Introduction
Astrocytes represent a highly heterogeneous glial population in the central nervous system (CNS) with diverse functions related to maintaining the integrity of the blood–brain barrier and the cerebral homeostasis of fluids, ions, pH, and neurotransmitters [
1,
2]. Astrocytes contribute to protecting from glutamate-mediated excitotoxicity by removing glutamate from the synaptic cleft and preventing hyperstimulation of postsynaptic receptors [
3,
4]. They also closely interact with neurons as well as blood vessels thereby regulating CNS metabolism [
5] and, by storing and release of glycogen, transfer of glucose metabolites and generation of lactate, dynamically support neural functions [
6‐
8]. Furthermore, astrocytes regulate lipid metabolism such as cholesterol efflux and oxidation of fatty acids [
9,
10]. Besides regulating essential processes in the CNS, astrocytes have been described to closely interact with microglia and respond dynamically to cytokines secreted [
11,
12]. This astro-microglial crosstalk was shown to result in conversion of astrocytes into a neurotoxic or neuroprotective state dependent on the stimuli (Fig.
2A) [
11,
13].
For decades, expression of glial fibrillary acidic protein (GFAP), an astrocytic-specific marker in the brain, is used to define astrocytic identity and reactivity [
1]. Expression of GFAP, however, varies among astrocyte subpopulations and is region-dependent; for example, grey matter regions demonstrate lower GFAP levels compared to white matter and hippocampus [
14]. Therefore, using ubiquitously expressed astrocyte markers such as aldehyde dehydrogenase 1 family member L1 (ALDH1L1), aldolase C, or glutamine synthetase (GS) are important for a comprehensive characterization of astrocytic populations [
1].
Rapid methodological developments resulted in the identification of transcriptional astrocytic expression profiles in physiological as well as pathological states [
15‐
17]. These assignments led to the translation from structural phenotyping to region- and context-specific functions thereby identifying astrocytic subpopulations with a spectrum of potential effector functions ranging from neurotoxic properties with impaired autophagy, synapse elimination, and dopamine regulation [
18,
19] to neuroprotective signatures promoting CNS recovery and repair by secreting transforming growth factor β (TGF-β) and brain derived neurotrophic factor (BDNF), angiopoietin-1 (ANG1), and expression of connexin-43 (CX43) [
20‐
22].
Multiple system atrophy (MSA), a rare sporadic oligodendroglial α-synucleinopathy, is a rapidly progressing neurodegenerative disease with an average survival rate of 6–9 years [
23‐
25]. Clinically, MSA is subdivided into a parkinsonian (MSA-P) and cerebellar phenotype (MSA-C) accompanied by an early and severe autonomic failure [
26]. In MSA-P, the dorsolateral putamen, caudate nucleus, and substantia nigra display pronounced neuronal loss whereas cortical areas such as the precentral gyrus are less affected [
27]. Reactive astrogliosis has been shown in the putamen of MSA patients demonstrating an inverse correlation between reactive astrocytes and the distance to glial cytoplasmic inclusions [
28,
29].
To shed more light into astrocytic reactivity in MSA, we first examined post mortem prototypical MSA-P brain regions to determine reactive astrogliosis in the cortex, the striatum, and the substantia nigra. Next, to test if astrocyte reactivity observed in post mortem tissue is recapitulated in a mouse model of MSA, we took advantage of the MBP29-hα-syn mouse model.
In 2005, Shults and colleagues generated a mouse model overexpressing human α-synuclein (hα-syn) in oligodendrocytes under the control of a myelin-basic-protein (MBP) specific promotor (MBP29-hα-syn) [
30]. These mice start to develop progressive motor deficits at the age of 10 weeks, prematurely die between 20 and 30 weeks of age, and show a severe neuronal loss accompanied by a widespread demyelination and region-specific microgliosis matching important functional and neuropathological features of MSA [
31,
32]. Due to the important role of astrocytes to alter neurodegenerative and inflammatory processes [
33‐
35] there is an urgent need to comprehensively characterize astrocytic subpopulations in MSA and its respective model for identifying potential targets suitable for future interventional strategies. After confirming a similar astrocytic phenotype in MBP29-hα-syn mice, we made use of an optimized state-of-the-art isolation protocol using magnetic activated cell sorting (MACS) approach to decipher the transcriptional profile of astrocytes in CNS regions differentially affected by the oligodendroglial synucleinopathy.
Material and methods
Human post mortem brain tissue
Post mortem cortex, putamen and substantia nigra of MSA-P patients and age- and sex-matched controls (each n = 4, Table
1) were obtained from the Netherlands Brain Bank (NBB), Netherlands Institute for Neuroscience, Amsterdam (open access:
http://www.brainbank.nl). MSA cases were clinically diagnosed and neuropathologically confirmed as MSA-P according to the second consensus criteria (Table
1; [
36]). Paraffin-embedded tissue of the cortex, the striatum, and the substantia nigra were sectioned at a thickness of 5 µm.
Table 1
Overview of MSA-P patients with clinical characteristics and controls
MSA-P | F | 66 | 485 | 88 | 1005 |
MSA-P | M | 67 | 370 | 77 | 1376 |
MSA-P | F | 67 | 435 | 97 | 1244 |
MSA-P | F | 59 | 400 | 54 | 1102 |
Control | F | 77 | 340 | – | 1111 |
Control | F | 60 | 450 | – | 1240 |
Control | M | 51 | 465 | – | 1450 |
Control | M | 55 | 435 | – | 1393 |
Animals
Transgenic MBP29-hα-syn mice overexpressing human α-syn under the control of the murine MBP promoter and corresponding non-transgenic littermates (NTG) serving as controls were anesthetized and sacrificed at an age of 4 weeks. For immunofluorescence staining, animals were perfused using 4% paraformaldehyde (PFA) and 0.9% sodium chloride solution according to the European and National Institute of Health guidelines for humane treatment. Afterwards, brains were transferred into 4% PFA for 2–4 h and subsequently stored in 30% sucrose solution. Whole forebrains were sagittally sectioned at a thickness of 40 µm and stored in cryoprotection solution (25% 0.2 mol/L phosphate buffer) at − 20 °C until further staining. For expression analysis of proteins and mRNA, by Western blot (WB) and quantitative PCR (qPCR), respectively, mice were perfused using phosphate-buffered saline (PBS). After the removal of the brain, the cortices and striata were subsequently micro-dissected, snap-frozen in liquid nitrogen, and stored at −80 °C until homogenization. For the isolation of astrocytes and oligodendrocytes, mice were sacrificed, brains were removed, dissected as described above and stored at 4 °C in Dulbecco’s Phosphate Buffered Saline containing MgCl2 and CaCl2 (D-PBS) until further processing for MACS.
Immunohistochemistry of human post mortem tissue
Immunohistochemistry on formalin-fixed and paraffin-embedded tissue was performed following a previously published protocol [
31]. In brief, sections at a thickness of 5 µm were deparaffinized. After microwave antigen retrieval using citrate buffer, sections were stained using the REAL EnVision Detection System peroxidase/DAB+, Rabbit/Mouse (Dako, # K5007) according to the manufacturer’s protocol. The kit uses 3,3′-diaminobenzidine tetrahydrochloride (DAB) as a chromogen. Finally, sections were counterstained using hematoxylin. Primary antibodies and staining conditions were as follows: antigen retrieval using citrate buffer (0.1 M) for 30 min followed by anti-GFAP (Agilent Cat# M0761, RRID:AB_2109952, 1:100) incubation overnight. As positive control, an intracerebral metastasis of a breast carcinoma with an adjacent reactive border zone was used. Analyzed areas of each region is given in Additional file
1: Table S2.
Immunofluorescence staining of human post mortem brain tissue
Brain sections were deparaffinized by heating at 60 °C for 1 h followed by incubation in xylene and re-hydrated with a descending ethanol series (100%/95%/70%). Antigen retrieval was performed using Citrate-EDTA buffer (10 mM Citric acid, 2 mM EDTA, 0.05 Tween-20, pH 6.2) at 95 °C for 15 min. Afterwards, sections were permeabilized with 1% Triton-X100 in PBS for 30 min. Sections were incubated with primary antibodies at 4 °C for 72 h: glial fibrillary acidic protein GFAP (Agilent Cat# GA524, RRID:AB_2811722, 1:500) and excitatory amino acid transporter 2 (Novus Cat# NBP1-20136, RRID:AB_2190752, 1:500). Primary antibodies were followed by the incubation of secondary antibodies: donkey anti-rabbit Alexa Fluor 488 (Molecular Probes Cat# A-21206, RRID:AB_2535792, 1:500) and donkey anti-mouse Cy3 (Millipore Cat# AP192C, RRID:AB_92642, 1:500) at 4 °C overnight. Nuclei were counterstained using 4′,6-diamidino-2-phenylindole (DAPI). Imaging was performed using the fluorescence Observer microscope (Carl Zeiss Microscopy GmbH, Jena, Germany).
Immunofluorescence staining of murine forebrain
For immunofluorescence staining, brain sections of six animals per genotype were utilized. Remaining cryoprotection solution was removed by rinsing with PBS. For improvement of antibody penetration, sections were incubated in 0.3 M glycine, 0.2% Triton-X100, and 10% DMSO in dH
2O at room temperature (RT) for 1 h. Furthermore, unspecific antibody binding was blocked using 0.3% donkey serum, 0.2% Triton-X100 in PBS at 37 °C for 1 h. For astrocyte labeling, primary antibodies were incubated at 4 °C for 72 h: GFAP (Abcam Cat# ab4674, RRID:AB_304558, 1: 1.000), ALDH1L1 (Abcam Cat# ab87117, RRID:AB_10712968, 1:500), and S100β, (Abcam Cat# ab52642, RRID:AB_882426, 1:200). Forebrain sections were rinsed using PBS and subsequently incubated with secondary antibodies at RT for 1 h: anti-chicken Alexa 488 (Jackson ImmunoResearch Labs Cat# 703-545-155, RRID:AB_2340375, 1:500) and anti-rabbit Alexa 647 (Jackson ImmunoResearch Labs Cat# 711-605-152, RRID:AB_2492288, 1:500). Secondary antibody was rinsed and nuclei were counterstained using DAPI. Imaging was performed using the Axio Observer fluorescence microscope (Carl Zeiss Microscopy GmbH, Jena, Germany). Images were processed and analyzed applying ImageJ and Fiji software [
37,
38].
Isolation of murine astrocytes and oligodendrocytes
Adult astrocytes and oligodendrocytes were isolated using MACS. Generation of single cell suspensions, myelin and red blood cell removal was performed using the Neural Tissue Dissociation Kit (P) (130-092-628, Miltenyi Biotec). For isolation of cortical and striatal glial cells, olfactory bulbs and hindbrain were removed and the respective regions of five animals were pooled. After adding papain to samples in C-tubes (130-093-237, Miltenyi Biotec), homogenization of the samples was performed with program 37C_ABDK_01 of the gentleMACS™ Octo Dissociator with Heaters (130-096-427, Miltenyi Biotec). To obtain a sufficient amount of RNA, five animals of mixed sex per genotype were pooled for each sample used for subsequent bulk RNA sequencing (for 3n in total: 15 animals/genotype per region). After removal of debris and red blood cells, Fc-receptors (FcR) were blocked by incubation of single cell suspension using FcR Blocking Reagent (mouse, 130-092-575, Miltenyi). Astrocytes were magnetically labelled using ACSA-2 MicroBeads (130-097-678, Miltenyi Biotec) according to the manufacturer’s protocol. O4
+ oligodendrocytes were pre-labeled with anti-O4 MicroBeads (human, mouse, rat, 130-094-543, Miltenyi Biotec). The generated cell suspensions were passed through MS columns twice (130-097-678, Miltenyi Biotec) by performing two runs applying a magnetic field by the OctoMACS™ manual separator (130-042-109, Miltenyi Biotec) modified from previously published protocols [
39]. ACSA2
+ astrocytes were flushed into Eppendorf Tubes
® in AstroMacs Separation Buffer. Cell pellets were collected and consecutively lysed in Buffer RLT Plus (1053393, Qiagen) for RNA isolation, resuspended in 2% bovine serum albumin in D-PBS (FC buffer) for flow cytometry, or snap-frozen in liquid nitrogen, and stored at − 80 °C until further use.
Flow cytometry of murine astrocytes and oligodendrocytes
Cell suspensions were centrifuged in FC buffer. Subsequently, cell pellets were resuspended in FC buffer containing fluorophore-conjugated antibodies: the astroglia recognizing antibody ACSA2, APC (Miltenyi Biotec Cat# 130-123-284, RRID:AB_2811488, 1:75) and the oligodendroglia recognizing antibody O4 Phycoerythrin (R and D Systems Cat# FAB1326P, RRID:AB_664169, 1:75). For proper labeling, cells were incubated with antibodies at 4 °C for 15 min. Pellets were resuspended in cold FC buffer and dead cells were stained using DAPI. ACSA2+ and O4+ cells were subsequently analyzed using the BD LSRFortessa™ Cell Analyzer (BD Biosciences GmbH). Quantification of ACSA2+ and O4+ populations were calculated using FlowJo™ Software (BD Biosciences GmbH).
RNAscope
Prior to RNAscope analysis, forebrains were embedded in Tissue Tek O.C.T Compound (SA-4583, Sakura), snap-frozen and sectioned at a thickness of 14 µm. For RNAscope, sections were heated at 60 °C for 30 min and post-fixed in 4% PFA at 4 °C. Sections were dehydrated using an ascending alcohol series (50/70/100% ethanol) and air-dried afterwards. Tissue was further dehydrated with hydrogen peroxide. Antigen-retrieval was performed using the RNA–Protein Co-detection Ancillary Kit (323180, ACD). Primary antibodies were diluted in Co-detection Antibody Diluent (323180, ACD) and incubated at 4 °C overnight: GFAP (Abcam Cat# ab4674, RRID:AB_304558, 1: 1.000) and glutamine synthetase (GS) (Santa Cruz Biotechnology Cat# sc-74430, RRID:AB_1127501, 1:500). Sections were fixed using 2% PFA after primary antibody incubation. Additionally, sections were treated with Protease III according to manufacturer’s protocol. Afterwards, probes were hybridized at 40 °C for 2 h: RNAscope Probe-Mm-Myrf-C2 (524061-C2, Biotecne), and RNAscope Probe-Mm-Sox10 (435931, Biotecne). Signal amplifiers (AMP1 and 2) and fluorophores (TSA Vivid Fluorophore 520/650, 7527/7523, TOCRIS) were linked to the probes at 40 °C for 30 min, respectively. Fluorophore-conjugated secondary antibodies were diluted in Co-detection Antibody Diluent and linked to primary antibodies at RT for 30 min. Nuclei were counterstained using DAPI and sections were mounted in Aqua-Poly/Mount.
Western blotting (WB)
For WB analysis, five animals per genotype were processed. Total protein amount was calculated to 15 µg in radioimmunoprecipitation assay (RIPA) buffer: 50 mM Tris–HCl (pH 8.0, 150 mM NaCl, 2 mM EDTA, 1% v/v Nonidet P-40 (Roche), 0.5% v/v Na-deoxycholate (Roth), 0.1% w/v SDS (Applichem)) complemented with protease inhibitor (11697498001, Roche), and phosphatase inhibitor (4906837001, Roche) for SDS-PAGE on 4–12% Bis–Tris gels (NP0322BOX, Invitrogen). Protein was blotted on polyvinylidene difluoride membranes for fluorescence applications (PVDF-FL, IPFL 00010, Millipore). Unspecific antibody binding was blocked using 1% bovine serum albumin followed by incubation at 4 °C overnight with the following primary antibodies: GFAP (Abcam Cat# ab4674, RRID:AB_304558, 1: 1.000), vimentin (VIM) (Abcam Cat# ab20346, RRID:AB_445527, 1:1.000), GLT-1 (Millipore Cat# AB1783, RRID:AB_90949, 1:500), GLAST (Abcam Cat# ab41751, RRID:AB_955879, 1:500), GS (Santa Cruz Biotechnology Cat# sc-74430, RRID:AB_1127501, 1:500), aquaporin-4 (AQP-4) (Sigma-Aldrich Cat# A5971, RRID:AB_258270, 1:500), and growth associated protein 43 (GAP-43) (Millipore Cat# MAB347, RRID:AB_94881, 1:1.000). Primary antibodies were rinsed and followed by secondary antibodies at RT for 1 h: anti-mouse Alexa Fluor 488 (Molecular Probes Cat# A-21202, RRID:AB_141607, 1:1.000), anti-rabbit Alexa Fluor 488 (Molecular Probes Cat# A-21206, RRID:AB_2535792, 1:1.000), anti-chicken Alexa Fluor 488 (Jackson ImmunoResearch Labs Cat# 703-545-155, RRID:AB_2340375, 1:500), anti-rabbit Alexa Fluor 647 (Jackson ImmunoResearch Labs Cat# 711-605-152, RRID:AB_2492288, 1:500), and anti-mouse Alexa Fluor 568 (Thermo Fisher Scientific Cat# A10037, RRID:AB_2534013, 1:1.000).
RNA isolation
RNA was isolated using the RNeasy Plus Micro Kit (74034, Qiagen). Cells were lysed in RLT Plus Buffer (+ β-mercaptoethanol, 1:100). Homogenized lysate was transferred to gDNA eliminator spin. RNA was transferred to a RNeasy spin column and eluted into new tubes according to manufacturer’s protocol. Total RNA concentration was measured using the NanoDrop® ND-1000 (Thermo Scientific) and stored at − 80 °C until bulk RNA sequencing or quantitative real-time PCR.
Quantitative real-time PCR (qRT-PCR)
Coding DNA (cDNA) was generated using GoScript™ Reverse Transcription System according to manufacturer’s protocol. For the analysis of gene expression, the SSoFast™ EvaGreen
® Supermix was mixed with respective primers (Additional file
1: Table S1) and amplified using the Light Cycler 480 system. For the analysis, gene expression levels were normalized to housekeeper genes Tyrosine 3-Monooxygenase (
Ywhaz) and Glucuronidase Beta (
Gusb). For qPCR, 4 animals per genotype were analyzed.
Bulk RNA sequencing
Samples were submitted to Genewiz GmbH (Leipzig, Germany) to prepare a standard and ultra-low input RNA-seq [
40] library prior sequencing using an Illumina HiSeq platform. Reads were delivered as fastq files. Subsequent pre-processing and quality control of fastq files was performed by the Core Unit of Bioinformatics, Data Integration and—Analysis (CuBiDa) of the University Hospital Erlangen, Erlangen, Germany.
In the first step of the bioinformatics data analysis of bulk RNA-seq data, the quality of the raw paired-end sequencing data sets (fastq) was assessed with the tool FastQC v0.11.9 [
41]. Samples that passed quality assessments were further optimized by a trimming step with the tool Trimmomatic v0.39 [
42]. Resulting samples were mapped with STAR aligner v2.7.9a [
43] utilizing the mouse reference genome (Mus_musculus.GRCm39) from Ensembl release 104. The aligned sequences were converted to raw read counts per exon/gene with the tool featureCounts which is included in the Rsubread software package v2.14.2 [
44]. Gene symbol annotation was performed using the AnnotationDbi package [
45]. Modelling, normalization, and differential gene expression analysis was performed using the DESeq2 package [
46] in combination with the R Version 4.1.2. Log2 fold-changes were set to > 1 and adj. p-values to 0.05. Differential expressed genes were pre-ranked by stats and gene set enrichment performed using the fgsea package. Results were visualized using the ggplot2 package. To determine cell lineages, the enricher-function of clusterProfiler package [
47] was applied on the 50 most highly expressed transcripts. For assessment of cellular identity, the gene list was enriched on a single-cell RNA sequencing based mouse cell-marker dataset: CellMarker 2.0 [
48].
P value threshold was set to 0.05 and
p value adjustment was performed using Benjamini–Hochberg (BH) procedure. Results were visualized using ggplot2 package.
Statistical analyses
Data were analyzed and visualized using GraphPad Prism 9.5.0 and CorelDraw X6 software. Shapiro–Wilk test was performed for determination of normal distribution of expression levels and cell counts. Welch’s t-test was conducted for determination of statistical significance. If no Gaussian distribution was determined, statistical analysis was performed using Mann–Whitney–U test. Unless otherwise stated, all graphs represent mean values and standard deviation. P values were defined as statistically significant at p < 0.05.
Discussion
Reactive astrogliosis is a prominent feature in multiple neurodegenerative disorders [
51]. Several studies taking advantage of microarray- and RNA sequencing technology and surface marker-based sorting of astrocytes emphasized region- and context-dependent diversity of astrocytes [
16,
33‐
35,
52]. Here, we analyzed GFAP expression as a proxy for reactive astrogliosis in
post mortem tissue of MSA-P patients in brain regions severely affected by disease pathologies such as the motor cortex, putamen, and substantia nigra. We detected an increase of GFAP
+ cells suggesting reactive astrogliosis in all three regions. Notably, the putamen and the substantia nigra, most prominently affected in MSA-P, displayed a proportionally higher number of GFAP
+ astrocytes compared to motor cortex. To extend the characterization of astrocytes, we analyzed expression of EAAT2, an important transporter for glutamate reuptake. A decrease of glutamate reuptake transporter expression has been linked to other neurodegenerative disorders, such as Alzheimer’s disease (AD) and amyotrophic lateral sclerosis (ALS) [
53‐
56]. Quantifying GFAP
+/EAAT2
+ astrocytes hinted towards decreased numbers in MSA-P patients compared to controls and revealed altered distribution of EAAT2 in both brain regions.
Analyzing the expression of EAAT2 in the cortex and the putamen of MSA-P patients provided evidence of a lower number of GFAP
+/EAAT2
+ astrocytes in MSA-P patients. Notably, the distribution pattern of EAAT2 is altered in MSA-P patients in cortex and striatum, demonstrating a re-distribution from a diffuse to a nuclear pattern. This altered localization of EAAT2 was already described in vitro in a model of AD upon exposure of astrocytes to Aβ
1-42 [
57]. The altered expression pattern results in a slower clearance time of excessive glutamate in the synaptic cleft and is therefore potentially linked to Aβ-related pathology. Based on this astrocytic phenotype in the human
post mortem tissue, we further characterized astrocytes in MBP29-hα-syn mice. Mirroring the findings in human
post mortem tissue, we also observed a substantial increase of GFAP
+ cells in the cortex, the putamen and the substantia nigra. An enhanced astrocytic response was previously observed in grey compared to white matter regions of MSA-P patients and/or in MBP29-hα-syn mice [
31]. The similarities in GFAP expression between MSA patients and the present transgenic MSA mouse model prompted us to a comprehensive analysis of astrocytes in distinct brain regions. Interestingly, a tendency of an increased total astrocyte number was present in the cortex of MBP29- hα-mice. Previous work showed an increased number of oligodendrocytes in the corpus callosum and preserved numbers of oligodendrocytes in the putamen of MBP29- hα-mice [
32]. These findings are further supported by the observations of Ge and colleagues, hinting towards the presence of a glial progenitor population at the age between 3–4 weeks postnatally [
58]. The increased levels of the reactivity-associated markers GFAP and VIM in the striatum suggest that these astrocytes may be much more responsive to α-syn pathology compared to other regions. This notion was accentuated by a decrease of GLT-1 and GLAST protein expression exclusively in the striatum potentially resulting in glutamate excess. The context-dependent expression levels of glutamate transporter were described in 6-hydroxydopamine-lesioned PD model [
59]. Alternatively, the reduced levels of GLT-1 and GLAST in the striatum compared to the cortex hint towards the intrinsic heterogeneity between cortical and striatal astrocytes thus supporting the findings of a distinct profile of astrocytes in cortex and striatum of MBP29-hα-syn mice. Assessing the expression of additional homeostasis-associated astrocyte markers, we observed an upregulation of GS in the cortex while there was no significant change in the striatum. Lack of GS expression was already shown to impact neurodegenerative processes [
60]. Furthermore, AQP-4 and GAP-43 protein levels were more increased in the cortex compared to the striatum suggesting that the capacity to maintain CNS homeostasis of striatal astrocytes is potentially impaired in MBP29-hα-syn mice.
To investigate the impact of MSA-related neuropathological processes on astrocytes on a cellular level, we developed an astrocyte isolation method with high purity modified from an existing surface marker-based sorting protocol using ATP1B2 as an unbiased astrocyte-specific surface marker [
61] and subsequent astrocyte RNA sequencing. The bioinformatic analysis of the gene expression patterns revealed astrocyte identity in the isolated cell population but also indicated the pattern of neural stem cells. On the one hand, this enrichment might result from the shared expression patterns of astrocytes and neural stem cells [
62]. On the other hand, since these mice are in a postnatal developmental period, this enrichment may be due to the presence of co-isolated neural stem cells residing in the subventricular zone close to the striatum [
63]. Improving the isolation approach, we were able to achieve a high purity of ACSA2
+ cells and high-quality outcome after bulk RNA sequencing. To evaluate astrocyte reactivity, we compared the astrocytic gene expression profile derived from the cortex and striatum MBP29-hα-syn mice to RNA-seq datasets previously published by Zamanian et al. [
16] which induced reactive astrogliosis in wild-type mice using intraperitoneal lipopolysaccharide injections. Interestingly, the transcriptome of both astrocyte gene expression profiles overlapped with Zamanian et al., however, striatal astrocytes displayed an enrichment of these markers.
The most striking result to emerge by analyzing the cortex and the striatum, was a pro-inflammatory signature in striatal compared to cortical astrocytes. Elevated levels of pro-inflammatory cytokines were observed in the cerebrospinal fluid (CSF) of MSA patients [
64]. Moreover, Valera and colleagues assessed pro-inflammatory cytokine levels in transgenic MSA mice [
65]. In this study, the authors provide evidence of elevated CSF levels of IL-1β, IL-1ra, IL-13, IL-17, MIP-1α, and G-CSF using a cytokine proteomic assay matching the findings of the present transcriptomic findings. Additionally, region-dependent responses have been described in mouse models of Huntington’s disease [
20,
66]. Striatal mouse astrocytes have been shown to respond context-specifically by G
i-GPCR signaling in vivo with an upregulation of neuroinflammatory responses but whether the inflammatory response is toxic or protective remained unclear [
66]. There is evidence that striatal astrocytes demonstrate altered transcriptomic profiles in inflammatory environment [
67] associated with impaired homeostatic functions for calcium and glutamate signaling [
68]. Distinct populations of astrocytes were identified in a HD mouse model exhibiting variable transcriptional phenotypes in the cortex reflecting a rather protective phenotype consistent with our findings [
69,
70]. Another study characterizing astrocytes by single nuclei sorting in the prefrontal cortex, most severely affected in AD patients, revealed decreased homeostatic and increased levels of inflammatory transcripts [
71]. In general, comparing astrocytes derived from AD patient’s
post mortem brains, revealed non-homogenous astrocyte population including subpopulations capable of upregulating supportive processes as stress and survival response, but also subsets with metabolic stress and defective clearance [
72]. Another remarkable result is that differentially expressed genes in striatal astrocytes are enriched in downregulated lipid-associated pathways and impaired energy homeostasis in the cortex of hereditary spastic paraplegia and AD patients [
73,
74]. Taken together, a more inflammatory astrocytic response is present in CNS regions predominantly affected by the respective pathology.
Here, we assessed a pronounced pro-inflammatory state of astrocytes accompanied by impaired homeostatic processes via regional isolation and subsequent bulk RNA sequencing. The alteration of electrophysiological characteristics in neuroinflammatory conditions induced by brain abscesses has already been described in striatal astrocytes [
75,
76]. Furthermore, an induction of neurotoxic astrocytes via release of microglial TNFα and IL1β has recently been shown [
11]. In contrast, astrocytes are the major source of chemokines (CCL2, CXCL1 [
77]) with corresponding receptors expressed by microglia potentially implicating a bi-directional communication. The upregulation of macrophage colony-stimulating factor (CSF-1) in astrocytes upon exposure to pro-inflammatory cytokines was observed which in turn suppressed microglial interferon- and MHC expression [
78]. Another potential route of astrocyte-microglia communication is the microglial secretion of vascular endothelial growth factor β (VEGFB) for modulation of pathogenic characteristics of astrocytes [
79]. In addition to the secretory component of astrocytes, components of the innate immune system play a pivotal role, such as toll-like receptors (TLR) and complement factors [
80‐
82], which were observed to be upregulated in the striatum of MBP29-hα-syn mice. Especially, TLR4 is known as an essential receptor for astrocyte and microglia activation by aSyn [
83] and could therefore play a crucial role in the crosstalk of both cell types. More detailed analysis of the complex interaction between microglia and astrocytes is necessary in future work to reveal potential mechanisms of cross-talk and maintenance of immune response in the CNS.
To our surprise, cortical astrocytes display an oligodendroglial and pro-myleinogenic expression pattern within the 20 most highly upregulated genes (ERMN, MAG, OPALIN, MYRF). Given the fact, that glutamate transporter expression is influenced by MSA pathology, we are not able to apply more specific isolation via the glutamate reuptake transporter GLAST (ACSA1). Nonetheless, we were able to determine high enrichment of astrocytes by bioinformatical enrichment algorithm. Moreover, we applied an RNA hybridization approach to confirm the presence of oligodendroglial transcription factors within or in proximity of astrocytic processes. A second limitation is due to the fact that the visualization of an entire astrocyte is difficult using commercially available antibodies against astrocyte markers, such as GFAP, ALDH1L1, S100β, and GS. Using RNAscope, we succeeded to visualize the presence of oligodendroglial transcripts in proximity of astrocytic processes counterstained with GFAP or GS. Combined with the enrichment of upregulated transcripts in cortical astrocytes involved in protein secretion, these results may potentially suggest that astrocytes support oligodendrocytes by energy supply, secreting trophic factors, and support of immune response as shown in other demyelinating diseases [
84‐
87]. Furthermore, intercellular RNA transfer might serve as a possible mechanism for astrocytes to provide support to neighboring cells [
88]. Future investigations are necessary to fully explore the specific potential of astrocytes to endogenously express distinct oligodendroglial transcripts thereby either promoting a change of cellular identity or supporting dysfunctional oligodendrocytes in the cortex of MBP29-hα-syn mice.
The present study provides a detailed region-specific characterization of the transcriptomic landscape of astrocytes in MSA. Our work contributes to a better understanding of astrocyte heterogeneity in distinct brain regions differentially affected by MSA-related neuropathology. The upregulation of reactivity-associated proteins in astrocytes, to certain limits, does not necessarily affect astrocytic capacity to support surrounding cells or the maintenance of CNS homeostasis in the affected regions. Moreover, despite reactive cortical astrogliosis, MBP29-hα-syn mice upregulate homeostasis related proteins such as AQP-4, GS, and GAP-43. In the striatum, severely affected by MSA pathology, we provided evidence for the presence of reactive astrocytes acquiring a more harmful phenotype by upregulation of pro-inflammatory transcripts and failing to maintain essential functions as energy and lipid homeostasis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.