Cellular fibronectin exacerbates α-synuclein aggregation via integrin alpha4beta1 mediated PARP1 and SCD elevation

Abstract

Mitochondrial dysfunction and lipid metabolic disturbance may promote pathologic α-synuclein (α-syn) aggregation, accelerating the progression of Parkinson's disease (PD). Whether extracellular matrices are associated with those pathological mechanisms in PD remains elusive. Here, we aimed to identify if cellular fibronectin (cFn), a component of extracellular matrices, contributes to α-syn abnormality via inducing mitochondrial energy depletion or disrupting lipid homeostasis. In Our study, 1-methyl-4-phenyl-1, 2,3,6-tetrahydropyridine (MPTP)-treated PD mice and human neuronal SH-SY5Y cells were used. Astrocyte-derived cFn protein delivery and AAV-mediated cFn knockdown mouse models were established to validate the functional role of cFn. Mitochondrial dysfunction was detected by transmission electron microscopy (TEM), and the level of poly (ADP‒ribose) (PAR) polymerase-1(PARP1), pathologic α-syn and cFn-induced lipid dysmetabolism was determined. We demonstrated that excessive cFn accumulated in the SNpc of MPTP-treated mice, and cFn rather than plasma Fn (pFn) exacerbated neuronal mitochondrial dysfunction and α-syn accumulation. Mechanically, cFn induced PARP1 activation via integrin α4β1, which contributed to neuronal NAD ​+ ​depletion and pathologic α-syn aggregation. Furthermore, cFn induced an increase in free fatty acids (FAs) and triglycerides (TAG) in neurons by binding to integrin α4β1, which synergistically contributed to α-syn abnormality. We revealed that cFn induced stearoyl-CoA desaturase (SCD) activation via integrin α4β1, which was interacted with SCD. Genetically depleting cFn suppressed PARP1 activation and SCD elevation, which further rescued the mitochondrial disruption and α-syn abnormalities in MPTP-treated mice. Overall, our findings suggest that cFn exacerbates α-syn aggregation via integrin α4β1-mediated PARP1 and SCD elevation. cFn-targeting therapy may be a promising strategy for treating PD.

Graphical abstract

Schematic representation of the possible mechanisms of cFn ​in PD.

Introduction

Parkinson's disease (PD) is associated with pathologic α-synuclein (α-syn) aggregation, which leads to the formation of Lewy bodies (LBs) and contributes to the loss of dopaminergic neurons in the pars compacta of the substantia nigra (SNpc) [[1], [2], [3]]. Recently, the pathological mechanisms underlying α-syn accumulation have remained elusive. Poly (ADP‒ribose) (PAR) polymerase-1 (PARP1), an enzyme that participates in DNA repair, induces mitochondrial energy depletion by reducing the levels of NAD+ and ATP [[4], [5], [6], [7]]. Moreover, PARP1 activation leads to the accumulation of PAR, which interacts with α-syn and accelerates α-syn fibrillation and phosphorylation [4]. Pathologic forms of α-syn can interact with mitochondrial fusion proteins such as MFN-1 and OPA-1 or be translocated into the inner mitochondrial membrane, subsequently exacerbating mitochondrial impairment and forming a vicious cycle [8,9]. Studies have also shown that an imbalance in lipid metabolites contributes to α-syn aggregation. For example, a deficiency of glucocerebrosidase (GCase), a lysosomal hydrolase, promotes α-syn aggregation by increasing the level of its substrate called glucosylceramide (GlcCer) [10,11]. Furthermore, an increase in unsaturated fatty acids (MUFAs), such as oleic acid (OA), enhances the neurotoxicity of α-syn [12,13], whereas depletion of stearoyl-CoA desaturase (SCD), a rate-limiting enzyme involved in OA production, decreases α-syn inclusions in neurons and rescues neurodegeneration in human α-syn^E46K-expressing mice [12]. Generally, α-syn abnormalities are closely related to mitochondrial dysfunction and lipid metabolism disturbances.

Previous reports have demonstrated that inflammation-induced astrocyte activation by lipopolysaccharide (LPS) or a TLR3 agonist leads to the accumulation of cellular fibronectin (cFn) in the central nervous system (CNS) [14,15]. Several studies have also revealed that excessive cFn aggregates in the brain exacerbate neuroinflammation by promoting an activated phenotype in macrophages and microglia [16]. In toxin-induced demyelinated lesions, cFn aggregates inhibit oligodendrocyte differentiation and remyelination [14,17]. Neuroinflammation also occurs in PD [[18], [19], [20], [21], [22]], however, as a component of the extracellular matrix in the brain, whether cFn is increased in the SNpc of PD models and contributes to pathological conditions beyond neuroinflammation requires further investigation. Here, through in vivo and in vitro studies, we revealed the critical role of cFn in neuronal mitochondrial dysfunction and lipid metabolism. We showed that cFn induced PARP1 activation and SCD elevation via integrin receptor α4β1 (or called integrin α4β1), which led to mitochondrial disruption and increased MUFAs, ultimately exacerbating pathological α-syn aggregation. Understanding the pathophysiologic functions of cFn can potentially help identify novel therapeutic targets in PD.

Materials and Methods

Animals

All animal experiments were approved by the Experimental Animal Ethics Committee of Zhujiang Hospital of Southern Medical University (Approval No. LAEC-2020–052). One hundred and sixty 8-week-old wild-type male C57BL/6 WT mice were purchased from Guangdong Experimental Animal Centre, China. Before the experiments, the animals were kept in a quarantine room for 7 days. The mice were housed in social groups of up to five individuals. The mice were housed in standard mouse cages with appropriate bedding at room temperature and standard humidity under a 12-h light/dark cycle. Food and water were provided ad libitum. To prevent injury, the mice that received stereoscopic brain injections were housed individually. The main text presents experiments conducted specifically with C57BL/6 mice.

Treatments with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine and stereotactic injection

The subacute MPTP PD mouse model [[23], [24], [25]] was used to determine the levels of Fn in the SNpc of the PD model. For the induction of the mouse subacute PD model, 8-week-old mice were intraperitoneally injected with 30 ​mg/kg MPTP (Sigma Catalogue Number M0896) daily for 5 consecutive days, while saline was intraperitoneally injected into the control mice. According to previous studies [[23], [24], [25]], MPTP or saline injections were initiated 21 days prior to the sacrifice of the mice. The knockdown of cFn produced by astrocyte or endothelial cells in the brain was induced by the stereotactic injection of AAV-Fn1-shRNA. Control shRNA (pHBAAV-U6-scramble-MCS-CMV-Luc-WPRE) constructs for luciferase and Fn1 shRNA were purchased from Hanheng Biology. All of the mice were sacrificed to obtain brain tissues after performing the behavioural tests. We used AAV9 serotype in our experiments, and the control shRNA sequence was performed in Table S1. For the AAV-Fn1 shRNA design, U6 was used as promoter, and Woodchuck Hepatitis Virus Posttranscriptional Regulatory Element (WPRE) was used as post-translational elements (pHBAAV- U6-Fn1-MCS-CMV-Luc-WPRE).

The stereotactic injection of cFn was performed as previously described [14,26]. cFn was injected 21 days with a 10 ​ml Hamilton syringe and a 33-gauge needle before the MPTP-treated or healthy mice were sacrificed. All of the mice were injected at the same time. cFn was dissolved to 1.5 ​mg/ml with ACSF and injected into the unilateral SNpc (2 ​μl per hemisphere at 0.5 ​μl/min) with the following coordinates: anteroposterior (AP) ​= ​+3.08 ​mm, mediolateral (ML) ​= ​−1.0 ​mm, and dorsoventral (DV) ​= ​−4.5 ​mm from bregma. After injection, the needle was held for 5 ​min to allow the tissues to fully absorb the solution. After the injection, the animals were monitored for 2 ​h, and postoperative care was performed. Intracerebroventricular injection of AAV-Fn1-shRNA to knock down cFn expression was performed 5 weeks before sacrifice, since Fn1 shRNA requires 3–5 weeks after injection to knock down Fn1 gene expression in the brain. AAV-Fn1-shRNA was injected into the lateral ventricle (2 ​μl per hemisphere at 0.5 ​μl/min) with the following coordinates: anteroposterior (AP) ​= ​+0.6 ​mm, mediolateral (ML) ​= ​−1.5 ​mm, and dorsoventral (DV) ​= ​+2.0 ​mm from bregma. Live imaging of the animal brain was used to determine the effect of intracerebroventricular injection. 3 sequences for mouse Fn1 shRNA were used for verification (Table S1). For PARP1 depletion in vivo, based on previous studies [27], mice were treated with Talazoparib (MCE, HY-16106) for consecutive 28 days before sacrifice (oral gavage at the dose of 0.33 ​mg/kg). The flowchart of our in vivo studies was performed in Fig. S1 a.

Behavioural tests

Rotarod test and pole test were used to determine the motor deficit of mouse models. For the rotarod test, mice were placed on a rotarod and trained to walk by gradually raising the speed from 5 ​rpm to 40 ​rpm 3 days before sacrifice. Training is 3 times a day for 3 days (up to 10 turns per training). After the 3-day adaptation training, the test began. The mice were placed in the centre of the rotarod, and the body axis was perpendicular to the cylinder axis. The speed gradually increased from 5 ​rpm to 40 ​rpm within 180 ​s, times that mice spent on the wheel was recorded (if the time exceeded 180s, it was recorded as 180s). Each mouse was tested 3 times, with an interval of 30min, and the average value was counted. For the pole test, all mice were given adaptive training 3 days before sacrifice. During the experiment, we placed the mice at the top of the pole (height 60 ​cm, diameter 2 ​cm) with their heads facing up, and the time of turning direction and climbing to the ground (the limbs must land completely) were recorded. For 5 consecutive experiments, the best result was recorded.

Cell culture, transfection and treatment

Human-derived SH-SY5Y cells were cultured in DMEM containing 10 ​% foetal bovine serum and 1 ​% penicillin/streptomycin/glutamine. The cells were cultured at 37 ​°C with 5 ​% CO2. The SH-SY5Y cells were transfected when their growth density was 70 ​%–80 ​%, which normally occurs 2 days after seeding. For the construction of α-syn overexpressing SH-SY5Y cells, GST-wild type α-syn plasmids (pcDNA3.1-SNCA-WT-GST-MYC-C) were used in transfection. In our experiments, the GST tag was removed after protein overexpression in cell lysates used for western blotting, as our aim was to accurately determine the molecular weight of α-syn oligomers. Specifically, SH-SY5Y cells overexpressing α-syn were lysed using RIPA lysis buffer (Sigma, Cat. No. 20–188). The lysates were then treated with PreScission protease (Beyotime, China) and incubated at 4 ​°C overnight. Subsequently, the samples were mixed with Glutathione-Sepharose beads and rotated at 4 ​°C for 2 ​h. The beads were then pelleted by centrifugation at 500×g for 2–3 ​min, and the supernatants were collected. Successful cleavage of the GST tag was confirmed via western blot analysis. Knockdown of PARP1 in SH-SY5Y cells was induced by 10 ​mM mouse PARP1 siRNA. PARP1 expression was verified by Western blotting 72 ​h after transfection. Three sequences for mouse or human PARP1 siRNA were used for verification. siRNAs were transfected into SH-SY5Y cells with the Lipofectamine 2000 reagent (Invitrogen Catalogue Number 116685000). After transfection, the culture medium without serum was changed within 4–6 ​h. All siRNA sequences were produced by Heyuan Biology, China. 3 sequences for human PARP1 siRNA were used for verification (Table S1). One day after transfection, the SH-SY5Y cells were coated with cFn (5 ​μg was added to the wells of 8- and 96-well plates, while 50 ​μg was added to the wells of a 6-well plate, according to previous report [14]) for 48 ​h before the experiments. For the control groups, the cells were incubated with the same volume of PBS. For NOS inhibition, the cells were pretreated with 10 ​μM NG-nitro-L-arginine methyl ester and hydrochloride (L-NAME) (Sigma, N5751) for 12 ​h [28]. To prevent interactions between Fn and different receptors, SH-SY5Y cells were pretreated with the integrin α5β1 receptor inhibitor ATN-161 trifluoroacetate salt (MCE HY-13535A), the integrin α4β1 receptor inhibitor TR-14035 (MCE HY-15770), or the TLR4 inhibitor TLR4-IN-C34 (MCE HY-107575) following the manufacturers' protocols. To inhibit the EIIIA-Fn association, SH-SY5Y cells were pretreated with 5 or 10 ​μg/ml AF38Pep for 48h, a blocking polypeptide mimics the binding site for EIIIA domain and was validated by previous reports [29] (amino acid sequence: VMPYISTTPAKPCTSENCGNSWYGGFKSKNENKIYFIN, constructed by Qiangyao Biology, China). To prevent SCD expression, SH-SY5Y cells were treated with MK-8245 (MCE, HY-13070) at a concentration of 1 ​μM, which is based on the Manufacturer's instructions.

Preparation of astrocyte-derived cellular fibronectin

Astrocyte-derived cFn, which is predominantly present in the deoxycholate (DOC)-insoluble fraction, was prepared from primary rat astrocytes according to methods described in previous studies [14]. Briefly, astroglial matrices were prepared by water lysis of primary rat astrocytes that were pretreated with LPS for 48 ​h. Subsequently, the deposits were scraped in ice-cold deoxycholate buffer and further solubilized on ice for 30 ​min. Protein concentrations were determined via Bradford's protein assay (Bio-Rad) with bovine serum albumin as the standard. Equal amounts of protein from the deoxycholate extracts were separated into soluble and insoluble fractions via centrifugation at 13000 ​rpm for 20 ​min at 4 ​°C. The deoxycholate-insoluble pellets were dissolved in a solution containing 2 ​% SDS in 20 ​mM Tris-HCl (pH 8.8). The presence of Fn in DOC-insoluble fractions and LPS-induced astroglial matrices was confirmed by Western blotting and immunofluorescence staining, respectively. For Stereotactic injections, the aggregates isolated from the DOC-insoluble fraction were dialysed against PBS for 24 ​h at 4 ​°C. Following dialysis, the protein concentration was quantified by Ultraviolet spectrophotometer, and the presence of cFn aggregates was verified by Western blot analysis. Using Ultraviolet spectrophotometer, we verified that A260/A280 value of the protein preparation was 0.58 while A260/A230 was 2.2, suggesting that the protein preparation contained minimal impurities such as nucleic acids, salts, or organic compounds. Moreover, the endotoxin level of protein preparation was determined by Chromogenic LAL Endotoxin Assay Kit (Beyotime, China) and demonstrated that the endotoxin level is under 1 EU.

Immunofluorescence staining and immunohistochemistry

The mice were anaesthetized with sodium pentobarbital and perfused with both normal saline and 4 ​% paraformaldehyde, and the brains were fixed with 4 ​% paraformaldehyde for 12 ​h. The thickness of the brain paraffin section was 6 ​μm. For immunofluorescence staining (IF), 6 ​μm brain slices were permeabilized in 0.1 ​% Triton X-100, blocked in 5 ​% bovine serum albumin (BSA), and incubated with the antibodies listed above (Table S2). The sections were then incubated with donkey anti-rabbit IgG (Alexa Fluor 488, Abcam, Catalogue number ab150073, 1:200 dilution) and/or donkey anti-mouse IgG (Alexa Fluor 594, Abcam, Catalogue number ab150108, 1:200 dilution) secondary antibodies, and the nuclei were counterstained with DAPI solution. For immunohistochemistry (IHC), the thickness of the brain paraffin section was 40 ​μm, four sections through the substantia nigra within the range covering the entire SNpc (between −2.8 and −3.88 ​mm AP from bregma) were processed for TH immunohistochemistry. the brain slices were incubated with 3 ​% H2O2, blocked with 10 ​% goat serum, and incubated with primary antibodies overnight at 4 ​°C. The signals were detected with a DAB staining kit (KeyGEN, China). For the quantification of TH ​+ ​cell numbers in each group, TH ​+ ​cells in SNpc were conducted at regular, predetermined intervals using a counting frame size of 80 ​μm ​× ​80 ​μm and a sampling grid spacing of 100 ​μm ​× ​100 ​μm in a Leica DM6 photomicroscope equipped with a digital camera and StereoInvestigator software [30]. The coefficient of error was calculated to assess precision, with values below 0.1 deemed acceptable.

Cell counting was performed with a Leica DM6 photomicroscope and corresponding software. Images were captured via Nikon fluorescence microscopy, and 30–50 images were acquired for each experimental condition. The intensity of the fluorescence signal or the number of positive cells was analysed with ImageJ software. We selected cells in at least 5 sections in the SNpc of each group (n ​= ​5) and measured the mean fluorescence intensity for statistical analysis. Statistical analysis was performed with GraphPad Prism 9, data with a normal distribution are presented as the mean ​± ​S.E.M.

Western blotting analysis and coimmunoprecipitation

The cells and brain tissues were collected for lysis with RIPA lysis buffer (Sigma Catalogue Number 20–188) mixed with a protease inhibitor mixture. After centrifugation at 12000×g for 15 ​min, the supernatant was retained, and the total protein concentration was determined via BCA protein assay (Abcam, Catalogue Number ab102536). For the brain tissues of the mice subjected to stereotactic injection, we only used tissues from the ipsilateral side of the injection. To observe the aggregated states of α-syn or Fn, TX-insoluble or DOC-insoluble samples were not boiled before western blotting and used non-reducing loading buffer. The extracted total proteins (50 μg/lane) from each group were transferred to PVDF membranes by SDS‒polyacrylamide gel electrophoresis. The membranes were blocked with 5 ​% BSA at 37 ​°C for approximately 2–4 ​h and incubated overnight with the outlined primary antibody (Table S2) at 4 ​°C. The next day, we incubated the membranes with specific anti-mouse (CST, Catalogue Number 14709) or anti-rabbit (CST, Catalogue Number 14708) secondary antibodies (1:10000) for 1–2 ​h. The results were ECL system and analysed with ImageJ software. The grey values of the target protein bands were normalized to that of β-actin and are expressed as relative ratios compared with those of the control groups.

For coimmunoprecipitation, we used a Pierce® Co-Immunoprecipitation Kit purchased from Thermo Fisher (Catalogue Number No. 24169). Following the manufacturer's protocol, the brain tissues of healthy mice or MPTP-treated mice were lysed in IP lysis buffer supplemented with a protease inhibitor mixture. After centrifugation for 15 ​min at 12000×g, the supernatants were incubated with protein G Plus-Agarose Immunoprecipitation reagent and IP antibodies. After incubation at 4 ​°C for 12 ​h, the magnetic beads were washed with elution buffer for 15 ​min. The immunoprecipitates were subsequently boiled at 100 ​°C in 5x SDS-loaded buffer for 15 ​min, followed by western blotting.

Flow cytometry

Cells were collected and resuspended in serum-free culture medium. Mitochondrial membrane depolarization was characterized by the tetraethylbenzimidazolylcarbocyanine iodide (JC-1) (Beyotime, China) red:green fluorescence ratio, which indicates the mitochondrial membrane potential and was visualized via flow cytometry. JC-1 was diluted with DMEM at a ratio of 1:1000, and the cells were washed with 1 ​× ​PBS 3 times and then incubated with JC-1 for 20 ​min in the dark at 37 ​°C. For the positive control groups in the experiments on mitochondrial membrane potential, the cells were treated with 100 ​μM carbonyl cyanide 3-chlorophenylhydrazone (CCCP) for 1 ​h prior to incubation with JC-1. The ratio of JC-1 polymers and monomers was calculated as the ratio of the PE-A fluorescence intensity and the FITC fluorescence intensity, which was quantified by flowjo X software. According to the manufacturer's protocol, PE-A fluorescence intensity represents the level of JC-1 polymers, while FITC fluorescence intensity represents the level of JC-1 monomer.

NO measurement, mitochondrial complex I activity evaluation, Oil Red O staining, free fatty acid and triglyceride assay

NO levels in SH-SY5Y cells were measured via an NO assay kit (Abcam ab65328 MA) according to the manufacturer's instructions. Mitochondrial complex I kits (Solarbio, China) were used to detect mitochondrial complex I activity, and the measurement was performed according to the manufacturer's instructions. For lipid droplet detection, we used an Oil Red O Staining kit (Solarbio, China). SH-SY5Y cells for Oil Red O staining were seeded in 6-well plates and washed with 1x PBS 3 times before staining. After the cells were perfused with 4 ​% paraformaldehyde for 10 ​min, they were stained within 10–20 ​min. Oil Red O Staining was performed according to the manufacturer's instructions, and images were captured in the bright field with a Zeiss AXIO Vert. A1 digital camera. The positive area of lipid droplets was analysed with ImageJ software. Free fatty acids and triglycerides were evaluated via free fatty acids and triglyceride assay kits purchased from Beyotime (China), respectively. For the experiments, the cells were lysed in IP lysis buffer with a protease inhibitor mixture. After centrifugation for 10 ​min at 12000×g, the free fatty acid or triglyceride levels in the samples were measured according to the manufacturer's instructions, and the results were determined with a Biotek multifunctional microplate reader.

Transmission electron microscopy

For mitochondrial ultrastructure detection, the mouse midbrain was harvested from 4 mm3 sections and prefixed in 2.5 ​% glutaraldehyde. Next, we washed the sample twice in cold acetone, wetted it in Epon 812 resin at room temperature, and polymerized it at 60 ​°C for 3 days. An ultrathin slice of the embedded sample was collected on an uncoated nickel mesh, sliced 70 ​nm thick, stained with lead citrate, and examined with H-7650 transmission electron microscopy. The sample was then polymerized at 60 ​°C for 3 days. An ultrathin slice of the embedded sample was collected on an uncoated nickel mesh. The slices were sliced 70 ​nm thick, stained with lead citrate, and examined via H-7650 transmission electron microscopy.

LC‒MS/MS analysis

UHPLC system (Vanquish, Thermo Fisher Scientific) was used for analysis, and an UPLC HSS T3 column (2.1 ​mm ​× ​100 ​mm, 1.8 ​μm) was coupled with a Q Exactive HFX mass spectrometer (Orbitrap MS, Thermo). Mobile phase A contained 40 ​% water, 60 ​% acetonitrile, and 10 ​mmol/L ammonium formate, and mobile phase B contained 10 ​% acetonitrile and 90 ​% isopropyl alcohol. Then, 10 ​mmol/L ammonium formate was added to 50–1000 ​mL of mixed solvent. The temperature of the column was 55 ​°C, and the automatic injector temperature was 4 ​°C. QE mass spectrometers were used for data-dependent acquisition of MS/MS spectra.

Data-Dependent Acquisition (DDA) mode control acquisition software (Xcalibur 4.0.27, Thermo), and the acquisition software continues to evaluate the full-scan mass spectrum. ESI source conditions were as follow: sheathing gas flow rate: 30 Arb, Aux gas flow rate: 10 Arb, the capillary temperature: 350 ​°C, the full mass spectrometry resolution: 120,000, the MS/MS resolution: 7500, the collision energy was 10/30/60 in NCE mode, and the injection voltage was 4 ​kV(positive) or −3.8 ​kV(negative).

Data files were converted to mzXML format by ProteoWizard. We used The CentWave algorithm in XCMS for peak detection, extraction, alignment, and integration. Lipid identification was achieved through LipidBlast library, which was developed based on XCMS.

Statistical analysis

All of the data are presented as the mean ​± ​s.e.m. of at least 3 independent experiments. Statistical analysis was performed with GraphPad Prism 9. The sample size for each experiment was estimated via preliminary experiments or previous experience with different experimental approaches. No algorithm or software was used when we randomized the animal subjects. Differences among 2 groups were analysed by unpaired two-tailed Student's t-test, whereas differences among multiple groups were analysed by ANOVA followed by Tukey's post hoc test. P values ​< ​0.05 were considered significant.

Results

Cellular fibronectin is elevated in MPTP-induced PD model mice

Firstly, we observed that Fn was increased in the SNpc of MPTP-treated PD model mice (Fig. 1 a). Fn in the brain has two variants: plasma Fn (pFn) and cellular Fn (cFn). pFn is produced by hepatocytes and expressed as a soluble dimer, which can enter the CNS when the blood‒brain barrier (BBB) is disrupted, whereas cFn is produced by resident cells in the CNS and acts as a component of the extracellular matrix [31]. Both pFn and cFn can assemble into aggregates that are insoluble in deoxycholate (DOC) [15]. We found that the DOC-insoluble fraction of Fn was also elevated in the SNpc of PD mice, which represented the Fn aggregates with high molecular weights (Fig. 1 b). As cFn in the SNpc can be produced by microglia, endothelial cells, and astrocytes in the brain [14], we further determined whether cFn is increased in the SNpc and confirmed the cellular distribution of Fn in MPTP-treated mice. Through double-labelling immunofluorescence staining, we observed that the expression of Fn in CD31-positive endothelial cells and GFAP-positive astrocytes, rather than IBA-1-positive microglia, was elevated in the SNpc of PD mice (Fig. 1 c-e), suggesting that increased Fn in the SNpc of PD mice may be produced by astrocytes and endothelial cells, which may assemble into aggregated forms and deposit in the brain. However, the staining of Fn in CD31-positive endothelial cells also label the plasma Fn in the SNpc since BBB disruption occurs in MPTP-treated models [3]. Therefore, both pFn and cFn may accumulate in the SNpc of PD mice.

Fig. 1.

Astrocyte-derived Fn was accumulated in the SNpc of MPTP-treated mice. (a) Representative immunoblots (left) and quantification (right) of Fn in the SNpc of healthy mice and MPTP-treated mice (n ​= ​5). (b) Representative immunoblots (left) and quantification (right) of DOC-insoluble fibronectin aggregates in the SNpc (n ​= ​5). (c–e) Representative images of the colocalization of Fn with CD31 (c), IBA-1 (d) and GFAP (e), and the relative fluorescence intensity of Fn in CD31⁻, IBA-1- or GFAP-positive cells was measured. DAPI (blue) was used for nuclear staining. Scale bar: 50 ​μm. Bars indicate the mean ​± ​s.e.m. Unpaired Student's t-test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, compared with the control group.

Cellular fibronectin induces α-syn aggregation and mitochondrial dysfunction in SH-SY5Y cells

To further determine whether pFn or cFn contributes to the pathogenesis of PD, dopaminergic neuronal SH-SY5Y cells were coated with pFn or cFn prepared from astroglial matrices of LPS-treated primary rat astrocytes since inflammatory mediators can induce cFn production in astrocytes [32]. Before cFn treatment in vitro, the presence of Fn in LPS-treated astrocytes was confirmed (Fig. S1 b) and we verified that Fn aggregates were elevated in LPS-induced astrocyte matrices (Fig. S1 c). Moreover, we found that IST-9, a marker of the EIIIA domain of cFn, which is absent in pFn [33], was elevated in the astrocyte matrices after LPS treatment, confirming that the Fn released from LPS-induced astrocytes is cFn rather than pFn. (Fig. S1 d). Subsequently, we found that cFn, rather than pFn, induced Triton (TX)-insoluble α-syn aggregation and phosphorylation at residue 129 in SH-SY5Y cells overexpressing wild-type α-syn (transfected with GST-α-syn plasmids) (Fig. 2 a). Astroglial matrices from untreated rat astrocytes (AMs), which presented a low level of cFn, did not increase the level of TX-insoluble pathological α-syn (Fig. 2 b). Moreover, cFn rather than pFn or AM led to the reduction of the mitochondrial marker COX IV (cytochrome c oxidase subunit IV) in SH-SY5Y cells (Fig. 2 b). cFn treatment also led to a decrease in the mitochondrial membrane potential and mitochondrial complex I activity in SH-SY5Y cells (Fig. 2 c-d). Overall, we hypothesized that cFn may contribute to α-syn abnormalities and mitochondrial dysfunction in PD. Furthermore, the protein‒protein interaction (PPI) network constructed via STRING analysis was used to predict the underlying mechanisms of cFn-induced α-syn aggregation and mitochondrial impairment. Using multiple proteins association option in STRING website (https://www.string-db.org/), which can integrate both known and predicted interactions from various sources including experimental data and curated pathway databases, we searched for the association between Fn and genes associated with α-syn aggregation and mitochondrial dysfunction in homo sapiens, such as PARK2, UCHL1, PINK1, DJ-1, LRRK2, ATP13A2, HTRA2 and PARP1. Interestingly, the PPI network revealed that FN1, the coding gene of Fn, is potentially associated with PARP1 rather than other genes (Fig. 2 e). Previous studies have demonstrated that PARP1 activation, which can be promoted by reactive oxygen species (ROS)- or nitric oxide (NO)-induced DNA damage, can accelerate α-syn aggregation by promoting PAR synthesis and inducing mitochondrial dysfunction via depletion of NAD+ and ATP [4]. Therefore, we concluded that cFn promotes PARP1 activation in neurons. Consistently, we confirmed that cFn, rather than pFn, increased the levels of PARP1 and its downstream member PAR in SH-SY5Y cells (Fig. 2 f). Moreover, the levels of NAD+ and ATP in SH-SY5Y cells decreased after coating with cFn (Fig. 2 g-h).

Fig. 2.

cFn, rather than pFn, exacerbated pathological α-syn aggregation and induced mitochondrial dysfunction and PARP1 activation in SH-SY5Y cells. (a) Representative western blots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in wild-type α-syn-overexpressing SH-SY5Y cells that were coated with PBS, cFn, pFn or untreated astroglial matrices (five independent replicates). (b) Representative images of COX IV in SH-SY5Y cells, and the relative fluorescence intensity was calculated (five independent replicates). Scale bar: 100 ​μm. (c) Mitochondrial membrane potential in SH-SY5Y cells was measured by JC-1 staining with flow cytometry, and the fluorescence intensity data are presented (three independent replicates). DAPI (blue) was used for nuclear staining. (d) Mitochondrial complex I activity in SH-SY5Y cells was measured via a Biotek multifunctional microplate reader (three independent replicates). (e) A PPI network constructed by the STRING online tool shows the predicted protein interaction between Fn and PARP1, which was analysed by multiple proteins option (Input: Fibronectin, PARK2, UCHL1, PINK1, DJ-1, LRRK2, ATP13A2, HTRA2 and PARP1; constraint: homo sapiens). (f) Representative western blots and quantification of PARP1 and PAR in SH-SY5Y cells that were coated with PBS, cFn, pFn or untreated astroglial matrices (five independent replicates). (g–h) The NAD+ (g) and ATP (h) levels in SH-SY5Y cells were measured via a Biotek multifunctional microplate reader (three independent replicates). Bars indicate the mean ​± ​s.e.m., ANOVA with Tukey's post hoc test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, and ns, not significant.

Stereotactic injection of cellular fibronectin induces α-syn aggregation and mitochondrial dysfunction in MPTP-treated mice

We next determined whether cFn influences mitochondrial impairment and α-syn aggregation in vivo. Firstly, we confirmed the presence of cFn aggregate in the protein preparation used for injection, which was isolated from LPS-treated astroglial matrices (Fig. S1 e). When different concentrations of cFn were injected in the SNpc of mice (2 ​μl per hemisphere), we firstly found that 1.5 ​mg/ml cFn rather than 0.1 ​mg/ml or 0.5 ​mg/ml significantly increased the level of S129-phosphorylated α-syn and PARP1 in the ipsilateral SNpc (injection side) of mice on Day 14 and Day 21 (Fig. S2 a-b). Therefore, we selected 1.5 ​mg/ml as the concentration of cFn injection in further studies. Furthermore, we observed the number of tyrosine hydroxylase (TH)-positive neurons in the ipsilateral SNpc of mice that were injected with 1.5 ​mg/ml cFn at different times. We found that the number of TH ​+ ​neurons decreased by 25 ​% on Day 14 and 41 ​% on Day 21 (Fig. S3 a). Therefore, Day 21 post injection was used as the time point for in vivo studies.

After stereotactic injection of cFn into the SNpc of MPTP-treated mice or healthy mice for 21 days, the levels of TX-insoluble phosphorylated α-syn (ser129) and α-syn oligomers were elevated in the ipsilateral SNpc of stereotactically injected mice compared with those of mice that had not received cFn injection (Fig. 3 a). Moreover, stereotactic injection of cFn further promoted the elevation of PARP1 and PAR in the SNpc of the MPTP-treated or healthy mice (Fig. 3 b). γ-H2AX, a marker of DNA strand damage, was elevated after cFn injection (Fig. 3 b), suggesting that cFn may induce DNA impairment and subsequently induce PARP1 activation.

Fig. 3.

cFn stereotactic injection exacerbated PARP1 activation, α-syn abnormalities and the loss of TH ​+ ​neurons. (a) Representative western blots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in the SNpc of mice treated with MPTP or cFn injection (n ​= ​5). (b) Representative western blots and quantification of PAR, PARP1 and γ-H2AX in the SNpc of each group (n ​= ​5). (c–d) Representative images of Ser129-phosphorylated α-syn (c) or PARP1 (d) and TH in the SNpc of each group. The fluorescence intensity of p-α-syn or PARP1 in TH ​+ ​cells was calculated and normalized to that of the control (n ​= ​5). Scale bar: 100 ​μm. (e) Representative immunohistochemical images of TH ​+ ​cells and the total number of TH ​+ ​cells in the SNpc of each group were counted (n ​= ​5). Scale bar: 100 ​μm. DAPI (blue) was used for nuclear staining. (f) Representative transmission electron microscopy (TEM) images of morphological changes in SNpc neuronal mitochondria. Scale bars: 1 ​μm (upper) and 500 ​nm (lower). (g) Representative images and quantification of the relative fluorescence intensity of COX IV in the SNpc (n ​= ​5). Scale bar: 100 ​μm. Bars indicate the mean ​± ​s.e.m., ANOVA with Tukey's post hoc test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, and ns, not significant.

In further support of our hypothesis, we found that Ser129 phosphorylated α-syn and PARP1 in TH ​+ ​neurons were increased in the cFn-injected side (ipsilateral side) of the mice with or without MPTP (Fig. 3 c-d), whereas the number of TH ​+ ​neurons was decreased in the SNpc of the cFn-injected sides (Fig. 3 e).

We next performed double immunofluorescence staining with anti-AIF and anti-TOM20 (mitochondrial marker) antibodies to detect AIF translocation to the nucleus, which can be induced by PARP1 activation. Notably, we discovered that the release of AIF from the mitochondria to the nucleus was greater on the cFn-injected side than on the contralateral side (Fig. S3 b). These findings suggest that excessive cFn triggered and exacerbated PARP1 activation, α-syn aggregation, and the loss of TH ​+ ​neurons in the SNpc. Through transmission electron microscopy (TEM), we also found that cFn injection induced neuronal mitochondrial cristae fracture and vacuolation on the ipsilateral side of the midbrain, which may disturb the respiratory function in neurons, indicating that cFn leads to abnormal mitochondrial morphology (Fig. 3 f). Moreover, cFn injection promoted the suppression of COX IV in the ipsilateral SNpc (Fig. 3 g), verifying that cFn contributes to mitochondrial disruption in vivo.

Cellular fibronectin facilitates PARP1 activation through integrin α4β1 mediation and promotes α-syn aggregation and mitochondrial dysfunction by increasing PARP1

We further investigated the underlying mechanisms of cFn-induced PARP1 activation. As our studies revealed that cFn increases the expression of PARP1 and its downstream members, we suppose that the alternatively spliced EIIIA and EIIIB domains that are contained in cFn rather than pFn may play critical roles in cFn-induced neuronal impairments [33]. Previous studies have demonstrated that the EIIIA domain binds to different receptors such as integrins α4β1, α4β7, α9β1 and Toll-like receptor 4 (TLR4); however, integrin α9β1 is absent from neurons, and the β7 subunit has not been found in the CNS [[34], [35], [36]]. Therefore, cFn may play a functional role through integrin α4β1 or TLR4. After cFn-coated SH-SY5Y cells were treated with different receptor inhibitors, we found that integrin α4β1 inhibition, rather than TLR4 or integrin α5β1 inhibition, rescued cFn-induced nNOS and PARP1 activation (Fig. 4 a). Moreover, integrin α4β1 inhibition decreased cFn-induced NO release in SH-SY5Y cells (Fig. 4 b). Integrin α4β1 inhibition also reduced TX-insoluble α-syn aggregation and phosphorylation at ser129 in cFn-coated α-syn-overexpressing SH-SY5Y cells (Fig. 4 c). Consistently, pretreatment with the NOS inhibitor L-NAME reversed the increase in PARP1 and γ-H2AX induced by cFn in vitro (Fig. 4 d-e). These results suggest that cFn promotes nNOS expression in neurons by binding to integrin α4β1, which leads to NO production and DNA oxidative damage, subsequently increasing PARP1 and inducing α-syn aggregation [37]. To further confirm whether cFn induced PARP1 elevation via its alternatively spliced EIIIA domains, SH-SY5Y cells were treated with AF38Pep, a polypeptide that mimics the integrin Binding Site for EIIIA domains before coated with cFn. We found that after 5 or 10 ​μg/ml AF38Pep treatment, which can inhibit the function of EIIIA domain, the increase of PAR and PARP1 in SH-SY5Y cells induced by cFn was eliminated (Fig. 4 f). Through co-immunoprecipitation, we validated that 10 ​μg/ml AF38Pep significantly reduced the interaction between Fn and integrin α4β1 (Fig. 4 g). Subsequently, we examined whether LPS-treated astrocyte matrices induced comparable effects on α-syn aggregation and PARP1 activation in vitro, as observed with cFn treatment. We demonstrated that LPS-treated astrocyte matrices increased the levels of PAR, PARP1, and γ-H2AX in SH-SY5Y cells (Fig. S3 c) and promoted TX-insoluble α-syn aggregation and phosphorylation at Ser129 (Fig. S3 d), effects similar to those induced by cFn treatment. Following treatment with 10 ​μg/ml AF38Pep, these elevations were attenuated, confirming that LPS-treated astrocyte matrices induce α-syn aggregation and PARP1 activation through the release of cFn. in Furthermore, after PARP1 in SH-SY5Y cells was knocked down by human PARP1 siRNA1 (sequence: 5′-GGGCAGAGGUGAAGGCAGATT-3′), which was verified by western blot (Fig. S4 a), we found that TX-insoluble phosphorylated α-syn (Ser129) and α-syn oligomers in cFn-coated SH-SY5Y cells were significantly decreased after PARP1 knockdown. However, compared with those in untreated cells, the level of insoluble pathological α-syn was still elevated (Fig. 5 a, c). Moreover, oral administration of Talazoparib, a novel PARP1 inhibitor for 28 days significantly decreased both Ser129-phosphorylated α-syn and α-syn oligomers in the ipsilateral SNpc of cFn injected mice (Fig. 5 b, d). Through immunofluorescence staining, we confirmed that Talazoparib treatment decreased PARP1 expression in the SNpc (Fig. 5 e). Meanwhile, the COX IV reduction induced by cFn in SH-SY5Y cells was rescued by PARP1 knockdown (Fig. 5 f). We also observed that the decrease in the mitochondrial membrane potential and mitochondrial complex I activity in SH-SY5Y cells induced by cFn was reversed by PARP1 depletion (Fig. 5 g-h). Overall, PARP1 activation plays an essential role in cFn-exacerbated α-syn abnormalities and mitochondrial dysfunction.

Fig. 4.

cFn induced PARP1 activation via EIIIA-integrin α4β1 interaction, which subsequently contributed to pathological α-syn aggregation and mitochondrial dysfunction in SH-SY5Y cells. (a) Representative western blots and quantification of nNOS and PARP1 in SH-SY5Y cells coated with cFn and treated with different receptor inhibitors (five independent replicates). (b) Quantification of NO release in SH-SY5Y cells treated with cFn or different receptor inhibitors (five independent replicates). (c) Representative western blots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in SH-SY5Y cells (five independent replicates). (d–e) Representative images and quantification of PARP1 (d) and γ-H2AX (e) staining in SH-SY5Y cells 12 ​h after L-NAME or PBS pretreatment (five independent replicates). DAPI (blue) was used for nuclear staining. Scale bar: 100 ​μm. (f) Representative western blots and quantification of PAR and PARP1 in SH-SY5Y cells coated with cFn or AF38Pep at different concentration (five independent replicates). (g) Representative coimmunoprecipitation images showing an interaction between Fn and the integrin α4 subunit in SH-SY5Y cells, and the quantification of Fn and α4 subunit level in the coimmunoprecipitation samples was performed (five independent replicates). All data are presented as the mean ​± ​s.e.m. Differences among 2 groups were analysed by Unpaired Student's t-test, while differences among multiple groups were analysed by ANOVA followed by Tukey's post hoc test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, and ns, not significant.

Fig. 5.

cFn induced α-syn aggregation and mitochondrial dysfunction via promoting PARP1 activation. (a) Representative western blots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in α-syn-overexpressing SH-SY5Y cells coated with cFn or transfected with PARP1 siRNA (five independent replicates). (b) Representative western blots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in the SNpc of mice injected with cFn or treated with PARP1 inhibitor Talazoparib (n ​= ​5). (c) Representative images and quantification of the relative fluorescence intensity of Ser129-phosphorylated α-syn in the SH-SY5Y cells (five independent replicates). Scale bar: 100 ​μm. (d–e) Representative images and quantification of the relative fluorescence intensity of Ser129-phosphorylated α-syn and PARP1 in the SNpc of mice with cFn injection or PARP1 inhibition (n ​= ​5). Scale bar: 100 ​μm. (f) Representative images and quantification of the relative fluorescence intensity of COX IV in SH-SY5Y cells coated with cFn or transfected with PARP1 siRNA (five independent replicates). Scale bar: 100 ​μm. DAPI (blue) was used for nuclear staining in each merged image. (g) The mitochondrial membrane potential of SH-SY5Y cells coated with cFn or transfected with PARP1 siRNA was measured by JC-1 staining and flow cytometry. The fluorescence ratio of JC-1 polymer/monomer was determined (three independent replicates). (h) The mitochondrial complex I activity of each group was measured with a Biotek multifunctional microplate reader (three independent replicates). All of the data are presented as the mean ​± ​s.e.m., ANOVA with Tukey's post hoc test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, and ns, not significant.

Brain depletion of cellular fibronectin in MPTP-treated mice rescues α-syn aggregation and mitochondrial dysfunction

We further investigated whether depletion of astrocyte derived-cFn in the brain has neuroprotective effects. To minimize only the effect of cFn rather than pFn, we injected pHBAAV-U6-Fn1-MCS-CMV-Luc-WPRE into the lateral ventricles of the mice to prevent cFn expression in various cell types such as astrocytes and endothelial cells in the brain. Since pFn in the brain originates from plasma, the level of pFn in the SNpc of PD mice cannot be eliminated via intraventricular injection. To determine the knockdown efficiency of AAV injection, 3 sequences of Fn1-shRNA were stereotactically injected into MPTP-treated mice. Mice in different groups were sacrificed 5 weeks post shRNA injection. Through western blotting, we verified that the level of Fn was reduced by approximately 70 ​% in the SNpc of MPTP-treated mice after shRNA3 injection (Fig. S4 b). Therefore, shRNA 3 were selected for further studies as the AAV-Fn1 shRNA. Furthermore, we confirmed that AAV-Fn1 shRNA injection (shRNA3) in MPTP-treated mice significantly reduced the expression of Fn in GFAP-positive astrocytes (Fig. S4 c). Meanwhile, the expression of Fn in CD31-positive endothelial cells decreased significantly after AAV-Fn1 shRNA injection (Fig. S4 d). However, compared to healthy mice, the fluorescence intensity of Fn co-localized with CD31⁺ cells in the SNpc of MPTP ​+ ​AAV-Fn1 shRNA mice remained significantly higher. The increase in this fluorescence intensity may be attributed to the elevated levels of pFn induced by BBB disruption in MPTP models, which cannot be depleted by AAV injection. We further observed that AAV-Fn1 shRNA injection significantly rescued the elevated levels of PARP1, PAR and γ-H2AX in the SNpc of MPTP-treated mice (Fig. 6 a). TX-insoluble α-syn oligomers and Ser129 phosphorylated α-syn in the SNpc were also reduced in MPTP model mice after cFn depletion (Fig. 6 b). In addition, the release of AIF into the nucleus was suppressed in the SNpc of MPTP-treated mice when cFn was depleted (Fig. S5 a).

Fig. 6.

cFn depletion protected MPTP-treated mice from mitochondrial dysfunction and α-synuclein phosphorylation. (a) Representative immunoblots and quantification of PARP1, γ-H2AX, and PAR in the SNpc of the control group, MPTP group, MPTP ​+ ​AAV-Fn1-shRNA group and MPTP ​+ ​AAV-NC (Negative control AAV, pHBAAV-U6-scramble-MCS-CMV-Luc-WPRE) group (n ​= ​5). (b) Representative immunoblots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in the SNpc of each group. AAV-NC was used as a control AAV (n ​= ​5). (c) Representative images of morphological changes in mitochondria in midbrain neurons from each group. Scale bar: 1 ​μm (upper), 500 ​nm (lower). (d) Representative immunohistochemical images and quantification of TH ​+ ​cells in the SNpc of each group (n ​= ​5). Scale bar: 100 ​μm. (e–g) Quantification of turn time (e) and total time (f) in the pole test, as well as the latency to fall in the rotarod test (g), in the control, MPTP, MPTP ​+ ​AAV-Fn1-shRNA and MPTP ​+ ​AAV-NC groups (n ​= ​10). All of the data are presented as the mean ​± ​s.e.m., ANOVA with Tukey's post hoc test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, and ns, not significant.

Immunofluorescence staining further confirmed the reduction in the number of Ser129 phosphorylated α-syn-containing and PARP1-positive cells in the SNpc of PD mice when cFn expression was knocked down in the brain (Fig. S5 b‒c). Moreover, cFn depletion decreased neuronal mitochondrial abnormalities such as mitochondrial cristae fracture and vacuolation and rescued the loss of TH ​+ ​neurons in the midbrains of MPTP-treated mice (Fig. 6 c-d).

The motor behaviour of the different groups was examined via the rotarod test and pole test. When MPTP-treated mice were injected with AAV-Fn1 shRNA, the total time spent on the rotarod test significantly increased, whereas both the turn time and total time spent on the pole test decreased (Fig. 6 e‒g). In general, we verified that reducing cFn production in the brain can rescue PARP1 activation, α-syn abnormalities and mitochondrial dysfunction in the SNpc, thus alleviating the loss of dopaminergic neurons and motor deficits.

Cellular fibronectin leads to the elevation of free fatty acids and stearoyl-CoA-desaturase expression via integrin α4β1 mediation and subsequently exacerbates α-syn abnormalities

Our study demonstrated that the level of pathological α-syn in cFn-coated α-syn-overexpressing SH-SY5Y cells can be significantly reduced by PARP1 knockdown. However, after PARP1 depletion, both the level of insoluble α-syn oligomers and the level of Ser129 phosphorylated α-syn induced by cFn were still 2-fold greater than those in untreated cells. Therefore, we propose that in addition to promoting PARP1 activation, cFn may contribute to pathologic α-syn accumulation through alternative mechanisms. Given that lipid metabolic disturbance is closely associated with α-syn aggregation, we further assessed the effects of cFn on lipid metabolism. Through LC‒MS/MS analysis, we observed that triacylglycerol (TAG) and free fatty acids (FAs), such as palmitoleic acid (16:1) and oleic acid (18:1), which are MUFAs, were significantly increased in the ipsilateral SNpc of cFn-injected mice (Fig. 7 a‒d). As excess FAs can be converted to TAGs and ultimately accumulate in lipid droplets (LDs) [12,38], we considered that cFn may promote the synthesis of FAs in neurons. To further verify our hypothesis, we conducted Oil Red O staining in vitro and found that coating with cFn significantly increased LDs accumulation in SH-SY5Y cells, which could be alleviated by integrin α4β1 inhibition (Fig. 7 e). Moreover, we confirmed that cFn significantly increased TAG and FAs levels in SH-SY5Y cells, which could be rescued by inhibiting integrin α4β1 (Fig. 7 f-g). These results suggested that cFn also induced FAs synthesis and subsequently increased TAG via integrin α4β1. Previous studies revealed that sterol CoA desaturase (SCD) facilitates the formation of MUFAs, primarily palmitoleic acid and oleic acid [13,39]. Moreover, excessive MUFAs lead to increased α-syn toxicity and the formation of α-syn aggregates [12]. Thus, on the basis of our results, in addition to increasing PARP1, cFn may induce SCD elevation by binding to integrin α4β1.

Fig. 7.

cFn induced FAs and TAGs elevation by increasing SCD, which contributed to pathological α-syn aggregation. LC‒MS/MS Analysis generated data were performed in A-D. (a–b) Changes in FAs in the SNpc of cFn-injected mice compared with those in healthy mice were determined by hierarchical clustering analysis (a) and matchstick analysis (b) (n ​= ​6), (c–d) Changes in TAGs in the SNpc of cFn-injected mice compared with healthy mice were determined by hierarchical clustering analysis (c) and matchstick analysis (d) (n ​= ​6). In the image of hierarchical clustering analysis, red plots represent upregulation, and blue plots represent downregulation. In the matchstick analysis, red lines represent upregulation, while blue lines represent downregulation. ∗P ​< ​0.05, ∗∗P ​< ​0.01. FAs species was described in Table S3; (e) Representative Oil Red O staining images and relative quantification of the positive area of lipid droplets in SH-SY5Y cells coated with cFn or treated with an integrin α4β1 inhibitor (five independent replicates). Oli Red O staining was detected in bright field channel. Scale bar: 25 ​μm. (f–g) The levels of TAGs (f) and FAs (g) in SH-SY5Y cells coated with cFn or treated with an integrin α4β1 inhibitor were measured (three independent replicates). (h) Representative immunoblots and quantification of SCD in the SNpc of the control, MPTP, MPTP ​+ ​AAV-Fn1-shRNA, and MPTP ​+ ​AAV-NC groups (n ​= ​5). (i) Representative immunoblots and quantification of SCD in SH-SY5Y cells coated with cFn or treated with an integrin α4β1 inhibitor (five independent replicates). (j) Representative western blots and quantification of SCD in the SNpc of mice injected with cFn or treated with AF38Pep (five independent replicates). (k–l) Representative coimmunoprecipitation images showing an interaction between SCD and the α4 subunit in the SNpc of healthy mice or cFn injected mice, and the quantification of α4 subunit or SCD level in the coimmunoprecipitation samples was performed (n ​= ​5). (m) Representative western blots and quantification of the TX-insoluble α-syn oligomer and Ser129-phosphorylated α-syn in α-syn-overexpressing SH-SY5Y cells coated with cFn or treated with an SCD inhibitor (five independent replicates). All data are presented as the mean ​± ​s.e.m. Differences among 2 groups were analysed by Unpaired Student's t-test, while differences among multiple groups were analysed by ANOVA followed by Tukey's post hoc test. ∗P ​< ​0.05, ∗∗P ​< ​0.01, ∗∗∗P ​< ​0.0005, ∗∗∗∗P ​< ​0.0001, and ns, not significant.

We further verified that cFn depletion through AAV-Fn1 shRNA injection in the SNpc of MPTP-treated mice significantly reduced the level of SCD in the ipsilateral SNpc (Fig. 7 h). In SH-SY5Y cells, the level of SCD was significantly elevated by coating with cFn, which was eliminated by integrin α4β1 inhibition or EIIIA-Fn interaction blockade by AF38Pep (Fig. 7 i-j). LPS-treated astrocyte matrices also elevated the level of SCD in SH-SY5Y cells, which can be attenuated by EIIIA-Fn interaction blockade (Fig. S3 c). Coimmunoprecipitation revealed that SCD interacted with the integrin α4 subunit rather than the downstream members induced by cFn, such as PARP1 and α-syn (Fig. 7 k-l), moreover, increased interaction between SCD and α4 subunit was performed in the SNpc of cFn-injected mice (Fig. 7 k-l), suggesting that integrin α4β1 may mediate the protein expression of SCD via direct interactions. Furthermore, we determined whether SCD contributes to cFn exacerbating α-syn abnormalities. After SCD inhibitor MK-8245 treatment, cFn-induced increases in the levels of TX-insoluble α-syn oligomers and Ser129 phosphorylated α-syn in WT α-syn-overexpressing SH-SY5Y cells were reduced by almost 40 ​% (Fig. 7 m). Overall, these results demonstrated that integrin α4β1 may interact with SCD, and the binding of cFn and integrin α4β1 may increase the level of SCD, subsequently leading to FAs and TAGs elevation and LDs accumulation, ultimately promoting pathologic α-syn accumulation.

Discussion

Here, our study demonstrated that cFn, which is produced mainly by inflammation-induced astrocytes, is increased in the SNpc of MPTP-treated models and enhances the level of PARP1 via integrin α4β1, subsequently aggravating pathological α-syn aggregation and mitochondrial dysfunction in neurons. Our results provide evidence that cFn mediates FA homeostasis in neurons by increasing the level of SCD and promoting the biosynthesis of MUFAs, which synergistically exacerbates α-syn abnormalities. Thus, therapies aimed at modulating cFn can potentially preserve mitochondrial function and attenuate α-syn aggregation and lipid metabolic disturbances.

Numerous studies have reported the accumulation of Fn in multiple sclerosis, chronic relapsing experimental autoimmune encephalomyelitis (EAE), and toxin-induced lesions, which can result from BBB disruption or neuroinflammation [14,40,41]. Astrocyte activation and BBB leakage also occur in MPTP-treated PD mice [2]; however, whether Fn accumulates in the SNpc of PD mouse models and the role of Fn in PD pathogenesis are unknown. Firstly, we verified that Fn aggregates were deposited in the SNpc of MPTP-treated mice. Moreover, both astrocyte-derived Fn and pFn or endothelial cell-derived Fn were elevated in the brains of PD model mice induced by MPTP, which can form aggregates. In agreement with these results, we further investigated whether Fn contributes to PD pathogenesis and defined the presence of the variants of Fn that are pathogenic in the brain. In vitro studies revealed that cFn, rather than pFn, induced neuronal mitochondrial dysfunction and exacerbated α-syn aggregation. Moreover, cFn, rather than pFn, induced PARP1 activation in SH-SY5Y cells, which impaired NAD ​+ ​pools and led to α-syn aggregation via increased PAR. Previous studies revealed that astrocyte-derived cFn inhibits remyelination [14,42]; however, the differences between the pathological roles of cFn and pFn remain elusive. As cFn rather than pFn contains alternatively spliced EIIIA and EIIIB domains [33], we considered that the binding of EIIIA and integrin receptors may lead to cFn-induced neuronal damage, which is not elicited by pFn. Consistent with our hypothesis, integrin α5β1 inhibition, which prevents the interaction between integrin α5β1 and the RGD motif in both cFn and pFn [43], cannot suppress the PARP1 activation induced by cFn. Interestingly, preventing the EIIIA domain of cFn from binding with integrin α4β1 rather than TLR4 rescued cFn-induced PARP1 and nNOS, suggesting that cFn induces nNOS expression and NO release via integrin α4β1, which contributes to DNA oxidative damage and PARP1 activation [37,44]. Previous studies reported that integrin α4β1, expressed by microglia, contributes to oxidative stress and the polarization of microglia when binding with vascular cell adhesion molecule 1 (VCAM1) [45]; however, whether integrin α4β1 in neurons participates in neuronal damage remains unknown. We found that the neuronal integrin α4β1 plays an essential role in neuronal NO production. Similarly, studies have reported that integrin α9β1, which is absent in neurons and interacts with cFn EIIIA domains, promotes NO biosynthesis in human colon adenocarcinoma cell lines [36,46]. V-containing Fn isoforms (V regions) in alternatively spliced IIICS regions are also recognized by α4β1 integrin, which is almost 50 ​% absent in pFn [33]. Therefore, future studies should distinguish the functional roles of the EIIIA domains and V regions in neuronal integrin α4β1-mediated pathology.

Consistent with our in vitro studies, stereotactically injecting cFn into the SNpc of MPTP-treated mice elevated the levels of PARP1 via DNA strand damage. cFn injection also promoted pathological α-syn aggregation, mitochondrial disruption and TH ​+ ​neuron damage in the SNpc of both PD mice and healthy mice, suggesting that cFn induced by neuroinflammation may amplify MPTP-induced pathology and trigger neurodegeneration. We further verified that PARP1 knockdown reduced cFn-induced exacerbation of α-syn accumulation and mitochondrial dysfunction in SH-SY5Y cells, demonstrating that cFn at least partially affects PARP1 activation in neurons. Previous studies revealed that Fn aggregates induce classically activated phenotypic features in microglia [16], which may also lead to neuronal damage via the inflammatory response in vivo. However, we confirmed that mitochondrial dysfunction and α-syn abnormalities in neurons can be induced by cFn without affecting other cell types.

To validate the neuroprotective effects of cFn depletion, we used AAV-Fn1-shRNA to block cFn production in the brains of MPTP-treated mice. Notably, these in vivo results are consistent with the findings of the in vitro experiments. PARP1 activation and α-syn abnormalities were rescued by genetically depleting cFn in the brain. Mitochondrial ultrastructure disruption, motor deficit and dopaminergic neuron impairments in MPTP-treated mice were also attenuated by cFn depletion, indicating that cFn deposition contributes to MPTP-induced neuronal impairments, which could be a potential target for delaying neurodegeneration. Preventing the deposition of cFn or promoting cFn clearance in the brain may effectively alleviate mitochondrial dysfunction and α-syn accumulation in PD patients.

Our results demonstrated that the levels of α-syn oligomers and Ser129 phosphorylated α-syn in cFn-coated SH-SY5Y cells depleted of PARP1 were still greater than those in untreated cells, suggesting that cFn also induces α-syn aggregation through other mechanisms. Recent studies have revealed a close association between α-syn aggregation and lipid metabolism, and our metabolomics analysis revealed changes in lipid metabolic components in the SNpc of cFn-injected mice. Therefore, we hypothesized that the imbalance in lipid metabolic homeostasis in neurons can be induced by cFn and further act in parallel with PARP1 activation to exacerbate α-syn aggregation. We found that cFn injection increased FA and TAG levels in the SNpc and further verified these results in SH-SY5Y cells. Notably, cFn injection increased OA levels in vivo, which enhances α-syn toxicity and promotes pathological α-syn aggregation [12,47]. Previous studies have shown that OA elevation may promote α-syn membrane binding, which induces the formation of aggregated α-syn inclusions [48]. The binding of α-syn to OA, which is integrated as fatty acyl side chains into membrane lipids, induces the local sequestration of α-syn monomers into an aggregated state [48]. Studies have also shown that α-syn aggregation occurs via liquid–liquid phase separation (LLPS), which can be promoted by lipid–membrane interactions [49]. However, whether FAs induced by cFN via SCD elevation promote α-syn LLPS needs further investigation. α-syn LLPS is regulated by the N-terminus and hydrophobic ‘nonamyloid-β component’ domain, which acts as an initial step towards α-syn aggregation [49]. Under normal physiological conditions, FAs such as OA are stored as TAGs in LDs to protect against FA toxicity¹². However, LDs biogenesis pathways can be overwhelmed when excessive FAs are produced and converted to TAGs [50]. Therefore, we believe that cFn may induce excessive FA synthesis and ultimately lead to LDs formation in neurons. Consistently, our results suggested that cFn elevated the level of SCD both in vivo and in vitro. As integrin α4β1 inhibition obviously blocked the increase in pathological α-syn, we hypothesized that cFn-induced SCD elevation may also be mediated by integrin α4β1. This hypothesis was further verified. Moreover, we found that the α4 subunit may interact with SCD, suggesting that the binding of cFn and integrin α4β1 may mediate the molecular interaction between the α4 subunit and SCD and subsequently change the level of SCD. Whether the interaction between integrin and SCD leads to a change in enzymatic activity or promotes protein modification needs further investigation.

In conclusion, as a component of the extracellular matrix, excessive cFn produced by neuroinflammation-induced astrocytes in PD may act as a trigger or amplifier of dopaminergic neuron damage, extending beyond the inflammatory response. Therefore, the extracellular microenvironment may play a critical role in regulating neuronal functions [51]. Our study identified cFn as a potential therapeutic target in PD progression. Moreover, a novel therapeutic strategy targeting the interaction between cFn and integrin α4β1 may be beneficial to patients with PD [52]. Since integrin α4β1 inhibition may disturb neuronal physiological functions such as neurodevelopment [53,54], specific competitive antagonists targeting the EIIIA domain warrant further investigation. Disrupting the biosynthesis of cFn in different cell types in the brain through genetic intervention may also effectively alleviate the loss of dopaminergic neurons, indicating its potential as a new therapeutic strategy for PD.

Data availability

All data are available in the main text or the supplementary materials. The data that support the findings of this study are available from the Zifeng Huang and corresponding author [Qing Wang], upon reasonable request.

Author contributions

Conceived and designed the study: ZFH, HZ, YQL, YYY, JLZ and QW. Performed the study: ZFH (allocated treatments), MWZ (performed the treatments and collected samples), HZ (prepared tissues), YYY (tested the samples), and QW (collected data). Revised the paper for intellectual content: ZFH, HZ, YYY, JLZ, YQL, MWZ, YHY, SZZ, BX, KLJ, CD, KPD, KRC, EK-T and QW. Data statistics and analysis: ZFH, CL and QW. Wrote the paper: ZFH, HZ, YYY, YQL, MWZ, BX, CD, KLJ, ZDZ, CL, XYC, KPD, KRC, EK-T and QW. ZFH and QW had accessed and verified the data reported in the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (NO: U24A20694, 82071414, 82471433, QW), Scientific Research Foundation of Guangzhou (NO: 202206010005, QW), Science and Technology Program of Guangdong of China ​(NO: 2020A0505100037, QW), Guangdong Basic and Applied Basic Research Foundation (2023A1515111058, ZFH), Natural Science Foundation of Guangdong Province (2025A1515012456, YQL), National Medical Research Council (STaR and PD LCG 000207, EKT).

Declaration of Competing Interest

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.

Appendix A. Supplementary data

The following is the Supplementary data to this article: