WGCNA package to analyze hub gene modules within the cuproptosis-related clusters. Co-expression modules were delineated with a soft power value of 18 and a scale-free R2 of 0.9 (Fig. 6A, B). Subsequently, the dynamic tree cutting algorithm yielded eleven distinct modules, each assigned a unique color. The interactions among the eleven co-expression modules were analyzed (Fig. 6C–F). Correlation analysis identified a strong association between the cuproptosis-related clusters and the brown, yellow, and green modules, collectively containing 546 protein-coding genes.

Fig. 5.

Fig. 5

Open in a new tab

Co-expression network architecture. (A, B) Scale-free topology analysis for optimal soft-threshold selection. (C) Module assignment heatmap with color-coded correlation patterns. (D) Eigengene-trait correlation matrix with coefficient/p value annotations.

Fig. 6.

Fig. 6

Open in a new tab

Module-cluster association analysis. (A, B) Network parameter optimization through soft-threshold determination. (C) Gene dendrogram with dynamic tree-cut module identification. (D) Eigengene clustering dendrogram. (E) Inter-module relationship heatmap. (F) Module-trait correlation matrix with dysregulation of cuproptosis-related genes score associations.

Establishment and validation of the machine learning prediction model

Through integrative analysis of pterygium-conjunctiva comparative transcriptomics and cuproptosis-associated clusters, we identified 15 pivotal hub genes (Fig. 7A). Four machine learning frameworks—Generalized Linear Model (GLM), Random Forest (RF), eXtreme Gradient Boosting (XGB), and Support Vector Machine (SVM)—were constructed to assess their predictive capabilities using these genetic signatures. The SVM algorithm showed competitive predictive performance with relatively low residual errors (Root Mean Square residual = 0.32) in comparison with other models (Fig. 7B, C). Receiver operating characteristic (ROC) analysis revealed varying discrimination capacities across models: GLM achieved the highest Area Under the Curve (AUC) of 0.865, followed by SVM and RF (both AUC = 0.84), and XGB (AUC = 0.82) (Fig. 7D). Although GLM yielded a marginally higher AUC, its residual error was substantially larger than those of SVM, XGB, and RF. Given that a large residual error may compromise the accuracy of a diagnostic model, we considered both discrimination and calibration performance in model selection. On balance, the SVM model provided a reasonable trade-off between AUC (0.84) and residual error (0.32), and was therefore selected as the preferred predictive model for subsequent analyses. Feature importance ranking based on permutation RMSE loss revealed model-specific predictive biomarkers (Fig. 7E).

Fig. 7.

Fig. 7

Open in a new tab

Machine learning model selection. (A) Venn diagram of module-cluster gene intersections. (B) Model performance comparison using reverse distribution curves. (C) Residual error distribution across algorithms. (D) ROC curve evaluation of RF/SVM/XGB/GLM classifiers. (E) Feature importance ranking across computational frameworks.

We developed a diagnostic model utilizing QPCR to validate the statistically significant differences in the expression levels of the SERTAD1, CSRNP1, and JMJD1C genes, which will be elaborated upon subsequently. We subsequently established a multivariate nomogram incorporating three core biomarkers (SERTAD1, JMJD1C, CSRNP1) for pterygium risk stratification (Fig. 8A). Model validation through calibration curves and decision curve analysis confirmed high predictive precision and clinical utility (Fig. 8B, C). Upon external validation using the GSE51995 cohort, the model demonstrated an impressive area under the curve (AUC) of 0.938 (with a 95% confidence interval [CI] ranging from 0.860 to 1.000), underscoring its robust generalizability across independent datasets (as depicted in Fig. 8D).

Fig. 8.

Fig. 8

Open in a new tab

Clinical prediction model development. (A) Multivariate nomogram incorporating three molecular biomarkers. (B, C) Clinical utility assessment through (B) DCA and (C) calibration plots. (D) External validation ROC curve in GSE51995 cohort.

Validation of hub gene mRNA expression

Analysis of RNA-Seq data from 68 clinical samples revealed significantly elevated expression levels (P < 0.05) of SERTAD1, JMJD1C, SMCHD1, and CSRNP1 in conjunctival tissues compared to pterygium, we found that the expressions of the SERTAD1, JMJD1C, SMCHD1, and CSRNP1 genes were downregulated in the pterygium tissue (Fig. 9A). Intriguingly, although CYP1B1 was detected in the sequencing dataset, its expression levels did not differ significantly between the two tissue types. QPCR validation subsequently confirmed the downregulation of SERTAD1, CSRNP1, and JMJD1C in pterygium specimens (P < 0.05 Fig. 9B). The 11 pterygium samples used for QPCR validation included 3 cases of Grade 1, 5 cases of Grade 2, and 3 cases of Grade 3 pterygiums. However, the expression trends of SMCHD1 and CYP1B1 were that their expression levels were low within the pterygium tissue, but there was no statistical difference between the pterygium group and the conjunctiva group (Fig. 9B). We conducted QPCR comparisons of the grade 1 pterygium tissue, the grade 2 pterygium tissue, and the grade 3 pterygium tissue with the conjunctival tissue. In the grade 1 pterygium tissue, the downregulation of SERTAD1, CSRNP1, and JMJD1C was observed. In the grade 2 and 3 pterygium tissues, there was no statistically significant difference in the expression of these five genes (Supplementary Fig. 1). In the GSE51995 dataset, the expression trends of SERTAD1, JMJD1C, SMCHD1, CSRNP1, and CYP1B1 all showed higher expression levels in conjunctival tissues compared to pterygium tissues (Supplementary Fig. 2).

Fig. 9.

Fig. 9

Open in a new tab

Validation of differential gene expression profiles. (A) Comparative RNA-Seq analysis revealed significantly reduced expression of SERTAD1, JMJD1C, SMCHD1, and CSRNP1 in pterygium tissues versus normal conjunctival controls (P < 0.05), whereas CYP1B1 exhibited comparable expression levels between groups. (B) QPCR validation confirmed transcriptional downregulation of SERTAD1, CSRNP1, and JMJD1C in pterygium (P < 0.05). The expression trends of SMCHD1 and CYP1B1 in pterygium tissue were lower than those in conjunctival tissue.

Discussion

Pterygium, a progressive ocular surface disorder, manifests as fibrovascular conjunctival encroachment onto the corneal surface. Chronic ultraviolet radiation exposure emerges as the predominant etiological factor, synergistically interacting with particulate pollutants and arid climatic conditions to drive disease progression40,41. Notably, epidemiological data from solar-intensive regions like Yunnan Province reveal disproportionately high prevalence rates (22.6–35.27%), establishing this condition as a priority ophthalmic public health concern8. The molecular pathogenesis remains elusive, particularly regarding the UV-induced epigenetic modifications and dysregulated Wnt/β-catenin signaling pathways implicated in disease initiation42. Current therapeutic modalities remain confined to surgical resection with adjunctive mitomycin C application, yet persistently high recurrence rates underscore the urgent need for targeted molecular therapies addressing the underlying pathobiology. This therapeutic gap necessitates accelerated translational research integrating multi-omics approaches to develop recurrence-preventive strategies43.

This study presents an investigation of the differential expression patterns of CuRGs in pterygium tissues and their potential immunomodulatory functions44. The scientific premise for this exploration derives from the well-documented roles of CuRGs in mediating programmed cell death pathways and oncogenic transformations, coupled with their emerging significance in ocular surface immunobiology. Through systematic molecular profiling, we identified distinct dysregulation patterns of multiple CuRGs in pterygium samples, speculating their pathobiological significance in disease manifestation. Notably, the observed transcriptional aberrations suggest that CuRGs dysregulation may be associated with immune remodeling in pterygium pathogenesis. These collective findings establish crucial conceptual frameworks for subsequent mechanistic studies employing transcriptomics approaches and preclinical models to dissect the copper-immunity axis, potentially informing the development of biomarker-driven therapeutic strategies for this visually debilitating disorder45. Our systematic investigation of CuRGs expression patterns in pterygium pathogenesis associated dysregulation in copper-associated transcriptional networks, potentially driving disease progression through distinct molecular mechanisms. Molecular profiling identified eight CuRGs with significant expression divergence, with ATP7B, SLC31A1, CDKN2A, and PDHB demonstrating marked upregulation, whereas PDHA1, LIPT1, NLRP3, and MTF1 exhibited significant downregulation compared to healthy conjunctival tissues. This coordinated dysregulation implicates CuRGs as critical mediators of copper-redox imbalance, potentially initiating a pathogenic cascade involving mitochondrial dysfunction and oxidative proteotoxicity-mechanisms previously documented in chronic inflammatory disorders17. Notably, the elevated ATP7B expression aligns with its established role in copper homeostasis, suggesting adaptive cellular responses to metal ion accumulation in limbal stem cell compartments20. Conversely, NLRP3 suppression reveals unexpected immune microenvironment modulation through cuproptosis pathways28. SLC31A1, as a key copper ion transporter, is primarily responsible for transporting copper into cells and serves as a crucial link in initiating cuproptosis46. CDKN2A is an important cell cycle regulator and tumor suppressor gene, and its expression changes are closely related to the prognosis of various cancers47. PDHB is the β-subunit of the pyruvate dehydrogenase complex, participating in the pyruvate oxidative decarboxylation process within the mitochondrial respiratory chain, and its expression level is associated with tumor pathological staging46. PDHA1 is the α-subunit of the pyruvate dehydrogenase complex and plays a central role in the initial step of the tricarboxylic acid cycle, with its expression being related to patient survival48. LIPT1 encodes lipoic acid transferase, which is involved in lipoylation modification, a process that serves as the direct molecular basis for cuproptosis48. MTF1 is a metal-responsive transcription factor that maintains intracellular metal ion (such as copper and zinc) homeostasis by regulating the expression of downstream genes, including metallothioneins, thereby influencing sensitivity to cuproptosis49. These paradoxical findings illuminate novel crosstalk between metal metabolism and immunoregulation.

Moreover, the immune infiltration analysis performed using the CIBERSORT algorithm revealed significant alterations in the composition of immune cell populations within pterygium tissues. Specifically, we observed an increase in regulatory resting dendritic cells, T cells and M0 macrophage indicating a state of local immune activation. This aligns with prior studies that have demonstrated the role of immune dysregulation in the pathogenesis of ocular surface diseases, including pterygium50. The significant correlation between the expression of DECuRGs and immune cell types, particularly the positive association with M1 macrophages, suggests that these genes may not only be involved in copper metabolism but also in modulating immune responses in pterygium. Such interactions highlight the complexity of the immune microenvironment in pterygium. Finally, our machine learning models demonstrated high predictive power for identifying pterygium patients based on the expression profiles of selected hub genes51. All the pterygium tissues incorporated in this study were primary nasal pterygium tissues originating from the Yunnan region. Among them, there were 5 cases of Grade 1 pterygium, 18 cases of Grade 2 pterygium, and 11 cases of Grade 3 pterygium. No recurrent pterygium tissues were included. Given that the average ultraviolet intensity in Yunnan is 260.75 MJ/m2, we observed that the expression levels of SERTAD1, JMJD1C, SMCHD1, CSRNP1, and CYP1B1 were higher in conjunctival tissues than in pterygium tissues. Consequently, we infer that these biomarkers are associated with primary nasal pterygium. Nevertheless, further research is warranted to explore the expression of key biomarkers in recurrent pterygium tissues. These findings suggest that integrating machine learning approaches with molecular data can provide valuable insights for understanding the disease and developing targeted interventions. Future research should focus on validating these predictive models in larger cohorts and exploring the mechanistic pathways through which CuRGs influence both copper homeostasis and immune responses in pterygium. The potential for utilizing these biomarkers in clinical practice may be associated to improved patient outcomes and reduced recurrence rates following surgical interventions.

Limitations

The limitations of this study include a relatively small sample size, which may restrict the generalizability of the findings. This research was centered around transcriptome sequencing of primary nasal lateral pterygium. Notably, we refrained from conducting separate comparisons among grade 1, grade 2, and grade 3 pterygium to identify markers specific to each grade. Looking ahead, the collection of a larger number of samples across different grades will facilitate more in-depth analysis. It is important to note that this study, based on the association analysis of transcriptome data, identified the abnormal expression patterns of cuproptosis-related genes dysregulation in pterygium. Additionally, the lack of extensive wet lab validation raises concerns regarding the reproducibility of the identified gene expressions in clinical settings. The limited sample size (n = 34) may compromise the precision of the estimated proportions of immune infiltration. Additionally, the CIBERSORT algorithm inherently has estimation biases, particularly with regarding in rare immune cell types. Futures studies should leverage single-cell sequencing and spatial transcriptomics to elucidate the underlying mechanisms of pterygium pathogenesis and delineate the composition of distinct cellular subpopulations within the tissue. In the future, further confirmation of the role of the cuproptosis-related pathway in diseases will be achieved through in vitro and in vivo experiments. The potential batch effects and discrepancies between different datasets may also introduce biases that impact the robustness of our conclusions. Furthermore, while our machine learning models demonstrated promising discriminative capabilities, the inherent complexity of biological systems necessitates caution in their clinical applicability. Lack of samples from recurrent cases, which prevents the assessment of the correlation between biomarkers and recurrence risk, and the single-center design may restrict the generalizability of the results. The current findings should not be directly used to guide clinical decision-making but rather to provide candidate biomarkers for future research. Future investigations should aim to incorporate larger cohorts and diverse populations to validate our results and refine the predictive models further.

Conclusion

In summary, this research provides valuable insights into the role of cuproptosis-related genes in the pathogenesis of pterygium, potentially correlates with their potential as novel therapeutic targets. The identification of key genes and their correlation with immune responses presents a foundation for future studies aimed at elucidating the underlying mechanisms of this disease. The developed predictive model, backed by external validation, demonstrates the feasibility of integrating genetic signatures into clinical practice, ultimately paving the way for more personalized approaches in managing pterygium. But these findings are hypothesis-generating and require validation in larger patient cohorts.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (724.3KB, jpg)
Supplementary Material 2 (634.6KB, jpg)
Supplementary Material 3 (8.6KB, xlsx)
Supplementary Material 4 (8.6KB, xlsx)
Supplementary Material 5 (14KB, xlsx)

Acknowledgements

We thank all research group members for the helpful discussion.

Author contributions

Hai Liu, Liwei Zhang and Ji Yang conceived this study. Ji Yang, Hai Liu, Ping Xiang designed the study. Ji Yang and BoYu Liang conducted experiments. Ji Yang, ChengYan Fang analyzed the data. Ji Yang and BoYu Liang wrote the manuscript. Clinical samples were collected and provided by BoYu Liang, ChengYan Fang, Jiajie Li and Peng Zhang. Ji Yang and BoYu Liang reviewed the manuscript and made critical corrections. Hai Liu, LiWei Zhang and Ji Yang provided the funding for the sequencing. All authors read and approved the final manuscript.

Funding

This work was carried out with the support of the National Natural Science Foundation of China (Grant No. 82460201 and 82560205). Key Project of the Provincial Clinical Center of Yunnan Province (2024YNLCYXZX0343). Yunnan University Medical Research Foundation (YDYXJJ2024-0003). National Clinical Key Specialty Ophthalmology Open Foundation (Grant No. ZKF2024041 and ZKF2024042) and the Yunnan Xingdian Youth Talent Program (YNQR-QNRC-2018-049, XDYC-MY-2024). Tianjin Natural Science Foundation (24JCQNJC02050).

Data availability

All the raw data disclosed in this article can be accessed upon request from the authors. Our RNA-seq data have been deposited in Sequence Read Archive (SRA https://dataview.ncbi.nlm.nih.gov/) with accession numbers PRJNA1147595 and the Gene Expression Omnibus (GEO https://www.ncbi.nlm.nih.gov/geo/) with accession numbers GSE51995.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The research protocol was approved by the Ethics Review Committee of the Affiliated Hospital of Yunnan University (approval number: 2022198).

Consent for publication

Not applicable.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Ji Yang, Bo-Yu Liang, Chen-Yan Fang have equally contributed as co-first authors.

Contributor Information

Li-Wei Zhang, Email: drzhangliwei@163.com.

Hai Liu, Email: herryhai@ynu.edu.cn.

References

  • 1.Oellers, P. et al. Prevalence, treatment, and outcomes of coexistent ocular surface squamous neoplasia and pterygium. Ophthalmology120, 445–450. 10.1016/j.ophtha.2012.08.010 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Chui, J. et al. Ophthalmic pterygium: A stem cell disorder with premalignant features. Am. J. Pathol.178, 817–827. 10.1016/j.ajpath.2010.10.037 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Hill, J. C. & Maske, R. Pathogenesis of pterygium. Eye (Lond.)3(Pt 2), 218–226. 10.1038/eye.1989.31 (1989). [DOI] [PubMed] [Google Scholar]
  • 4.Lingham, G. et al. Conjunctival ultraviolet autofluorescence area decreases with age and sunglasses use. Br. J. Ophthalmol.107, 614–620. 10.1136/bjophthalmol-2021-320284 (2023). [DOI] [PubMed] [Google Scholar]
  • 5.Rezvan, F. et al. Prevalence and risk factors of pterygium: A systematic review and meta-analysis. Surv. Ophthalmol.63, 719–735. 10.1016/j.survophthal.2018.03.001 (2018). [DOI] [PubMed] [Google Scholar]
  • 6.Sun, L. P. et al. The prevalence of and risk factors associated with pterygium in a rural adult Chinese population: The Handan eye study. Ophthalmic Epidemiol.20, 148–154. 10.3109/09286586.2013.763991 (2013). [DOI] [PubMed] [Google Scholar]
  • 7.Weber, I., Liao, K., Dang, T., Shah, M. & Wehner, M. R. Sunburn and cutaneous squamous cell carcinoma: A meta-analysis. JAMA Dermatol.161, 1148–1156. 10.1001/jamadermatol.2025.3473 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Ke, H.-q. et al. Study on prevalence of pterygium and its influencing factors in rural population in 6 regions of Yunnan Province. 58, 9 (2022).
  • 9.Lu, C. W. et al. Impacts of air pollution and meteorological conditions on dry eye disease among residents in a northeastern Chinese metropolis: A six-year crossover study in a cold region. Light Sci. Appl.12, 186. 10.1038/s41377-023-01207-1 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Tandon, R. et al. The association of sun exposure, ultraviolet radiation effects and other risk factors for pterygium (the SURE RISK for pterygium study) in geographically diverse adult (≥ 40 years) rural populations of India -3rd report of the ICMR-EYE SEE study group. PLoS ONE17, e0270065. 10.1371/journal.pone.0270065 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Lee, B. W. H., Ip, M. H., Tat, L., Chen, H. & Coroneo, M. T. Modified limbal-conjunctival autograft surgical technique: Long-term results of recurrence and complications. Cornea42, 1320–1326. 10.1097/ico.0000000000003337 (2023). [DOI] [PubMed] [Google Scholar]
  • 12.Hirst, L. W. Other considerations in pterygium surgery. Ophthalmology120, e60. 10.1016/j.ophtha.2013.06.011 (2013). [DOI] [PubMed] [Google Scholar]
  • 13.Hirst, L. W. Recurrence and complications after 1000 surgeries using pterygium extended removal followed by extended conjunctival transplant. Ophthalmology119, 2205–2210. 10.1016/j.ophtha.2012.06.021 (2012). [DOI] [PubMed] [Google Scholar]
  • 14.Zhang, M. C. & Wang, Y. [Pay attention to basic and clinical research of pterygium]. [Zhonghua yan ke za zhi] Chin. J. Ophthalmol.43, 868–871 (2007). [PubMed] [Google Scholar]
  • 15.Lev Ari, O. et al. Association between pterygium and ocular, periocular, and systemic inflammatory conditions: A large-scale national study. Cornea10.1097/ico.0000000000003697 (2024). [DOI] [PubMed] [Google Scholar]
  • 16.Böhm, E. W. et al. Oxidative stress in the eye and its role in the pathophysiology of ocular diseases. Redox Biol.68, 102967. 10.1016/j.redox.2023.102967 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zhao, R. et al. Cuproptosis, the novel type of oxidation-induced cell death in thoracic cancers: Can it enhance the success of immunotherapy?. Cell Commun. Signal.22, 379. 10.1186/s12964-024-01743-2 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Qi, N., Wang, B., Xing, W., Ge, F. & Liu, J. The protective role of quercetin against copper-induced female reproductive toxicity: Insights from transcriptome analysis. Food Chem. Toxicol.192, 114934. 10.1016/j.fct.2024.114934 (2024). [DOI] [PubMed] [Google Scholar]
  • 19.Huang, X. A concise review on oxidative stress-mediated ferroptosis and cuproptosis in Alzheimer’s disease. Cells10.3390/cells12101369 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Członkowska, A. et al. Wilson disease. Nat. Rev. Dis. Primers4, 21. 10.1038/s41572-018-0018-3 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Chevalier, K. et al. Eye involvement in Wilson’s disease: A review of the literature. J. Clin. Med.10.3390/jcm11092528 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Zhang, M. et al. Identification of cuproptosis-related circRNA-miRNA-mRNA network in laser-induced choroidal neovascularization models and in peripheral blood mononuclear cells of patients with neovascular age-related macular degeneration. Ophthalmic Res.66, 1417–1432. 10.1159/000535170 (2023). [DOI] [PubMed] [Google Scholar]
  • 23.Pan, Z. et al. Prevalence and risk factors for pterygium: A cross-sectional study in Han and Manchu ethnic populations in Hebei, China. BMJ Open9, e025725. 10.1136/bmjopen-2018-025725 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Zhang, L. W. et al. Identification of biomarkers and immune microenvironment associated with pterygium through bioinformatics and machine learning. Front. Mol. Biosci.11, 1524517. 10.3389/fmolb.2024.1524517 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Aihua, H. et al. Evaluation of global differential gene and protein expression in primary pterygium: S100A8 and S100A9 as possible drivers of a signaling network. PLoS ONE9, e97402 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Wong, Y. M. et al. Expression of insulin-like growth factor binding protein-3 in pterygium tissue. Br. J. Ophthalmol.90, 769–772 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Hou, A. et al. Evaluation of global differential gene and protein expression in primary pterygium: S100A8 and S100A9 as possible drivers of a signaling network. PLoS ONE9, e97402. 10.1371/journal.pone.0097402 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Xue, Q. et al. Copper metabolism in cell death and autophagy. Autophagy19, 2175–2195. 10.1080/15548627.2023.2200554 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Xie, J., Yang, Y., Gao, Y. & He, J. Cuproptosis: Mechanisms and links with cancers. Mol. Cancer22, 46. 10.1186/s12943-023-01732-y (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Ritchie, M. E. et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res.43, e47. 10.1093/nar/gkv007 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Kanehisa, M. & Sato, Y. KEGG mapper for inferring cellular functions from protein sequences. Protein Sci.29, 28–35. 10.1002/pro.3711 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Kanehisa, M., Furumichi, M., Sato, Y., Matsuura, Y. & Ishiguro-Watanabe, M. KEGG: Biological systems database as a model of the real world. Nucleic Acids Res.53, D672-d677. 10.1093/nar/gkae909 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Kanehisa, M. Toward understanding the origin and evolution of cellular organisms. Protein Sci.28, 1947–1951. 10.1002/pro.3715 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Szklarczyk, D. et al. The STRING database in 2017: Quality-controlled protein-protein association networks, made broadly accessible. Nucleic Acids Res.45, D362-d368. 10.1093/nar/gkw937 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Newman, A. M. et al. Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods12, 453–457. 10.1038/nmeth.3337 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: Gene set variation analysis for microarray and RNA-seq data. BMC Bioinform.14, 7. 10.1186/1471-2105-14-7 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Langfelder, P. & Horvath, S. WGCNA: An R package for weighted correlation network analysis. BMC Bioinformatics9, 559. 10.1186/1471-2105-9-559 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Nukui, T. & Onogi, A. An R package for ensemble learning stacking. Bioinform. Adv.3, vbad139. 10.1093/bioadv/vbad139 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Jia, F. et al. Exploring the cuproptosis-related molecular clusters in the peripheral blood of patients with amyotrophic lateral sclerosis. Comput. Biol. Med.168, 107776. 10.1016/j.compbiomed.2023.107776 (2024). [DOI] [PubMed] [Google Scholar]
  • 40.Ding, P., Wang, R. & He, Y. Risk factors for pterygium: Latest research progress on major pathogenesis. Exp. Eye Res.243, 109900. 10.1016/j.exer.2024.109900 (2024). [DOI] [PubMed] [Google Scholar]
  • 41.Lee, K. W. et al. Outdoor air pollution and pterygium in Korea. J. Korean Med. Sci.32, 143–150. 10.3346/jkms.2017.32.1.143 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Li, J. et al. Expression profiling suggests the involvement of hormone-related, metabolic, and Wnt signaling pathways in pterygium progression. Front. Endocrinol.13, 943275. 10.3389/fendo.2022.943275 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Huang, Y., He, H., Sheha, H. & Tseng, S. C. Ocular demodicosis as a risk factor of pterygium recurrence. Ophthalmology120, 1341–1347. 10.1016/j.ophtha.2013.01.001 (2013). [DOI] [PubMed] [Google Scholar]
  • 44.Sun, Z. et al. Cuproptosis and immune-related gene signature predicts immunotherapy response and prognosis in lung adenocarcinoma. Life10.3390/life13071583 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Yoon, C. H., Seol, B. R. & Choi, H. J. Effect of pterygium on corneal astigmatism, irregularity and higher-order aberrations: A comparative study with normal fellow eyes. Sci. Rep.13, 7328. 10.1038/s41598-023-34466-4 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Ge, X. et al. Clinical significance and immune microenvironment association of cuproptosis-related genes in pan-cancer. Exp. Physiol.10.1113/ep092310 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Wu, Z. et al. Comprehensive analysis of cuproptosis genes and cuproptosis-related genes as prognosis factors in esophageal squamous cell carcinoma. Genomics115, 110732. 10.1016/j.ygeno.2023.110732 (2023). [DOI] [PubMed] [Google Scholar]
  • 48.Liu, J. et al. A comprehensive analysis and validation of cuproptosis-associated genes across cancers: Overall survival, the tumor microenvironment, stemness scores, and drug sensitivity. Front. Genet.13, 939956. 10.3389/fgene.2022.939956 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Luan, M. et al. MTF1 attenuates ferroptosis and cuproptosis synergistic potentiation in gastric cancer. Cell Death Differ.10.1038/s41418-025-01641-1 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Labbé, A. et al. An in vivo confocal microscopy and impression cytology evaluation of pterygium activity. Cornea29, 392–399. 10.1097/ICO.0b013e3181bd44ce (2010). [DOI] [PubMed] [Google Scholar]
  • 51.Handelman, G. S. et al. eDoctor: Machine learning and the future of medicine. J. Intern. Med.284, 603–619. 10.1111/joim.12822 (2018). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (724.3KB, jpg)
Supplementary Material 2 (634.6KB, jpg)
Supplementary Material 3 (8.6KB, xlsx)
Supplementary Material 4 (8.6KB, xlsx)
Supplementary Material 5 (14KB, xlsx)

Data Availability Statement

All the raw data disclosed in this article can be accessed upon request from the authors. Our RNA-seq data have been deposited in Sequence Read Archive (SRA https://dataview.ncbi.nlm.nih.gov/) with accession numbers PRJNA1147595 and the Gene Expression Omnibus (GEO https://www.ncbi.nlm.nih.gov/geo/) with accession numbers GSE51995.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES