Cell cultures in this study included transfectants derived from human H4 neuroglioma: “H4/V1S:SV2,” “H4/CBD-V1S:SV2-CBD,” “H4/V1S:SV2/LAMP1-eCFP/mCherry-galectin-3,” and a transfectant from BE(2)-M17 neuroblastoma cells—“3D5.” The purpose of using each cell line was described in the section of Results. All cell lines were maintained in OPTI-MEM (Invitrogen) medium containing 10% fetal bovine serum (Invitrogen) at 37°C with 5% CO₂ and 100% humidity. For live cell imaging with confocal microscopy, cells were cultured in Nunc^® Lab-Tek^® II chambered coverglass (Sigma-Aldrich). For the differentiation of human dopaminergic cell line BE(2)-M17-derived cells, the medium was replaced with Neurobasal medium (Invitrogen), 2% B-27 supplement (Invitrogen), 2 mM L-glutamine (Sigma-Aldrich), and 10 μM retinoic acid (Sigma-Aldrich).

Lentiviral plasmids carrying LAMP1-eCFP and mCherry-galectin-3 were described previously (Jiang et al., 2017). The lentiviral vector for CRISPR-Cas9 knockout of GBA1 was designed by VectorBuilder Inc. (Supplementary Figure 2). The sequences of two single guide RNAs (sgRNAs) were “AGACCAATGGAGCGGTGAAT” and “TGTGGTGAGTACT GTTGGCG,” respectively. The protocols used for the preparation of lentivirus carrying genes of interest were the same as described previously (Jiang et al., 2013).

Commercially available nanoparticles were Gold (Nanocs), TiO₂ (US Research Nanomaterials), ZnO (Inframat), Fe₃O₄ (US Research Nanomaterials), and SiO₂ (US Research Nanomaterials). The Alexa Fluor^TM 647-labeled SiO₂ nanoparticles were customized product (30 nm) from Nanocs. The chitin nanoparticles were customized (80 nm) from Nanoshel. To prepare fresh concentrated nanoparticle solutions, the original powder was weighed and added into 1 × PBS, then sonicated for 5 min at maximal power (Sonicator 3000, Misonix). The concentrated nanoparticle solution was further diluted into culture media to make final nanoparticle-supplemented media, which were sonicated for at least 1 min before cell treatment.

Cells were plated on coverglass with 8-well culture chamber at the same density overnight, then treated with different nanoparticles. The next day, the cells were exposed to 1 mM lysosomotropic detergent [L-leucyl-L-leucine methyl ester (LLME)] (Cayman Chemical) for 1–2 h to induce endolysosomal rupture. In parallel, sibling cultures without nanoparticles or LLME treatment were set as two different negative controls. After fixation with 4% paraformaldehyde (PFA), cells with and without αS inclusions were evaluated and photographed under a confocal microscope (Zeiss LSM 510, Carl Zeiss MicroImaging). For measuring the ratio of cells bearing αS inclusions, five fields (upper right, upper left, center, lower right, and lower left) with at least 90 cells were selected from each group for cell count.

Neuroblastoma BE(2)-M17D-derived cell model—3D5 (Ko et al., 2008) were differentiated and induced to express human wild-type αS, then exposed to media supplemented with and without SiO₂ nanoparticles for 1 day, followed by the induction of endolysosomal membrane rupture for 1–2 h. Cells were then harvested for protein extraction by extraction buffer [1% Triton X-100 (Tx) and 1% (v/v) protease inhibitor cocktail (Sigma) in 1 × PBS] to obtain Tx-soluble and Tx-insoluble fractions following a previous protocol (Bae et al., 2015). The same amount of proteins from different groups was mixed with loading buffer and then resolved by sodium dodecyl–sulfate polyacrylamide gel electrophoresis (SDS–PAGE) followed by the Western blotting.

The method for detecting the binding affinity of nanoparticles to αS was similar to that reported in a previous study (Hata et al., 2014) with minor modifications. Freshly prepared recombinant αS solution was mixed with different nanoparticles to get a final concentration of 0.5 μg/μl for αS and 200 μg/ml for both SiO₂ and chitin, respectively. A tube of αS solution with the same concentration of αS but without mixing with any nanoparticle was included as a control (Con). All samples were incubated at 37°C for 1 h with constant rotating. The αS protein bound to nanoparticles was isolated by centrifugation at 30,000 × g for 20 min at 4°C. The top layer of supernatant was removed to a new tube for measuring the concentration remained in solutions. The pellets containing a mixture of nanoparticles and particle-bound αS were washed with 1 ml PBS and centrifuged again for three times to remove residual unbound αS protein, then mixed with 10% SDS buffer, followed by the addition of an equal amount of Laemmli sample buffer (Bio-Rad Laboratories) and boiled for 5 min at 95°C. The boiled samples were centrifuged, and the supernatants were used for SDS–PAGE. The gels were subjected to silver staining to show the bound αS in each group. A tube containing αS solution was saved before mixing with nanoparticles and used as a negative control.

Pellets of cultured cells (comparable cell number per group) were resuspended in 1 × PBS and then lysed by ultrasonication four times, followed by centrifugation at 1,500 × g for 10 min at 4°C. Supernatants collected from cultured cells were used for measuring the concentration of sphingosine according to the manufacturer’s instructions (Sphingosine ELISA Kit Lifespan Biosciences). Briefly, samples were added to a plate, followed by the addition of detection reagent A and 1 h of incubation at 37°C. After incubation, the reagents in the plate were removed and the plate was washed completely with buffer and then loaded with detection reagent B for 45 min of incubation at 37°C. At the end of incubation, sample wells were emptied and washed again, then loaded with TMB substrate for 10–20 min of incubation at 37°C, followed by the addition of stop solution and the measurement of optical density by a microplate reader (SpectraMax Paradigm, Molecular Devices).

Cells were rinsed with 1 × PBS, fixed in 4% PFA, and permeabilized with 0.1 M Tris-buffered saline (TBS; pH 7.6) containing 0.5% Triton X-100 for 5 min, then blocked with 3% goat serum in TBS, incubated with the primary antibody in TBS containing 1% goat serum overnight at 4°C and then with the secondary antibody for 1 h at room temperature. Immunolabeled cells were mounted in VECTASHIELD^® antifade mounting media with or without DAPI (Vector Laboratories), then examined under a confocal microscope. Primary antibodies included mouse against GBA (Abcam) and mouse against HA (Sigma), and the secondary antibodies include the Alexa Fluor 568 and the 647 anti-mouse (Thermo Fisher Scientific).

H4/V1S:SV2/LAMP1-eCFP/mCherry-galectin-3 cells with GBA1 deletion (GBA1-) or without (WT) were treated with mature fibrils derived from recombinant αS fused with HA tag (αSHA) as described previously (Jiang et al., 2008). Such αSHA fibrils were preincubated at 4°C for 2 days to facilitate the endocytosis seeding pathway because such incubation can significantly reduce the capability of αS fibrils in direct penetration of cell membrane according to our previous study (Jiang et al., 2017). Cells were fixed when GBA- cells formed enough seeded αS inclusions under confocal microscope. A portion of sibling cells from each group was subjected to immunocytochemical staining with antibody against HA tag to demonstrate the distribution of exogenous αSHA-fibrils (shown as Alexa Fluor^TM 647). Another portion of cells was subjected to electron microscopy (EM) samples process and subsequent conventional EM and immuno-EM. Immuno-EM was performed by immunolabeling Venus (orb334993, Biorbyt) and HA tag with a gold of 15 nm (25806, EMS) and 2 nm (25125, EMS) in samples, respectively.

Cells for transmission EM were fixed with 2% glutaraldehyde, 2% PFA in 0.1 M PBS; cells for Immunoelectron Electron Microscopy (immunoEM) were fixed with 4% PFA in 0.1 M PBS. For EM, cells were postfixed in 1% OsO₄; washed three times in distilled water; stained with 1% uranyl acetate in 50% ethanol; and dehydrated with 70, 80, 95, and 100% ethanol sequentially. The cells were then treated with propylene oxide, infiltrated, and embedded in Epon 812 (Polysciences). For immunoEM, the cells were dehydrated in 30, 50, 70, and 90% ethanol, sequentially, then 90% ethanol-LR White (1:1) and 90% ethanol-LR White resin (1:2). They were then infiltrated and embedded in pure LR White. Ultrathin sections were cut from the Epon 812 or LR White-embedded samples by Leica Ultramicrotome. Ultrathin sections were examined after counterstaining with uranyl acetate and lead citrate. The sections were examined and photographed with a Philips 208S electron microscope.

Data from at least three sets of independent experiments were analyzed by one-way ANOVA with Dunnett’s post hoc test or Student’s t-test for the comparison of groups >3 and (=2, respectively, to determine statistical significance.

First, we want to know if nanoparticles from different materials are able to induce the formation of αS aggregates within cells. For this purpose, we tested nanoparticles that have been shown to promote αS to aggregate in salt solutions through direct interaction or to accumulate in cells through an indirect mechanism (Alvarez et al., 2013; Joshi et al., 2015; Xie and Wu, 2016; Mohammadi and Nikkhah, 2017; Tahaei Gilan et al., 2019; Khodabandeh et al., 2020). The nanoparticles tested were as follows: SiO₂, Ti₂O₃, Fe₂O₃, and ZnO. To maximize the comparability of results from different nanoparticles, only those with similar size (∼30 nm) were used.

The cell model used for the evaluation of nanoparticle-induced αS aggregation was derived from H4 neuroglioma cell line (ATCC^® HTB-148^TM). H4 cells were transfected to stably express the N-terminal half of Venus YFP tagged to αS (V1S) and C-terminal half of Venus YFP tagged to αS (SV2). This transfectant, referred to as H4/V1S:SV2, is useful for monitoring the aggregation of αS in live cells in real time. Because binding between V1S and SV2 will reconstitute YFP fluorescence, the brightness of fluorescence emitted can be used to estimate the extent of αS aggregation (Jiang et al., 2017). H4/V1S:SV2 cells growing in eight-well-chambered culture coverglass were treated with the same concentration of different nanoparticles (30–40 μg/ml) and then observed under confocal microscope daily to monitor the formation of αS inclusions.

Our results showed no evidence of αS inclusions in cells after 1 week of nanoparticle treatment. Since using H4/V1S:SV2 cells to visualize αS aggregation induced by exogenous seeds is a well-established experiment in our laboratory (Jiang et al., 2017), and the time for the induced αS inclusion to appear in this cell line has never been more than 3 days, we deduced that such negative results could be due to insufficient nanoparticles escaping the endocytic pathway to interact with cytoplasmic αS. If this is the case, cells with impaired endolysosomes should allow more nanoparticles to enter the cytoplasm. Since lysosomal dysfunction has been suggested to be an important pathogenic mechanism in PD (Dehay et al., 2010, 2012; Moors et al., 2016), we wondered if αS aggregation could be induced by nanoparticles in cells with endolysosomal impairment. For this purpose, the lysosomotropic detergent LLME that impairs endolysosomal function by irreversible accumulation in acidic compartments, leading to damage of endolysosomal membranes (Uchimoto et al., 1999) was chosen.

Cells treated with different nanoparticles for 1 day were exposed to 1 mM LLME to disrupt the integrity of endolysosomal membranes. As expected, after 1–2 h in LLME, all nanoparticle-treated cells developed αS inclusions. We observed differences in the ability of different nanoparticles to induce the formation of αS inclusions (Figures 1A,B). Moreover, nanoparticle-treated cultures exposed to LLME had fewer cells compared to those without such exposure, and the difference was statistically significant. Exposing cells to LLME in the absence of nanoparticles for 1–2 h did not result in either formation of αS inclusions or cell death, indicating that the nanoparticle-induced αS inclusions are cytotoxic (Figure 1C).

Next, we explored if nanoparticle-induced αS inclusions were associated with the rupture of endolysosomal membranes. For this study, we focused on SiO₂ nanoparticles, because cells treated with SiO₂ had the most αS inclusions (Figures 1A,B). Moreover, SiO₂ nanoparticles are less expensive, and they are the most common nanoparticle used in humans, such as in cosmetics (Napierska et al., 2010; Murugadoss et al., 2017).

Since the accumulation of galectin-3 on endolysosomal membrane is an indicator of endolysosomal rupture (Flavin et al., 2017), we introduced mCherry-tagged galectin-3 (mCherry-galectin-3) and eCFP-tagged LAMP1 (LAMP1-eCFP) into H4/V1S:SV2 cells to generate a new cell line referred to as H4/V1S:SV2/LAMP1-eCFP/mCherry-galectin-3. Cells from these transfectants were treated with Alex Fluor^TM 647-labeled SiO₂ nanoparticles (SiO₂-AF647) for 1 day, then exposed to LLME. Under confocal microscopy, we found that all induced αS inclusions were closely associated with galectin-3 and LAMP1 as reflected by the colocalization of Venus, mCherry, and eCFP (arrows in Figure 2). In contrast, cells treated with SiO₂ nanoparticles alone showed the retention of nanoparticles in endolysosomes, no rupture of endolysosomes, and no αS inclusions, as reflected by colocalization between Alex Fluor^TM 647 and eCFP without the accumulation of mCherry and Venus (denoted by arrow heads in Figure 2). These results strongly supported that endolysosomal membrane rupture may play a role in nanoparticle-induced formation of αS inclusions.

It is known that the size of nanoparticles can highly influence their in vivo pharmacokinetics and cellular interaction (e.g., cellular uptake, biodistribution, and circulation half-life) (Hoshyar et al., 2016). Therefore, we studied the influence of nanoparticle size on the formation of αS inclusions. H4/V1S:SV2 cells were treated with SiO₂ nanoparticles of three different sizes (8, 25, and 65 nm), respectively, then exposed to LLME. For this experiment, the concentration of nanoparticles was 200 μg/ml, which is over-saturated because nanoparticles of all three different sizes at lower concentration did not show a consistent proportion of αS inclusion-bearing cells. Using saturation levels of nanoparticles excludes the possibility that observed differences between different-sized nanoparticles were due to the presence of more small nanoparticle particles than large ones at a given concentration. Our results showed that the ratio of cells containing αS inclusions to total cells was about 16, 58, and 85%, respectively, for cells treated with nanoparticles of 65, 25, and 8 nm in size (Figures 3A,B). In addition, we found that in cultures treated with smaller nanoparticles, more cells contained multiple αS inclusions. The ratios of cells containing more than 10 inclusions of αS to total inclusion-bearing cells were 0, 13, and 55%, respectively, for those treated with nanoparticles of 65, 25, and 8 nm in size (Figures 3A,C). These results indicated that nanoparticles induce the formation of intracellular αS inclusions in a size-dependent manner.

Next, we determined if nanoparticle-induced aggregates contain detergent-insoluble αS. Because H4/V1S:SV2 cells express a Venus tag-fused αS, to rule out the possible promotive effect of Venus on the formation of detergent-resistant αS, we used a cell line, 3D5 (Ko et al., 2008), expressing unlabeled αS. Cells of this model were derived from a neuroblastoma BE(2)-M17 cell line. They inducibly express wild-type human αS through a Tetoff mechanism and display neuronal phenotypes upon retinoic acid-induced differentiation (Ko et al., 2008). 3D5 cells were differentiated and induced to express αS in media with retinoic acid, but without Tet (see Figure 4A). On the 5th day, half of the cultures were treated with SiO₂ nanoparticles for 24 h. Subsequently, cultures with and without nanoparticle treatment were treated with LLME for 1–2 h. The other half served as control. The four groups of cells were referred to as Con (without any treatment), LLME (treated with LLME only), SiO₂ (treated with SiO₂ nanoparticles only), and SiO₂/LLME (treated with both SiO₂ nanoparticles and LLME), respectively. They were harvested for protein extraction to separate Triton detergent (Tx)-soluble and insoluble fractions. Both fractions were then analyzed by SDS–PAGE and Western blotting for the detection of αS. The results showed that SiO₂/LLME had the most αS oligomers in Tx-soluble fraction; and only SiO₂/LLME contained αS aggregates in Tx-insoluble fractions (Figure 4C). We further tested if phosphorylation on those αS aggregates occurs during the treatment. Results showed that the form of αS phosphorylated at serine 129 was also evidently increased in the group of SiO₂/LLME in both fractions (Figure 4C). Therefore, in the presence of LLME, nanoparticles can induce our cell model to form detergent-insoluble αS aggregates, which is accompanied by an increase of pathological form of αS (phosphorylation at serine 129) (Bernal-Conde et al., 2019). Moreover, immunoblotting (Figure 4C) demonstrated the presence of higher cleaved Caspase 3 in SiO₂/LLME than other samples, suggesting the formation of αS aggregates was associated with apoptotic cell death.

Since previous studies have shown that the effects of nanoparticles on αS aggregation are associated with their mutual binding affinity (Alvarez et al., 2013; Mohammadi and Nikkhah, 2017; Tahaei Gilan et al., 2019), we investigated if binding affinity plays a role in the formation of αS inclusions in cells after the internalization of nanoparticles. To answer this question, we designed an experiment that used nanoparticles with and without strong binding affinity to αS as positive and negative controls. Therefore, we compared the affinity to αS between nanoparticles from different materials and chose SiO₂ and chitin nanoparticles with similar size (80 nm) as the two controls because binding studies showed that there was considerable αS bound to SiO₂ nanoparticles, while only negligible αS bound to chitin nanoparticles (Figure 5A).

Next, we establish a new cell line that stably co-expresses chitin binding domain (CBD)-tagged V1S at N-terminus (CBD-V1S) and SV2 at C-terminus (SV2-CBD), referred to as “H4/CBD-V1S:SV2-CBD” (see Supplementary Figure 1). Due to the nature of CBD (Hashimoto et al., 2000), both CBD-V1S and SV2-CBD expressed in this cell line can specifically bind to exogenous chitin nanoparticles. If binding affinity to αS is a determining factor for the induction of αS inclusions by nanoparticles, H4/CBD-V1S:SV2-CBD cells with endolysosomal impairment should form αS inclusions upon the treatment with chitin nanoparticles and LLME because CBD fused αS is able to specifically bind chitin through its CBD tag. In contrast, the number of αS inclusions induced by chitin nanoparticles in H4/V1S:SV2 cell line with lysosomal impairment should be significantly less because there is only negligible binding between αS and chitin due to the absence of CBD. However, SiO₂ nanoparticles should induce αS inclusions in both cell lines due to their significantly higher affinity to αS. Results from the cell-based study were exactly consistent with our expectations (see Figures 5B,C), strongly supporting the hypothesis that binding affinity of nanoparticles to αS determines their capability to induce αS inclusions in cells.

The effect of LLME on endolysosomal impairment prompted us to study if there were endogenous lysosomotropic substances in human cells that are associated with PD. We found sphingosine to be a potential candidate. Sphingosine is an endogenous biomolecule significantly increased in patients with Gaucher disease (GD) due to the Glucocerebrosidase (GCase) deficiency in this disease (Mistry et al., 2014). It is also a lysosomotropic reagent similar to LLME as its accumulation in cells leads to the formation of dilated endolysosomes (Lima et al., 2017). GD is the most common of the lysosomal storage diseases, and it is caused by a hereditary deficiency of the enzyme GCase, which is encoded by a gene named GBA1. Interestingly, mutation of GBA1 gene recently emerged as a common genetic risk associated with PD. Approximately 5% of patients with PD carry a GBA1 mutation, compared to <1% of the control population (Stoker et al., 2018). Moreover, a decrease in Gcase activity has been detected in idiopathic brain tissue of PD (Chiasserini et al., 2015; Parnetti et al., 2017). Therefore, we hypothesized that GCase deficiency may render cells more susceptible to the formation of αS inclusions in response to nanoparticle treatment.

To test this hypothesis, cells with GBA1 gene deletion were generated and used to assess αS aggregation in response to nanoparticles. As shown in Figure 6A, H4/V1S:SV2/LAMP1-eCFP/mCherry-galectin-3 cells were infected with lentivirus carrying GBA1-knockout and control vectors, referred to as “GBA1-” and “WT.” After 3 days of infection, a portion of sibling cells from WT and GBA1- were harvested to evaluate the effect of GBA1 deletion on the level of GCase expression (Figures 6B,C) and sphingosine production (Figure 6D). A part of sibling cultures from “WT” and “GBA1-” groups were then treated with Alex Fluor^TM 647-labeled SiO₂ nanoparticles to derive two subgroups, referred to as “WT/SiO₂” and “GBA1-/SiO₂.” After 1 day, “WT” cells with and without SiO₂ nanoparticle treatment were exposed to sphingosine (Sph, 20 μM) to induce endolysosomal impairment, which further derived two more subgroups referred to as “WT/Sph” and “WT/SiO₂/Sph.” An hour after sphingosine exposure, cells in all groups were fixed with 4% PFA and evaluated by confocal microscopy for the presence of αS inclusions. As expected, Sph treatment or GBA1 deletion both induced endolysosomal membrane rupture, reflected by colocalization of LAMP-eCFP and punctate mCherry-galectin-3 signals in cells. No αS inclusions were observed in cells with either endolysosomal rupture induction (GBA1- or WT/Sph) or nanoparticle treatment alone (WT/SiO₂). In contrast, cells with both nanoparticle treatment and endolysosomal impairment (WT/SiO₂/Sph and GBA1-/SiO₂) developed αS inclusions in numbers significantly different from the groups of WT/SiO₂, GBA1-, and WT/Sph (Figures 6E,F). Since cells with GBA1 deletion had a significantly higher level of sphingosine (Figure 6D), and either GBA1 deletion or exogenous sphingosine treatment can induce endolysosomal membrane rupture and facilitate nanoparticles to induce the formation of αS inclusions, it is reasonable to conclude that GCase deficiency may render cells more susceptible to the formation of αS inclusions in response to nanoparticle treatment due to the high risk of impairment in endolysosomal system.

To understand the ultrastructure of nanoparticle-induced αS inclusions in cells with Gcase deficiency, cultures belonging to the groups WT/SiO₂, GBA1-/SiO₂, and GBA1- were processed for EM examination. Our observations (Figure 7) revealed that only cells in GBA1-/SiO₂ group contained abundant nanoparticle-associated inclusions. In contrast, cells in WT/SiO₂ group contained abundant aggregated nanoparticles within endolysosomes (denoted by red cross in Figure 7), consistent with the view that nanoparticles are internalized via endocytosis (Behzadi et al., 2017). Moreover, those in GBA1- group had abnormally swollen membranous structures (denoted by red star in Figure 7), similar to those observed in neurons from animals with GBA1 knockout (Uemura et al., 2015; Schondorf et al., 2018). The inclusions detected in GBA1-/SiO₂ group consisted of a mixture of congregated nanoparticles and fragmented membranous structures in the absence of intact encircling membrane and filamentous structure (denoted by red diamond in Figure 7), indicating that the endolysosomes in cells were ruptured and no αS fibril was formed.

Subsequent immuno-EM staining revealed that the nanoparticle-associated membranous structures were immunoreactive to antibody against αS (denoted by red arrowhead in Figure 7), suggesting that the membrane-bound αS is a constituent of the nanoparticle-induced inclusion.

As we know, αS fibrils can recruit unfolded αS to amplify aggregates in buffer system; moreover, exogenous αS fibrils can seed the formation of Lewy body-like intracellular inclusions in cultured cells (Luk et al., 2009). If membrane-bound αS is a constituent of nanoparticle-induced αS inclusions, it should also exist in αS fibril-induced inclusions due to the mutual affinity between αS fibrils and αS molecules. To find the answer, we used mature fibrils derived from recombinant αS fused with HA tag (αSHA) to treat H4/V1S:SV2/LAMP1-eCFP/mCherry-galectin-3 cells with GBA1 deletion (GBA1-) or without (WT). Cells of GBA1- should be more susceptible to the formation of αS inclusions due to the endolysosomal impairment compared with those of WT. Results from confocal imaging showed that seeded inclusions were associated with membrane proteins, such as LAMP1 and galectin-3 reflected by their fused fluorescent protein eCFP and mCherry (in Figure 8A), respectively, suggesting the involvement of membrane structures and membrane rupture in seeding. In contrast, seeds in WT cells were only colocalized with endolysosomal marker LAMP1, suggesting its endocytic pathway of cellular uptake, but did not induce αS inclusion due to the confinement of nanoparticles in endolysosomes. Results from EM showed that αSHA fibrils were internalized by cells. In WT cells, they were only detained in endolysosomes; In GBA- cells, they formed membrane-associated inclusions in which the seeds were enclosed by clustered membrane structures. Moreover, immuno-EM revealed that the inclusions consisted of Venus-immunopositive membranous outer layer and HA-immunopositive inner core (denoted by red asterisk), suggesting that these inclusions were formed by the seeding between exogenous αSHA fibrils and cytoplasmic membrane-bound αS-Venus.