Introduction
In face of aging world population and the absence of effective cures, central nervous system (CNS) diseases pose a significant economic burden on society. Alzheimer’s disease (AD), Parkinson’s disease (PD), and multiple sclerosis (MS) are chronic and progressive diseases of the CNS, characterized by the loss of neurons in the brain or the spinal cord [
1]. Although a variety of efforts at the molecular level have attempted to elucidate the basic biological pathologies contributing to these diseases, the principal causes and cures for these diseases remain elusive, which hinders the discovery of disease-modifying therapies.
AD is the most common neurodegenerative disease, identified by the presence of extracellular aggregates of amyloid β (Aβ) peptides and intraneuronal tau neurofibrillary tangles in the brain. Patients with AD primarily exhibit the impairments in short-term memory and cognitive processing [
2]. PD is the second most common neurodegenerative disease, and is clinically defined by the symptoms of akinesia, rigidity, and tremor. Cytosolic Lewy bodies aggregated by α-synuclein and the loss of dopaminergic neurons in the substantia nigra pars compacta are the major neuropathological features [
3]. MS is a progressive neuroinflammatory disease with distinct lesion characteristics in the cortical grey versus subcortical white matter, and neurodegeneration at chronic stages. Clinically, the symptoms of MS are diverse, involving the impairments of movement, sensation, vision, and cognition [
4]. Although AD, PD, and MS differ in many clinical and pathological aspects, it is possible that they share cross-molecular features. Growing evidence indicates that brain endothelial dysfunction might play a significant role in the neurobiology of AD, PD, and MS [
5,
6], which causes derangement of the mitochondrial machinery [
7], suppressing glutamate reuptake by astrocytes and resulting in glutamate-mediated toxicity [
8]. Immune cell infiltration from the periphery into the CNS as well as the inflammatory responses mediated by reactive astrocytes and by activated microglia in the CNS also have been implicated in AD, PD, and MS [
9]. Elucidating the shared and distinct molecular crosstalk could provide a biological basis for the treatments of AD, PD, and MS, which remains controversial in the field.
The establishment of single-nucleus RNA-sequencing (snRNA-seq) databases over the last decade permits global bioinformatics analyses of gene expression in different cells. Transcriptomic profiling, through snRNA-seq of patient-derived tissues, can address confounding by cellular composition, providing previously unattainable insight into cell-type-specific transcriptomic pathology [
10,
11]. Existing data resources have not yet been fully exploited to understand the causal disease pathways in AD, PD, and MS. With this in mind, we integrated several existing human snRNA-seq datasets to gain a comprehensive view of the cell-type-specific transcriptional changes of these diseases in the CNS. Our results showed the shared and distinct transcriptional changes in multiple cell types among AD, PD, and MS. We hypothesized that
HSPB1 may be the core molecule of the shared pathological mechanism of the three diseases, which is the result of the induction by the blood–brain barrier (BBB) in response to cellular stress, providing insight into the nervous system diseases with unique pathogenic processes. Additionally, arctigenin has shown potential as a therapeutic drug for AD, PD, and MS.
Discussion
AD, PD, and MS are CNS diseases that differ clinically. Previous studies have noted the importance of the common pathological changes in these diseases, such as small vessel disease [
5] and inflammation [
9]. However, the molecular crosstalk between these three diseases remains largely unknown. This study aimed to identify the key molecules which regulate the pathogenic pathways in AD, PD, and MS.
We observed that the dataset obtained from the white matter of MS patients had a higher proportion of glial cells compared to neuronal populations. The nature of the tissue and the inherent heterogeneity of the cell type present pose challenges in achieving a completely unbiased representation. The result of the higher proportion of glial populations may lead to an overemphasis on glial-specific gene expression patterns while potentially masking or downplaying certain neuronal-specific signals. Furthermore, the altered proportions in the dataset could impact the functional interpretation of the results. It is important to consider the potential contributions of glial cells to the observed gene expression patterns and their implications for the underlying biological processes.
There was significant overlap in differentially expressed genes and pathways between AD and PD in our study. In the GSEA analysis, the genes of AD and PD were co-enriched into two identical pathways, chaperone-mediated protein folding and protein folding, which has previously been shown to be associated with the abnormal protein folding in AD and PD. In our study, we found that
HSPA1A, which encodes the major heat shock protein in the HSP70 family, was upregulated in certain cell types, including oligodendrocytes, OPCs, pericytes, and astrocytes in AD, as well as oligodendrocytes, pericytes, and endothelial cells in PD. Our findings align with prior research on AD and PD. Some studies showed that HSPA1A is upregulated in human entorhinal cortex samples and layer III pyramidal cells of AD, contributing to protein folding abnormalities, and altered synaptic transmission [
28,
29]. A proteomics analysis of extracellular vesicles isolated from cerebrospinal fluid in AD also revealed elevated expression of HSPA1A compared to the mild cognitive impairment and control groups [
30]. Single-cell transcriptomics also uncovered the upregulation of HSPA1A (HSP70) in endothelial cells of patients with PD, which was further confirmed in PD patients’ blood specimens and peripheral blood mononuclear cells [
31,
32]. Furthermore, HSPA1A protein regulates the processing and generation of APP (amyloid precursor protein) and the production and aggregation of Aβ [
33,
34]. HSPA1A exerted a significant restorative effect on neuronal morphology and functional status in the temporal cortex and hippocampal region in transgenic mouse [
35]. The upregulation of HSPA1A can also play a preventive and decelerating role in PD-like neurodegeneration through its chaperone activity, which effectively inhibits α-synuclein aggregation and microglia activation [
36,
37]. Taken together, our findings strongly support that HSPA1A could serve as a potential therapeutic target for AD and PD.
In our study, MS-related biological processes were mainly related to the transport of proteins to the endoplasmic reticulum (ER) (GO:0006613, GO:0072599, GO:0045047, GO:0006614), which is responsible for the proper folding and processing of polypeptide chains into functional proteins in cells. ER stress occurs when misfolded or unfolded proteins accumulate in the ER due to exogenous or endogenous factors. In MS, the destruction of the BBB leads to an escalation of pro-inflammatory damage, leading to cell damage. This is followed by oxidative damage and ER stress [
38], leading to apoptosis or repair of CNS cells. These results are largely controlled by the unfolded protein response [
39], which is consistent with our hypothesis. The preliminary findings indicate a strong association between ER function and MS, highlighting the need for additional investigation in this area.
The most prominent discovery arising from our analysis was the consistent upregulation of
HSPB1 (
HSP27) across among AD, PD, and MS. An increase of HSPB1 protein in the chronic active lesions of MS brains was reported [
40], as we found in the present study. HSPB1 phosphorylation also was reported to be elevated in MS and AD [
41,
42].
HSPB1 codes for a small heat shock protein that probably maintains folding competence in denatured proteins [
43,
44]. In response to environmental stresses such as heat shock, HSPB1 protein is upregulated [
45]. The molecular chaperone activity of HSPB1 may regulate a broad spectrum of biological processes, including the phosphorylation of neurofilament proteins and their transport along axons [
46]. In AD, amyloid plaques and neurofibrillary tangles (composed of hyperphosphorylated tau) are present extracellularly and intracellularly, respectively [
47]. A transgenic mouse model of AD showed HSPB1 localization in plaques [
48]. By crossing the APPswe/PS1dE9 mouse model with the mouse model of HSPB1 overexpressed, spatial learning and electrophysiological parameters improved significant lye [
49]. HSPB1 can delay, but not prevent fibril formation by interacting with hyperphosphorylated tau [
50]; this interaction is enhanced with increased tau phosphorylation [
51]. Moreover, overexpression of HSPB1 reduces the cellular deposition and cytotoxicity of α-synuclein in PD [
52]. HSPB1 binds to the surface of α-synuclein fibrils, thereby reducing their hydrophobicity [
53]. Based on the cumulative evidence from previous studies, we put forth the hypothesis that the observed upregulation of HSPB1 in our study strongly suggests its protective function in mitigating the formation of pathological protein aggregates, particularly in response to toxic stimuli or stressful conditions associated with neurological disorders.
We observed that the upregulation of
HSPB1 occurred in different cells in the three diseases.
HSPB1 exhibited upregulation in endothelial cells in PD and MS, as well as in astrocytes, pericytes, OPCs, and excitatory cells in AD. Initially, we attributed it to a non-specific cellular response. However, we have observed that the upregulation of
HSPB1 is consistently present in the constituent cells of the BBB, including endothelial cells, pericytes, and astrocytes, across AD, PD, and MS. This finding indicates that HSPB1 plays a role in the disruption of the BBB in three diseases. A previous study suggested that the endothelium-targeted overexpression of HSPB1 ameliorated BBB disruption after ischemic brain injury [
54,
55]. The immune response of HSPB1 was also found to be present in the temporal cortex of patients with epilepsy and was mainly confined to vascular walls and glial cells [
56]. As a selective physical barrier, the BBB plays a protective role in maintaining the environmental balance in the brain. One hypothesis we consider is that conceptual BBB damage may play an important role in the pathogenesis of AD, PD, and MS. BBB dysfunction contributes to the onset and progression of AD [
57,
58], PD [
59], and MS [
60] as an upstream or downstream events. However, clinical studies in this area have shown that the BBB is less pronounced in patients with AD, PD, and MS. The BBB damage is more associated with cerebrovascular disease [
61]. The recent interest in AD immunotherapy has provided a reassessment of the BBB damage, as intact BBB is a potential barrier to effective treatment of anti-amyloid immunoglobulin [
62]. At the same time, dysfunctional BBB may be a risk factor for immune-mediated toxicity, including autoimmune encephalitis. Another reason for revisiting the role of BBB is that the current understandings of the pathogenesis of AD, PD, and MS are incomplete. Our findings suggest that
HSPB1 is initially induced in regions where BBB is disrupted in response to cellular stress, particularly in endothelial cells and pericytes associated with blood vessels, as well as astrocytes. This endogenous response may serve as a protective mechanism against vascular and BBB injury. Further investigation is required to understand the mechanism by which HSPB1 maintains BBB structure and function.
PPI analysis of DEGs revealed that the upregulated members of the HSP family may engage in interactions with specific proteins within various cells implicated in neurodegenerative diseases. In our study, the pericytes of AD brains appeared to be a crucial focal point. HSP family genes were upregulated in pericytes and interacted with the upregulated ribosomal protein family in AD brains. Different ribosomal proteins within the ribosome bear distinct roles and functions, collaborating to ensure accurate and efficient protein synthesis. Interestingly, previous study has also reported the elevated expression of ribosomal proteins in the capillaries of AD brains, which is remarkably like what we observed in pericytes [
63]. The researchers proposed that ribosomal function is augmented in the cerebral blood vessels of AD patients, leading to an aberrant protein translation network. The clinically used anti-AD drugs donepezil and tacrine were reported to inhibit ribosome biosynthesis [
64]. These viewpoints suggested the significant involvement of ribosome biosynthesis in the pathogenesis of AD and the importance of it as a therapeutic target. Increased ribosome biosynthesis also occurred in the microglia of MS, which indirectly interacted with
HSPB1. This suggests the potential involvement of microglial protein synthesis in the pathogenesis of MS. In the endothelial cells of PD, we observed a concurrent upregulation and interaction between the HSP family and the inflammation-related gene
IRF1.
IRF1 is involved in immune responses, cell apoptosis, and tumorigenesis. Previous study has found that IRF1 protein is upregulated in α-Syn overexpressed SH-SY5Y cells and the substantia nigra of A53T α-Syn transgenic mice. Furthermore, α-Syn overexpression facilitates the translocation of IRF1 from the cytoplasm to the nucleus, mediating neuroimmune responses [
65]. Our research demonstrated the upregulation of
IRF1 and HSP family genes in endothelial cells, revealing that besides immune cells, protein processing and immune-inflammatory responses in endothelial cells are also noteworthy. In this study, although we focused on the HSP family genes and their interacting genes, their prominent roles in distinct cell types appeared to vary across different diseases. This observation suggested that different cells may play diverse roles in the context of various diseases.
The transcriptional regulatory analysis identified several modules that are highly related to the pathogenesis of AD, PD, and MS, mainly in microglial, oligodendrocytes, and pericytes. We found that multiple top 20 specific TFs in microglia, oligodendrocytes, and astrocytes were shared across the AD, PD, and MS. Combined with the hub genes identified above, we found that these shared TFs can regulate multiple hub genes, including PSAP and CNTN2 in AD, and CHI3L1 in PD. The identification of these shared TFs has opened new avenues for studying their underlying pathological mechanisms.
We further identified that arctigenin could be a potential therapeutic drug for AD, PD, and MS. Arctigenin has also been shown having neuroprotective effect in vivo and in vitro of AD and PD models [
24,
26]. Arctigenin could obviously attenuate the decrease of cell survival rates in SH-SY5Y cells with PD phenotypes by acting against cell apoptosis through the decrease of Bax/Bcl-2 and caspase-3, and by reducing the surplus reactive oxygen species production and downregulating the mitochondrial membrane potential [
26]. These results shed light on the potential therapeutic value of arctigenin as a treatment option for AD, PD, and MS.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.