Single‐cell atlas of neuroglial dynamics in SNCA‐A53T Parkinson's disease mouse model

Abstract

Background

Parkinson's disease (PD) is a neurodegenerative disorder characterized by progressive degeneration of midbrain substantia nigra dopaminergic neurons, resulting in striatal dopamine depletion and motor dysfunction. While this pathological cascade is well‐established, its underlying mechanisms remain elusive.

Methods

To further investigate the pathological mechanisms of PD, we performed single‐cell RNA sequencing of the midbrain and striatum from Hua‐Syn (SNCA*A53T) transgenic (A53T) mice as a PD model.

Results

Analysis of 22 865 midbrain and 32 117 striatal cells revealed cell‐type‐specific risk association. Glial populations (astrocytes, microglia, oligodendrocytes) showed significant enrichment for PD‐risk genes. Variance‐based clustering identified PD‐enriched subclusters exhibiting upregulated inflammatory pathways, apoptotic pathways, proteostasis disruption, glutamatergic signaling dysregulation, and mitochondrial respiratory chain defects. Transcriptional regulation analysis identified genes associated with PD specific activity, including Rorb and Foxc1 in the midbrain and Dbx2 and Klf13 in the striatum. Cell–cell interactions showed that cell‐to‐cell signaling was enhanced, and the SEMA and CCL neuroinflammatory axes were specifically activated in the PD group.

Conclusions

Our integrative analysis delineates the cellular and molecular architecture of the pathological process triggered by the expression of A53T mutant α‐synuclein, and provides a framework for targeted therapeutic development.

We performed single‐cell RNA sequencing of midbrain and striatal tissues from SNCA‐A53T Parkinson's disease (PD) mice, revealing glia‐enriched PD‐risk gene signatures and disease‐specific subpopulations. Transcriptional dysregulation of key TFs (e.g., Rorb, Foxc1) and enhanced neuroinflammatory signaling (SEMA, CCL, MIF) were identified. Our study delineates the cellular and molecular architecture of A53T‐driven PD pathology, highlighting glial‐neuronal crosstalk as a central mechanism.

1. INTRODUCTION

Parkinson's disease (PD) is a common neurodegenerative disorder that affects millions of people worldwide. It is characterized by the loss of dopamine‐producing neurons in the brain, resulting in tremors, rigidity, and other motor symptoms. ¹ The pathological mechanisms underlying PD remain poorly understood; however, it is believed that a combination of genetic and environmental factors contributes to the development of the disease. Several genes, including SNCA, LRRK2, and PARKIN, have been linked to PD development, with their mutations disrupting cellular processes such as protein degradation and mitochondrial function, ultimately leading to the accumulation of harmful protein aggregates and oxidative stress. ² , ³ , ⁴ Despite extensive research, uncertainty persists regarding the molecular intricacies of PD pathogenesis.

The advent of single‐cell/nucleus RNA sequencing (sc/snRNA‐seq) technology has positioned it as the preeminent tool for comprehensively evaluating cell‐type heterogeneity. This technique has been widely used in neuroscience, enabling the unbiased characterization of distinct cell populations within the substantia nigra and other brain regions affected by risk factors. ⁵ , ⁶ Conventional methodologies predominantly depend on bulk sequencing, which amalgamates gene expression profiles across entire tissue samples, consequently masking the distinct contributions of individual cell types. In contrast, scRNA‐seq provides a transformative approach, enabling the dissection of cellular heterogeneity that underpins essential neurobiological processes. ⁷ Notably, alpha‐synuclein (α‐syn), a protein intricately implicated in the PD pathogenesis, serves as a pivotal focus in this context. The A53T mutation in the human SNCA gene has been implicated in developing familial PD. ⁸ Although previous studies have provided valuable insights into the effects of the A53T mutation on the cellular phenotypes of the mouse brain, how it affects cell state at the single‐cell transcription level has not been elucidated. In this study, we applied scRNA‐seq to brain regions, including the midbrain and striatum, of 13‐week‐old Hua‐Syn (SNCA*A53T) transgenic (A53T) mice and their matched controls (wild‐type, WT). Firstly, we performed cell clustering and cell type annotation based on marker genes. Secondly, using the UCell method, ⁹ we compared the gene set scores between A53T mice and WT mice to determine the cell types affected by the expression of the α‐syn. Additionally, we utilized the graph signal processing‐based MELD method ¹⁰ to obtain cell density at single‐cell resolution in the midbrain and striatum regions, with a focus on cell type differences. We further subdivided key cell types, including astrocytes, microglia, and oligodendrocytes, based on cell density and transcriptome, and identified differentially expressed genes within these specific subpopulations for gene set enrichment analysis. Finally, we analyzed changes in transcriptional regulatory networks and communication patterns among PD‐specific cell types, aiming to explore the impact of early α‐syn aggregation on cellular communication modes.

2. METHODS

2.1. Animals

Male A53T mice and wild‐type mice with a C57BL/6J background were used in this study. A53T mice were donated by Professor Zhang Zhentao of Wuhan University [SCXK(E)2019‐0004] and housed in a pathogen‐free barrier facility at Xiangyang No. 1 People's Hospital [SCXK(E)2022‐0093]. At 13 weeks of age, A53T mice (A53T group, n = 3) and wild‐type littermates (WT group, n = 3) were euthanized, and the brain tissue was quickly removed. The study was approved following the China Public Health Service Guide for the Care and Use of Laboratory Animals. Experiments involving mice and protocols were approved by the Ethical Committee of Biomedical Basic Research of Xiangyang No. 1 People's Hospital (XYYYE20220030).

2.2. Tissue isolation

Mouse midbrain and striatum sites were taken, washed three times with pre‐chilled Hanks Balanced Salt Solution (HBSS), and then cut into 2–3 mm blocks. After digestion, the tissue blocks were spin‐digested with GEXSCOPE® Tissue Dissociation Solution (Singleron) at 37°C for 2.5 h; they were then filtered through 70 and 40 μm sterile filters and centrifuged at 360 × g for 5 min. Subsequently, the supernatant was discarded, and the cell pellet was suspended in 600 μL of phosphate‐buffered saline (PBS, HyClone). Samples were stained with Tissue Blue (Biorad, USA) and cell viability was estimated using a phase contrast light microscope (Nikon, Japan).

2.3. Single‐cell RNA sequencing and the primary analysis of sequencing data

Single‐cell suspensions were prepared in PBS (HyClone) with 1 × 10⁵ cells/mL. The single‐cell suspensions were then loaded onto microfluidic devices, and scRNA‐seq libraries were constructed according to the Singleron GEXSCOPE® protocol via the GEXSCOPE® Single Cell RNA Library Kit (Singleron Biotechnologies). Individual libraries were diluted to 4 nM and combined for sequencing on an Illumina HiSeq X with 150 bp paired‐end reads. After scRNA‐seq, fastQC and fastp were used to remove low‐quality raw reads and splice sequences. Subsequently, reads were localized to the reference genome mm10 using STAR. Thereafter, gene counts and UMI counts obtained by FeatureCounts software were used to construct expression matrix files.

2.4. Quality control, integration, and clustering

Computational analysis was performed using Seurat (v 3.1.2). ¹¹ For each sample dataset, we filtered expression matrix by the following criteria: (1) cells with gene count <200 or with top 2% gene count were excluded; (2) cells with top 2% UMI count were filtered out; (3) cells with mitochondrial content >20% were removed; (4) genes expressed in less than 5 cells were excluded. Data were normalized and scaled using NormalizeData and ScaleData, respectively. We identified the top 2000 highly variable genes via FindVariableFeatures for principal component analysis (PCA). To account for batch effects, the first 50 principal components (PCs) were utilized as input for Harmony (v1.2.0) ¹² integration. Subsequently, the first 20 Harmony‐adjusted components were used for downstream cell clustering and dimensionality reduction. Cell clusters were identified using FindNeighbors and FindClusters with the resolution parameter set at 1.2, yielding a total of 35 distinct clusters. Cluster visualization was performed using Uniform Manifold Approximation and Projection (UMAP) via the RunUMAP function. ¹³

2.5. PD‐related likelihood scoring and differential cell abundance

The PD‐risk genes (PDRG) were obtained from DisGeNET ¹⁴ and filtered using a Score_gda threshold of >0.1 (Table S1), and then human gene symbols were converted to mouse gene symbols using the biomaRt package. ¹⁵ Subsequently, UCell ⁹ algorithm was applied to evaluate the enrichment score of PDRG in each single cell. Within each cell type, a two‐sample t‐test was used to compare the PDRG scores of cells from the A53T group and the WT group.

To explore the change of cell abundance in PD at the single‐cell level, we used MELD ¹⁰ to calculate the PD relative likelihood for every cell in the dataset. Then, the vertex frequency clustering (VFC) algorithm ¹⁶ was applied to identify the PD‐enriched or WT‐enriched cell subclusters.

2.6. Differential expression and functional enrichment

Differentially expressed genes (DEGs) were identified by the Wilcoxon rank sum test with a significance level of 0.05. Although this method does not explicitly model zero inflation, it is a non‐parametric rank‐based test that is commonly used for single‐cell RNA‐seq data and is relatively robust to sparsity and dropout events. We controlled false‐discovery rates (FDRs) using the Benjamini‐Hochberg procedure. The DEGs between the PD and WT groups were identified using thresholds of adjusted p‐value < 0.05 and |log2FC| > 1. To interpret the biological functions and pathways of these DEGs, we performed functional enrichment annotation against the MSigDB gene sets ¹⁷ and KEGG pathways ¹⁸ using the GSEApy ¹⁹ , a Python package for gene set enrichment analysis.

2.7. Regulatory network inference

We adopted pySCENIC ²⁰ to infer the regulatory networks in the dataset and explore the differences in regulatory effects between A53T and WT samples. The cell‐type‐specific regulons were identified using the regulon specificity scores (RSS). ²¹ If the regulon activity is highly different among cell clusters, the RSS approaches 1; otherwise, the RSS will be equal to 0. Combining the RSS of each regulon, we explored the regulons with different activities between the two groups within each cell type. To further verify whether the regulatory elements that were upregulated in the A53T group were functionally associated with the condition of PD, we performed the hypergeometric test using the PDRG and the genes regulated by each regulon.

2.8. Cell–cell communication analysis

CellChat ²² (version 5.0) was used to infer cell–cell communication networks between cell types associated with PD, and the total number of interactions and interaction strength of the inferred cell–cell communication networks among different cell types from A53T and WT were compared using the “compareInteractions” function. The communication pathways that turn off, decrease, turn on, or increase in A53T group compared to WT group were identified by comparing the information flow for each signaling pathway based on the “rankNet” function. Finally, the ligand‐receptor pairs level to detect the up‐ and down‐regulated signaling ligand‐receptor pairs were zoomed in A53T compared to WT. In this analysis, all the plots are derived from the visualization functions in the CellChat package.

3. RESULTS

3.1. Single‐cell transcriptomic profiling of midbrain and striatum in A53T mice

To systematically characterize the complex transcriptional features of PD, we collected the midbrain and striatum samples from six mice (including three A53T mice and three wild‐type mice), and processed them for single‐cell RNA sequencing (scRNA‐seq). By using 10× Genomics Chromium Single Cell 3′ Solution, we obtained 65 494 droplet‐based scRNA‐seq profiles from these mouse brain tissues. After quality control, 54 982 high‐quality profiles were retained, with comparable representation from the midbrain (22 865 cells) and striatum (32 117 cells). To identify major cell types, we integrated cells from A53T model and WT mice using Harmony and performed unsupervised clustering following batch correction.

In the midbrain, the resulting clusters showed clear and distinct marker expression patterns. Based on established markers, we assigned these clusters into 14 cell types, including astrocytes (Aqp4, Aldoc, Slc1a3), choroid plexus cells (Ttr, Sostdc1, Kl), endothelial cells (ECs) (Pecam1, Cdh5, Cldn5), ependymal cells (Ccdc153, Hdc, Tmem212), erythrocytes (Hbb‐bs, Hba‐a1, Alas2), fibroblasts (Dcn, Col1a2, Col1a1), MPs (macrophages; Lyz2, Pf4, Mrc1), microglial cells (P2ry12, Cx3cr1, Tmem119), mural cells (Rgs5, Acta2, Pdgfrb), neuroblasts (Sox11, Sox4, Tubb3), neutrophils (Lyz2, Csf3r, Cxcr2), oligodendrocytes (Olig1, Mbp, Plp1), T cells (Cd3d, Cd2, Trac), and tanycytes (Scn7a, Prdx6) (Figure 1A,B). Similarly, in the striatum, these cell clusters were annotated into 11 cell types. The marker expression patterns for these populations were consistent with those observed in the midbrain and aligned with established cell type signatures in the literature (Figure 1C,D). In the midbrain and striatum, we counted the number of cells of each cell type between the A53T group and the control group. The distribution of each cell type in the brain region was relatively balanced, and the cell production of A53T mice and wild type mice was similar (Figure 1E,F).

FIGURE 1.

Single‐cell profiling of midbrain and striatum. (A) UMAP plots of midbrain with cells colored by cell types. (B) Heatmap of mean expression in the midbrain, representative markers across different cell types. (C) UMAP plots of striatum with cells colored by cell types. (D) Heatmap of mean expression in the striatum. (E, F) Heatmap with number of cells of each cell type in the midbrain and striatum. (G, H) Density plots with the distribution of PD risk gene scores within each cell type. *p < 0.05, **p < 0.01.

To evaluate the relevance of this model to PD, we retrieved human PD‐risk genes from DisGeNET, ¹⁴ selecting those with a Score_gda >0.1 (Table S1), and mapped them to mouse orthologs by biomaRt. ¹⁵ Using Ucell ⁹ scoring, we quantified PD‐risk gene activity across cell types. Both midbrain and striatum of A53T mice showed significantly higher PD‐risk scores than controls. Notably, the increase was most pronounced in astrocytes, microglia, and oligodendrocytes, whereas neuronal populations showed relatively modest changes (Figure 1G,H). These results indicate that glial compartments exhibit the strongest enrichment of PD‐associated transcriptional signatures in the A53T model, aligning with the view that early and cell‐type‐specific responses in glial lineages may contribute to disease‐related vulnerability.

3.2. Cellular states and the molecular characteristics in the midbrain of A53T mice

To interpret the cell states and molecular characteristics in the midbrain of PD, we calculated the PD‐related likelihood scores across all cells in our dataset by using MELD ¹⁰ (Figure 2A). Utilizing VFC analysis, we identified the heterogeneity within cell populations and detected subset of cells exhibiting differential responses to PD. Compared with wild‐type mice, A53T samples showed a higher number of PD‐related likelihood cells in astrocytes and choroid plexus cells, while the overall number of neuroblasts and microglia was lower (Figure 2B). The PD‐related likelihood scores across the three glial cell types varied significantly, consistent with PD‐associated enrichment and depletion across specific cell subpopulations.

FIGURE 2.

Analysis of cell abundance and gene expression changes in the midbrain. (A) PD relative likelihood estimated by MELD. (B) Jitter plots showing PD relative likelihood for each cell type. Gray dots indicate the fold change in cell numbers between conditions within each cell type. (C–E) Vertex frequency clustering identifies PD‐enriched subpopulations in microglia (C), oligodendrocytes (D), and astrocytes (E). (F–H) Enrichment analysis of marker genes for PD‐enriched subclusters: Cluster 3 in microglia (F), cluster 3 in oligodendrocytes (G), and cluster 2 in astrocytes (H).

To systematically evaluate the pathological impact of PD on the three principal glial cell lineages, we initially quantified the sample‐associated relative likelihood for each subpopulation of glial cells using advanced computational approaches (Figure 2C–E). Our rigorous comparative analysis demonstrated significant PD‐associated enrichment patterns in specific subpopulations, in particular, the third microglial subcluster (MG3; Figure 2C), the third oligodendrocyte subcluster (OL3; Figure 2D), and the second astrocytic subcluster (AS2; Figure 2E).

To elucidate the functional relevance of the signature genes defining these PD‐enriched subclusters, we performed gene set enrichment analysis (GSEA) using EnrichR. Comparison of MG3 signature genes with upregulated genes in other microglial subclusters showed striking enrichment for pathways including oxidative phosphorylation, mTORC1 signaling, unfolded protein response, and ribosome biogenesis (Figure 2F). For OL3, comparison with other oligodendrocyte subclusters revealed enrichment of cell death pathways, including the P53 pathway, apoptosis, and ferroptosis (Figure 2G). As key regulators of neuronal integrity, oligodendrocytes contribute to clearing damaged cells and proteins and participate in neuroinflammatory responses. The enrichment of cell death pathways in OL3 suggests this subpopulation may contribute to neuronal loss and PD pathogenesis. For AS2, enrichment analysis indicated significant association with pathways related to dopaminergic synapse and glutamatergic synapse (Figure 2H). This pattern is consistent with a PD‐associated transcriptional state that may affect synaptic function in the model.

Across most cell types, we observed significant differences in transcriptional regulation between the A53T and the WT samples. In total, we identified 1329 up‐regulated and 2128 down‐regulated DEGs in the midbrain. Under pathological conditions, the transcriptomic changes in neuroblasts and neutrophils are not significant, with only two upregulated genes and no downregulated genes (Table 1). It is noteworthy that we observed that the DEGs in ECs, ependymal cells, microglial cells, mural cells and oligodendrocytes have higher scores (Figure 3A), indicating that the introduction of the SNCA gene with the A53T mutation has a more significant impact on these cell types.

TABLE 1.

The number of DEGs that are upregulated and downregulated in each cell type.

Cell type	Midbrain tissue			Striatal tissue
	Up	Down	Total	Up	Down	Total
Astrocytes	428	210	638	17	17	34
Choroid plexus cells	–	–	–	141	14	155
B cells	0	0	0	–	–	–
ECs	384	187	571	664	77	741
Ependymal cells	2	2	4	176	118	294
Erythrocytes	–	–	–	0	0	0
MPs	289	142	431	22	14	36
Microglial cells	984	311	1295	105	150	255
Mural cells	396	151	547	85	154	239
Neuroblasts	189	78	267	2	0	2
Neutrophils	5	3	8	2	0	2
Oligodendrocytes	626	350	976	99	241	340
T cells	19	34	53	4	23	27
Tanycytes	–	–	–	3	0	3

FIGURE 3.

Analysis of the expression levels and functions of differentially expressed genes in specific cell types of the midbrain. (A) Expression patterns of upregulated and downregulated DEGs in the midbrain. (B–E) Boxplots of expression levels of selected DEGs in the indicated cell types (red, A53T; blue, WT). (F) Barplot of DEG by counting the number of upregulated or downregulated genes (y‐axis) and the number of cell types in which each gene shows differential expression (x‐axis). (G) Functional enrichment of DEGs significantly upregulated in at least three cell types.

Neuroblasts in A53T mice showed elevated expression of Clu, Prnp, and S100a8 (Figure 3B). Clu is linked to Alzheimer's disease, ²³ , ²⁴ Prnp is a cellular receptor for soluble amyloid beta (Aβ) oligomers, ²⁵ and S100a8 is involved in neuroinflammation and can co‐accumulate with α‐syn. ²⁶ The significantly upregulated genes in astrocytes include Slc1a2, Slc1a3 and Glul (Figure 3C). These genes encode key glutamatergic transporters and are involved in glutamate metabolism. Their dysregulation may contribute to altered glutamate homeostasis and calcium‐dependent gliotransmitter release in astrocytes. Microglial cells revealed the downregulation of Fth1, which encodes ferritin light/heavy chain, implicated in neurodegenerative diseases due to iron accumulation. Zfp36, encoding Tristetraprolin (TTP), indicated neuroinflammation (Figure 3D). For oligodendrocytes from A53T mice, Adipor2, Slc38a2, and Sgk1 were upregulated (Figure 3E). Adipor2 is associated with neuroinflammation, Slc38a2 encodes a glutamatergic transporter and receptor, and Sgk1 is a serine/threonine protein kinase involved in various cellular processes. ²⁷ These findings provide insights into gene expression patterns in various cell types associated with PD. ²⁸ By calculating the reproducibility of DEGs in different cell types, we found that most DEGs are cell type specific (Figure 3F). The functional enrichment analysis of genes that were significantly upregulated in three or more cell types revealed that pathways related to protein processing in endoplasmic reticulum, antigen processing and presentation, mineral absorption, lysosome, apoptosis, and IL‐17 signaling pathway were enriched (Figure 3G).

3.3. Cellular states and molecular signatures in the striatum of A53T mice

Given the critical role of striatal dopamine depletion in PD pathogenesis, we characterized cellular states and molecular alterations in the striatum of A53T mice to examine how the A53T mutation affects cellular states. Using the MELD algorithm, we quantified the relative likelihood score of PD association per cell, then applied VFC to analysis of cell enrichment. This revealed a significant decrease in the proportion of PD‐associated astrocytes and mural cells, alongside an increase in PD‐associated neuroblasts. Substantial heterogeneity in PD‐likelihood scores within glial cell populations indicated the existence of cell groups with abundant PD pathogenic genes and cell groups lacking such genes (Figure 4A,B). To resolve these subpopulations, we applied VFC analysis. This classified glial cells into distinct subclusters (Figure 4C–E). Notably, subcluster 0 within astrocytes (Figure 4C), oligodendrocytes (Figure 4D), and microglia (Figure 4E) showed predominant enrichment in WT mice.

FIGURE 4.

Analysis of cell abundance and gene expression changes in the striatum. (A) PD relative likelihood estimated by MELD. (B) Jitter plots showing PD relative likelihood for each cell type. Gray dots indicate the fold change in cell numbers between conditions within each cell type. (C–E) Vertex frequency clustering identifies PD‐enriched subpopulations in astrocytes (C), oligodendrocytes (D), and microglia (E). (F–H) Enrichment analysis of marker genes for WT‐enriched subclusters: Cluster 0 in astrocytes (F), cluster 0 in oligodendrocytes (G), and cluster 0 in microglia (H).

Compared to other astrocyte subclusters, subcluster 0 exhibited significant downregulation of genes involved in key inflammatory pathways (TNF‐α signaling via NF‐κB, lysosomal pathways). Its overall expression profile remained consistent with core midbrain astrocyte signatures (Figure 4F). Similar to findings in the midbrain, pathways related to apoptosis and the P53 pathway were impacted. Additionally, subcluster 0 showed significant alterations in pathways governing protein biosynthesis (ribosome, spliceosome, proteasome, lysosome), consistent with known roles of protein metabolism dysregulation in PD. MAPK signaling pathway activation was also observed (Figure 4G). MAPK activation, potentially triggered by axon guidance proteins and neuroinflammatory factors, ²⁹ , ³⁰ can drive cellular stress, inflammation, dopaminergic neuron vulnerability, and influence α‐syn aggregation/clearance and cellular homeostasis. ³¹ In addition to shared enrichment in mTORC1 signaling with subcluster 0 of oligodendrocytes, this subcluster also showed specific enrichment for pathways directly linked to Parkinson's disease (Figure 4H).

DEG analysis identified 3322 upregulated and 1468 downregulated DEGs in the striatum. Cell types significantly impacted by PD in the striatum mirrored those in the midbrain (Table 1). A heatmap of the top 70 DEGs demonstrated consistent expression patterns across nearly all cell types (Figure 5A). Ccl family genes (induced by IL‐1β, TNF‐α, LPS, or viral stimuli) were broadly elevated, ³² indicating systemic neuroinflammatory activation. Upregulation of ubiquitin genes Ubb and Ubc suggested dysregulated protein homeostasis. ³³ The downregulation of mt‐Rnr1 and mt‐Rnr2 expression indicates an abnormal disruption of mitochondrial function, leading to oxidative stress and damage to mitochondrial DNA, which is associated with the pathogenesis of PD. The number of cell types with an upregulated or downregulated intersection of DEGs (Figure 5A) indicates that more cell types are involved in the pathological progression related to neuroinflammation and mitochondrial dysfunction in PD.

FIGURE 5.

Analysis of the expression levels and functions of differentially expressed genes in specific cell types of the striatum. (A) Expression patterns of upregulated and downregulated DEGs within each cell type of the striatum. (B–D) Boxplots showing expression levels of selected DEGs in the indicated cell types (red, A53T; blue, WT). (E) Number of upregulated and downregulated genes (y‐axis) as a function of the counts of cell types in which differential expression (x‐axis) is observed. (F) Enrichment analysis of DEGs significantly upregulated in at least three cell types.

Given that mitochondrial complex I dysfunction and reduced NADH dehydrogenase activity are strongly implicated in PD, ³⁴ this alteration suggests potential metabolic stress. Reflecting observations in A53T mice, ³⁵ we detected decreased expression of Mt1 and Mt3 (Figure 5B), indicating diminished neuroprotective capacity associated with the disease. Significant upregulation of chemokines Ccl3 and Ccl4 (MIP‐1α) was observed in the A53T group (Figure 5C), supporting their established role in driving neuroinflammation in PD. ³³ Expression of genes involved in mitochondrial complex I assembly (Ndufb9, Ndufa13, Ndufs6) was significantly increased in oligodendrocytes (Figure 5D). Consistent with the analysis results of the characteristics of cells in the midbrain region, most of the DEGs are cell‐type specific (Figure 5E). KEGG enrichment analysis of striatal DEGs highlighted significant involvement in ribosome, antigen processing and presentation, protein processing in endoplasmic reticulum, phagosome, apoptosis, lysosome, and ferroptosis pathways (Figure 5F).

3.4. Transcriptional regulation of disease‐specific cells in the midbrain

SCENIC analysis identified 396 transcription factor (TF) regulons, with 120 showing significant upregulation in A53T mice versus 75 in WT controls (FDR<0.05, Figure 6A). Cell‐type specificity analysis revealed, PD‐associated regulons primarily targeted glial cells and ECs, and WT‐enriched regulons predominantly regulated ECs and choroid plexus cells (Figure 6A). Regulon Specificity Score (RSS) analysis identified two TF regulators with pronounced activity changes: RORB activity significantly increased in astrocytes of A53T mice (Figure 6B), and FOXC1 activity markedly elevated in the same cell types (Figure 6C). Using RcisTarget, we defined target genes of FOXC1 regulons. Comparative expression analysis confirmed significant dysregulation of these targets in A53T versus WT (Figure 6D,E). Hypergeometric testing between PD‐risk genes (DisGeNET) and targets of 120 PD‐upregulated regulons identified 34 TFs with significant overlap (p < 0.05, Figure 6F). Cell‐type mapping revealed JUNB and MEF2C regulons preferentially modulated microglial transcription, SOX10 regulons targeted oligodendrocytes, and FOXC1 regulons governed ECs and mural cells (Figure 6F). This further validates the cell types associated with PD obtained from gene set scoring analysis, abundance difference analysis, and differential gene analysis.

FIGURE 6.

Comprehensive analysis of midbrain‐specific transcriptional regulation in PD. (A) Pie charts showing the proportions of upregulated regulons in WT (left) and PD (right) across midbrain cell types, with the central bar indicating the number of specific differentially regulated regulons. (B, C) Ranking of regulons in astrocytes (B) and ECs (C) based on regulon specificity scores. (D) Expression levels of RORB‐regulated genes in astrocytes. (E) Expression levels of FOXC1‐regulated genes in ECs. (F) Overlap between target genes of regulons with significantly increased activity in PD and PD risk genes assessed by a hypergeometric test. Thirty‐four regulons with p < 0.05 are shown, including their activity (heatmap) and corresponding log‐transformed p values (bar plot).

3.5. Transcriptional regulation of disease‐specific cells in the striatum

SCENIC analysis identified 357 transcription factor regulons in the striatum. Comparative activity assessment revealed 101 regulons significantly upregulated in A53T mice (FDR <0.05) and 110 regulons significantly enriched in WT controls (Figure 7A). Cell‐type‐resolved regulon activity mapping demonstrated that 15.2% of A53T‐upregulated regulons targeted astrocytes, and 18.5% preferentially modulated oligodendrocytes (Figure 7A). To investigate the specific regulons of astrocytes and oligodendrocytes and their regulatory roles in PD, we sought to evaluate the cell‐specific regulons activation states. The RSS chat of astrocytes and oligodendrocyte regulators shows that DBX2 and KLF13 are upregulated in the WT group and the PD group, respectively (Figure 7B,C). Target genes of DBX2 and KLF13 regulons (annotated via RcisTarget) exhibited concordant expression changes. DBX2 targets showed preferential upregulation in WT astrocytes (Figure 7D), and KLF13 targets demonstrated significant induction in A53T oligodendrocytes (Figure 7E). Hypergeometric testing between PD‐risk genes and targets of 101 A53T‐upregulated regulons identified 26 TFs with significant overlap (p < 0.05; Figure 7F). Cell‐type‐resolved activity profiling of these TFs revealed distinct regulatory networks across striatal populations (Figure 7F). Through the enrichment of neuroglial‐specific regulatory elements, the cell type‐specific transcription factor activity, and the convergence of PD risk genes, we confirmed the central role of astrocytes and oligodendrocytes in the pathological changes of Parkinson's disease in the striatum. This transcriptional dysregulation pattern indicates that glial cells are involved in the pathological progression of PD.

FIGURE 7.

Analysis of PD transcriptional regulation in the striatum region. (A) Pie charts showing the proportions of upregulated regulons in WT (left) and A53T (right) across midbrain cell types, with the central bar indicating the number of specific differentially regulated regulons. (B, C) Ranking of regulons in astrocytes (B) and oligodendrocytes (C) based on regulon specificity scores. (D) Expression levels of DBX2‐regulated genes in astrocytes. (E) Expression levels of KLF13‐regulated genes in oligodendrocytes. (F) Overlap between target genes of regulons with significantly increased activity in A53T and PD risk genes assessed by a hypergeometric test. Twenty‐six regulons with p < 0.05 are shown, including their activity (heatmap) and corresponding log‐transformed p values (bar plot).

3.6. The alterations in the intercellular signal network of the midbrain region in A53T mice

To further investigate the intercellular communication changes induced by the A53T mutation, we applied CellChat ²² (version 5.0) to infer cell–cell communication networks between cell types associated with PD. We explored the communications in signaling pathways and ligand‐receptor pairs in A53T and WT cells separately, and compared the differences in the communication networks between A53T and WT. Cell–cell interaction revealed enhanced intercellular signaling in A53T mice versus WT controls, increased ligand‐receptor pair interactions (p < 0.01), and augmented overall communication strength (Figure 8A). The analysis of cell‐type‐specific signal transduction revealed that all cell types exhibited an increase in signal emission/reception capabilities, but neuroblast cells showed a decrease in signal output (Figure 8B). The common and condition‐specific signaling pathways shared under different conditions are shown in Figure 8C. Subsequently, we further compared the information flow for each signaling pathway between A53T group and WT group, and found that some pathways, such as the EPHB and EPHA pathways, were turned off in A53T mice, while the THBS, ncWNT, PROS, BMP, and PTPRM pathways were turned on only in A53T mice (Figure 8D).

FIGURE 8.

Cell–cell communication analysis in the midbrain region. (A) Number and strength of intercellular ligand–receptor interactions in WT and A53T. (B) Heat maps showing interaction number (left) and interaction strength (right) between cell populations in WT and PD. Blue indicates decreased communication in PD, whereas red indicates increased communication relative to WT. (C) Identification and visualization of conserved and condition‐specific signaling pathways in WT and PD. Top signaling pathways (blue labels) are enriched in WT, whereas bottom pathways (red labels) are enriched in PD. (D) Heat maps showing outgoing and incoming signaling patterns across cell types in WT and PD. (E, F) Signaling network of the SEMA3 pathway in WT (E) and PD (F). (G, H) Signaling network of the SEMA4 pathway in WT (G) and PD (H). (I, J) Signaling networks of THBS (I) and PROS (J) pathways. Arrows are colored by the source cell type, and line thickness represents communication strength.

We analyzed the common and condition‐specific signaling pathways that exist under different conditions (Figure 8E–J). The core function of the SEMA3 signal is to cause the collapse of the growth cone and the rejection of axons in neurons. This effect is particularly active in PD. The core of the SEMA3 signal is usually located on endothelial cells, but in PD models, the participation of neuroblasts, choroid plexus cells, and fibroblasts has increased (Figure 8E,F). We also discovered another SEMA4 signal, namely SEMA4D, which encodes a membrane‐bound protein and is involved in processes such as neuronal migration and synapse formation. Mutations or abnormal expression of SEMA4D may harm neuronal function, thereby leading to the occurrence of PD. Different from the physiological conditions, in the A53T mouse model, SEMA4D shows a completely different change pattern (Figure 8G,H). The SEMA4D signal is also emitted from microglia, and tanycytes, choroid plexus cells, and ependymal cells receive these signals and enhance the signal strength of oligodendrocytes in PD (Figure 8G,H). In the specific signaling pathway exclusively activated in PD, the THBS signaling pathway mainly involves signal emission from fibroblasts to interact with various cells (Figure 8I). In the PROS signaling pathway, both fibroblasts and microglial cells emit signals that are received by oligodendrocytes (Figure 8J).

3.7. The alterations in the intercellular signal network of the striatum region in A53T mice

CellChat analysis demonstrated significant augmentation of striatal quantity and strength signaling in A53T mice (Figure 9A). Except for ependymal cells, astrocytes, and neuroblasts, all cell types show an increase in the intensity of outgoing or incoming communication. The analysis of cell type‐specific signal showed that most cell types exhibited enhanced signal output/input capacity. However, the signal reception ability of ependymal cells decreased by 41.2%, the input signal of astrocytes decreased by 33.7%, and the signal integration ability of neuroblasts decreased by 29.8% (Figure 9B). We further compared the information flow for each signaling pathway between A53T and WT and found that some pathways, such as the IFN‐II, CEACAM, and DESMOSOME pathways, were turned off in A53T, while the BMP, WNT, and TWEAK pathways were turned on only in A53T group (Figure 9C). The BMP level remained persistently elevated in the urine of patients with LRRK2 G2019S, R1441G/C, and VPS35 D620N mutations. ³⁶ , ³⁷

FIGURE 9.

Cell–cell communication analysis in the striatum region. (A) Number and strength of intercellular ligand‐receptor interactions in WT mice and A53T mice. (B) Heat maps showing interaction number (left) and interaction strength (right) between cell populations in WT and A53T mice. Blue indicates decreased communication in A53T, whereas red indicates increased communication relative to WT. (C) Identification and visualization of conserved and condition‐specific signaling pathways in WT and A53T. Top signaling pathways (blue labels) are enriched in WT, whereas bottom pathways (red labels) are enriched in A53T. (D, E) Signaling network of the CCL pathway in WT (D) and A53T (E). (F, G) Signaling network of the MIF pathway in WT (F) and A53T mice (G). Arrows are colored by the source cell type, and line thickness represents communication strength.

To focus on alterations most likely to be relevant to PD, we distinguished the signaling pathways that involve CCL and the MIF pathway. In the CCL WT signaling pathway network, MPs are essential signal emitters, while T cells and mast cells are significant signal emitters. Microglia and neutrophils are the primary signal receptors. Under pathological conditions, the interactions between cells increase, leading to enhanced intercellular communication. Furthermore, the signal reception of neutrophils and microglial cells is enhanced (Figure 9D,E).

In the MIF signaling pathway network, the primary signal receivers are mural cells, ependymal cells, ECs, and B cells among all cells of WT group. However, the overall communication among all cells was enhanced, with particular activation of neuroblasts, astrocytes, and oligodendrocytes, leading to the occurrence of neuroinflammation in the A53T group (Figure 9F,G).

3.8. Regulation of RORB and FOXC1 in the human midbrain of Parkinson's disease

To characterize the single‐cell transcriptome features of the midbrain in human Parkinson's disease patients and controls, we reanalyzed 41 435 high quality cells of midbrain from 5 PD donors and 6 controls, which were obtained from the gene expression omnibus database (https://www.ncbi.nlm.nih.gov/geo, GSE157783). We used the CCA integration method of Seurat to remove batch effects, align PD and control cells into a shared low‐dimensional space and reduce technical variation across samples. After integration, PD and control cells were well mixed in the UMAP space without gross group separation, indicating that signals of markers are better displayed at the level of specific cell types (Figure 10A). We annotated these single cells based on the canonical cell type‐specific genes from a published study, ⁶ they were identified as 12 cell types, including astrocytes (AQP4), CADPS2+ neurons (CADPS2), dopaminergic neurons (DaNs, TH), endothelial cells (CLDN5), ependymal (FOXJ1), excitatory (SLC17A6), GABA (GRIK1), inhibitory (GAD2), microglia (CD74), oligodendrocytes (MOBP), oligodendrocyte precursor cells (OPCs, VCAN), and pericytes (PDGFRB) (Figure 10B,C). Using Seurat, we assessed the expressing‐cell proportions and expression levels of canonical marker genes across cell types. The expected enrichment patterns of these markers provide support for the accuracy of our cell‐type annotations (Figure 10D).

FIGURE 10.

Human midbrain scRNA‐seq validates glial and endothelial transcriptional remodeling in PD. (A) UMAP of all 41 435 cells from 5 PD and 6 control donors; red dot indicated PD group, and blue dot indicated control group. (B) Feature plots of representative markers in the UMAP used for annotation, the deeper the orange color, the higher the gene expression level. (C) Cell‐type annotation of the single cell atlas, different colors represent different cell types. (D) Bubble plot summarizing marker expression level (color) and the percentage of expressing cells (dot size) across annotated types. (E) Boxplot with target genes of RORB in astrocytes comparing PD (red) and controls (blue); ***p < 0.001, **p < 0.01. (F) Boxplot with target genes of FOXC1 in endothelial cells comparing PD and controls; ***p < 0.001.

Based on our previous findings in mice, we extracted astrocytes and endothelial cells from the human midbrain, and then compared them between the two groups to screen for conserved genes across species. In human astrocytes, numerous target genes of RORB showed differential expression between PD and control donors. Several of these changes displayed the same directionality as in our mouse dataset, with GJA1 significantly overexpressed in Parkinson's patients. In endothelial cells, FOXC1 target genes also exhibited coordinated alterations in PD (Figure 10E). Similarly, in human endothelial cells, multiple FOXC1 target genes displayed PD‐associated increases, PTMS and ITGB1, in particular, showed elevated expression in both human and mouse samples (Figure 10F). The recurrence of these gene‐level shifts in two independent species supports a conserved remodeling of RORB‐ and FOXC1‐regulated pathways in glial and vascular compartments of the PD midbrain.

4. DISCUSSION

Our integrated scRNA‐seq analysis of midbrain and striatal tissues in A53T mice reveals multifaceted cellular perturbations underlying PD pathogenesis. By combining transcriptional profiling with advanced computational approaches, we demonstrated that the alterations in PD related cell subset, transcriptional reprogramming, and intercellular communication in astrocytes, microglia, and oligodendrocytes play a centric role in the pathological progression of PD caused by the A53T mutation. Astrocytes, oligodendrocytes, and microglia exhibit the strongest enrichment for PD‐risk genes (DisGeNET) and show subtype‐specific alterations in both brain regions (Figures 1, 2, 3, 4, 5). Variance‐based clustering identified PD‐enriched glial subclusters with distinct functional impairments. Including microglial subclusters showed upregulated oxidative phosphorylation and UPR pathways, oligodendrocyte subclusters displayed activation of p53‐mediated cell death pathways, and astrocytic subclusters were enriched for synaptic signaling pathways (Figures 2 and 4). Regulatory network analysis revealed 34 dysregulated TFs in the midbrain and 26 in the striatum, with RORB (astrocytes/ECs) and FOXC1 (ECs/mural cells) emerging as key PD‐associated regulators (Figures 6 and 7). CellChat analysis detected enhanced global signaling strength, PD‐specific pathway activation (SEMA3/4, THBS, CCL, MIF), and neuroinflammatory rewiring via glial‐neuronal crosstalk (Figures 8 and 9). Furthermore, we analyzed the expression of target genes in astrocytes and endothelial cells of the human midbrain from PD patients and controls; some genes may serve as candidate targets for diagnosis or treatment (Figure 10).

We note that while the previous studies ⁶ identified glial alterations in human substantia nigra, our work provides complementary insights into early‐stage transcriptomic changes in the A53T mouse model, particularly the enrichment of PD‐risk genes in specific glial subclusters. Our findings position neuroglial cells as central orchestrators of PD pathology. The early activation of astrocytes, observed through both abundance shifts and pathway enrichment, aligns with human evidence of astrocytic glutamate dysregulation precipitating dopaminergic neuron loss. ³⁸ Notably, the PD‐enriched astrocytic subcluster AS2 demonstrated upregulated glutamatergic synapse pathways, potentially explaining the excitotoxicity mechanisms in our model. The discovery of Rorb as a key dysregulated TF warrants special attention. Interestingly, RORB has also been implicated in Alzheimer's disease (AD), where it regulates synaptic homeostasis and amyloid‐β clearance. ³⁹ This raises the possibility that Rorb dysfunction may represent a common mechanism across neurodegenerative disorders, potentially involving shared transcriptional programs in glial cells. Our study newly implicates its role in glial transcriptional reprogramming during PD pathogenesis. Similarly, FOXC1 dysregulation may disrupt neurovascular integrity, given its established function in endothelial development. ⁴⁰ Single‐cell sequencing of postmortem brain tissues donated by PD patients (Figure 10) further suggests that dysregulation of these TFs ultimately impacts the pathways governing neuroinflammation, protein homeostasis, and metabolic stress.

Unlike conventional cluster‐level comparisons, our MELD‐based perturbation mapping quantified disease effects along continuous transcriptional manifolds. ¹⁰ This approach revealed subtle but biologically significant subpopulations masked in standard analyses, including the ferroptosis‐prone OL3 oligodendrocyte subcluster. ⁴¹ The integration of VFC clustering with pathway enrichment further enabled functional annotation of these cryptic disease states. The identification of PD‐specific signaling hubs (CCL, MIF) provides mechanistic grounding for emerging neuroinflammatory targets. Our data corroborate human studies showing CCL2 elevation in substantia nigra ⁴² while demonstrating cell‐type‐resolved communication networks (Figure 9D–G). Particularly compelling is the microglia‐to‐ependyma SEMA4D axis – a previously unrecognized pathway that potentiates oligodendrocyte activation. Our regional analysis reveals both shared and distinct glial responses. MG3 and OL3 were PD‐enriched in both midbrain and striatum, indicating a coordinated pan‐regional response. Regional‐specific signal transduction was also observed. For instance, midbrain microglia showed enhanced SEMA4D signaling, while striatal astrocytes exhibited downregulation of TNF‐α/NF‐κB pathways. These results indicate that the activation of SEMA4D in the midbrain glial cells may spread along the substantia nigra‐striatum pathway, thereby affecting the targets of the striatum. On the contrary, alterations in the specific striatal CCL and MIF pathways may exacerbate neuroinflammation.

The model used in this study achieved the overexpression of the human mutant SNCA gene through the regulation of a heterologous promoter. Thirteen‐week‐old A53T mice did not exhibit significant loss of tyrosine hydroxylase (TH)‐positive dopaminergic neurons, nor did they show motor dysfunction related to PD. However, the single‐cell sequencing results of this model indicated that early transcriptional changes were observed before the onset of neurodegenerative lesions. Which helps fill the gap in early‑stage pathological data of human Parkinson's disease. It is important to emphasize that the identification of disease‐associated subclusters (MG3, OL3, AS2) is descriptive and does not establish causation. Further functional experiments are required to test whether these subclusters actively drive pathology. For example, lineage tracing using *Cx3cr1‐CreER* mice could determine whether MG3 microglia originate from resident microglia or infiltrating monocytes. Additionally, conditional knockout of the transcription factor Rorb in oligodendrocytes could directly assess its contribution to PD‐related transcriptional changes. Sex differences in PD risk, onset, and glial‐mediated neuroinflammation are well established. However, whether the enrichment of PD risk genes in glial subpopulations and the identified transcriptional factor networks are gender‐dependent remains unclear. This is one of the limitations of this study. Future studies should determine the gender‐specific characteristics of these glial dynamics.

5. CONCLUSION

This study establishes a comprehensive atlas of A53T‐driven pathology, highlighting three critical dimensions: neuroglial subpopulations as primary carriers of PD genetic risk, RORB/FOXC1‐mediated transcriptional dysregulation, and reconfigured intercellular communication networks. While confirming established PD mechanisms, such as mitochondrial dysfunction and proteostasis failure, ⁴³ , ⁴⁴ we newly implicate semaphorin signaling and TF‐specific regulatory cascades in disease propagation. These findings provide a framework for developing cell‐type‐targeted interventions to prevent the progression of PD.

AUTHOR CONTRIBUTIONS

Binqing Qin: Conceptualization; data curation; writing – original draft. Zichu Fu: Formal analysis; visualization. Xuanxuan Zou: Methodology; validation. Senmao Chai: Conceptualization; software. Jingjing Weng: Data curation; software. Puqing Wang: Funding acquisition; methodology. Xiaodong Sun: Funding acquisition; project administration; writing – review and editing. Ming Sang: Funding acquisition; project administration; writing – review and editing.

FUNDING INFORMATION

This work was supported by Hubei Province Innovation Development Joint Fund (Xiangyang) (2025AFD038), Hubei Science and Technology Project (2023BCB140), Hubei Provincial Technology Innovation Project (2025CFC027), Faculty Development Grants from Hubei University of Medicine (2025QDJZR02), and Natural Science Foundation of Hubei Provincial Department of Education (2026AFB019).

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

ETHICS STATEMENT

All animal procedures followed the China Public Health Service Guide for the Care and Use of Laboratory Animals and were approved by the Ethical Committee of Biomedical Basic Research of Xiangyang No. 1 People's Hospital (approval number: XYYYE20220030).

Supporting information

Table S1. PD‐risk genes.

DATA AVAILABILITY STATEMENT

The data presented in the study are deposited in the NCBI Sequence Read Archive repository, accession number GSE306642.