Single-Cell Analyses Reveal Necroptosis's Potential Role in Neuron Degeneration and Show Enhanced Neuron-Immune Cell Interaction in Parkinson's Disease Progression

Parkinson's disease (PD) is a common neuron degenerative disease among the old, characterized by uncontrollable movements and an impaired posture. Although widely investigated on its pathology and treatment, the disease remains incompletely understood. Single-cell RNA sequencing (scRNA-seq) has been applied to the area of PD, providing valuable data for related research. However, few works have taken deeper insights into the causes of neuron death and cell-cell interaction between the cell types in the brain. Our bioinformatics analyses revealed necroptosis-related genes (NRGs) enrichment in neuron degeneration and selecting the cells by NRGs levels showed two subtypes within the main degenerative cell types in the midbrain. NRG-low subtype was largely replaced by NRG-high subtype in the patients, indicating the striking change of cell state related to necroptosis in PD progression. Moreover, we carried out cell-cell interaction analyses between cell types and found that microglia (MG)'s interaction strength with glutamatergic neuron (GLU), GABAergic neuron (GABA), and dopaminergic neuron (DA) was significantly upregulated in PD. Also, MG show much stronger interaction with NRG-high subtypes and a stronger cell killing function in PD samples. Additionally, we identified CLDN11 as a novel interaction pattern specific to necroptosis neurons and MG. We also found LEF1 and TCF4 as key transcriptional regulators in neuron degeneration. These findings suggest that MG were significantly overactivated in PD patients to clear abnormal neurons, especially the NRG-high cells, explaining the neuron inflammation in PD. Our analyses provide insights into the causes of neuron death and inflammation in PD from single-cell resolution, which could be seriously considered in clinical trials.


Introduction
Parkinson's disease (PD) is the second most reported neurodegenerative disorder, afecting millions of people worldwide every year and is getting more and more prevalent these years [1][2][3][4].It is a neurodegenerative disorder characterized by the death of dopaminergic neurons in the brain, leading to motor and cognitive impairments.Te exact cause of the disease is not fully understood, where both genetic and environmental factors are believed to play a role.
Based on former research, several cellular pathways have been implicated in the pathogenesis of PD, including oxidative stress, mitochondrial dysfunction, infammation, and protein misfolding and aggregation.First, oxidative stress has been shown to contribute to dopaminergic neuron death and the accumulation of α-synuclein, a protein that is a hallmark of the disease [5].Second, mitochondrial dysfunction can lead to increased production of ROS as well as impaired energy production and calcium signaling, all of which contribute to cell death [6].Moreover, many misfolded proteins such as α-synuclein are abundant in the brain and comprise the pathological hallmark of PD such as Lewy bodies.Te misfolding and aggregation of these misfolded proteins are thought to contribute to the death of dopaminergic neurons in PD [7].
Single-cell RNA sequencing (scRNA-seq) reveals the presence and abundance of RNA at a given moment from a single-cell resolution and has already become a powerful method to understand the pathology of diverse diseases, such as cancers, Alzheimer's disease (AD), and infammatory bowel disease (IBD) [8][9][10].Its high resolution allows for the identifcation of cell subpopulations and their gene expression signatures, enabling a deeper understanding of cellular heterogeneity and the complex regulatory networks that underlie cellular functions and diseases.One of the main applications of scRNA-seq is in the feld of developmental biology, where it has been used to study the diferentiation of stem cells into various cell lineages [11].scRNA-seq has also been used in neurobiology to study the heterogeneity of brain cells and their role in diseases.For example, a study by Lake et al. used scRNA-seq to identify the molecular diversity of diferent cell types in the brain and to analyze the efects of aging on the gene expression profles of these cells [12].scRNA-seq has also been applied in immunology, infectious diseases, and regenerative medicine.As technology continues to improve and become more accessible, it is expected to have a major impact on a wide range of research felds.Among these research studies, many works used scRNA-seq to build an atlas of diseases, producing a lot of data worth further analyzing.
We analyzed published snRNA-seq data from PD patients' samples in this work.3 groups of degenerative neurons were identifed.Diferentially expressed genes were analyzed on these cells followed by GO and KEGG enrichment.We found a wide presence of necroptosis in these cells and then carried out cell interaction analysis to study the interactions of MG with these cells.In a word, our work reveals the importance of necroptosis in PD progression from the single-cell aspect and predicts a novel interaction pattern between MG and necroptosis neuron cells.

Data Collection.
We acquired midbrain single-nucleus RNA sequencing (snRNA-Seq) data from NCBI (https://www.ncbi.nlm.nih.gov/) using the accession code GSE140231 [13], which include seven health human donors' substantia nigra, GSE157783 [14], which is composed of fve PD patients' and six health donors' midbrain, and GSE148434 [15], consisting of six health samples and six PD samples, all of these samples are substantia nigra (Table S1).Because substantia nigra is a part of midbrain, we called our integrated dataset as midbrain dataset.We also defned the healthy or normal sample as term "control," which means no clinical symptoms related to PD (absence of movement disorder).

Data Preprocessing and Integration.
Raw sequencing data were processed using CellRanger-6.1.2.Te expression matrix was then imported into R (4.2.1) using Seurat (4.9.9) [16].We used Seurat package default parameters if no parameters are declared.We removed cells that had fewer than 200 genes or over 4000 genes, as well as those with over 3% of mitochondria genes expressed in them, to ensure that only cells in good condition and not doublets were included.After separately normalizing the data, we merged the Patient and Control expression matrixes together by using Seurat build in function called "FindIntegrationAnchors" with selecting the top 2000 highly variable genes (HVGs).We then scaled the matrix and performed principal component analysis (PCA) using the top 2000 HVGs.We chose the top 15 PCs for downstream cluster identifcation and visualization.We identifed 7 clusters under 0.04 resolution using the "FindClusters" function.All the steps taken after importing the data into R were processed using Seurat internal functions.
Using the Seurat function "FindAllMarkers," we identifed the top 30 markers for each cell type.We then used the clusterProfler (4.6.2) [18,19] and genome-wide annotation database to perform GO enrichment analysis between clusters.

Diferential Expression Gene (DEG) Analysis in Sub-Clusters between Control and Patient.
To analyze a certain cell type, we generated a volcano plot of diferentially expressed genes (log2Foldchange >0.58, adjusted P Value <0.01) using the Seurat function "FindMarkers."(Table S2) We then logarithmically transformed the normalized expression matrix by column to make the diferences between the two groups clearer in the heatmap.KEGG enrichment analysis was conducted using clusterProfler [18,19].KEGG dotplot and cnetplot were generated using the R package enrichplot.To separate cells in one cluster into two groups with high or low expression of necroptosis-associated genes, we calculated the mean expression level for each gene and assigned cells with more than 60% of genes expressed above the mean as "high" and those with less than 60% as "low." We then performed PCA of the two groups using the R package "FactoMineR" (2.8) [20] and visualized the results using the R package "factoextra" (1.0.7).

Protein and Protein Interaction Analysis.
We found transcription factors which exist in top 200 diferent expressed genes by using human transcription factors database, humanTFDB [21].Ten, we used STRING [22] database to obtain the protein and protein interaction information with minimum required interaction score >0.4.To make the result clearer and more readable, we used Cytoscape (3.9.1) [23] for further processing.Default mode was used in the analysis.

2
Parkinson's Disease 2.6.Cell-Cell Interaction Analysis.We used the R package CellChat (1.6.1)[24] to quantitatively infer and analyze intercellular communication networks from the data.CellChat employs network analysis and pattern recognition approaches to predict major signaling inputs and outputs for cells and how those cells and signals coordinate for functions.CellChat classifes signaling pathways and delineates conserved and context-specifc pathways through manifold learning and quantitative contrasts.First, we found identify overexpressed ligands or receptors by using functions called identifyOverExpressedGenes and identifyOver-ExpressedInteractions.Ten, we computed the communication probability via computeCommunProb function.If the interactions appeared in few cells (usually <10), we removed them by flterCommunication function.In the next step, we used computeCommunProbPathway to predict the pathways according to the probable interations.Finally, we calculated the aggregated cell-cell communication network by using aggregateNet function.

Single-Cell Landscape of Midbrain in PD Patients.
Te main lesions of Parkinson's disease occur in the midbrain.Terefore, in this study, we utilized published singlenucleus sequencing data of midbrain.A total of 14,011 midbrain cells were collected and clustered into seven groups using unsupervised clustering methods.Based on cell markers reported in previous studies, we identifed seven distinct cell types, including GABAergic neuron (GABA), glutamatergic neuron (GLU), oligodendrocyte (OLG), oligodendrocyte precursor cell (OPC), astrocyte (AST), microglia (MG), and dopaminergic neuron (DA) (Figure 1(a)).Te annotated cell markers are listed in Figure 1(b).We performed GO functional enrichment analysis to identify the main pathways in each of these cell clusters (Figure 1(c)).Te enriched pathways corresponded to the known pathways of each cell type, such as myelination of OLG and homeostasis of MG.Tis demonstrates the accuracy and reliability of our data processing and cluster annotation.
Next, we investigated the changes in cell ratios between PD patient and control samples.Our fndings revealed that three major cell types, including GLU, GABA, and DA, were signifcantly reduced in PD, while MG were the most boosted cells in PD (Figure 1(d)).Tese intriguing observations piqued our curiosity to study the underlying reasons for these cell proportion changes, and the potential cellular interactions involved.

DEG Analysis and Pathway Enrichment Show Upregulated Necroptosis in GLU Degeneration
. GLU is the largest cluster in the midbrain and is signifcantly reduced in PD development.Terefore, we frst focused our analysis on the GLU group to understand the mechanisms driving GLU degeneration in PD progression.We analyzed the diferentially expressed genes (DEGs) between GLU cells from patient samples and control normal samples (Figure 2(a)).
We found that some heat shock proteins, such as HSP90AB1, were signifcantly upregulated in the PD samples, indicating a higher burden of misfolded abnormal proteins.We also plotted the signifcant DEGs in single-cell resolution and classifed them as upregulated or downregulated genes (Figure 2(b)).To further elucidate the drivers of GLU degeneration, we performed KEGG enrichment analysis for the upregulated genes in PD (Figure 2(c)).Except the reactive oxygen species (ROS) pathway, the necroptosis pathway was also enriched in PD, indicating the necroptosis's role in GLU degeneration.We identifed individual genes enriched in the ROS, necroptosis, and PD pathways (Figure 2(d)), and the PD pathway still showed upregulated genes associated with abnormal protein folding.

GLU of Diferent Necroptosis States Shows Distinct Distribution in PD and Control Samples.
In greater detail, we selected the top six enriched genes in the necroptosis pathway and compared their expression levels between patient and control samples (Figure 3(a)).Subsequently, we classifed GLU into two subtypes based on their necroptosis state, utilizing the genes we had selected.Cells that exhibited lower expression levels for at least four out of the six genes, relative to the average, were identifed as NRG-low subtype, while the remaining cells were identifed as NRG-high subtype.Te distribution of these subtypes was displayed on a heatmap (Figures 3(b) and 3(c)).We employed PCA to cluster the cells and analyzed the relative proportions of each subtype.Our analysis revealed that NRG-high GLU cells were predominant in the PD brain, whereas NRG-low cells were the primary subtype observed in control samples (Figures 3(d) and 3(e)).Tese fndings suggest that necroptosis may contribute to GLU degeneration.

3.4.
Necroptosis in DA and GABA Degeneration.In addition to GLU, the other two primary clusters of neurons in the midbrain-dopaminergic neurons (DA) and GABAergic neurons (GABA)-also exhibited a signifcant reduction in the PD model, which is consistent with the former notion that degeneration of DA neurons is a leading cause of PD [25].To further investigate changes in these two cell types, we conducted a DEG analysis of DA and GABA (Figures 4(a), S1A).Surprisingly, we also identifed necroptosisrelated genes, such as HSP90AB1, FTH1, SLC25A4, and PPIA.KEGG analysis also revealed enrichment of the necroptosis pathway (Figures 4(b), S1B).Subsequent classifcation of DA or GABA neurons based on NRG expression showed a higher proportion of NRG-high cells in the PD group (Figures 4(c) and 4(d), S1C and S1D).Tese fndings suggest that necroptosis is a common occurrence in degenerated neurons during PD progression.

CellChat Analysis Shows Enhanced Cell Interactions between MG and Neuron Cells in PD.
Microglia are resident macrophage cells of the brain that serve as the primary immune defense system in the central nervous system.Given Parkinson's Disease  Parkinson's Disease the signifcant number of neuron degeneration observed in PD progression, we postulated that microglia were activated in PD and began attacking neurons, potentially leading to neuron infammation.To investigate this, we conducted CellChat analysis to study the interactions between neurons and microglia.We discovered that the total interactions were markedly stronger in patient than in control (Figure 5(a)).Additionally, both incoming and outgoing interactions of microglia were higher, indicating an active MG state in PD (Figure 5(b)).We then analyzed the potential molecular interactions between microglia and the three groups of degenerated cells (Figures 5(c) and 5 (d)).We detected signifcantly more possible molecular interactions, indicating enhanced microglia-neuron interactions in PD samples.Among these molecular pathways in PD, CX3CL1 was specifcally upregulated in DA, which may play a role in recruiting immune cells.Furthermore, the phagocytosis pathway, such as Gas6-Mertk, was specifc to PD-DA with microglia.Tese results suggest that, in PD progression, DA becomes dysfunctional and releases chemokines, such as CX3CL1, to recruit microglia and is more susceptible to attack by microglia through phagocytosis.

NRG-High Cells Show Stronger Interaction with MG.
To further address the necroptosis's efect on neuron infammation, we examined the interactions between NRGhigh neurons and MG.NRG-high cells exhibited a signifcantly greater number of total cell interactions with MG (Figure 6(a)).MG's outgoing interactions with NRG-high cells were particularly higher compared to NRG-low cells (Figure 6(b)).Detailed interaction pathways were analyzed and the larger number of molecular interactions was identifed between MG and NRG-high cells (Figure 6(c)).In addition, we discovered a potential pathway, CLDN11, which was specifcally associated with NRG-high cells and microglia (Figure 6(d)).Tis gene is not well-studied and may play an important role in the enhanced MG-necroptosis cells interactions in the PD pathological process.Furthermore, the CX3C pathway was also stronger between NRGhigh cells and MG (Figure 6(e)), indicating that cells undergoing necroptosis have a greater ability to attract microglia.Notably, DA did not show the same pattern here, which may result from the small cell number in the NRGlow DA.Collectively, these fndings showed robust necroptosis cell-MG interactions in PD tissues.

Microglia Functions Were Enhanced in PD Patients.
Upon activation, microglia upregulate CD86, TMEM119, CD11B, and CD45, and they exert their cytotoxic efects on target cells through the release of various substances, including TNF-α, glutamate, cathepsin B, superoxide, or nitric oxide (Figure 8) [26,27].To further validate the microglia's excessive activation in the progression of Parkinson's disease

Discussion
Necroptosis is a type of programmed cell death that has gained signifcant attention in recent years due to its distinct molecular mechanisms and implications in various physiological and pathological processes.Unlike apoptosis, which is a well-known and regulated form of cell death characterized by controlled dismantling of the cell, necroptosis is an infammatory and lytic form of cell death.It represents an alternative cell death pathway that can be activated when cells are unable to undergo apoptosis, often triggered by various stress signals, such as infection, infammation, or cellular damage [28][29][30].
Necroptosis has been implicated in various pathological conditions, including ischemic injuries, viral infections, and infammatory disorders.Its dysregulation has been linked to the pathogenesis of conditions such as Alzheimer's disease, and infammatory bowel diseases.Consequently, understanding the molecular mechanisms underlying necroptosis may hold therapeutic potential for developing novel treatments to modulate cell death and infammation in these disorders [31].However, it is not widely investigated how necroptosis will infuence Parkinson's disease.Some clinical studies have observed signifcant increases in necroptosis-related genes (NRGs) in postmortem examinations of Parkinson's disease (PD) patients, such as MLKL [25].However, our understanding of the expression profles of these genes across diferent cell types within the brain and how their expressions can infuence neighboring cells remains limited.Although a recent study reported the enrichment of NRGs in PD samples using bulk RNA-seq data, it lacked deeper insights into cell interaction changes and robust verifcation at the single-cell level [32].
In our study, we comprehensively examined the expression patterns of NRGs among midbrain neurons in PD samples from a single-cell perspective and identifed some NRGs upregulated in PD patients.Among these genes, GLUL encodes glutamate-ammonia ligase which is important in RIPK3-mediated metabolic enzyme regulation [33].It helps the production of reactive oxygen species (ROS) and enhanced necroptosis [34].HSP90 increases MLKL oligomerization and plasma membrane translocation to trigger necroptosis and is required for TNF-induced necroptosis [35,36].FTH1 and FTL1 encode the heavy and light chain of ferritin, respectively.Te acccumulation of ferritin can elicit oxidative infammation and NF-κB-TNFα pathway, which may further trigger necroptosis [37].PPIA (cyclophilin A) acts as an immune infammatory mediator that secretes proinfammatory cytokines induced by oxidative stress [38][39][40][41].SLC25A4 encodes the adenine nucleotide translocator-1 (ANT1), which is involved in metabolism via the regulation of ATP/ADP release from mitochondria and in regulated cell death as part of the mitochondrial permeability transition pore [42].We further used these genes to study the cell subtypes distribution within the brain and found NRG-low subtype's replacement by NRG-high subtypes.Our cell-cell interaction analyses showed more activated microglia and more interaction of microglia with the degenerative neurons.Tese fndings shed light on the signifcant role that necroptosis may play in neuron degeneration and microglia (MG)-mediated immune clearance.Additionally, we identifed novel cell interaction pathways and central transcription factors in PD that warrant further investigation.
However, our analysis has potential limitations due to the limited sample number and heterogeneity between samples.Also, the interaction predicted by CellChat may not

Conclusion
We analyzed published data on PD patients and found widespread necroptosis genes upregulated in three main degenerative neuron types in the midbrain.Our cell-cell interaction analysis showed that MG, the immune cells of the brain, have much stronger interactions with the neurons undergoing necroptosis.All these results implied that the main neurons in the midbrain turn to a dysfunctional state through a necroptosis-related mechanism, which activates MG to clear the dysfunctional neurons in PD, leading to neuron degeneration and infammation.We also identifed CLDN11 as a potential interaction pathway specifc between MG and NRG-high cells.It may play a role in mediating the enhanced cell interactions in PD.In a word, our results suggest that the necroptosis cell-microglia axis may play a crucial role in PD progression.Our results showed broad application prospects in drug development targeting necroptosis [33].

Figure 1 :
Figure 1: Single cell landscape of midbrain in PD patients.(a) Cell clusters in midbrain between control and PD patients' samples.(b) Cell markers for each cluster.(c) GO enrichment of the clusters.(d) Clusters' proportion changes between PD and control samples.MG ratio increases in PD samples while GLU, GABA, and DA ratios decrease.(CTR: control ratio, PDR: PD ratio).

Figure 2 :
Figure 2: DEG analysis and pathway enrichment of GLU.(a) Volcano plot showing DEGs of GLU between PD and control samples.(b) Heatmap of DEGs in single cells.(c) KEGG enrichment of GLU DEGs.(d) Genes enriched in ROS, necroptosis, and PD pathways.

3. 7 .
Analysis of Transcription Factors Driving the Necroptosis in Degenerated Neurons.We then conducted an analysis of the transcriptional regulation in the PD samples to elucidate the mechanisms of neuron necroptosis.Diferentially expressed genes (DEGs) and transcription factors (TFs) were compared to identify the distinctly expressed transcription factors (Figure 7(a)).Additionally, we performed a proteinprotein interaction analysis to examine the downstream genes regulated by these TFs (Figure 7(b)).Our fndings indicated the involvement of 23 central TFs in the degeneration process in PD, controlling approximately 100 downstream genes.Furthermore, most of the enriched NRGs were regulated by these TFs, with LEF1, TCF4, and GATA3 playing signifcant roles in the regulatory network (Figure 7(c)).

Figure 3 :
Figure 3: GLU subcluster distribution.(a) NRGs expression between PD and control samples.(b) Heatmap showing two subclusters distribution of GLU in PD samples.(c) Heatmap showing two subclusters of GLU in control samples.(d) PCA of two GLU NRG subclusters in PD samples.(e) PCA of two GLU NRG subclusters in control samples.

Figure 4 :Figure 5 Figure 5
Figure 4: DEG analysis and pathway enrichment of GABA and DA.(a) Volcano plot showing DEGs of DA and GABA between PD and control samples.(b) Heatmap of DEGs enriched in the pathway.(c) Heatmap of DA NRG subclusters distribution.(d) Heatmap of GABA NRG subcluster distribution.

Figure 6 :
Figure 6: CellChat of NRG subtypes and MG.(a) Total interactions between MG and NRG subtypes.(b) MG outgoing interactions with NRG subtypes showing higher interactions between MG and NRG-high groups.(c) Interaction pathways between MG and NRG subtypes.(d) CLDN pathway between MG and NRG subtypes.(e) CX3C pathway of subtypes.

Figure 7 :Figure 8 :
Figure 7: Analysis of hub transcription factors driving the necroptosis in degenerated neurons.(a) Regulated transcription factors were obtained from DEGs; 23 TFs were identifed.(b) PPI network of the TFs and DEGs in degenerated neurons.NRGs were centered.(c) PPI network of the NRGs and their related TFs in degenerated neurons.TFs were centered.