Identification of UBE2N as a biomarker of Alzheimer's disease by combining WGCNA with machine learning algorithms

Identification of UBE2N as a biomarker of Alzheimer's disease by combining WGCNA with machine learning algorithms | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Identification of UBE2N as a biomarker of Alzheimer's disease by combining WGCNA with machine learning algorithms Gangyi Feng, Manli Zhong, Hudie Huang, Pu Zhao, Xiaoyu Zhang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3904783/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 22 Feb, 2025 Read the published version in Scientific Reports → Version 1 posted 15 You are reading this latest preprint version Abstract Alzheimer’s disease (AD) is the most common neurodegenerative disorder leading to progressive cognitive decline. With the development of machine learning analysis, screening biomarkers based on existing clinical data is becoming conducive to understanding the pathogenesis of AD and discovering new treatment targets. Our study integrated three AD datasets in the GEO database for differential expression analysis. After constructing a WGCNA network, 109 key genes were obtained and 48 core genes were analyzed from 109 genes using a protein-protein interaction network. The least absolute shrinkage and selection operator, support vector machine recursive feature elimination, and Random Forest methods were applied to obtain the features associated with the 48 core genes and 13 potentially related AD biomarkers were selected. By intersecting InnateDB database with them, we found a potential immune-related marker, UBE2N. MFUZZ cluster analysis revealed that UBE2N is closely related to T cell and B cell functions and the synaptic vesicle cycle signaling pathways. In addition, the expression levels of UBE2N were decreased in the temporal cortex and hippocampus of Tau P301S mice but not APP/PS1 mice. Our findings are the first comprehensive identification of UBE2N as a biomarker for AD, paving the way for much-needed early diagnosis and targeted treatment. Biological sciences/Computational biology and bioinformatics Biological sciences/Neuroscience Health sciences/Diseases/Neurological disorders Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction Alzheimer's Disease (AD) is the most common cause of dementia, responsible for 60–70% of cases 1 . The neuropathologic examination is considered to provide the gold standard for AD diagnosis based on its hallmarks: extracellular depositions of senile plaques generated by amyloid β (Aβ) and neurofibrillary tangles formed by hyperphosphorylated tau in different brain regions 2 . Many risk factors have been identified for AD, including age, genetic risk variants, stress, immune system dysfunction, and infectious diseases 3 . Based on the current knowledge of the pathogenesis of AD, academic and pharmaceutical industries are conducting R&D to achieve a breakthrough in the generation of effective AD drugs 4 . Aβ has been investigated as the primary therapeutic target for many years. Aducanumab and lecanemab are anti-amyloid antibodies approved by the US Food and Drug Administration (FDA) for AD treatment 5 , 6 . At the same time, increasing studies focus on therapies that target the tau protein 7 . AD biomarkers are confirmatory in clinical decision-making, which is particularly important with advancing anti-Aβ/tau protein disease-modifying therapies. Besides, biomarkers can be targeted to improve the diagnosis and prognosis 8 . Recent studies have shown that innate immune genes and immune cells directly or indirectly affect AD. Even without crossing the blood-brain barrier (BBB), T-cells regulate brain homeostasis through a cascade of immune signals and secretory molecules. A recent study showed that CD8 + T cells are abnormally expanded in the brains of patients with mild cognitive impairment (MCI) and AD, indicating that CD8 + T cells may affect neurodegeneration and cognitive impairment in AD 9 . CD4 + T cells infiltrate the brain to promote Aβ clearance and neuronal repair 10 . Removing B cells can significantly reduce Aβ and reverse memory deficits in a 3xTg AD mouse model 11 . Blocking the transforming growth factor in the peripheral macrophage β (TGF-β)-mediated signaling pathway can reduce Aβ in the brain of the Tg2576 mouse model, which may be a potential treatment for AD 12 . However, the roles of innate and adaptive immune cells have not been fully clarified, and there is an urgent need to identify new immune-related biomarkers to explain the neuroinflammation and pathogenesis of AD further. The analysis and prediction of biomarkers of AD would increase our understanding of the pathology and enhance the development of new drug targets, clinical trials, and overall diagnosis 13 . Recently, weighted gene co-expression network analysis (WGCNA) has been used to screen potential biomarkers of various diseases 14 , 15 . The advantage of WGCNA is that it can analyze the correlation between genes rather than being limited to a single gene 16 . Machine learning is a branch of computer science and statistics that focuses on detecting, diagnosing, and treating diseases 17 . The combination of WGCNA and machine learning can improve the accuracy of identifying potential biomarkers of the disease 18 . Using this approach, SNRPG was identified as a critical gene in AD and metabolic syndrome 19 . Furthermore, PLXNB1, GRAMD3, and GJA have been screened from different cortex and cerebellum regions related to the Braak NFT stage in AD 20 . Similar genomic transcription patterns in different cortex regions, according to the Braak 0-VI phases, may participate in the pathological progression of AD through the oxidation pathway 21 . Therefore, specific biomarkers have been screened for different research purposes using a combination of these two methodologies. In this study, we first identified 239 differentially expressed genes (DEGs) in the middle temporal gyrus (MTG) of patients with AD based on the Gene Expression Omnibus (GEO) database. Then, 109 key DEGs were identified using WGCNA. A total of 48 core genes were analyzed from the 109 genes through a protein-protein interaction (PPI) network. Finally, 13 potential biomarkers that may be related to AD were identified and verified in the GSE109887 dataset through joint machine-learning analysis. Immune-related genes selected via the InnateDB database intersected with 13 potential AD-related genes, and the potential immune-related marker, UBE2N, was obtained. Transcription factor prediction and GSEA analyses were used to investigate further the biological processes and pathways of UBE2N related to AD occurrence. The results of the MFUZZ cluster analysis showed that UBE2N was involved in T cell and B cell functions and the synaptic vesicle cycle signaling pathways. Additionally, UBE2N expression was verified in Tau P301S transgenic mice as decreased. These results suggest that UBE2N may be a novel biomarker with implications for AD treatment. Results Screening of DEGs in the brain of patients with AD To screen for DEGs in the brain of patients with AD, we first removed the batch effect of genes between the AD and control group and crossed the Principal Component Analysis (PCA) dataset that showed separation from each other, laying the foundation for subsequent analysis (Fig. 1 A and B ). The results included 124 samples from normal individuals and 157 samples of patients with AD. By setting the screening criteria of | LogFC | ≥ 0.5, p < 0.05, 86 upregulated and 153 downregulated genes were identified. The expression of DEGs is shown in heat and volcano maps (Fig. 1 C and D ). GO analysis revealed that DEGs were mainly enriched in vesicle-mediated synaptic transport, synaptic vesicle cycle, and other related pathways ( Fig. S1 A ). KEGG enrichment analysis showed that DEGs were mainly enriched in the Alzheimer's disease, cAMP, and MAPK signaling pathways ( Fig. S1 B ). Screening Characteristic Genes of AD by Construction of WGCNA A total of 2626 genes with expression variance in the top 25% were included in the WGCNA. The analysis of soft threshold selection showed that the average connectivity was high and the scale-free network distribution reached its optimal level when β = 7 (R 2 = 0.85) (Fig. 2 A). Subsequently, we obtained nine independent modules (with the lowest number of genes in the module set to 30 and genes not included in the module shown in gray) (Fig. 2 B) by setting the clustering height to 0.25, merging highly correlated modules, and confirming the independence between each module (Fig. 2 C). Next, the correlation between the module genes and AD was analyzed. The results showed that the highest correlation module was the turquoise module, which positively correlated with the control group (r = 0.43, p = 2e-14) and negatively correlated with the AD group (r = -0.43, p = 2e-14) (Fig. 2 D). The genes inside the turquoise module were screened according to the standard in which GS ≥ 0.35 and ME ≥ 0.8 were set as the values closer to r, and 195 genes were obtained (Table S1 ). In total, 109 characteristic genes related to AD were identified by overlapping 195 genes with DEGs (Fig. 2 E). GO analysis showed that the characteristic genes were mainly enriched in exocytosis, glomerular development, and synaptic vesicle cycle ( Fig. S2 A ). KEGG analysis showed that the characteristic genes were mainly enriched in GABAergic synapses, synaptic vesicle cycles, and cAMP signaling pathways ( Fig. S2 B ). Screening for potential biomarkers of AD by machine learning algorithms PPI networks were constructed based on the 109 potential targets. A total of 48 gene nodes with 80 lines were obtained (Fig. 3 A). First, 19 genes predicting the incidence of AD were obtained from 48 genes analyzed using the least absolute shrinkage and selection operator regression model (Fig. 3 B). Then, 48 genes were analyzed using the support vector machine recursive feature elimination, of which 36 showed a high accuracy rate (0.787) and a low error rate (0.213) (Fig. 3 C). The RF results showed 33 AD-related genes were selected from the 48 (Fig. 3 D and E ). Three machine learning algorithms obtained 13 overlapping genes (Fig. 3 F). ATP6V1E1, CCKBR, DYNC1I1, NRN1, SV2B, SYT1, TUBB2A , and UBE2N were expressed at low levels in the brains of patients with AD, whereas INPPL1, ITPKB, ITSN1, RAPGEF3 , and TBL1X were highly expressed ( Fig. S3 ). The validation set exhibited similar results ( Fig. S4 ). Identifying UBE2N as a biomarker for AD The diagnostic value of the 13 biomarkers analyzed by Receiver Operating Characteristic (ROC) curves was greater than 0.7 for all areas under the curve (AUC) (Fig. 4 C). A total of 1696 immune genes were retrieved from the InnateDB database and overlapped with 13 AD biomarkers, identifying UBE2N as an overlapping gene (Fig. 4 B). Biomarker correlation analysis revealed that UBE2N expression was positively correlated with seven genes ( TUBB2A, SV2B, NRN1, CCKBR, DYNC1I1, ATP6V1E1, and SYT1 ) and negatively correlated with five genes ( INPPL1, ITSN1, ITPKB, RAPGEF3, and TBL1X ) (Fig. 4 A). The UBE2N column line graph was modeled by analyzing the calibration curves, and the differences between the normal and predicted values were small, indicating that the model was accurate (Fig. 4 D and E ). In the decision curve analysis (DCA), the model curve was above the gray line, implying that patients could benefit from the model within the threshold (Fig. 4 F). The clinical impact curve (CIC) also demonstrated a better overall net benefit in the threshold range (Fig. 4 G). This suggests that the UBE2N columnar line graph model constructed in this study can be used to assess AD prognosis. Analysis of potential regulatory mechanisms of UBE2N Based on the pooled median expression values, data from patients with AD were divided into two groups of high and low expression of UBE2N, and KEGG analysis was performed on the differential genes. The results showed that UBE2N might participate in the activation of five pathways, including nicotine addiction, calcium reabsorption regulated by endocrine and other factors, synaptic vesicle circulation, oxidative phosphorylation, and alanine, aspartate, and glutamate metabolism; UBE2N may be involved in the inhibition of five signaling pathways, including the interaction of the viral protein with cytokines and cytokine receptors, graft rejection, and malaria and Staphylococcus aureus infections (Fig. 5 A). Meanwhile, GSEA analysis showed that the remaining six genes that were positively correlated with UBE2N were mainly enriched in the synaptic vesicle circulation pathway ( Fig. S5 ). These results indicated that UBE2N may be related to vesicular function. In addition, the potential regulatory network analysis of UBE2N showed that hsa-miR-128-3p, hsa-miR-149-5p, hsa-miR-221-3p, hsa-miR-222-3p, hsa-miR-5010-5p, hsa-miR-522-3p, and hsa-miR-96-5p regulated UBE2N expression and the transcription factor ATF1 was also involved in the transcription regulation process of UBE2N (Fig. 5 B). Cluster Analysis of MFUZZ Expression Patterns According to the expression patterns of UBE2N, 50 different clustering results were obtained. The obtained clustering results for ssGSEA scoring were analyzed with the control and AD groups for correlation analysis, and cluster 27 showed the closest module related to UBE2N (Fig. 6 A-C). Functional enrichment analysis showed that the cluster 27 module genes were mainly enriched in the T cell receptor, B cell receptor and synaptic vesicle signaling pathways (Fig. 6 D). Subsequently, the genes in the cluster 27 module overlapped with 13 AD biomarkers, and seven genes ( ATP6V1E1, CCKBR, DYNC1I1, NRN1, SV2B, SYT1, and TUBB2A ) were highly correlated with UBE2N expression (Fig. 6 E). Immune cell infiltration analysis To evaluate the infiltration status of immune cells in control and AD groups, we first compared the expression levels of 28 types of immune cells. Compared to the control group, the infiltration rate of T cells in the AD group was higher, including natural killer, gamma delta, central memory CD8 + , factor memory CD8 + , and central memory CD4 + T cells. However, the infiltration rate of factor memory CD4 + T cells was lower in the AD group than in the control group (Fig. 7 A). Subsequently, correlations between the eight genes and immune cells were analyzed. UBE2N and DYNC1I1 positively correlated with Activated CD4 + T cells, SYT1 negatively correlated with Activated CD8 + T cells, and CCKBR negatively correlated with immature dendritic cells (Fig. 7 B). In addition, UBE2N showed the highest correlation with effector memory CD4 + T cells (r = 0.7) (Fig. 7 C and D ). Expression levels of UBE2N were decreased in the brain of an AD mouse model RT-qPCR results revealed that the gene expression levels of UBE2N, ATP6V1E1, CCKBR, SV2B and TUBB2A in the cerebral cortex of Tau P301S mice were significantly reduced, while the expression levels of DYNC1I1 , NRN1 , and SYT1 did not change compared to the control group (Fig. 8 A). The protein levels of UBE2N decreased significantly in the cerebral cortex and hippocampus of Tau P301S mice (Fig. 8 B and C ). However, there was no significant change in UBE2N expression in the cerebral cortex of APP/PS1 mice ( Fig. S6 ). We also observed that UBE2N was co-localized with NeuN in the hippocampus and temporal cortex. Compared to C57BL/6 mice, Tau P301S mice showed decreased UBE2N fluorescence intensity in the hippocampus and cortex (Fig. 8 D and E ). Discussion AD is a neurodegenerative disease that impairs cognitive function and mainly involves changes in brain regions related to learning and memory, such as the temporal lobe and hippocampus 22 . In the current study, we performed a comprehensive and in-depth analysis of temporal lobe gene expression profile data to explore AD-specific related genes, and 13 HUB genes were identified. Among them, UBE2N has been validated in the cerebral cortex and hippocampus of Tau P301S mice and is the most valuable potential diagnostic marker of AD in our study. By analyzing microarray data from the temporal lobes of AD patients in the GEO database, 239 DEGs were identified, including 86 up-regulated and 153 down-regulated genes. Overlapping the key modular genes obtained by WGCNA analysis with DEGs yielded 109 key DEGs, which were mainly enriched in GABAergic synaptic, B-cell receptor, and synaptic vesicle cycle signaling pathways, all of which are key pathological changes in the pathogenesis of AD 23 . Subsequently, a machine learning algorithm obtained 13 HUB genes. Overlapping with the immune genes in the InnateDB database yields the UBE2N gene. Furthermore, gene correlation analysis revealed that ATP6V1E1, CCKBR, SV2B, DYNC1I1, NRN1, SYT1, and TUBB2A are positively correlated with UBE2N. The AUC areas under the ROC curves are all greater than 0.7, indicating that the constructed model could accurately predict the onset of AD 24 . UBE2N plays an important role in many neurodegenerative diseases. Overexpression of UBE2N increases the aggregation of mutant Huntington's proteins 25 , whereas knockdown of the E2 enzymes UBE2N, UBE2L3, UBE2D2 and UBE2D3 (UBE2D2/3) significantly reduces autophagic clearance of depolarized mitochondria, and furthermore, UBE2N, UBE2L3, and UBE2D2/3 synergistically promote Parkin-mediated mitochondrial autophagy 26 . Interestingly, the abnormal downregulation of UBE2N causes in vivo immunosuppressive dysfunction of regulatory T cells, leading to abnormal activation of T cells and inducing various inflammatory responses 27 . Therefore, we analyzed the infiltration of 28 immune cells and found that the infiltration rate of T cells was significantly higher in the AD brains, suggesting that the balance of T cells may be dysregulated in AD. One reason could be that there is a decrease in tight junction molecules in the vascular endothelium during AD progression, leading to an increase in the permeability of the blood-brain barrier. Because of increased chemokines for T cells in the brain of AD patients, these changes together promote T cell infiltration 28 , 29 . The accumulation of activated T cells has been demonstrated to induce neuronal death and exacerbate neuroinflammation 30 . Activated T cells can also promote the release of TNF-α, IL-1, and IL-6 pro-inflammatory factors from peripheral blood mononuclear cells (PBMC), exacerbating the inflammatory response 31 . In addition, a large number of CD8 + T cells are found in the hippocampus of AD patients, and tau-specific CD4 T cells are widely distributed in the peripheral blood of AD and PD patients, suggesting that T cells may be closely associated with AD progression, particularly in terms of tau pathology 32 , 33 . Therefore, we speculate that UBE2N may affect AD pathology by regulating T cells, which needs to be confirmed by further experiments. Furthermore, MFUZZ cluster analysis showed that the cluster consisting of the 27th modular gene had the highest correlation with UBE2N, and the functions of the modular genes were mainly related to immunity and synapses, as expected. Cluster 27 overlapped with 13 HUB genes in 8 genes (UBE2N, ATP6V1E1, CCKBR, DYNC1I1, NRN1, SV2B, SYT1, and TUBB2A), seven of which are positively associated with UBE2N. Our RT-qPCR results showed that the mRNA levels of UBE2N were significantly reduced in the cerebral cortex of Tau P301S mice. We observed that the protein levels of UBE2N were significantly reduced in the cortex and hippocampus of Tau P301S mice, but not in the APP/PS1 mice, indicating that UBE2N may be involved in AD pathogenesis in tau-related pathways. Apart from UBE2N, we also found that the mRNA levels of ATP6V1E1, CCKBR, SV2B and TUBB2A were decreased in the Tau P301S cortex. ATP6V1E1 is a large multi-subunit complex divided into a peripheral structural domain (V1) and a proton transmembrane translocation structural domain (V0) that is up-regulated in early AD and down-regulated in late-stage AD 34 . ATP6V1E1 acts as a proton pump and mediates the acidification of endosomes, lysosomes, the Golgi and synaptic vesicles 35 . Its dysfunction, therefore, disrupts PH homeostasis, affecting organelle acidification and, consequently, contributing to AD. ATP6V1E1 is reduced not only in the brain but also in the peripheral blood of AD patients, suggesting that ATP6V1E1 may play an important role in the diagnosis and treatment of AD 36 . Cholecystokinin (CCK) mediates its action through two G-protein-coupled receptors, CCKAR and CCKBR. Its absence leads to abnormalities in the cerebral cortex and corpus callosum development and further affects the migration of cortical interneurons 37 . Synthetic CCK analogs can effectively reduce Aβ load in the brain and normalize the levels of PKA, CREB, BDNF and TrkB receptors, thereby improving APP/PS1 mice cognition 38 . Synaptic vesicle glycoprotein 2B (SV2B) is a synaptic protein that is involved in APP/Aβ metabolism 39 . There is evidence that SV2B knockout protects against Aβ-induced memory deficits and ameliorates cholinergic system dysfunction caused by injection of Aβ 40 . However, another study found that Aβ levels were significantly elevated in the hippocampus of SV2B knockout mice as compared to WT mice 40 . Here, we found downregulation of SV2B in Tau P301S mice, suggesting that SV2B may also be related to tau pathology. TUBB2A (Tubulin Beta 2A Class IIa) is a microtubule protein. Tau from the AD brains increased endogenous Tau in cortical neurons; simultaneously, transcriptome sequencing results showed that TUBB2A is remarkably present in neurons 41 . Here, we first demonstrated that TUBB2A is reduced in the cerebral cortex of Tau P301S mice, indicating that TUBB2A might play a role in the tau-related pathway in AD. Conclusion In this study, we identified UBE2N as a diagnostic marker for AD by combining WGCNA with machine learning approaches. In addition, GSEA analysis indicated that UBE2N may modulate AD onset/development by affecting synaptic vesicles and inflammation, and immune cell correlation analysis revealed that UEB2N may regulate T-cell infiltration in AD pathogenesis. Our analysis provides a new perspective for further exploring the underlying mechanisms by which UEB2N impacts neuroinflammation and synaptic function in the context of AD, and we will subsequently perform relevant experiments. Materials and methods Data acquisition and pre-processing Four microarray datasets (GSE5281, GSE84422 and GSE132903, GSE109887) related to AD were obtained from the GENE EXPRESSION OMNIBUS database ( https://www.ncbi.nlm.nih.gov/geo/ ) using the “SVA” package in R (4.2.1) to remove batch effects among data sets, and the information about the datasets is in Table 1 42 . The gene expression differences (GEDs) were analyzed by the “Limma” software package based on the screening criteria “Adjusted p < 0.05 and | logFC | ≥ 0.5” 43 . The volcanic and thermal maps were created by the "ggplot2" software package and the “pheatmap” software package, respectively. Weighted gene co-expression network analysis (WGCNA) to screen target genes WGCNA was performed to identify co-expression modules using the R package of “WGCNA” (version 1.72.1). The top 25% of genes with the highest variance were applied for subsequent WGCNA analyses to guarantee the accuracy of quality results by checking the missing values and clustering the samples. The “soft” threshold power (β) is calculated to construct a biologically meaningful scale-free topological network. In addition, the topological overlap matrix (TOM) is constructed on the basis of the adjacency matrix, and the dynamic tree-cutting algorithm is used to merge similar modules. Additionally, gene saliency (GS), module affiliation (MM), and correlation coefficients between gene modules and clinical features were calculated to visualize the characteristic gene network. Finally, the potential gene targets for Alzheimer's disease were obtained by the intersection of DEGs and genes within the significant gene module 14 . protein-protein interaction networks Construction String database ( https://string-db.org/ ) was used to construct protein-protein interaction networks for the AD potential target by setting a confidence level (0.7), followed by the Cytascape (3.8.2) software to view this graph. Based on this network, the genes were selected as the biomarker genes in the pathological process of AD patients for the subsequent screening. Screening of AD markers by the Machine learning algorithm We use machine learning algorithms to analyze the central genes in the PPI network to obtain characteristic markers of AD. Firstly, the “glmnet” (4.1.6) R software package was used for Lasso regression analysis to obtain 19 important genes 44 . Next, 34 important genes were obtained by the “e1071”(1.7.13) R software package, which was used for SVM-REF analysis 45 . The “random forest” (4.7–1.1) R software package was performed for random forest analysis, and the genes with scores greater than 2 were retained 46 . The genes obtained by combining three methods were considered biomarkers of AD. Curve Analysis of Receiver Operating Characteristics (ROC) The “Corrplot” (0.92) R software package was to analyze the correlation of AD biomarkers screened through machine learning. Then, we used the “pROC” (1.18.0) R software package to create Receiver Operating Characteristic (ROC) curves and calculated the area under the curve (AUC) to evaluate the clinical diagnostic value of biomarkers 47 . Diagnostic Column Line Graph Construction and Validation The immune gene dataset was obtained from the InnateDB database ( http://www.innatedb.com ), and 1696 immune genes were obtained 48 . The obtained immune genes and potential biomarkers of AD were intersected and analyzed to screen the AD biomarkers related to immunity. Then, we used the “RMS” (6.5.0) R software package to construct a column line graph model to predict the incidence rate of AD 49 . Finally, calibration curves were used to evaluate the accuracy of the column line graph model; decision curve analysis and clinical impact curves were used to evaluate the clinical utility of the model 50 . Enrichment and regulatory mechanism analysis of UBE2N We performed GSEA analysis on the selected immune biomarker-UBE2N and used the Enrichr database ( https://maayanlab.cloud/Enrichr/ ) to analyze the transcription factor (TF) of UBE2N. miRTarBase ( https://mirtarbase.cuhk.edu.cn/ ), Starbase ( https://starbase.sysu.edu.cn/ ), and TargetScan ( www.targetscan.org ) databases were used to predict miRNAs that regulate UBE2N translation. Then the regulatory network diagram of UBE2N was constructed by Cytoscape (3.8.2). Animals Tau P301S transgenic mice [B6C3-Tg (Prnp-MAPT*P301S) PS19 Vle/J] were originally purchased from the Jackson Laboratory (Bar Harbor, ME, United States) and C57BL/6 mice were obtained from Beijing HuaFuKang Bioscience Co., Ltd. (Beijing, China) for animal mating. In offspring, Tau P301S transgenic mice and wild type mice were obtained in the same month through genotype identification. They (n = 8 of each group) were housed under a light/dark cycle of 8:00/20:00 and controlled temperature (24 ± 2°C) and humidity (40–70%) conditions for 9 months. All authors complied with the ARRIVE guidelines. All treatments and experimental procedures were performed under the National Institutes of Health guidelines and approved by the Northeastern University Laboratory Animal Ethical Committee (EC-2023A012). Immunohistochemistry The mice in the two experimental groups (n = 8 in each group) were anesthetized and half brains were removed and fixed using 4% paraformaldehyde and then embedded in paraffin. Serial 5 µm coronal sections were incubated with blocking solution (5% bovine serum albumin [BSA] and 1% normal goat serum) for 1 h and then incubated with rabbit anti-UBE2N (1:200, Abcam) overnight at 4°C. Next day, the sections were incubated with biotinylated goat anti-rabbit IgG (1:500) for 1 h at room temperature (RT) and then with the avidin–biotin–peroxidase (ABC) complex (1:100) for 30 min at RT. After washing with PBS, the sections were immersed in 3,3’-diaminobenzidine for development. One section was incubated within normal rabbit serum (1:100) for nonspecific staining and served as a negative control. Images of immunohistochemical staining were captured using a light microscope (DM4000B; Leica, Wetzlar, Germany). Immunofluorescence Staining and Confocal Laser Scanning Microscopy The 5 µm coronal sections were preincubated with blocking buffer for 1 h and then with rabbit anti-UBE2N (1:200, Abcam) and mouse monoclonal anti-NeuN antibodies (1:200, Thermo) incubated overnight at 4°C. Alexa Fluor® 488-and Alex Fluor® 594-conjugated secondary antibodies were mixed together and treated to sections for 2 h and finally labeled using DAPI (1:500). After mounted with an antifade mounting medium, the mages were taken using the laser scanning confocal microscope (Leica, TCS, SP8, Wetzlar, Germany). Western blot The temporal cortex and hippocampus of half-brain of C57BL/6 mice and Tau p301s mice were lysed in RIPA buffer and centrifuged to extract protein supernatant. 10 µg proteins were separated by 4–12% SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Burlington, MA, USA). the membranes were incubated in 5% BSA solution at room temperature for 1h. Subsequently, the membranes were incubated overnight at 4°C in rabbit anti-UBE2N antibody (1:2000, Abcam, Cambridge, UK) and mouse anti-GAPDH (1:10000, A1978, Sigma, Burlington, MA, USA). Finally, the membranes were incubated with the horseradish peroxidase (HRP)-conjugated secondary antibodies for 2 hours after washing. Bands were detected using a chemiluminescence imaging analysis system (Tanon, 5500, Shanghai, China) and enhanced chemiluminescence (ECL) Kits (EMD Millipore, Burlington, MA, USA). Each experiment was repeated at least three times. Quantitative Real-Time Polymerase Chain Reaction (RT-qPCR) Total RNA was extracted from the cortex of C57BL/6 mice and Tau p301s mice using Total RNA KIT I (R6834-02, OMGEA, USA), and 500 ng of template RNA was reverse transcribed into cDNA using the GoScript™ Reverse Transcription System (Promega, A5001) according to the manufacturer's instructions. PCR reactions were performed with 20 ng of cDNA template at a volume of 10 µL reaction mixture using the Bio-Rad CFX PCR system. The sequences of the genes encoding GAPDH and selected differential genes were obtained from the GenBank database, and specific primers were designed using Primer Premier 5.0 (Table 2 ). The mRNA expression was calculated according to Eq. 2 −∆∆CT . Table 1 Data source. GEO datasets Platform Sample normal Sample AD Publication years Regions GSE5281 GPL570 12 16 2006 USA GSE84422 GPL96 14 44 2016 USA GSE132903 GSE109887 GPL10558 GPL10904 98 32 97 46 2019 2019 USA Germany Table 2 Primer sequences for RT-qPCR. Gene Forward Reverse UBE2N CCGCACAGTTCTGCTATCAA AGTCCATGCTCTCGCTGTTT ATP6V1E1 CTTGTACCAGCTGCTGGAGCC AGGCCTCCTGGTCAATCTGGA CCKBR GATGGCTGCTACGTGCAACT CGCACCACCCGCTTCTTAG SV2B GCGGCCTGGCTGATAAACT AGAGGAAGGCTCCATATCCCT TUBB2A TGCCCTCACCCAAGGTCTCTG GGCAGGTGGTCACTCCACTCA SYT1 DYNC1I1 NRN1 CTGTCACCACTGTTGCGAC GTCGTCATGGAAGCAAAGCA GCGGTGCAAATAGCTTACCTG GGCAATGGGATTTTATGCAGTTC AAGGAGTAGAGCGGCTTGTT TGATGTTCGTCTTGTCGTCCA 2.13. Statistical Analysis Data are presented as mean ± SEM. Student’s t test were used to analyze differences between groups, as appropriate. The analyses were performed using ImageJ and GraphPad Prism 9.0 software. p < 0.05 or p < 0.01 were considered statistically significant. Declarations Data availability statement The datasets analyzed in this study are available from the corresponding authors upon reasonable request. Additional information The authors declare that they have no competing interests. Author Contribution G.F. performed the experimental procedures and statistical analysis. M.Z. and H.H. assisted with data analysis. P.Z., X.Z., and T.W. generated and validated the mouse model and performed animal experiments. H.G. conceived the experiments and wrote the manuscript. H.X. conceived the experiments, supervised the project, and wrote and revised the manuscript. All authors reviewed and approved the final version of the manuscript. Acknowledgments This study was supported by Shenzhen Natural Science Foundation-The Stable Support Program (20220810144826003), the Research Start-up Fund for Young Investigators in Shenzhen University (QNJS0384), the Construction Project of Liaoning Provincial Key Laboratory, China (2022JH13/10200026), the Special Projects of the Central Government in Guidance of Local Science and Technology Development (2022JH6/100100025) and the National Natural Science Foundation of China (81771174, 81971015). We would like to thank Editage ( www.editage.com ) for English language editing. References Scheltens, P. et al. Alzheimer's disease. The Lancet 397, 1577–1590 (2021). Long, J. M. & Holtzman, D. M. Alzheimer Disease: An Update on Pathobiology and Treatment Strategies. Cell 179, 312–339 (2019). R, A. A. Risk factors for Alzheimer's disease. Folia Neuropathol 57, 87–105, doi: 10.5114/fn.2019.85929 (2019). Cummings, J. et al. Drug development in Alzheimer’s disease: the path to 2025. Alzheimer's Research & Therapy 8, 39 (2016). Mahase, E. Alzheimer's disease: Lecanemab gets full FDA approval and black box safety warning. BMJ 382, 1580, doi: 10.1136/bmj.p1580 (2023). Dickson, S. P., Hennessey, S., Nicodemus Johnson, J., Knowlton, N. & Hendrix, S. B. Avoiding future controversies in the Alzheimer's disease space through understanding the aducanumab data and FDA review. Alzheimers Res Ther 15, 98, doi: 10.1186/s13195-023-01238-1 (2023). Qureshi, I. A. et al. A randomized, single ascending dose study of intravenous BIIB092 in healthy participants. Alzheimers Dement (N Y) 4, 746–755, doi: 10.1016/j.trci.2018.10.007 (2018). Cummings, J. The Role of Biomarkers in Alzheimer’s Disease Drug Development. Reviews on Biomarker Studies in Psychiatric and Neurodegenerative Disorders, 29–61 (2019). Gate, D. et al. Clonally expanded CD8 T cells patrol the cerebrospinal fluid in Alzheimer’s disease. Nature 577, 399–404 (2020). Monsonego, A., Nemirovsky, A. & Harpaz, I. CD4 T cells in immunity and immunotherapy of Alzheimer's disease. Immunology 139, 438–446 (2013). Kim, K. et al. Therapeutic B-cell depletion reverses progression of Alzheimer’s disease. Nature Communications 12, 2185 (2021). Town, T. et al. Blocking TGF-β–Smad2/3 innate immune signaling mitigates Alzheimer-like pathology. Nature medicine 14, 681–687 (2008). Wang, T. et al. Integrated bioinformatic analysis reveals YWHAB as a novel diagnostic biomarker for idiopathic pulmonary arterial hypertension. Journal of Cellular Physiology 234, 6449–6462 (2019). Langfelder, P. & Horvath, S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9, 559 (2008). Langfelder, P. & Horvath, S. Fast R Functions for Robust Correlations and Hierarchical Clustering. Journal of Statistical Software 46, i11 (2012). Ozsolak, F. & Milos, P. M. RNA sequencing: advances, challenges and opportunities. Nature Reviews. Genetics 12, 87–98 (2011). Peiffer-Smadja, N. et al. Machine learning for clinical decision support in infectious diseases: a narrative review of current applications. Clinical Microbiology and Infection: The Official Publication of the European Society of Clinical Microbiology and Infectious Diseases 26, 584–595 (2020). Chen, Y., Liao, R., Yao, Y., Wang, Q. & Fu, L. Machine learning to identify immune-related biomarkers of rheumatoid arthritis based on WGCNA network. Clinical Rheumatology 41, 1057–1068 (2022). Li, J. et al. Identification of diagnostic genes for both Alzheimer's disease and Metabolic syndrome by the machine learning algorithm. Frontiers in Immunology 13, 1037318 (2022). Zhao, X., Yao, H. & Li, X. Unearthing of Key Genes Driving the Pathogenesis of Alzheimer's Disease via Bioinformatics. Frontiers in Genetics 12, 641100 (2021). Chai, K. et al. Application of weighted co-expression network analysis and machine learning to identify the pathological mechanism of Alzheimer's disease. Frontiers in Aging Neuroscience 14, 837770 (2022). DeTure, M. A. & Dickson, D. W. The neuropathological diagnosis of Alzheimer’s disease. Molecular neurodegeneration 14, 1–18 (2019). Liu, C. et al. Identification of Immune Cells and Key Genes associated with Alzheimer's Disease. International Journal of Medical Sciences 19, 112–125 (2022). von Dadelszen, P. et al. Prediction of adverse maternal outcomes in pre-eclampsia: development and validation of the fullPIERS model. Lancet (London, England) 377, 219–227 (2011). Yin, P. et al. Aged monkey brains reveal the role of ubiquitin-conjugating enzyme UBE2N in the synaptosomal accumulation of mutant huntingtin. Hum Mol Genet 24, 1350–1362, doi: 10.1093/hmg/ddu544 (2015). Geisler, S., Vollmer, S., Golombek, S. & Kahle, P. J. The ubiquitin-conjugating enzymes UBE2N, UBE2L3 and UBE2D2/3 are essential for Parkin-dependent mitophagy. J Cell Sci 127, 3280–3293, doi: 10.1242/jcs.146035 (2014). Chang, J.-H. et al. Ubc13 maintains the suppressive function of regulatory T cells and prevents their conversion into effector-like T cells. Nature immunology 13, 481–490 (2012). Carrano, A. et al. Amyloid Beta induces oxidative stress-mediated blood-brain barrier changes in capillary amyloid angiopathy. Antioxidants & Redox Signaling 15, 1167–1178 (2011). Goldeck, D. et al. Enhanced Chemokine Receptor Expression on Leukocytes of Patients with Alzheimer's Disease. PLoS ONE 8, e66664 (2013). Giuliani, F., Goodyer, C. G., Antel, J. P. & Yong, V. W. Vulnerability of human neurons to T cell-mediated cytotoxicity. Journal of Immunology (Baltimore, Md. : 1950) 171, 368–379 (2003). Mietelska-Porowska, A. & Wojda, U. T Lymphocytes and Inflammatory Mediators in the Interplay between Brain and Blood in Alzheimer's Disease: Potential Pools of New Biomarkers. Journal of Immunology Research 2017, 4626540 (2017). Lindestam Arlehamn, C. S. et al. Widespread tau-specific CD4 T cell reactivity in the general population. The Journal of Immunology 203, 84–92 (2019). Merlini, M., Kirabali, T., Kulic, L., Nitsch, R. M. & Ferretti, M. T. Extravascular CD3 + T Cells in Brains of Alzheimer Disease Patients Correlate with Tau but Not with Amyloid Pathology: An Immunohistochemical Study. Neuro-Degenerative Diseases 18, 49–56 (2018). Chang, S.-H. et al. Proteomic profiling of brain cortex tissues in a Tau transgenic mouse model of Alzheimer's disease. Biochemical and Biophysical Research Communications 430, 670–675 (2013). Paunescu, T. G., Jones, A. C., Tyszkowski, R. & Brown, D. V-ATPase expression in the mouse olfactory epithelium. American Journal of Physiology-Cell Physiology 295, C923-C930 (2008). Yu, H. et al. Integrated transcriptomics reveals the brain and blood biomarkers in Alzheimer's disease. CNS Neuroscience & Therapeutics (2023). Nishimura, S. et al. Functional synergy between cholecystokinin receptors CCKAR and CCKBR in mammalian brain development. PloS One 10, e0124295 (2015). Zhang, Z. et al. Cholecystokinin signaling can rescue cognition and synaptic plasticity in the APP/PS1 mouse model of Alzheimer’s disease. Molecular Neurobiology, 1–23 (2023). Miyamoto, M. et al. Synaptic Vesicle Protein 2B Negatively Regulates the Amyloidogenic Processing of AβPP as a Novel Interaction Partner of BACE1. Journal of Alzheimer's disease: JAD 75, 173–185 (2020). Detrait, E., Maurice, T., Hanon, E., Leclercq, K. & Lamberty, Y. Lack of synaptic vesicle protein SV2B protects against amyloid-β25–35-induced oxidative stress, cholinergic deficit and cognitive impairment in mice. Behavioural brain research 271, 277–285 (2014). Ficulle, E. et al. A human tau seeded neuronal cell model recapitulates molecular responses associated with Alzheimer’s disease. Scientific Reports 12, 2673 (2022). Leek, J. T., Johnson, W. E., Parker, H. S., Jaffe, A. E. & Storey, J. D. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics (Oxford, England) 28, 882–883 (2012). Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Research 43, e47 (2015). Wang, S. et al. An Eight-CircRNA Assessment Model for Predicting Biochemical Recurrence in Prostate Cancer. Frontiers in Cell and Developmental Biology 8, 599494 (2020). Sanz, H., Valim, C., Vegas, E., Oller, J. M. & Reverter, F. SVM-RFE: selection and visualization of the most relevant features through non-linear kernels. BMC bioinformatics 19, 432 (2018). Yang, Y., Xu, L., Qiao, Y., Wang, T. & Zheng, Q. Construction of a neural network diagnostic model and investigation of immune infiltration characteristics for Crohn's disease. Frontiers in Genetics 13, 976578 (2022). Cheng, S. et al. Novel blood test for early biomarkers of preeclampsia and Alzheimer’s disease. Scientific Reports 11, 15934 (2021). Breuer, K. et al. InnateDB: systems biology of innate immunity and beyond–recent updates and continuing curation. Nucleic Acids Research 41, D1228-1233 (2013). Zhang, J.-A. et al. Development of an Immune-Related Gene Signature for Prognosis in Melanoma. Frontiers in Oncology 10, 602555 (2020). Liu, C. et al. Identification of a glycolysis- and lactate-related gene signature for predicting prognosis, immune microenvironment, and drug candidates in colon adenocarcinoma. Frontiers in Cell and Developmental Biology 10, 971992 (2022). Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterials.pdf UncroppedblotsPCRrawdataandIFimages.zip Cite Share Download PDF Status: Published Journal Publication published 22 Feb, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 28 Oct, 2024 Reviews received at journal 04 Sep, 2024 Reviewers agreed at journal 14 Aug, 2024 Reviewers agreed at journal 14 Aug, 2024 Reviewers agreed at journal 14 Aug, 2024 Reviewers agreed at journal 07 Jun, 2024 Reviewers agreed at journal 05 Jun, 2024 Reviewers agreed at journal 11 Apr, 2024 Reviews received at journal 31 Mar, 2024 Reviewers agreed at journal 31 Mar, 2024 Reviewers invited by journal 30 Mar, 2024 Editor assigned by journal 19 Mar, 2024 Editor invited by journal 06 Feb, 2024 Submission checks completed at journal 06 Feb, 2024 First submitted to journal 28 Jan, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3904783","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":271429061,"identity":"72f097ba-6f50-419c-b8f2-4cba1edf57f0","order_by":0,"name":"Gangyi Feng","email":"","orcid":"","institution":"Northeastern University","correspondingAuthor":false,"prefix":"","firstName":"Gangyi","middleName":"","lastName":"Feng","suffix":""},{"id":271429062,"identity":"4a2cb33c-e80c-4ed2-8455-eef52b7761f1","order_by":1,"name":"Manli Zhong","email":"","orcid":"","institution":"Northeastern University","correspondingAuthor":false,"prefix":"","firstName":"Manli","middleName":"","lastName":"Zhong","suffix":""},{"id":271429063,"identity":"421812cd-f711-4f4b-aefc-a022e0332279","order_by":2,"name":"Hudie Huang","email":"","orcid":"","institution":"Shenzhen University","correspondingAuthor":false,"prefix":"","firstName":"Hudie","middleName":"","lastName":"Huang","suffix":""},{"id":271429064,"identity":"41686089-6253-48e9-a06e-85b1803c9757","order_by":3,"name":"Pu Zhao","email":"","orcid":"","institution":"Northeastern University","correspondingAuthor":false,"prefix":"","firstName":"Pu","middleName":"","lastName":"Zhao","suffix":""},{"id":271429065,"identity":"30ead4d6-ce1a-43ce-8103-add7a748f09a","order_by":4,"name":"Xiaoyu Zhang","email":"","orcid":"","institution":"Dalian Institute of Chemical Physics","correspondingAuthor":false,"prefix":"","firstName":"Xiaoyu","middleName":"","lastName":"Zhang","suffix":""},{"id":271429066,"identity":"d0d5e212-2383-4950-ad40-1f7ee09da9cc","order_by":5,"name":"Tao Wang","email":"","orcid":"","institution":"Northeastern University","correspondingAuthor":false,"prefix":"","firstName":"Tao","middleName":"","lastName":"Wang","suffix":""},{"id":271429067,"identity":"215e1ff1-faa5-4b49-b5d1-94ed1b139cc1","order_by":6,"name":"Huiling Gao","email":"","orcid":"","institution":"Northeastern University","correspondingAuthor":false,"prefix":"","firstName":"Huiling","middleName":"","lastName":"Gao","suffix":""},{"id":271429068,"identity":"3c680fb0-0221-404e-8ae6-da0581c0d6c1","order_by":7,"name":"He Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYDACdsYGhgQgzQ/hMhOhhRmqRbKBeC1Q2uAAsVr4mZnbJB7UHE7cfP74MwmGCuvEBvazB/BqkWxmbJNIOJaWuO1GQpoEw5n0xAaevAS8WgwOMzYbJLDZALUwHJNgbDuc2CDBY0CEln8SiZv7D7ZJMP4jTkvjg8Q2m8QNDMlsEowNRGgB+gWopS/NeMaNNGaLhGPpxm08Ofi18LO3Pzj449th2f7+4w9vfKixlu1nP4NfCypIAGI2EtSPglEwCkbBKMABAHRhQq4+/kM+AAAAAElFTkSuQmCC","orcid":"","institution":"Shenzhen University","correspondingAuthor":true,"prefix":"","firstName":"He","middleName":"","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2024-01-28 06:04:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3904783/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3904783/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-90578-z","type":"published","date":"2025-02-22T15:57:18+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":50832993,"identity":"0e7ced53-0a0d-493d-85a0-0970322c3f6c","added_by":"auto","created_at":"2024-02-08 04:35:51","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":126606,"visible":true,"origin":"","legend":"\u003cp\u003eData preprocessing for DEG.\u003cstrong\u003e (A) \u003c/strong\u003eRaw PCA showed the analysis of GSE13903, GSE5281 and GSE844223 data sets. \u003cstrong\u003e(B)\u003c/strong\u003e Combat PCA showed the three datasets after being removed from the batch. \u003cstrong\u003e(C) \u003c/strong\u003eThe heat map showed the top 30 up-regulated genes and the top 30 down-regulated genes after logFC sequencing. \u003cstrong\u003e(D)\u003c/strong\u003e The volcano map showed DEGs of | logFC | \u0026gt; 0.5, and the significant DEGs were marked. PCA:\u003cstrong\u003e \u003c/strong\u003ePrincipal component analysis; DEGs: the differential genes.\u003c/p\u003e","description":"","filename":"image1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/591bfbbbb894dddcd5e5c779.jpg"},{"id":50833191,"identity":"438e1886-0a1c-48a0-84d4-48aeab2ada75","added_by":"auto","created_at":"2024-02-08 04:43:51","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":133115,"visible":true,"origin":"","legend":"\u003cp\u003eConstruction of WGCNA co-expression network. \u003cstrong\u003e(A) \u003c/strong\u003eSoft threshold (β = 7) and scale-free topological fit index (R\u003csup\u003e2\u003c/sup\u003e). \u003cstrong\u003e(B) \u003c/strong\u003eOriginal and combined modules under the clustering tree. Cluster dendrogram was the result before module cutting, while Merged Dynamic was the result after module cutting. \u003cstrong\u003e(C) \u003c/strong\u003eCluster tree plot of module feature genes. \u003cstrong\u003e(D)\u003c/strong\u003e Heat map of module-trait correlations. The values inside the brackets represented the p-value of significance, while the values outside the brackets represented the values of correlation between Control and AD groups. \u003cstrong\u003e(E)\u003c/strong\u003e Venn diagram of key module genes versus DEGs. The overlapping regions represented the same 109 genes in DEGs and WGCNA. AD: Alzheimer’s Disease; DEGs: the differential genes; WGCNA: Weighted gene co-expression network analysis.\u003c/p\u003e","description":"","filename":"image2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/f90ae27708291caecde3fad4.jpg"},{"id":50832996,"identity":"c6e1aa8d-fc36-454e-a7e5-458a9c8b3ef0","added_by":"auto","created_at":"2024-02-08 04:35:51","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":172679,"visible":true,"origin":"","legend":"\u003cp\u003eScreening for potential biomarkers of AD by machine learning algorithm.\u003cstrong\u003e (A)\u003c/strong\u003e PPI network diagram, red and green represented the upregulated and downregulated genes, respectively. \u003cstrong\u003e(B) \u003c/strong\u003eThe minimum absolute contraction and selection operator models (Lasso) selected characteristic genes. \u003cstrong\u003e(C) \u003c/strong\u003eSVM-RFE algorithm selected biomarker feature genes. The red circle at the lowest point in the left image indicated a minimum error rate of 0.213 for 36 genes, while the red circle at the highest point in the right image represented a maximum accuracy rate of 0.787 for 36 genes. \u003cstrong\u003e(D)\u003c/strong\u003e Random forest tree algorithm was used to evaluate characteristic genes.\u003cstrong\u003e (E) \u003c/strong\u003eThe top 20 important genes were obtained by random forest tree algorithm. \u003cstrong\u003e(F)\u003c/strong\u003e Venn diagram of the three algorithms used to screen genes. The overlapping parts of the three circles represented 13 common genes obtained from the three machine learning algorithms. AD: Alzheimer’s Disease; SVM-RFE: the support vector machine recursive feature elimination.\u003c/p\u003e","description":"","filename":"image3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/c91244328ee95964800bdeb3.jpg"},{"id":50832995,"identity":"55965d5b-4bcd-419f-aaa8-ce5f9eed68b6","added_by":"auto","created_at":"2024-02-08 04:35:51","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":330381,"visible":true,"origin":"","legend":"\u003cp\u003eEvaluating UBE2N as a biomarker for AD.\u003cstrong\u003e (A) \u003c/strong\u003eVenn plot of immune genes and diagnostic markers. \u003cstrong\u003e(B) \u003c/strong\u003eCorrelations between genes. Red squares showed the positive correlation of genes, and blue squares represented the negative correlation of genes. \u003cstrong\u003e(C)\u003c/strong\u003e The training focused on ROC curves for diagnostic markers. \u003cstrong\u003e(D) \u003c/strong\u003eNorman diagrams were used to predict the incidence of AD. \u003cstrong\u003e(E)\u003c/strong\u003e The ROC curve evaluated the clinical application value of the Norman diagram model. \u003cstrong\u003e(F) \u003c/strong\u003eThe DCA curve evaluated the clinical application value of the Norman diagram model. \u003cstrong\u003e(G) \u003c/strong\u003eClinical impact curve: the red curve (number of high-risk individuals) represents the number of individuals classified as positive (high risk) by the model at each threshold probability; the blue curve (the number of at-risk individuals with results) was the number of true positives at each threshold probability. ROC: Receiver Operating Characteristic; DCA: the decision curve analysis.\u003c/p\u003e","description":"","filename":"image4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/9b12d711f4e9b53f88e1b987.jpg"},{"id":50832997,"identity":"bf505b8c-2e6e-4b4e-89eb-c2d57aeffc9d","added_by":"auto","created_at":"2024-02-08 04:35:51","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":86713,"visible":true,"origin":"","legend":"\u003cp\u003eAnalysis of potential regulatory mechanisms of UBE2N. \u003cstrong\u003e(A) \u003c/strong\u003eButterfly diagram showing the first five up- and down-regulated pathways of GSEA. \u003cstrong\u003e(B) \u003c/strong\u003eRegulatory network of UBE2N. The blue graphs represented microRNAs associated with UBE2N; ATF1 as the transcription factor (the green triangle) was associated with UBE2N. GSEA: Gene Set Enrichment Analysis.\u003c/p\u003e","description":"","filename":"image5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/5f4c5ed96b2689873c135eac.jpg"},{"id":50833000,"identity":"ba2d24b3-3e8a-42ff-8a77-88c7cdf9439e","added_by":"auto","created_at":"2024-02-08 04:35:52","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":374962,"visible":true,"origin":"","legend":"\u003cp\u003eCluster Analysis of MFUZZ Expression Patterns. \u003cstrong\u003e(A)\u003c/strong\u003e MFUZZ expression pattern clustering results. \u003cstrong\u003e(B) \u003c/strong\u003essGSEA scores and expression characteristics of clustering modules between control groups (green column) and AD (red column). \u003cstrong\u003e(C) \u003c/strong\u003eCorrelation of clustering modules with UBE2N. Red indicated the positive correlation with the cluster; green indicated the negative correlation with the cluster. The darker the color, the stronger the correlation. \u003cstrong\u003e(D) \u003c/strong\u003eKEGG enrichment analysis of genes in Cluster 27, The darker the color, the more genes enriched in the pathway, and the longer the column, the smaller the \u003cem\u003ep\u003c/em\u003e-value. \u003cstrong\u003e(E) \u003c/strong\u003eCore genes co-expressed with UBE2N in AD. 8 core genes were obtained by crossing with Cluster27 and AD dataset. GSEA: Gene Set Enrichment Analysis.\u003c/p\u003e","description":"","filename":"image6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/904649eb670dc6d3df899f2b.jpg"},{"id":50833192,"identity":"d56a985e-7d9b-4d13-920e-ef3a84c374e6","added_by":"auto","created_at":"2024-02-08 04:43:52","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":331533,"visible":true,"origin":"","legend":"\u003cp\u003eImmune cell infiltration analysis. \u003cstrong\u003e(A) \u003c/strong\u003eComparison of 28 immune cell types, where green and orange represent the control and AD groups, respectively. \u003cstrong\u003e(B)\u003c/strong\u003e Correlation between characteristic genes and immunity. \u003cstrong\u003e(C) \u003c/strong\u003eCorrelation of UBE2N with 28 types of immune cells. \u003cstrong\u003e(D)\u003c/strong\u003e Correlation of UBE2N with effector memory CD4\u003csup\u003e+\u003c/sup\u003e T cells.\u003c/p\u003e","description":"","filename":"image7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/472f828eec75a3d2329ed05c.jpg"},{"id":50832999,"identity":"0e215bfa-d179-414d-bbfd-304cf2251d7f","added_by":"auto","created_at":"2024-02-08 04:35:52","extension":"jpg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":148007,"visible":true,"origin":"","legend":"\u003cp\u003eExpressions of UBE2N in the brain of Tau\u003csup\u003eP301S\u003c/sup\u003e mice. \u003cstrong\u003e(A)\u003c/strong\u003e Differential expression of \u003cem\u003eUBE2N\u003c/em\u003e and related genes in the cortex of control (n = 8) and Tau\u003csup\u003eP301S\u003c/sup\u003e mouse brains (n = 8) analyzed by RT-qPCR using GAPDH as a control.\u003cstrong\u003e (B) \u003c/strong\u003eUBE2N expression in the temporal cortex of control (n=8) and Tau\u003csup\u003eP301S\u003c/sup\u003e mice (n = 8). \u003cstrong\u003e(C) \u003c/strong\u003eUBE2N expression in the hippocampus of control (n=8) and Tau\u003csup\u003eP301S\u003c/sup\u003e mice (n = 8). Positive staining of UBE2N was observed in cortex \u003cstrong\u003e(D)\u003c/strong\u003e and hippocampus \u003cstrong\u003e(E)\u003c/strong\u003e, and it colocalized with red-labeled NeuN in cortex and hippocampus in immunofluorescence images. The scale bar for immunohistochemistry images is 50 μm; the scale bar for cortical and hippocampal immunofluorescence images is 200 μm and 500 μm, respectively. All data from at least three independent experiments are represented as the means ± SEM. * \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05, ** \u003cem\u003ep \u003c/em\u003e\u0026lt; 0.01 with respect to the control group.\u003c/p\u003e","description":"","filename":"image8.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/1f8690ceffeec571285a24e1.jpg"},{"id":77052736,"identity":"8eb06b9f-cc14-4ef9-a47d-6dc096555441","added_by":"auto","created_at":"2025-02-24 16:24:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2907928,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/9d3f502a-d80b-41be-b2ad-ac61ce4d9302.pdf"},{"id":50833002,"identity":"67d2e59e-f4a5-4048-9805-9546ff562399","added_by":"auto","created_at":"2024-02-08 04:35:53","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":11827990,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterials.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/fcb6845053ba1f8d61084adb.pdf"},{"id":50833003,"identity":"9aad1b7f-fbbc-4a88-9720-2830b252d768","added_by":"auto","created_at":"2024-02-08 04:35:55","extension":"zip","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":36121143,"visible":true,"origin":"","legend":"","description":"","filename":"UncroppedblotsPCRrawdataandIFimages.zip","url":"https://assets-eu.researchsquare.com/files/rs-3904783/v1/a3aaadf93e79544827a65840.zip"}],"financialInterests":"No competing interests reported.","formattedTitle":"Identification of UBE2N as a biomarker of Alzheimer's disease by combining WGCNA with machine learning algorithms","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAlzheimer's Disease (AD) is the most common cause of dementia, responsible for 60\u0026ndash;70% of cases\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. The neuropathologic examination is considered to provide the gold standard for AD diagnosis based on its hallmarks: extracellular depositions of senile plaques generated by amyloid β (Aβ) and neurofibrillary tangles formed by hyperphosphorylated tau in different brain regions\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Many risk factors have been identified for AD, including age, genetic risk variants, stress, immune system dysfunction, and infectious diseases\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Based on the current knowledge of the pathogenesis of AD, academic and pharmaceutical industries are conducting R\u0026amp;D to achieve a breakthrough in the generation of effective AD drugs\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Aβ has been investigated as the primary therapeutic target for many years. Aducanumab and lecanemab are anti-amyloid antibodies approved by the US Food and Drug Administration (FDA) for AD treatment\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. At the same time, increasing studies focus on therapies that target the tau protein\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. AD biomarkers are confirmatory in clinical decision-making, which is particularly important with advancing anti-Aβ/tau protein disease-modifying therapies. Besides, biomarkers can be targeted to improve the diagnosis and prognosis\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eRecent studies have shown that innate immune genes and immune cells directly or indirectly affect AD. Even without crossing the blood-brain barrier (BBB), T-cells regulate brain homeostasis through a cascade of immune signals and secretory molecules. A recent study showed that CD8\u003csup\u003e+\u003c/sup\u003e T cells are abnormally expanded in the brains of patients with mild cognitive impairment (MCI) and AD, indicating that CD8\u003csup\u003e+\u003c/sup\u003e T cells may affect neurodegeneration and cognitive impairment in AD\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. CD4\u003csup\u003e+\u003c/sup\u003e T cells infiltrate the brain to promote Aβ clearance and neuronal repair\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Removing B cells can significantly reduce Aβ and reverse memory deficits in a 3xTg AD mouse model\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. Blocking the transforming growth factor in the peripheral macrophage β (TGF-β)-mediated signaling pathway can reduce Aβ in the brain of the Tg2576 mouse model, which may be a potential treatment for AD\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. However, the roles of innate and adaptive immune cells have not been fully clarified, and there is an urgent need to identify new immune-related biomarkers to explain the neuroinflammation and pathogenesis of AD further.\u003c/p\u003e \u003cp\u003eThe analysis and prediction of biomarkers of AD would increase our understanding of the pathology and enhance the development of new drug targets, clinical trials, and overall diagnosis\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Recently, weighted gene co-expression network analysis (WGCNA) has been used to screen potential biomarkers of various diseases\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. The advantage of WGCNA is that it can analyze the correlation between genes rather than being limited to a single gene\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Machine learning is a branch of computer science and statistics that focuses on detecting, diagnosing, and treating diseases\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. The combination of WGCNA and machine learning can improve the accuracy of identifying potential biomarkers of the disease\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Using this approach, SNRPG was identified as a critical gene in AD and metabolic syndrome\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Furthermore, PLXNB1, GRAMD3, and GJA have been screened from different cortex and cerebellum regions related to the Braak NFT stage in AD\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Similar genomic transcription patterns in different cortex regions, according to the Braak 0-VI phases, may participate in the pathological progression of AD through the oxidation pathway\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Therefore, specific biomarkers have been screened for different research purposes using a combination of these two methodologies.\u003c/p\u003e \u003cp\u003eIn this study, we first identified 239 differentially expressed genes (DEGs) in the middle temporal gyrus (MTG) of patients with AD based on the Gene Expression Omnibus (GEO) database. Then, 109 key DEGs were identified using WGCNA. A total of 48 core genes were analyzed from the 109 genes through a protein-protein interaction (PPI) network. Finally, 13 potential biomarkers that may be related to AD were identified and verified in the GSE109887 dataset through joint machine-learning analysis. Immune-related genes selected via the InnateDB database intersected with 13 potential AD-related genes, and the potential immune-related marker, UBE2N, was obtained. Transcription factor prediction and GSEA analyses were used to investigate further the biological processes and pathways of UBE2N related to AD occurrence. The results of the MFUZZ cluster analysis showed that UBE2N was involved in T cell and B cell functions and the synaptic vesicle cycle signaling pathways. Additionally, UBE2N expression was verified in Tau\u003csup\u003eP301S\u003c/sup\u003e transgenic mice as decreased. These results suggest that UBE2N may be a novel biomarker with implications for AD treatment.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eScreening of DEGs in the brain of patients with AD\u003c/h2\u003e \u003cp\u003eTo screen for DEGs in the brain of patients with AD, we first removed the batch effect of genes between the AD and control group and crossed the Principal Component Analysis (PCA) dataset that showed separation from each other, laying the foundation for subsequent analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA \u003cb\u003eand B\u003c/b\u003e). The results included 124 samples from normal individuals and 157 samples of patients with AD. By setting the screening criteria of | LogFC | \u0026ge; 0.5, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, 86 upregulated and 153 downregulated genes were identified. The expression of DEGs is shown in heat and volcano maps (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC \u003cb\u003eand D\u003c/b\u003e). GO analysis revealed that DEGs were mainly enriched in vesicle-mediated synaptic transport, synaptic vesicle cycle, and other related pathways (\u003cb\u003eFig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA\u003c/b\u003e). KEGG enrichment analysis showed that DEGs were mainly enriched in the Alzheimer's disease, cAMP, and MAPK signaling pathways (\u003cb\u003eFig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eB\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eScreening Characteristic Genes of AD by Construction of WGCNA\u003c/h2\u003e \u003cp\u003eA total of 2626 genes with expression variance in the top 25% were included in the WGCNA. The analysis of soft threshold selection showed that the average connectivity was high and the scale-free network distribution reached its optimal level when β\u0026thinsp;=\u0026thinsp;7 (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.85) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Subsequently, we obtained nine independent modules (with the lowest number of genes in the module set to 30 and genes not included in the module shown in gray) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) by setting the clustering height to 0.25, merging highly correlated modules, and confirming the independence between each module (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Next, the correlation between the module genes and AD was analyzed. The results showed that the highest correlation module was the turquoise module, which positively correlated with the control group (r\u0026thinsp;=\u0026thinsp;0.43, p\u0026thinsp;=\u0026thinsp;2e-14) and negatively correlated with the AD group (r = -0.43, p\u0026thinsp;=\u0026thinsp;2e-14) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). The genes inside the turquoise module were screened according to the standard in which GS\u0026thinsp;\u0026ge;\u0026thinsp;0.35 and ME\u0026thinsp;\u0026ge;\u0026thinsp;0.8 were set as the values closer to r, and 195 genes were obtained (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). In total, 109 characteristic genes related to AD were identified by overlapping 195 genes with DEGs (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). GO analysis showed that the characteristic genes were mainly enriched in exocytosis, glomerular development, and synaptic vesicle cycle (\u003cb\u003eFig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eA\u003c/b\u003e). KEGG analysis showed that the characteristic genes were mainly enriched in GABAergic synapses, synaptic vesicle cycles, and cAMP signaling pathways (\u003cb\u003eFig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eB\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eScreening for potential biomarkers of AD by machine learning algorithms\u003c/h2\u003e \u003cp\u003ePPI networks were constructed based on the 109 potential targets. A total of 48 gene nodes with 80 lines were obtained (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). First, 19 genes predicting the incidence of AD were obtained from 48 genes analyzed using the least absolute shrinkage and selection operator regression model (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Then, 48 genes were analyzed using the support vector machine recursive feature elimination, of which 36 showed a high accuracy rate (0.787) and a low error rate (0.213) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). The RF results showed 33 AD-related genes were selected from the 48 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD \u003cb\u003eand E\u003c/b\u003e). Three machine learning algorithms obtained 13 overlapping genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF). \u003cem\u003eATP6V1E1, CCKBR, DYNC1I1, NRN1, SV2B, SYT1, TUBB2A\u003c/em\u003e, and \u003cem\u003eUBE2N\u003c/em\u003e were expressed at low levels in the brains of patients with AD, whereas \u003cem\u003eINPPL1, ITPKB, ITSN1, RAPGEF3\u003c/em\u003e, and \u003cem\u003eTBL1X\u003c/em\u003e were highly expressed (\u003cb\u003eFig. S3\u003c/b\u003e). The validation set exhibited similar results (\u003cb\u003eFig. S4\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eIdentifying UBE2N as a biomarker for AD\u003c/h2\u003e \u003cp\u003eThe diagnostic value of the 13 biomarkers analyzed by Receiver Operating Characteristic (ROC) curves was greater than 0.7 for all areas under the curve (AUC) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). A total of 1696 immune genes were retrieved from the InnateDB database and overlapped with 13 AD biomarkers, identifying UBE2N as an overlapping gene (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Biomarker correlation analysis revealed that UBE2N expression was positively correlated with seven genes (\u003cem\u003eTUBB2A, SV2B, NRN1, CCKBR, DYNC1I1, ATP6V1E1, and SYT1\u003c/em\u003e) and negatively correlated with five genes (\u003cem\u003eINPPL1, ITSN1, ITPKB, RAPGEF3, and TBL1X\u003c/em\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). The UBE2N column line graph was modeled by analyzing the calibration curves, and the differences between the normal and predicted values were small, indicating that the model was accurate (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD \u003cb\u003eand E\u003c/b\u003e). In the decision curve analysis (DCA), the model curve was above the gray line, implying that patients could benefit from the model within the threshold (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eF). The clinical impact curve (CIC) also demonstrated a better overall net benefit in the threshold range (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eG). This suggests that the UBE2N columnar line graph model constructed in this study can be used to assess AD prognosis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eAnalysis of potential regulatory mechanisms of UBE2N\u003c/h2\u003e \u003cp\u003eBased on the pooled median expression values, data from patients with AD were divided into two groups of high and low expression of UBE2N, and KEGG analysis was performed on the differential genes. The results showed that UBE2N might participate in the activation of five pathways, including nicotine addiction, calcium reabsorption regulated by endocrine and other factors, synaptic vesicle circulation, oxidative phosphorylation, and alanine, aspartate, and glutamate metabolism; UBE2N may be involved in the inhibition of five signaling pathways, including the interaction of the viral protein with cytokines and cytokine receptors, graft rejection, and malaria and Staphylococcus aureus infections (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). Meanwhile, GSEA analysis showed that the remaining six genes that were positively correlated with UBE2N were mainly enriched in the synaptic vesicle circulation pathway (\u003cb\u003eFig. S5\u003c/b\u003e). These results indicated that UBE2N may be related to vesicular function. In addition, the potential regulatory network analysis of UBE2N showed that hsa-miR-128-3p, hsa-miR-149-5p, hsa-miR-221-3p, hsa-miR-222-3p, hsa-miR-5010-5p, hsa-miR-522-3p, and hsa-miR-96-5p regulated UBE2N expression and the transcription factor ATF1 was also involved in the transcription regulation process of UBE2N (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eCluster Analysis of MFUZZ Expression Patterns\u003c/h2\u003e \u003cp\u003eAccording to the expression patterns of UBE2N, 50 different clustering results were obtained. The obtained clustering results for ssGSEA scoring were analyzed with the control and AD groups for correlation analysis, and cluster 27 showed the closest module related to UBE2N (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA-C). Functional enrichment analysis showed that the cluster 27 module genes were mainly enriched in the T cell receptor, B cell receptor and synaptic vesicle signaling pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). Subsequently, the genes in the cluster 27 module overlapped with 13 AD biomarkers, and seven genes (\u003cem\u003eATP6V1E1, CCKBR, DYNC1I1, NRN1, SV2B, SYT1, and TUBB2A\u003c/em\u003e) were highly correlated with \u003cem\u003eUBE2N\u003c/em\u003e expression (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eE).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eImmune cell infiltration analysis\u003c/h2\u003e \u003cp\u003eTo evaluate the infiltration status of immune cells in control and AD groups, we first compared the expression levels of 28 types of immune cells. Compared to the control group, the infiltration rate of T cells in the AD group was higher, including natural killer, gamma delta, central memory CD8\u003csup\u003e+\u003c/sup\u003e, factor memory CD8\u003csup\u003e+\u003c/sup\u003e, and central memory CD4\u003csup\u003e+\u003c/sup\u003e T cells. However, the infiltration rate of factor memory CD4\u003csup\u003e+\u003c/sup\u003e T cells was lower in the AD group than in the control group (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA). Subsequently, correlations between the eight genes and immune cells were analyzed. \u003cem\u003eUBE2N\u003c/em\u003e and \u003cem\u003eDYNC1I1\u003c/em\u003e positively correlated with Activated CD4\u003csup\u003e+\u003c/sup\u003e T cells, \u003cem\u003eSYT1\u003c/em\u003e negatively correlated with Activated CD8\u003csup\u003e+\u003c/sup\u003e T cells, and \u003cem\u003eCCKBR\u003c/em\u003e negatively correlated with immature dendritic cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). In addition, \u003cem\u003eUBE2N\u003c/em\u003e showed the highest correlation with effector memory CD4\u003csup\u003e+\u003c/sup\u003e T cells (r\u0026thinsp;=\u0026thinsp;0.7) (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC \u003cb\u003eand D\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eExpression levels of UBE2N were decreased in the brain of an AD mouse model\u003c/h2\u003e \u003cp\u003eRT-qPCR results revealed that the gene expression levels of \u003cem\u003eUBE2N, ATP6V1E1, CCKBR, SV2B\u003c/em\u003e and \u003cem\u003eTUBB2A\u003c/em\u003e in the cerebral cortex of Tau\u003csup\u003eP301S\u003c/sup\u003e mice were significantly reduced, while the expression levels of \u003cem\u003eDYNC1I1\u003c/em\u003e, \u003cem\u003eNRN1\u003c/em\u003e, and \u003cem\u003eSYT1\u003c/em\u003e did not change compared to the control group (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA). The protein levels of UBE2N decreased significantly in the cerebral cortex and hippocampus of Tau\u003csup\u003eP301S\u003c/sup\u003e mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB \u003cb\u003eand C\u003c/b\u003e). However, there was no significant change in UBE2N expression in the cerebral cortex of APP/PS1 mice (\u003cb\u003eFig. S6\u003c/b\u003e). We also observed that UBE2N was co-localized with NeuN in the hippocampus and temporal cortex. Compared to C57BL/6 mice, Tau\u003csup\u003eP301S\u003c/sup\u003e mice showed decreased UBE2N fluorescence intensity in the hippocampus and cortex (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eD \u003cb\u003eand E\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eAD is a neurodegenerative disease that impairs cognitive function and mainly involves changes in brain regions related to learning and memory, such as the temporal lobe and hippocampus\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. In the current study, we performed a comprehensive and in-depth analysis of temporal lobe gene expression profile data to explore AD-specific related genes, and 13 HUB genes were identified. Among them, UBE2N has been validated in the cerebral cortex and hippocampus of Tau\u003csup\u003eP301S\u003c/sup\u003e mice and is the most valuable potential diagnostic marker of AD in our study.\u003c/p\u003e \u003cp\u003eBy analyzing microarray data from the temporal lobes of AD patients in the GEO database, 239 DEGs were identified, including 86 up-regulated and 153 down-regulated genes. Overlapping the key modular genes obtained by WGCNA analysis with DEGs yielded 109 key DEGs, which were mainly enriched in GABAergic synaptic, B-cell receptor, and synaptic vesicle cycle signaling pathways, all of which are key pathological changes in the pathogenesis of AD\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Subsequently, a machine learning algorithm obtained 13 HUB genes. Overlapping with the immune genes in the InnateDB database yields the UBE2N gene. Furthermore, gene correlation analysis revealed that ATP6V1E1, CCKBR, SV2B, DYNC1I1, NRN1, SYT1, and TUBB2A are positively correlated with UBE2N. The AUC areas under the ROC curves are all greater than 0.7, indicating that the constructed model could accurately predict the onset of AD\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eUBE2N plays an important role in many neurodegenerative diseases. Overexpression of UBE2N increases the aggregation of mutant Huntington's proteins\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, whereas knockdown of the E2 enzymes UBE2N, UBE2L3, UBE2D2 and UBE2D3 (UBE2D2/3) significantly reduces autophagic clearance of depolarized mitochondria, and furthermore, UBE2N, UBE2L3, and UBE2D2/3 synergistically promote Parkin-mediated mitochondrial autophagy\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Interestingly, the abnormal downregulation of UBE2N causes in vivo immunosuppressive dysfunction of regulatory T cells, leading to abnormal activation of T cells and inducing various inflammatory responses\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Therefore, we analyzed the infiltration of 28 immune cells and found that the infiltration rate of T cells was significantly higher in the AD brains, suggesting that the balance of T cells may be dysregulated in AD. One reason could be that there is a decrease in tight junction molecules in the vascular endothelium during AD progression, leading to an increase in the permeability of the blood-brain barrier. Because of increased chemokines for T cells in the brain of AD patients, these changes together promote T cell infiltration\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. The accumulation of activated T cells has been demonstrated to induce neuronal death and exacerbate neuroinflammation\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Activated T cells can also promote the release of TNF-α, IL-1, and IL-6 pro-inflammatory factors from peripheral blood mononuclear cells (PBMC), exacerbating the inflammatory response\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. In addition, a large number of CD8\u003csup\u003e+\u003c/sup\u003e T cells are found in the hippocampus of AD patients, and tau-specific CD4 T cells are widely distributed in the peripheral blood of AD and PD patients, suggesting that T cells may be closely associated with AD progression, particularly in terms of tau pathology\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. Therefore, we speculate that UBE2N may affect AD pathology by regulating T cells, which needs to be confirmed by further experiments.\u003c/p\u003e \u003cp\u003eFurthermore, MFUZZ cluster analysis showed that the cluster consisting of the 27th modular gene had the highest correlation with UBE2N, and the functions of the modular genes were mainly related to immunity and synapses, as expected. Cluster 27 overlapped with 13 HUB genes in 8 genes (UBE2N, ATP6V1E1, CCKBR, DYNC1I1, NRN1, SV2B, SYT1, and TUBB2A), seven of which are positively associated with UBE2N. Our RT-qPCR results showed that the mRNA levels of UBE2N were significantly reduced in the cerebral cortex of Tau\u003csup\u003eP301S\u003c/sup\u003e mice. We observed that the protein levels of UBE2N were significantly reduced in the cortex and hippocampus of Tau\u003csup\u003eP301S\u003c/sup\u003e mice, but not in the APP/PS1 mice, indicating that UBE2N may be involved in AD pathogenesis in tau-related pathways.\u003c/p\u003e \u003cp\u003eApart from UBE2N, we also found that the mRNA levels of ATP6V1E1, CCKBR, SV2B and TUBB2A were decreased in the Tau\u003csup\u003eP301S\u003c/sup\u003e cortex. ATP6V1E1 is a large multi-subunit complex divided into a peripheral structural domain (V1) and a proton transmembrane translocation structural domain (V0) that is up-regulated in early AD and down-regulated in late-stage AD\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. ATP6V1E1 acts as a proton pump and mediates the acidification of endosomes, lysosomes, the Golgi and synaptic vesicles\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Its dysfunction, therefore, disrupts PH homeostasis, affecting organelle acidification and, consequently, contributing to AD. ATP6V1E1 is reduced not only in the brain but also in the peripheral blood of AD patients, suggesting that ATP6V1E1 may play an important role in the diagnosis and treatment of AD\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Cholecystokinin (CCK) mediates its action through two G-protein-coupled receptors, CCKAR and CCKBR. Its absence leads to abnormalities in the cerebral cortex and corpus callosum development and further affects the migration of cortical interneurons\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Synthetic CCK analogs can effectively reduce Aβ load in the brain and normalize the levels of PKA, CREB, BDNF and TrkB receptors, thereby improving APP/PS1 mice cognition\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Synaptic vesicle glycoprotein 2B (SV2B) is a synaptic protein that is involved in APP/Aβ metabolism\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. There is evidence that SV2B knockout protects against Aβ-induced memory deficits and ameliorates cholinergic system dysfunction caused by injection of Aβ\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. However, another study found that Aβ levels were significantly elevated in the hippocampus of SV2B knockout mice as compared to WT mice\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. Here, we found downregulation of SV2B in Tau\u003csup\u003eP301S\u003c/sup\u003e mice, suggesting that SV2B may also be related to tau pathology. TUBB2A (Tubulin Beta 2A Class IIa) is a microtubule protein. Tau from the AD brains increased endogenous Tau in cortical neurons; simultaneously, transcriptome sequencing results showed that TUBB2A is remarkably present in neurons\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. Here, we first demonstrated that TUBB2A is reduced in the cerebral cortex of Tau\u003csup\u003eP301S\u003c/sup\u003e mice, indicating that TUBB2A might play a role in the tau-related pathway in AD.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study, we identified UBE2N as a diagnostic marker for AD by combining WGCNA with machine learning approaches. In addition, GSEA analysis indicated that UBE2N may modulate AD onset/development by affecting synaptic vesicles and inflammation, and immune cell correlation analysis revealed that UEB2N may regulate T-cell infiltration in AD pathogenesis. Our analysis provides a new perspective for further exploring the underlying mechanisms by which UEB2N impacts neuroinflammation and synaptic function in the context of AD, and we will subsequently perform relevant experiments.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eData acquisition and pre-processing\u003c/h2\u003e \u003cp\u003eFour microarray datasets (GSE5281, GSE84422 and GSE132903, GSE109887) related to AD were obtained from the GENE EXPRESSION OMNIBUS database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) using the \u0026ldquo;SVA\u0026rdquo; package in R (4.2.1) to remove batch effects among data sets, and the information about the datasets is in \u003cb\u003eTable\u0026nbsp;1\u003c/b\u003e\u003csup\u003e42\u003c/sup\u003e. The gene expression differences (GEDs) were analyzed by the \u0026ldquo;Limma\u0026rdquo; software package based on the screening criteria \u0026ldquo;Adjusted \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and | logFC | \u0026ge; 0.5\u0026rdquo;\u003csup\u003e43\u003c/sup\u003e. The volcanic and thermal maps were created by the \"ggplot2\" software package and the \u0026ldquo;pheatmap\u0026rdquo; software package, respectively.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eWeighted gene co-expression network analysis (WGCNA) to screen target genes\u003c/h2\u003e \u003cp\u003eWGCNA was performed to identify co-expression modules using the R package of \u0026ldquo;WGCNA\u0026rdquo; (version 1.72.1). The top 25% of genes with the highest variance were applied for subsequent WGCNA analyses to guarantee the accuracy of quality results by checking the missing values and clustering the samples. The \u0026ldquo;soft\u0026rdquo; threshold power (β) is calculated to construct a biologically meaningful scale-free topological network. In addition, the topological overlap matrix (TOM) is constructed on the basis of the adjacency matrix, and the dynamic tree-cutting algorithm is used to merge similar modules. Additionally, gene saliency (GS), module affiliation (MM), and correlation coefficients between gene modules and clinical features were calculated to visualize the characteristic gene network. Finally, the potential gene targets for Alzheimer's disease were obtained by the intersection of DEGs and genes within the significant gene module\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eprotein-protein interaction networks Construction\u003c/h2\u003e \u003cp\u003eString database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://string-db.org/\u003c/span\u003e\u003cspan address=\"https://string-db.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to construct protein-protein interaction networks for the AD potential target by setting a confidence level (0.7), followed by the Cytascape (3.8.2) software to view this graph. Based on this network, the genes were selected as the biomarker genes in the pathological process of AD patients for the subsequent screening.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eScreening of AD markers by the Machine learning algorithm\u003c/h2\u003e \u003cp\u003eWe use machine learning algorithms to analyze the central genes in the PPI network to obtain characteristic markers of AD. Firstly, the \u0026ldquo;glmnet\u0026rdquo; (4.1.6) R software package was used for Lasso regression analysis to obtain 19 important genes\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. Next, 34 important genes were obtained by the \u0026ldquo;e1071\u0026rdquo;(1.7.13) R software package, which was used for SVM-REF analysis\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. The \u0026ldquo;random forest\u0026rdquo; (4.7\u0026ndash;1.1) R software package was performed for random forest analysis, and the genes with scores greater than 2 were retained\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e. The genes obtained by combining three methods were considered biomarkers of AD.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eCurve Analysis of Receiver Operating Characteristics (ROC)\u003c/h2\u003e \u003cp\u003eThe \u0026ldquo;Corrplot\u0026rdquo; (0.92) R software package was to analyze the correlation of AD biomarkers screened through machine learning. Then, we used the \u0026ldquo;pROC\u0026rdquo; (1.18.0) R software package to create Receiver Operating Characteristic (ROC) curves and calculated the area under the curve (AUC) to evaluate the clinical diagnostic value of biomarkers\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eDiagnostic Column Line Graph Construction and Validation\u003c/h2\u003e \u003cp\u003eThe immune gene dataset was obtained from the InnateDB database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.innatedb.com\u003c/span\u003e\u003cspan address=\"http://www.innatedb.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and 1696 immune genes were obtained\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. The obtained immune genes and potential biomarkers of AD were intersected and analyzed to screen the AD biomarkers related to immunity. Then, we used the \u0026ldquo;RMS\u0026rdquo; (6.5.0) R software package to construct a column line graph model to predict the incidence rate of AD\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e. Finally, calibration curves were used to evaluate the accuracy of the column line graph model; decision curve analysis and clinical impact curves were used to evaluate the clinical utility of the model\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eEnrichment and regulatory mechanism analysis of UBE2N\u003c/h2\u003e \u003cp\u003eWe performed GSEA analysis on the selected immune biomarker-UBE2N and used the Enrichr database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://maayanlab.cloud/Enrichr/\u003c/span\u003e\u003cspan address=\"https://maayanlab.cloud/Enrichr/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to analyze the transcription factor (TF) of UBE2N. miRTarBase (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://mirtarbase.cuhk.edu.cn/\u003c/span\u003e\u003cspan address=\"https://mirtarbase.cuhk.edu.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), Starbase (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://starbase.sysu.edu.cn/\u003c/span\u003e\u003cspan address=\"https://starbase.sysu.edu.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and TargetScan (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e\u003ca href=\"https://www.ncbi.nlm.nih.gov/geo/\" target=\"_blank\"\u003ewww.targetscan.org\u003c/a\u003e\u003c/span\u003e\u003cspan address=\"http://www.targetscan.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) databases were used to predict miRNAs that regulate UBE2N translation. Then the regulatory network diagram of UBE2N was constructed by Cytoscape (3.8.2).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eAnimals\u003c/h2\u003e \u003cp\u003eTau\u003csup\u003eP301S\u003c/sup\u003e transgenic mice [B6C3-Tg (Prnp-MAPT*P301S) PS19 Vle/J] were originally purchased from the Jackson Laboratory (Bar Harbor, ME, United States) and C57BL/6 mice were obtained from Beijing HuaFuKang Bioscience Co., Ltd. (Beijing, China) for animal mating. In offspring, Tau\u003csup\u003eP301S\u003c/sup\u003e transgenic mice and wild type mice were obtained in the same month through genotype identification. They (n\u0026thinsp;=\u0026thinsp;8 of each group) were housed under a light/dark cycle of 8:00/20:00 and controlled temperature (24\u0026thinsp;\u0026plusmn;\u0026thinsp;2\u0026deg;C) and humidity (40\u0026ndash;70%) conditions for 9 months. All authors complied with the ARRIVE guidelines. All treatments and experimental procedures were performed under the National Institutes of Health guidelines and approved by the Northeastern University Laboratory Animal Ethical Committee (EC-2023A012).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eImmunohistochemistry\u003c/h2\u003e \u003cp\u003eThe mice in the two experimental groups (n\u0026thinsp;=\u0026thinsp;8 in each group) were anesthetized and half brains were removed and fixed using 4% paraformaldehyde and then embedded in paraffin. Serial 5 \u0026micro;m coronal sections were incubated with blocking solution (5% bovine serum albumin [BSA] and 1% normal goat serum) for 1 h and then incubated with rabbit anti-UBE2N (1:200, Abcam) overnight at 4\u0026deg;C. Next day, the sections were incubated with biotinylated goat anti-rabbit IgG (1:500) for 1 h at room temperature (RT) and then with the avidin\u0026ndash;biotin\u0026ndash;peroxidase (ABC) complex (1:100) for 30 min at RT. After washing with PBS, the sections were immersed in 3,3\u0026rsquo;-diaminobenzidine for development. One section was incubated within normal rabbit serum (1:100) for nonspecific staining and served as a negative control. Images of immunohistochemical staining were captured using a light microscope (DM4000B; Leica, Wetzlar, Germany).\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eImmunofluorescence Staining and Confocal Laser Scanning Microscopy\u003c/h2\u003e \u003cp\u003eThe 5 \u0026micro;m coronal sections were preincubated with blocking buffer for 1 h and then with rabbit anti-UBE2N (1:200, Abcam) and mouse monoclonal anti-NeuN antibodies (1:200, Thermo) incubated overnight at 4\u0026deg;C. Alexa Fluor\u0026reg; 488-and Alex Fluor\u0026reg; 594-conjugated secondary antibodies were mixed together and treated to sections for 2 h and finally labeled using DAPI (1:500). After mounted with an antifade mounting medium, the mages were taken using the laser scanning confocal microscope (Leica, TCS, SP8, Wetzlar, Germany).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eWestern blot\u003c/h2\u003e \u003cp\u003eThe temporal cortex and hippocampus of half-brain of C57BL/6 mice and Tau\u003csup\u003ep301s\u003c/sup\u003e mice were lysed in RIPA buffer and centrifuged to extract protein supernatant. 10 \u0026micro;g proteins were separated by 4\u0026ndash;12% SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Burlington, MA, USA). the membranes were incubated in 5% BSA solution at room temperature for 1h. Subsequently, the membranes were incubated overnight at 4\u0026deg;C in rabbit anti-UBE2N antibody (1:2000, Abcam, Cambridge, UK) and mouse anti-GAPDH (1:10000, A1978, Sigma, Burlington, MA, USA). Finally, the membranes were incubated with the horseradish peroxidase (HRP)-conjugated secondary antibodies for 2 hours after washing. Bands were detected using a chemiluminescence imaging analysis system (Tanon, 5500, Shanghai, China) and enhanced chemiluminescence (ECL) Kits (EMD Millipore, Burlington, MA, USA). Each experiment was repeated at least three times.\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eQuantitative Real-Time Polymerase Chain Reaction (RT-qPCR)\u003c/h2\u003e \u003cp\u003eTotal RNA was extracted from the cortex of C57BL/6 mice and Tau\u003csup\u003ep301s\u003c/sup\u003e mice using Total RNA KIT I (R6834-02, OMGEA, USA), and 500 ng of template RNA was reverse transcribed into cDNA using the GoScript\u0026trade; Reverse Transcription System (Promega, A5001) according to the manufacturer's instructions. PCR reactions were performed with 20 ng of cDNA template at a volume of 10 \u0026micro;L reaction mixture using the Bio-Rad CFX PCR system. The sequences of the genes encoding GAPDH and selected differential genes were obtained from the GenBank database, and specific primers were designed using Primer Premier 5.0 (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The mRNA expression was calculated according to Eq.\u0026nbsp;2\u003csup\u003e\u0026minus;∆∆CT\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eData source.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGEO\u003c/p\u003e \u003cp\u003edatasets\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePlatform\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSample normal\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSample\u003c/p\u003e \u003cp\u003eAD\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePublication years\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eRegions\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGSE5281\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGPL570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e2006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGSE84422\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGPL96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e2016\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGSE132903\u003c/p\u003e \u003cp\u003eGSE109887\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGPL10558\u003c/p\u003e \u003cp\u003eGPL10904\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e98\u003c/p\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e97\u003c/p\u003e \u003cp\u003e46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e2019\u003c/p\u003e \u003cp\u003e2019\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003cp\u003eGermany\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePrimer sequences for RT-qPCR.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGene\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eForward\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eReverse\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eUBE2N\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCCGCACAGTTCTGCTATCAA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAGTCCATGCTCTCGCTGTTT\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eATP6V1E1\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCTTGTACCAGCTGCTGGAGCC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAGGCCTCCTGGTCAATCTGGA\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eCCKBR\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGATGGCTGCTACGTGCAACT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCGCACCACCCGCTTCTTAG\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eSV2B\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGCGGCCTGGCTGATAAACT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAGAGGAAGGCTCCATATCCCT\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eTUBB2A\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTGCCCTCACCCAAGGTCTCTG\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGGCAGGTGGTCACTCCACTCA\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eSYT1\u003c/em\u003e\u003c/p\u003e \u003cp\u003e\u003cem\u003eDYNC1I1\u003c/em\u003e\u003c/p\u003e \u003cp\u003e\u003cem\u003eNRN1\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCTGTCACCACTGTTGCGAC\u003c/p\u003e \u003cp\u003eGTCGTCATGGAAGCAAAGCA\u003c/p\u003e \u003cp\u003eGCGGTGCAAATAGCTTACCTG\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGGCAATGGGATTTTATGCAGTTC\u003c/p\u003e \u003cp\u003eAAGGAGTAGAGCGGCTTGTT\u003c/p\u003e \u003cp\u003eTGATGTTCGTCTTGTCGTCCA\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003e2.13. Statistical Analysis\u003c/h2\u003e \u003cp\u003eData are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SEM. Student\u0026rsquo;s t test were used to analyze differences between groups, as appropriate. The analyses were performed using ImageJ and GraphPad Prism 9.0 software. \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 or \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01 were considered statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":" \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003eData availability statement\u003c/h2\u003e \u003cp\u003eThe datasets analyzed in this study are available from the corresponding authors upon reasonable request.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\u003cp\u003e \u003ch2\u003eAdditional information\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eG.F. performed the experimental procedures and statistical analysis. M.Z. and H.H. assisted with data analysis. P.Z., X.Z., and T.W. generated and validated the mouse model and performed animal experiments. H.G. conceived the experiments and wrote the manuscript. H.X. conceived the experiments, supervised the project, and wrote and revised the manuscript. All authors reviewed and approved the final version of the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgments\u003c/h2\u003e \u003cp\u003e This study was supported by Shenzhen Natural Science Foundation-The Stable Support Program (20220810144826003), the Research Start-up Fund for Young Investigators in Shenzhen University (QNJS0384), the Construction Project of Liaoning Provincial Key Laboratory, China (2022JH13/10200026), the Special Projects of the Central Government in Guidance of Local Science and Technology Development (2022JH6/100100025) and the National Natural Science Foundation of China (81771174, 81971015). We would like to thank Editage (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e\u003ca href=\"https://www.ncbi.nlm.nih.gov/geo/\" target=\"_blank\"\u003ewww.editage.com\u003c/a\u003e\u003c/span\u003e\u003cspan address=\"http://www.editage.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) for English language editing.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eScheltens, P. \u003cem\u003eet al.\u003c/em\u003e Alzheimer's disease. The Lancet 397, 1577\u0026ndash;1590 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLong, J. M. \u0026amp; Holtzman, D. M. Alzheimer Disease: An Update on Pathobiology and Treatment Strategies. Cell 179, 312\u0026ndash;339 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR, A. A. Risk factors for Alzheimer's disease. Folia Neuropathol 57, 87\u0026ndash;105, doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5114/fn.2019.85929\u003c/span\u003e\u003cspan address=\"10.5114/fn.2019.85929\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCummings, J. \u003cem\u003eet al.\u003c/em\u003e Drug development in Alzheimer\u0026rsquo;s disease: the path to 2025. Alzheimer's Research \u0026amp; Therapy 8, 39 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMahase, E. Alzheimer's disease: Lecanemab gets full FDA approval and black box safety warning. BMJ 382, 1580, doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.p1580\u003c/span\u003e\u003cspan address=\"10.1136/bmj.p1580\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDickson, S. P., Hennessey, S., Nicodemus Johnson, J., Knowlton, N. \u0026amp; Hendrix, S. B. Avoiding future controversies in the Alzheimer's disease space through understanding the aducanumab data and FDA review. Alzheimers Res Ther 15, 98, doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13195-023-01238-1\u003c/span\u003e\u003cspan address=\"10.1186/s13195-023-01238-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQureshi, I. A. \u003cem\u003eet al.\u003c/em\u003e A randomized, single ascending dose study of intravenous BIIB092 in healthy participants. Alzheimers Dement (N Y) 4, 746\u0026ndash;755, doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.trci.2018.10.007\u003c/span\u003e\u003cspan address=\"10.1016/j.trci.2018.10.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCummings, J. The Role of Biomarkers in Alzheimer\u0026rsquo;s Disease Drug Development. Reviews on Biomarker Studies in Psychiatric and Neurodegenerative Disorders, 29\u0026ndash;61 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGate, D. \u003cem\u003eet al.\u003c/em\u003e Clonally expanded CD8 T cells patrol the cerebrospinal fluid in Alzheimer\u0026rsquo;s disease. Nature 577, 399\u0026ndash;404 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMonsonego, A., Nemirovsky, A. \u0026amp; Harpaz, I. CD4 T cells in immunity and immunotherapy of Alzheimer's disease. Immunology 139, 438\u0026ndash;446 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim, K. \u003cem\u003eet al.\u003c/em\u003e Therapeutic B-cell depletion reverses progression of Alzheimer\u0026rsquo;s disease. Nature Communications 12, 2185 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTown, T. \u003cem\u003eet al.\u003c/em\u003e Blocking TGF-β\u0026ndash;Smad2/3 innate immune signaling mitigates Alzheimer-like pathology. Nature medicine 14, 681\u0026ndash;687 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, T. \u003cem\u003eet al.\u003c/em\u003e Integrated bioinformatic analysis reveals YWHAB as a novel diagnostic biomarker for idiopathic pulmonary arterial hypertension. Journal of Cellular Physiology 234, 6449\u0026ndash;6462 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangfelder, P. \u0026amp; Horvath, S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9, 559 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangfelder, P. \u0026amp; Horvath, S. Fast R Functions for Robust Correlations and Hierarchical Clustering. Journal of Statistical Software 46, i11 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOzsolak, F. \u0026amp; Milos, P. M. RNA sequencing: advances, challenges and opportunities. Nature Reviews. Genetics 12, 87\u0026ndash;98 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeiffer-Smadja, N. \u003cem\u003eet al.\u003c/em\u003e Machine learning for clinical decision support in infectious diseases: a narrative review of current applications. Clinical Microbiology and Infection: The Official Publication of the European Society of Clinical Microbiology and Infectious Diseases 26, 584\u0026ndash;595 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, Y., Liao, R., Yao, Y., Wang, Q. \u0026amp; Fu, L. Machine learning to identify immune-related biomarkers of rheumatoid arthritis based on WGCNA network. Clinical Rheumatology 41, 1057\u0026ndash;1068 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, J. \u003cem\u003eet al.\u003c/em\u003e Identification of diagnostic genes for both Alzheimer's disease and Metabolic syndrome by the machine learning algorithm. Frontiers in Immunology 13, 1037318 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, X., Yao, H. \u0026amp; Li, X. Unearthing of Key Genes Driving the Pathogenesis of Alzheimer's Disease via Bioinformatics. Frontiers in Genetics 12, 641100 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChai, K. \u003cem\u003eet al.\u003c/em\u003e Application of weighted co-expression network analysis and machine learning to identify the pathological mechanism of Alzheimer's disease. Frontiers in Aging Neuroscience 14, 837770 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeTure, M. A. \u0026amp; Dickson, D. W. The neuropathological diagnosis of Alzheimer\u0026rsquo;s disease. Molecular neurodegeneration 14, 1\u0026ndash;18 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, C. \u003cem\u003eet al.\u003c/em\u003e Identification of Immune Cells and Key Genes associated with Alzheimer's Disease. International Journal of Medical Sciences 19, 112\u0026ndash;125 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evon Dadelszen, P. \u003cem\u003eet al.\u003c/em\u003e Prediction of adverse maternal outcomes in pre-eclampsia: development and validation of the fullPIERS model. Lancet (London, England) 377, 219\u0026ndash;227 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYin, P. \u003cem\u003eet al.\u003c/em\u003e Aged monkey brains reveal the role of ubiquitin-conjugating enzyme UBE2N in the synaptosomal accumulation of mutant huntingtin. Hum Mol Genet 24, 1350\u0026ndash;1362, doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/hmg/ddu544\u003c/span\u003e\u003cspan address=\"10.1093/hmg/ddu544\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGeisler, S., Vollmer, S., Golombek, S. \u0026amp; Kahle, P. J. The ubiquitin-conjugating enzymes UBE2N, UBE2L3 and UBE2D2/3 are essential for Parkin-dependent mitophagy. J Cell Sci 127, 3280\u0026ndash;3293, doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1242/jcs.146035\u003c/span\u003e\u003cspan address=\"10.1242/jcs.146035\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang, J.-H. \u003cem\u003eet al.\u003c/em\u003e Ubc13 maintains the suppressive function of regulatory T cells and prevents their conversion into effector-like T cells. Nature immunology 13, 481\u0026ndash;490 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarrano, A. \u003cem\u003eet al.\u003c/em\u003e Amyloid Beta induces oxidative stress-mediated blood-brain barrier changes in capillary amyloid angiopathy. Antioxidants \u0026amp; Redox Signaling 15, 1167\u0026ndash;1178 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoldeck, D. \u003cem\u003eet al.\u003c/em\u003e Enhanced Chemokine Receptor Expression on Leukocytes of Patients with Alzheimer's Disease. PLoS ONE 8, e66664 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiuliani, F., Goodyer, C. G., Antel, J. P. \u0026amp; Yong, V. W. Vulnerability of human neurons to T cell-mediated cytotoxicity. \u003cem\u003eJournal of Immunology (Baltimore, Md.\u003c/em\u003e: 1950) 171, 368\u0026ndash;379 (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMietelska-Porowska, A. \u0026amp; Wojda, U. T Lymphocytes and Inflammatory Mediators in the Interplay between Brain and Blood in Alzheimer's Disease: Potential Pools of New Biomarkers. \u003cem\u003eJournal of Immunology Research\u003c/em\u003e 2017, 4626540 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLindestam Arlehamn, C. S. \u003cem\u003eet al.\u003c/em\u003e Widespread tau-specific CD4 T cell reactivity in the general population. The Journal of Immunology 203, 84\u0026ndash;92 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMerlini, M., Kirabali, T., Kulic, L., Nitsch, R. M. \u0026amp; Ferretti, M. T. Extravascular CD3\u0026thinsp;+\u0026thinsp;T Cells in Brains of Alzheimer Disease Patients Correlate with Tau but Not with Amyloid Pathology: An Immunohistochemical Study. Neuro-Degenerative Diseases 18, 49\u0026ndash;56 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang, S.-H. \u003cem\u003eet al.\u003c/em\u003e Proteomic profiling of brain cortex tissues in a Tau transgenic mouse model of Alzheimer's disease. Biochemical and Biophysical Research Communications 430, 670\u0026ndash;675 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaunescu, T. G., Jones, A. C., Tyszkowski, R. \u0026amp; Brown, D. V-ATPase expression in the mouse olfactory epithelium. American Journal of Physiology-Cell Physiology 295, C923-C930 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu, H. \u003cem\u003eet al.\u003c/em\u003e Integrated transcriptomics reveals the brain and blood biomarkers in Alzheimer's disease. CNS Neuroscience \u0026amp; Therapeutics (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNishimura, S. \u003cem\u003eet al.\u003c/em\u003e Functional synergy between cholecystokinin receptors CCKAR and CCKBR in mammalian brain development. PloS One 10, e0124295 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, Z. \u003cem\u003eet al.\u003c/em\u003e Cholecystokinin signaling can rescue cognition and synaptic plasticity in the APP/PS1 mouse model of Alzheimer\u0026rsquo;s disease. Molecular Neurobiology, 1\u0026ndash;23 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiyamoto, M. \u003cem\u003eet al.\u003c/em\u003e Synaptic Vesicle Protein 2B Negatively Regulates the Amyloidogenic Processing of AβPP as a Novel Interaction Partner of BACE1. Journal of Alzheimer's disease: JAD 75, 173\u0026ndash;185 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDetrait, E., Maurice, T., Hanon, E., Leclercq, K. \u0026amp; Lamberty, Y. Lack of synaptic vesicle protein SV2B protects against amyloid-β25\u0026ndash;35-induced oxidative stress, cholinergic deficit and cognitive impairment in mice. Behavioural brain research 271, 277\u0026ndash;285 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFiculle, E. \u003cem\u003eet al.\u003c/em\u003e A human tau seeded neuronal cell model recapitulates molecular responses associated with Alzheimer\u0026rsquo;s disease. Scientific Reports 12, 2673 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeek, J. T., Johnson, W. E., Parker, H. S., Jaffe, A. E. \u0026amp; Storey, J. D. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics (Oxford, England) 28, 882\u0026ndash;883 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRitchie, M. E. \u003cem\u003eet al.\u003c/em\u003e limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Research 43, e47 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, S. \u003cem\u003eet al.\u003c/em\u003e An Eight-CircRNA Assessment Model for Predicting Biochemical Recurrence in Prostate Cancer. Frontiers in Cell and Developmental Biology 8, 599494 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanz, H., Valim, C., Vegas, E., Oller, J. M. \u0026amp; Reverter, F. SVM-RFE: selection and visualization of the most relevant features through non-linear kernels. BMC bioinformatics 19, 432 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, Y., Xu, L., Qiao, Y., Wang, T. \u0026amp; Zheng, Q. Construction of a neural network diagnostic model and investigation of immune infiltration characteristics for Crohn's disease. Frontiers in Genetics 13, 976578 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng, S. \u003cem\u003eet al.\u003c/em\u003e Novel blood test for early biomarkers of preeclampsia and Alzheimer\u0026rsquo;s disease. Scientific Reports 11, 15934 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreuer, K. \u003cem\u003eet al.\u003c/em\u003e InnateDB: systems biology of innate immunity and beyond\u0026ndash;recent updates and continuing curation. Nucleic Acids Research 41, D1228-1233 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, J.-A. \u003cem\u003eet al.\u003c/em\u003e Development of an Immune-Related Gene Signature for Prognosis in Melanoma. Frontiers in Oncology 10, 602555 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, C. \u003cem\u003eet al.\u003c/em\u003e Identification of a glycolysis- and lactate-related gene signature for predicting prognosis, immune microenvironment, and drug candidates in colon adenocarcinoma. Frontiers in Cell and Developmental Biology 10, 971992 (2022).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3904783/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3904783/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAlzheimer\u0026rsquo;s disease (AD) is the most common neurodegenerative disorder leading to progressive cognitive decline. With the development of machine learning analysis, screening biomarkers based on existing clinical data is becoming conducive to understanding the pathogenesis of AD and discovering new treatment targets. Our study integrated three AD datasets in the GEO database for differential expression analysis. After constructing a WGCNA network, 109 key genes were obtained and 48 core genes were analyzed from 109 genes using a protein-protein interaction network. The least absolute shrinkage and selection operator, support vector machine recursive feature elimination, and Random Forest methods were applied to obtain the features associated with the 48 core genes and 13 potentially related AD biomarkers were selected. By intersecting InnateDB database with them, we found a potential immune-related marker, UBE2N. MFUZZ cluster analysis revealed that UBE2N is closely related to T cell and B cell functions and the synaptic vesicle cycle signaling pathways. In addition, the expression levels of UBE2N were decreased in the temporal cortex and hippocampus of Tau\u003csup\u003eP301S\u003c/sup\u003e mice but not APP/PS1 mice. Our findings are the first comprehensive identification of UBE2N as a biomarker for AD, paving the way for much-needed early diagnosis and targeted treatment.\u003c/p\u003e","manuscriptTitle":"Identification of UBE2N as a biomarker of Alzheimer's disease by combining WGCNA with machine learning algorithms","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-02-08 04:35:47","doi":"10.21203/rs.3.rs-3904783/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-10-29T03:11:32+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-09-04T22:57:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"79626785151232718825786824914725670200","date":"2024-08-14T14:34:16+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"38678789085398724205534109048425655082","date":"2024-08-14T13:15:19+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"294040106279884780496690705810182478639","date":"2024-08-14T12:27:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"194386911601365454497640304302511728234","date":"2024-06-07T19:28:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"329693784423128883334197189224826194430","date":"2024-06-05T20:15:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"37100877-1f51-4b5d-bfaf-99d02a62dcd9","date":"2024-04-11T06:35:57+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-03-31T11:18:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"8ddc55ab-1553-49ce-bafb-694bcd546a71_SNPRID","date":"2024-03-31T06:17:08+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-03-30T19:37:56+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-03-19T05:19:54+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2024-02-06T12:57:23+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-02-06T12:54:50+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2024-01-28T05:54:02+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7560d8e1-6ed2-4967-b8be-3448bdd45c45","owner":[],"postedDate":"February 8th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":28615507,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":28615508,"name":"Biological sciences/Neuroscience"},{"id":28615509,"name":"Health sciences/Diseases/Neurological disorders"}],"tags":[],"updatedAt":"2025-02-24T16:05:27+00:00","versionOfRecord":{"articleIdentity":"rs-3904783","link":"https://doi.org/10.1038/s41598-025-90578-z","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-02-22 15:57:18","publishedOnDateReadable":"February 22nd, 2025"},"versionCreatedAt":"2024-02-08 04:35:47","video":"","vorDoi":"10.1038/s41598-025-90578-z","vorDoiUrl":"https://doi.org/10.1038/s41598-025-90578-z","workflowStages":[]},"version":"v1","identity":"rs-3904783","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3904783","identity":"rs-3904783","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}