Introduction

The incidence of stroke, the second leading cause of death globally, is increasing due to the dramatic rise in the elderly population aged 65 years and older1. Ischemic stroke (IS) is primarily classified as either ischemic or hemorrhagic, with IS accounting for 87% of all cases and exhibiting strong heritability1,2. IS is characterized by intravascular thrombosis leading to brain tissue necrosis and focal neuronal functional deficits3. Furthermore, diabetes, hypertension, hyperlipidemia, smoking, and genetic factors have been implicated in the pathogenesis of IS. Due to the complexity of brain structure and cellular diversity within this organ, our understanding of molecular mechanisms underlying IS remains limited. Currently, clinical symptoms and imaging tests serve as primary means for screening and diagnosing IS; however early diagnosis and prognosis assessment are challenging when relying solely on these approaches. Without timely improvement in clinical strategies, both prevalence and mortality rates will continue to rise.

Neurotrophic factors (NTFs) are a group of secreted proteins that play crucial roles in the development, growth, survival, and differentiation of the nervous system. They also contribute significantly to maintaining central nervous system homeostasis4,5. This primarily includes Nerve Growth Factor (NGF), Brain-Derived Neurotrophic Factor (BDNF), Neurotrophin-3 (NT-3), and Neurotrophin-4/5 (NT-4/5)6. Among these factors, brain-derived neurotrophic factor (BDNF) is the most abundant and widely distributed in the nervous system7. Previous studies have demonstrated that BDNF plays a pivotal role in ischemia and its deficiency is associated with severe pathophysiological symptoms in stroke patients8.

Notably, Zhu ZH et al. demonstrated that BDNF-loaded exosomes derived from neural stem cells possess the potential to treat IS, effectively mitigating cerebral ischemia-induced damage9. Additionally, BDNF is implicated in mood disorders like post-stroke depression (PSD), as evidenced by lower serum levels of BDNF in PSD patients compared to those without depression10. Antidepressant medications can increase brain expression of BDNF, thereby alleviating depressive symptoms11. Additionally, NT-3 and NT-4 have shown important roles in regulating the survival and function of neurons, particularly in the context of spinal cord injuries and neural regeneration12,13. The signaling mechanisms of neurotrophins and their receptors are also being continually uncovered. BDNF and NT-4 regulate the activity of TrkB receptors through different endocytic pathways, leading to distinct biological effects14. Moreover, the expression of the p75 neurotrophin receptor is upregulated after neural injury, potentially participating in the processes of neuronal apoptosis and regeneration15. Therefore, we believe that NTFs play a significant role in IS’s development and further investigation into their underlying mechanisms is warranted.

In order to gain a deeper understanding of the role of NTFs in IS, we identified biomarkers associated with NTFs based on IS transcriptome data from public database and NTFs-related genes (NFRGs) through differential expression analysis, WGCNA and three machine learning models. We also explored the involvement of these biomarkers in the pathogenesis and progression of IS through immunoassays, GSEA analysis, construction of regulatory networks, and expression validation to provide new strategies and targets for the prevention, diagnosis, and treatment of IS.

Materials and methods

Data acquisition

The IS’s transcriptome sequencing datasets GSE16561 and GSE58294 were downloaded from the GEO database. According to the probe correspondence of the platform, the probe was converted into Symbol. The final GSE16561 dataset comprised transcriptome sequencing data of 39 individuals with IS and 24 healthy peripheral blood samples. The GSE58294 dataset included transcriptome sequencing data of 69 individuals with IS and 23 normal peripheral blood samples. We used the GSE16561 dataset for analysis and the GSE58294 dataset for validation. The 2040 NFRGs (correlation score ≥ 7) were from GeneCards database.

WGCNA identified co-expressed gene modules and obtained NFRGs in IS

In this study, WGCNA16 was utilized to identify co-expressed gene modules in IS, whereby correlation coefficient between genes was calculated and the hierarchical clustering tree was constructed. The significance of the modules was calculated and used to analyze the correlation between the disease and different modules. The key genes selected in IS were intersected with NFRGs and DEGs to identify NFRGs in IS by Venn diagram.

Functional enrichment analysis and construction PPI network of IS’s key NFRGs

GO and KEGG17,18,19 analysis for the key NFRGs of IS was conducted by R package “clusterProfiler”20 to explore the key pathways of action. The threshold screening condition was p < 0.05. The “GOplot” package21 was used to display the analysis results respectively in bubble chart and circle graph. The PPI network of the key NFRGs of IS were established using the STRING database.

Machine learning and prognosis evaluation

To reduce potential bias in identifying diagnostic genes, NFRGs in IS were analysed using several machine learning algorithms: SVM-RFE analysis using the ‘caret’ package (v 1.6–14) with 10-fold cross-validation, LASSO analysis using the The ‘glmnet’ package (v 4.1-8)22 is used for LASSO analysis, and the optimal value of the regularisation parameter lambda is determined by a grid search. The RF algorithm was analysed using the ‘randomForest’ package (v 4.6–14). The intersection genes were screened as IS-related biomarkers for diagnostic capability and expression verification. For diagnostic evaluation, the AUC was calculated and the ROC curve was plotted. Finally, box plots of gene expression were drawn.

GSEA enrichment and immunoinfiltration analysis

According to the median expression level of IS samples’s biomarkers, they were divided into high and low expression group, and then GSEA analysis was performed using clusterProfiler. The criteria for defining a statistically significant pathway were set as q < 0.25 and p < 0.05. The ssGSEA algorithm quantified the relative abundance of cell infiltration in IS patients and normal control samples. Spearman correlations between diagnostic marker genes and immune cell were analyzed using the ‘linkET’ package.

ceRNA network analysis and drug prediction evaluation

The regulatory relationship between TFs, miRNAs and biomarkers in IS was obtained from NetworkAnalyst database and visualized via Cytoscape (v.3.8.2)23.We used the DGIdb database to predict potential drugs or molecular compounds that interact with diagnostic markers.

RT-qPCR

We collected 15 IS and 5 normal peripheral blood samples from rehabilitation department of Shanxi Bethune hospital. The RNA was extracted and reverse-transcribed into cDNA for qRT-PCR to identify the expression level of diagnostic genes. RT-qPCR was then performed on the ABI 7300 real-time fluorescence quantifier. The results were analyzed by 2−△△Ct, and the data were analyzed by unpaired t-test. Primers were shown in Table 1. All patients were informed of the purpose of specimen collection according to the guidelines of the biomedical research ethics committee of the Shanxi Bethune hospital (approval number: YXLL-2024-056). This study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all individual participants included in the study.

Table 1 Primer sequence.
Full size table

Statistical analysis

All analyses were conducted using R version 4.2.1. Significance analysis employed the wilcoxon rank sum test to compare the differences between the two groups of samples.The statistical significance was demonstrated at a level of p < 0.05.

Result

Screening of DEGs in patients with IS and normal samples

Differential expression analysis was used to screen out DEGs between IS patients and normal samples. Figure 1A (volcano plot) showed 344 up-regulated genes and 318 down-regulated genes. Heatmaps of 662 DEGs was shown in Fig. 1B.

Fig. 1
figure 1

Identification of DEGs. A Volcano plot of DEGs; B Heatmap of DEGs. DEGs, differentially expressed genes.

Full size image

Construction of IS’s WGCNA co-expression network

The WGCNA algorithm was employed to construct the IS co-expression module, with a soft threshold of 8 (Fig. 2A). In this study, 33 co-expression modules were constructed, and the gene clustering tree (Fig. 2B) effectively clustered genes exhibiting same expression patterns. We found that the black and grey60 modules exhibited a positive correlation with IS (Fig. 2C), thus establishing them as pivotal modules of IS. The black module comprised 507 genes, while the grey60 module contained 164 genes (Fig. 2D & E). Finally, we intersected the genes in NFRGs, DEGs, and modules to obtain 25 important NFRGs (Fig. 2F).

Fig. 2
figure 2

WGCNA network construction. A The scale-free fit index of the soft threshold and the average connection of the soft threshold power; B Gene clustering tree; C Correlation between different modules and IS; D Scatter plot of feature genes in grey60 module; E Scatter plot of feature genes in black module; F Identification of IS-NFRGs. WGCNA, weighted gene co-expression network analysis; NFRGs, neurotrophic factors related genes.

Full size image

Functional enrichment and PPI analysis of 25 key NFRGs

GO analysis result found important NFRGs were mainly distributed in response to peptide hormone, cellular response to light stimulus, response to oxygen levels, aging, cellular response to environmental stimulus, response to insulin, hormone catabolic process, glucose import across plasma membrane, etc., as shown in Fig. 3A.While KEGG analysis revealed key genes were significantly enriched in estrogen signaling pathway, microRNAs in cancer, kaposi sarcoma-associated herpesvirus infection, serotonergic synapse, growth hormone synthesis, secretion and action, leishmaniasis, IL-17 signaling pathway and prostate cancer etc., as shown in Fig. 3B. PPI analysis showed that STAT3, FOS, DUSP1, PTGS2, MMP9 and TLR4 had high connectivity in the network, as shown in Fig. 3C.

Fig. 3
figure 3

Enrichment analysis of key genes and construction of PPI network. A Circle diagram of GO enrichment analysis for key genes; B Circle diagram of KEGG enrichment for key genes; C PPI network for key genes. GO, gene ontology; KEGG, kyoto encyclopedia of genes and genomes; PPI, protein-protein interaction.

Full size image

Machine learning screening for biomarkers of IS

We used the LASSO regression algorithm to identify seven genes that were identified as potential diagnostic biomarkers for IS (Fig. 4A&B). The SVM-RFE algorithm identified 10 genes as candidate biomarkers for IS (Fig. 4C). Besides, a total of diagnostic genes were identified by RF algorithm (Fig. 4D&E). The LASSO and AVM feature gene importance ranking was demonstrated in Supplementary Fig. 1. Five genes (MMP9, MARCKS, IGF2R, HECW2 and CYBRD1) were screened by Venn diagrams (Fig. 4F) and used as stable diagnostic biomarkers. We next verified the expression of key genes in the training set and validation set, and we found that MMP9, MARCKS, IGF2R, HECW2, and CYBRD1 were significantly different in the normal and IS groups (Fig. 5A&B). The result depicted in Fig. 5C demonstrate that MMP9 (AUC = 0.853), MARCKS (AUC = 0.913), IGF2R (AUC = 0.869), HECW2 (AUC = 0.872) and CYBRD1 (AUC = 0.840) exhibit AUC values exceeding 0.8. It further demonstrated the high diagnostic efficiency of these genes. The result of the verification set were consistent with those of the training set (Fig. 5D).

Fig. 4
figure 4

Identification of biomarkers. A, B Regression coefficient path graph and cross verification curve of LASSO logistic regression algorithm; C Variation curve of gene prediction value and error value in SVM-RFE algorithm; D, E RF feature importance identification; F The biomarkers’s intersection of three algorithms. LASSO, least absolute shrinkage and selection operator; SVM-RFE, support vector machine-recursive feature elimination; RF, random forest.

Full size image
Fig. 5
figure 5

Biomarker expression validation and ROC analysis. A Expression of biomarkers in training set; B Expression of biomarkers in validation set; C ROC curve of the training set; D ROC curve of the validation set. ROC, receiver operating characteristic.

Full size image

GSEA enrichment analysis

As shown in Fig. 6, the GSEA analysis revealed that pathways such as NOD_LIKE_RECEPTOR_SIGNALING_PATHWAY, GALACTOSE_METABOLISM, and CHRONIC_MYELOID_LEUKEMIA were significantly enriched among the IS samples exhibiting high expression of CYRBD1. For HECW2, NICOTINATE_AND_NICOTINAMIDE_METABOLISM, P53_SIGNALING_PATHWAY, FC_GAMMA_R_MEDIATED_PHAGOCYTOSIS and other pathways were enriched. In samples with high expression of IGF2R, LEUKOCYTE_TRANSENDOTHELIAL_MIGRATION, and CHEMOKINE_SIGNALING_PATHWAY were activated. In addition, we found that GALACTOSE_METABOLISM and TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY were related to MARCKS. LEISHMANIA_INFECTION, ERBB_SIGNALING_PATHWAY and GALACTOSE_METABOLISM were associated with MMP9.

Fig. 6
figure 6

GSEA enrichment analysis for MMP9, MARCKS, IGF2R, HECW2, and CYBRD1.

Full size image

Immune cell infiltration analysis of IS-related biomarker

We obtained scores of 28 kinds of immune cells in IS patients and normal patients. Firstly, the immune infiltration of different samples was shown in histogram (Fig. 7A). Next, the boxplot showed the significantly different immune cells in the IS and normal groups, including neutrophil, activated B cell, activated CD8 T cell, effector memory CD8 T cell, natural killer cell, etc., as shown in Fig. 7B. Analyzing the correlation between biomarkers and immune cells with significant differences, we observed a significant positive correlation between MMP9, MARCKS, IGF2R, HECW2, CYBRD1 and macrophage, activated dendritic cell, and neutrophil cell populations (Fig. 7C). Conversely, these biomarkers displayed a significant negative correlation with activated CD8 T cell, activated B cell, effector memory CD8 T cell populations as well as other immune cells (Fig. 7C).

Fig. 7
figure 7

Immunoinfiltration analysis. A Percentage of immune cell infiltration in normal and IS samples; B Differences in immune cell infiltration between normal and IS groups; C Correlation analysis of biomarkers and immune cells.

Full size image

ceRNA network and potential drug analysis

By analyzing the ceRNA network constructed in Fig. 8A, it is found that MARCKS were mainly regulated by miRNAs such as hsa-miR-138, hsa-miR-25 and hsa-miR-200b. The TFs regulating IGF2R include USF2, E2F4, CREG1, E2F3 and USF1, while miRNAs include hsa-miR-124, hsa-miR-16 and so on. CYBRD1 was regulated by more miRNAs, such as hsa-miR-372, hsa-miR-520a-3p and hsa-miR-577. Additionally, numerous TFs have been observed to regulate MMP9 expression, including FOSL1, JUND, RELA, FOS and ETS1. For HECW2, we can observe that the miRNAs regulating its expression include hsa-miR-892a, hsa-miR-454 and hsa-miR-301b, etc. In addition, we obtained 29 drug-gene interaction pairs in the DSigDB database, including genes (MMP9, IGF2R and MARCKS) and 29 drugs. The majority of drugs exhibited interactions with MMP9, as shown in Fig. 8B.

Fig. 8
figure 8

Multifactor regulatory network and drug prediction analysis. A Multifactor regulatory network and drug prediction analysis; B Drug-gene interaction.

Full size image

RT-qPCR verification results

As shown in Fig. 9, compared with the normal group, the expression levels of CYBRD1, MARCKS and MMP9 in the IS group were significantly increased. We found that there was no statistical difference in the expression levels of HECW2 and IGF2R between the control group and the IS group, but HECW2 and IGF2R showed a significant up-regulation trend in the IS group.

Fig. 9
figure 9

RT-qPCR analysis results. ns, not significant; *p < 0.05; **p < 0.01.

Full size image

Discussion

IS is a prevalent neurological disorder associated with elevated rates of disability and mortality24. Due to the limited therapeutic options and high risk involved, there is an urgent need to identify novel biomarkers and therapeutic targets. According to the study, NTFs, particularly BDNF, play a crucial role in the development, maintenance, and recovery of the central nervous system25. Tuwar MN’s group also believes that BDNF has the potential to be a potential therapeutic target for IS26. Therefore, we conducted bioinformatics analysis to explore the molecular mechanisms and potential biomarkers associated with IS in order to enhance diagnostic accuracy and improve therapeutic strategies. Through comprehensive analysis, we identified biomarkers closely implicated in IS pathogenesis, namely MMP9, MARCKS, IGF2R, HECW2, and CYBRD1. These genes are primarily enriched in pathways related to neuroprotection and inflammatory response, offering novel molecular targets for early diagnosis and treatment of IS.

The key biomarkers selected in this study, namely MMP9, MARCKS, IGF2R, HECW2, and CYBRD1, have been previously reported in early literature to be closely associated with the occurrence and development of IS. Matrix metalloproteinase 9 (MMP9) plays a crucial role in brain tissue remodeling following stroke27. Overexpression of MMP9 is linked to damage to the blood-brain barrier, brain edema, and increased inflammation. These processes may indirectly contribute to an elevated risk of IS by exacerbating brain injury27,28. This finding aligns with our results indicating high expression levels of MMP9 in IS samples as confirmed by RT-qPCR validation. Unfortunately, the clinical approval for the treatment of IS has not been granted to MMP9 inhibitors, possibly due to their lower specificity and stronger side effects28. Therefore, further research is required for the clinical application of MMP9. Myelin acid regulated protein kinase C substrates (MARCKS) exhibit highest expression in the brain during embryogenesis and are widely expressed throughout adulthood29, playing diverse roles in embryonic development, adult brain plasticity, and inflammatory response30. As cytoskeletal proteins, MARCKS play a crucial role in the transmission of nerve cell signals, and their abnormal expression is associated with nerve repair and regeneration following stroke29. Insulin-like growth factor 2 receptor (IGF2R) is a transmembrane glycoprotein of type 1 that plays a pivotal role in binding to and regulating circulating and tissue levels of the mitotropic peptide IGF231. IGF2 contributes to neuronal plasticity, memory formation and enhancement, aging, as well as neurological diseases32. Studies have demonstrated that serum levels of IGF2 are higher in cases of ischemic stroke compared to controls, while low levels of IGF2 during the acute phase serve as a risk marker for increased mortality33. HECT, C2 and WW ___domain-containing E3 ubiquitin protein ligase 2 (HECW2) is primarily expressed in brain, lung, and heart tissues34. While the role of HECW2 in IS has not been extensively investigated, multiple studies have demonstrated its significance in neurodevelopmental disorders and intellectual disabilities34,35. Through the qRT-PCR validation experiment we conducted, no expression differences were detected between the IS group and the control group for the two biomarkers, HECW2 and IGF2R. We speculate that this lack of difference may be due to limitations related to sample specificity and sample size. Cytochrome B Reductase 1 (CYBRD1) plays a crucial role in iron metabolism regulation36. In recent years, there have been notable advancements in understanding the involvement of iron in IS pathogenesis. Specifically, iron-dependent lipid peroxidation occurring during IS may contribute to ferroptosis37.The biomarkers we investigated may provide more specific and sensitive indications than traditional stroke diagnostic markers such as C-reactive protein (CRP) or brain natriuretic peptide (BNP). MMP9, as a matrix metalloproteinase, plays a key role in vascular remodelling and neurological injury; MMP9, a matrix metalloproteinase, plays a key role in vascular remodelling and nerve injury; MARCKS is involved in cell signalling and cell morphology; and IGF2R, HECW2 and CYBRD1 are involved in a number of important processes such as cell proliferation, apoptosis, and iron metabolism, respectively. We believe that these newly discovered biomarkers have significant potential for early diagnosis in the clinical setting. MMP9 and MARCKS levels can be assessed by simple blood tests (such as ELISA etc.), and their use in combination with medical imaging techniques (e.g. MRI or CT) can greatly improve the efficiency of stroke risk assessment and early intervention. Considering the potential importance of IGF2R, HECW2 and CYBRD1 in other related diseases (e.g., cardiovascular diseases and cancer), this further emphasises the need for in-depth research and development of their applications as multifunctional diagnostic tools. In conclusion, in our future work, we plan to work closely with clinicians to validate the effectiveness of these biomarkers in real clinical settings and to optimise the assays. Through continuous improvement and validation, we look forward to bringing these revolutionary biomarkers into everyday medical practice to improve outcomes for ischaemic stroke patients. They all have high diagnostic value (AUC > 0.8). However, it is important to note that demographic characteristics such as age and gender may have a significant impact on biomarker expression levels. In addition, comorbidities such as diabetes mellitus are also associated with chronic inflammation and may affect biomarker expression. Therefore, patient-specific conditions should be considered when applying this biomarker. Future studies should aim to validate these findings and explore other potential confounders to improve the applicability of the biomarker in different populations.

To further explore the potential mechanisms of these biomarkers, we conducted a GSEA. The analysis revealed significant enrichment of CYRBD1 in multiple signaling pathways, including the NOD-like receptor signaling pathway, galactose metabolism pathway, and chronic myeloid leukemia pathway. The NOD-like receptor signaling pathway plays a crucial role in innate immune responses38. A study on obese and non-obese IS indicated that obesity-induced strokes lead to more changes in the NOD-like receptor signaling pathway and monocyte infiltration39. Therefore, CYRBD1 may be involved in the pathogenesis of IS by modulating this pathway. For HECW2, we found significant enrichment in the nicotinate and nicotinamide metabolism pathways, P53 signaling pathway, and FcγR-mediated phagocytosis pathways. Previous studies have shown that activation of the P53 signaling pathway can induce cell cycle arrest and apoptosis40. Depending on the context and timing of activation, the P53 pathway may either protect neurons or exacerbate neuronal damage41. High expression of HECW2 with enriched P53 pathways suggests that HECW2 could potentially influence neuronal fate during IS through interactions with the P53 pathway. Additionally, the Toll-like receptor signaling pathway occupies a central position in immune responses. The association with MARCKS suggests that MARCKS might be involved in regulating Toll-like receptor-mediated immune reactions. Research indicates that activation and dysregulation of Toll-like receptors during IS can lead to excessive inflammatory responses42. Thus, interactions between the Toll-like receptor signaling pathway and MARCKS may play a crucial role in immune-inflammatory responses during IS. Overall, this GSEA analysis has unveiled complex relationships between these genes and multiple biological pathways. It not only provides valuable insights into the pathogenic mechanisms of IS but also offers potential directions for identifying diagnostic markers and developing targeted therapeutic strategies.

Due to the crucial role of immune dysfunction in the pathogenesis and progression of stroke, we also investigated the association between biomarkers and immune cells. The findings revealed a close correlation between the biomarkers and the extent of infiltration by macrophages, activated dendritic cells, neutrophils, as well as various B and T cells. Turner RJ et al. discovered that neutrophils not only serve as a source of MMP9 following IS, leading to injury and disruption of the blood-brain barrier, but they are also believed to be responsible for MMP9 production after tPA treatment27. Furthermore, both macrophages and monocytes exert an influence on the central nervous system during stroke43. Therefore, we posit that these immune cells play pivotal roles in IS pathology by participating in inflammatory responses and neuroprotection. Our findings offer novel insights into comprehending immune mechanisms underlying IS. In IS, neutrophils are rapidly recruited to the site of injury and release inflammatory mediators that exacerbate neuroinflammation; macrophages are both pro- and anti-inflammatory and participate in tissue repair and phagocytic clearance; and dendritic cells regulate the immune response. Understanding their roles in neuroinflammation and recovery mechanisms is essential for in-depth investigation of stroke pathology and therapeutic strategies. In the acute phase, these immune cells participate in the initial inflammatory response, and their effects depend on their activation state, which may either promote or alleviate tissue damage. In the long term, their persistent presence and activity can impact the repair and regeneration processes of neurons, ultimately affecting the overall recovery and quality of life of patients. The results of our current study further emphasize the importance of these immune cells in the pathophysiological process of ischemic stroke. The close association between biomarkers and the infiltration of these immune cells suggests that they could serve as key indicators for disease progression and prognosis. For example, changes in biomarker levels corresponding to alterations in immune cell infiltration could potentially predict the severity during the acute phase, such as the extent of brain tissue damage or the likelihood of complications. In the long term, these relationships might also provide valuable clues about patients’ potential for recovery and risk of recurrence.

PSD is the most prevalent psychiatric disorder that occurs following a stroke, resulting in diminished patient quality of life and recovery capacity10. Recent research suggests that BDNF may represent a promising therapeutic target for PSD; however, the underlying mechanism remains unclear10,44. Furthermore, our functional enrichment analysis of key important NFRGs related to IS revealed significant enrichments in processes such as insulin response, hormone catabolism, glucose transport across the plasma membrane, growth hormone synthesis and secretion, as well as estrogen signaling pathway. Previous studies have demonstrated that insulin modulates emotional behavior through a serotonin-dependent mechanism45. Koenigsberg HW et al. demonstrated a significant association between emotional expression and glycemic control in patients with insulin-dependent diabetes mellitus46. Karachaliou FH et al. revealed that anxiety and depression are prevalent emotional issues among children and adolescents, particularly in those with growth hormone deficiency (GHD)47. Yen JY et al. found a negative correlation between premenstrual estrogen levels and anxiety/stress in patients with PMDD48. Schechter D et al. conducted a review on the role of estrogen and progesterone in regulating female mood49. Therefore, our findings provide additional evidence supporting the relationship between NFRGs and the development process of PSD.

Regarding miRNA regulation, miRNAs like hsa-miR-138 and hsa-miR-25 play crucial roles. For instance, hsa-miR-138 has been shown to potentially target genes involved in NTFs, thereby influencing their expression levels. This, in turn, could impact the normal functioning of neurons and contribute to IS pathogenesis. Similarly, hsa-miR-25 may interact with key molecules in the NTF pathway, altering the cellular environment and potentially exacerbating the development of IS. Understanding these precise regulatory mechanisms is essential for devising targeted therapeutic interventions. In terms of identified drugs, those interacting with MMP9 hold significant promise. MMP9 is a key player in IS, and drugs targeting it could potentially disrupt the pathological processes. However, integrating these drugs into future therapeutic strategies demands careful consideration. Known side effects, such as potential impacts on normal tissue remodeling or interference with other physiological processes, need to be thoroughly evaluated.

Since NFRGs, IS and PSD are closely interconnected, we also investigated the correlation between biomarkers and PSD. Through a comprehensive literature review, we discovered that levels of MMP9 and VEGF protein exhibit fluctuations during stroke patient rehabilitation, which can serve as predictive indicators for cognitive function improvement and alleviation of depressive symptoms during the recovery process50. Although direct studies on MARCKS and PSD are limited, considering the pivotal role of neuroplasticity in depression development, it is plausible to suggest that MARCKS may indirectly contribute to the onset of PSD by influencing nerve repair and functional recovery. Furthermore, aberrant insulin signaling has been implicated in depression pathogenesis, while dysfunction of IGF2R may impact mood regulation. Despite scarce direct evidence available currently, given the involvement of E3 ubiquitin protein ligase in neurodegenerative disorders, HECW2 might potentially participate indirectly in post-stroke depression development through modulation of neuronal death and inflammatory response mechanisms. In summary, these identified biomarkers not only pertain to IS progression but also likely play a regulatory role in PSD. Despite the promising results, this study has limitations. On one hand, the samples of the selected dataset may have limitations of heterogeneity and potential batch effects, so for the two biomarkers in RT-qPCR were not significant. Therefore, caution is needed in extrapolating the results, and larger and more diverse cohorts should be used for validation in future studies. On the other hand, this study focuses solely on IS and does not incorporate a comprehensive analysis of other stroke subtypes. Given the substantial differences among stroke subtypes, focusing on just one subtype may limit the overall understanding of the disease and potentially overlook key information. Future research should therefore aim to broaden the scope of study.

Conclusion and future perspective

In summary, our study employs bioinformatics methods to identify biomarkers and signaling pathways closely associated with the pathogenesis of IS, which have the potential to serve as novel diagnostic biomarkers. While providing new insights into the molecular mechanisms underlying IS, further analysis of the upstream and downstream regulatory mechanisms of these biomarkers is necessary, along with validation on a larger sample set. Our future focus will be on conducting prognostic studies for IS in order to provide more effective strategies for early diagnosis and treatment.