Prion protein and synuclein localize in TIA1-positive stress granules

Aberrant stress granules have been linked with the aggregation of many neurodegenerative disease-associated proteins including tau protein [7, 23]. Stress granules are formed via a process called liquid–liquid phase separation (LLPS). Prion protein has most of the physicochemical properties to carry out LLPS, including a significant level of intrinsic disorder and the ability to interact with RNA [24].

So, firstly we sought to explore the granule formation propensity of PrP and synuclein, using the catGRANULE algorithm [25]. A score of 2.13 was identified for PrP and 1.13 for synuclein protein (Fig. 1E & F). To ascertain whether endogenous PrP and synuclein protein can actually form granules, we co-stained them with a well-known stress granule marker, TIA1 (Cytotoxic granule associated RNA binding protein TIA1) in stress-induced and control (untreated) cells. In the current study, sodium aresnite, a well know inducer of oxidative stress was used for stress induction [26].

The PrP was localized both in the cytoplasmic and nuclear compartments in the control HeLa cells (Fig. 1A), in agreement with previous findings [27,28,29,30]. Although, it is well know that PrP is a cell surface protein, it has been also localized in the nucleus of neuronal and endocrine cells [27,28,29,30]. PrP also interacts with several intracellular proteins, most of them are found in the cytosol, mitochondria, and nucleus [30, 31].

Fig. 1

Localization of PrP and α/ꞵ synuclein in TIA1-positive stress granules after oxidative stress induction in HeLa cells. A-D Representative immunofluorescence results. Localization of endogenous TIA1, PrP, and α/β synuclein was investigated in sodium-arsenite-treated and untreated (control) HeLa cells using co-immunofluorescence and confocal microscopy. Counter-staining was performed to visualize cell nuclei, scale bar = 50 μm. White arrows show yellow cytoplasmic dots with TIA1 and PrP co-localization in granules. Three different antibodies were used against prion protein (SAF 32, SAF 70, and 12F10). E and F The granule formation propensity of PrP, and synuclein was assessed by the catGRANULE algorithm. G-J Quantification of the number of TIA1-positive stress granules

Stress-treatment (oxidative stress, 0.6 mM sodium aresnite, 60 min) induced cytoplasmic foci of PrP (Fig. 1A). To find out whether or not the PrP granules were actually stress granules, we co-stained HeLa cells with TIA1. There was partial colocalization between PrP and TIA1, in the control cells. After oxidative stress treatment, co-localization was particularly observed between PrP and TIA1 in the form of granules (yellow foci in the cytoplasm) (Fig. 1A, enlarged view). The amount of cytoplasmic TIA1 signal was low because labeling only detected endogenous TIA1, and TIA1 that was present in the cytoplasm was largely in the form of inclusions. Three different antibodies were used against prion protein (SAF32, SAF70, 12F10) to confirm the localization of PrP in stress granules (Fig. 1A-C, G-I).

The synuclein protein was relatively highly enriched in the nucleus with a punctate pattern in the cytoplasm in control cells, in agreement with previous findings in Hela and SH-SY5Y cells [17, 32,33,34,35]. Alpha-synuclein can be found in the nucleus [36] and it’s nucleus localization is regulated by numerous factors, including post-translational modifications [34, 37] and oxidative stress [32, 33, 38]. The synuclein protein exhibited a change in its sub-cellular localization upon sodium aresnite treatment (stress), forming cytoplasmic foci that overlapped with stress granules (Fig. 1D and J).

Overall, these findings indicate that stress induces cytoplasmic PrP and synuclein inclusions, which co-localize with TIA1-containing stress granules.

Comparative interactome mapping identifies distinct and converging molecular pathways

Next, to investigate stress-induced alterations in the interactome of these prion/prion-like proteins, we performed comparative interactomics under control and oxidative stress conditions. Parallel analysis of three prion/prion-like proteins (PrP, synuclein, and tau) using an identical workflow, provided a unique opportunity to compare their interactomes. To find out the physiologically relevant interactor of bait proteins endogenous, native, and untagged proteins were successfully immunoprecipitated (Additional file 1: Fig. S1 A & B). In immunoblotting analysis of tau-immunoprecipitates, cleavage fragments of tau protein were also observed (Additional file 1: Fig. S1B) in agreement with previous studies [39,40,41]. A significant portion of endogenous tau protein is present in the form of proteolytic fragments (< 45 kDa) in the human brain [39,40,41]. These bands were not observed in the input as the expression of endogenous tau is quite low, while immunoprecipitation led to enrichment of tau species and the truncated forms of tau were detectable in the immnunoblotting analysis (Additional file 1: Fig. S1B)’’.

Proteins were considered specific interaction partners when they [1] were only identified in the interactome of target proteins and absent in the negative control or [2] showed statistically significant enrichment in comparison with negative controls (Additional file 2). In total, 597 proteins were classified as potential interacting partners (passed our cutoff criteria) (Additional file 1: Fig. S1C & Additional file 2). It should be noted that some of the identified potential interaction partners may be indirect interactors that require other proteins for binding to the bait proteins.

There were 280 proteins that were specifically identified under stressful conditions (sodium aresnite, 0.6 mM, 60 min), in all three bait proteins (Fig. 2A & Additional file 2). Among these potential interactors, two proteins (ARL1, MAP4) were shared in all three investigated bait proteins (Fig. 2A).

Fig. 2

Differential interactome of tau, α/ꞵ synuclein, and PrP protein identified by co-immunoprecipitation combined with mass spectrometry analysis in HeLa cells. A A Venn diagram is showing unique and shared proteins in all three bait proteins identified under oxidative stress stimuli. B Gene ontology (GO) analysis in the domain of ‘’Biological Process (BP)’’ from ShinyGO webtool, in all three, bait proteins. The significance of the gene ontology enrichment is represented by the negative logarithm of the FDR (FDR < 0.05). The blue highlighted terms are shared between tau and PrP interactome and the red highlighted terms are between tau and synuclein interactome

Functional enrichment analyses in the domain of biological process (BP) showed that the protein networks from each individual bait protein organized into several distinct and converging functional categories (Fig. 2B). Based on functional profiles of enriched GO-terms, the major similarity was discovered between interactome make-up of tau and prion protein after oxidative stress induction, in vitro. Five protein categories including ‘‘ribonucleoprotein complex biogenesis’’, ‘‘ribosome biogenesis’’ and ‘‘rRNA processing’’ were specifically enriched in tau- and PrP-interactomes. The protein categories related to metabolism including ‘’carboxylic acid metabolic process, oxoacid metabolic processes’’ were shared between tau and synuclein interactomes (Fig. 2B).

Stress-induced remodeling of tau interactome to regulate oxidation–reduction and RNA-processing

To understand network rearrangements in tau interactome after stress induction, we performed a comprehensive comparative analysis between control (non-stressed) and stress-induced cells (0.6 mM sodium aresnite). Overall, we identified 143 potential interaction partners of tau protein (Additional file 1: Fig. S3 A & Additional file 2), (i) stress-dependent partners (77 proteins) that associate with tau protein only upon oxidative stress induction (ii) stress-independent interactors (8 proteins), which associate with tau protein independently of stress, and (iii) stress-sensitive interactors (58 proteins), which were lost after stress induction (Additional file 2). The identified proteins include known tau-interacting partners (53 proteins), validating our results. Importantly, a majority of the other identified proteins represent previously uncharacterized tau interacting factors (Additional file 2).

To further explore systematically, if a particular molecular function, biological process, or cellular component was enriched in the interactome that preferentially co-isolated with tau protein under stressful conditions; a comprehensive GO search was conducted using ToppCluster (Fig. 3A). The top-ranked categories include ‘‘oxidoreductase activity, ribonucleoprotein complex biogenesis’’, ‘‘translational termination’’ in stress-induced cells (Fig. 3A); consistent with previously characterized functions of tau [7]. Interestingly, one category that was exclusively enriched in control was “DNA repair” indicating an important physiological role of tau in the cell (Fig. 3A).

Fig. 3

Stress-induced alterations in the tau-interactome A Functional enrichment analysis of GO-terms in the domains of biological process (BP), molecular function (MF), and cellular component (CC) was performed using the ToppCluster web tool. An ‘abstracted’ network option was chosen to generate a Cytoscape-compatible network, containing the top ten enriched terms in the tau-interactome from each control (con.) and stress-induced (Str.) condition. Some of the specific terms are labeled in the figure. B Post-translation modifications (acetylation, methylation, and phosphorylation) were identified in the tau-interactome of control and stress-induced (stress) cells by MS analysis

In the current study, we also investigated three post-translational modifications ꟷphosphorylation, methylation, and acetylationꟷby MS analysis. Among these modifications, phosphorylation, and acetylation were the most frequent modifications. A map of those post-translational modifications, which were exclusively identified in either control or stress-induced (stress) cells is shown (Fig. 3B). Methylation (at residue K702) on X-ray repair cross-complementing protein 5, acetylation (G2) and methylation (K34) on Myosin-7 and acetylation (A2) on COMM domain-containing protein 9 was exclusively identified in control cells. Methylation (A2) and phosphorylation (S1943) on Myosin-9 protein, acetylation (M1) on polypyrimidine tract-binding protein 1 and acetylation (A2) on scaffold attachment factor B1 was exclusively identified in stress cells (Fig. 3B).

Oxidative stress-induced changes in the synuclein-interactome related to cell-redox and metabolic processes

For the identification of synuclein-interactome, we used an antibody that recognizes both isoforms (α/β) of synuclein protein. We identified 224 potential interaction partners of synuclein protein, (i) stress-dependent partners (51 proteins), (ii) stress-independent interactors (28 proteins), and (iii) stress-sensitive interactors (145 proteins) (Additional file 1: Fig. S3 B & Additional file 2). The identified proteins include known interacting partners of synuclein (e.g. ARP2, 6PGD, G3BP1 among others). Importantly, a majority of the other identified proteins represent previously uncharacterized interacting candidates (Additional file 2).

Intriguingly, stress treatment induced significant alterations in the interactome of synuclein protein. The particular enrichment of ‘‘cell redox homeostasis’’ in the interactome of synuclein protein under stressful conditions indicates the role of synuclein in oxidative stress response (Fig. 4A). Additionally, enrichment of metabolic and actin cytoskeleton-related proteins in the stress-dependent interactome of synuclein protein indicates remodeling of metabolic and cytoskeleton activities as a result of stress stimuli. The significantly enriched categories under control conditions were ‘‘ribosome biogenesis’’, ‘‘protein import into the nucleus’’ and ‘‘ncRNA processing’’ (Fig. 4A). Strikingly, 14 interacting partners of synuclein protein showed stress-mediated changes in PTMs including Heat shock protein HSP 90-beta (HSP90B: phosphorylation-S255), RNA-binding motif protein X chromosome (RBMX: acetylation-V2, phosphorylation-S208), Small ribosomal subunit protein eS4, X isoform (RS4X: methylation-K62 & K168) among others (Fig. 4B).

Fig. 4

Oxidative stress-induced changes in synuclein interactome. A Comparative functional profile of control (con.) and stress (str.) induced cells, in the domains of biological process (BP), molecular function (MF) and cellular component (CC) using ToppCluster. An ‘abstracted’ network option was chosen to generate a Cytoscape-compatible network, containing the top ten enriched terms. Some of the specific terms are labeled in the figure. B A map of three posttranslational modifications (Acetylation, methylation, and phosphorylation) identified by mass spectrometry analysis, in both control and stress-dependent interactome of synuclein protein

Oxidative stress induces changes in the interactome of PrP related to protein localization and DNA-metabolic processes

In total, we identified 230 potential interaction partners of the prion protein, (i) stress-dependent partners (85 proteins), (ii) stress-independent interactors (31 proteins), and (iii) stress-sensitive interactors (114 proteins) (Additional file 1: Fig. S3 C & Additional file 2).

Functional enrichment analysis of interacting partners of prion protein showed two major themes related to the ‘’DNA-metabolic process’’ and ‘’protein localization’’ that were enriched in control cells. Stress treatment induced significant changes in the interactors of the prion protein. Top-enriched functional categories in the stress-induced cells were related to ‘’RNA localization’’ and ‘’ncRNA processing’’ (Fig. 5A). A map of post-translational modifications that were exclusively identified in either unstressed (control) or stress-induced (stress) cells including Tubulin alpha-1C chain (TUBA1C: phosphorylation-S48), Heat shock protein beta-1 (HSPB1: phosphorylation-S15, S83) and Chromobox protein homolog 3 (CBX3: phospho-S93, S95, methylation-K143) among others. A detailed description is given in Fig. 5B.

Fig. 5

Oxidative stress-induced changes in the prion protein interactome. A Functional enrichment analysis for GO-molecular function (MF), -biological process (BP), and -cellular component (CC) in unstressed (con: control) and oxidative stress (str) induced HeLa cells. An abstracted network is showing enriched functional categories in the interactome of PrP. Cytoscape software was used as a visualization tool for the network. The middle circle is showing shared terms. B A map of post-translation modifications (phosphorylation, methylation, and acetylation) identified in stress-independent (control) or stress-dependent (stress) interactome of PrP

Validation of PrP interaction with exportin-5 in SH-SY5Y Cells

To independently validate the PrP–exportin-5(XPO-5) interaction observed from PrP IP-MS (Additional file 2), we used PrP and XPO5 as bait proteins in immunoprecipitation and reverse co-immunoprecipitation (CO-IP). Exportin-5 was an interesting candidate as cellular stress has been linked with nucleocytoplasmic transport defects due to sequestration of transport components into stress granules [42]. To explore this, we used Co-IP and Co-immunofluorescence under control and stress induced conditions.

Endogenous PrP (bait) was immunoprecipitated from a human neuroblastoma cell line (SH-SY5Y) and XPO5 (prey) was probed by western blotting. Exportin-5 robustly interacted with PrP in both unstressed (control) and stressed cells (stress) (Fig. 6A) but did not interact with the negative control (NC) IgG precipitate (Fig. 6A). We employed reversed CO-IP to confirm this interaction. Endogenous exportin-5 (bait) was immunoprecipitated and probed with PrP (prey) by western blotting (Fig. 6B).

Fig. 6

Validation of PrP and XPO5 interaction. A Endogenous PrP (bait) was immunoprecipitated from a human neuroblastoma cell line (SH-SY5Y) and exportin-5 (prey) was probed by western blotting in unstressed cells (control: C), stressed cells (stress: S) and negative control (NC) IgG precipitates. B Reverse CO-IP: endogenous XPO5 (bait) was immunoprecipitated and probed with PrP (prey) by western blotting. C Abundance of XPO5 in HeLa and SH-SY5Y cells under control (cont.) and stress (str.) condition by immunoblotting. *We are not sure about the identity of these bands. D Fractionation in SH-SY5Y cells under control and stress-induced condition. W: whole cell fraction, C: cytoplasmic fraction, N: nuclear fraction. E and F Densitometry analysis of XPO5 in HeLa cells (Welch’s t-test p-value = 0.0052) and SH-SY5Y cells (Welch’s t-test p-value = 0.0004). G Densitometry analysis of all fractions in control and stress conditions. H Localization of endogenous PrP and XPO5 in stress and control (cont.) HeLa cells using confocal microscopy. Counter-staining was performed to visualize cell nuclei, scale bar = 50 μm

There were slight differences in the co-purified proteins under control and stress conditions. This prompted us to investigate their expression under control and oxidative stress conditions (Additional file 1: Fig. S2). The abundance of PrP (SAF70) was significantly reduced after stress induction (Additional file 1: Fig. S2B & D) in SH-SY5Y cells. There was also a significant reduction in the mono-glycosylated form of PrP (SAF 32) after stress induction (Additional file 1: Fig. S2B and F). The abundance of exportin-5 was significantly reduced after stress treatment in both HeLa (Fig. 6C and E) and SH-SY5Y (Fig. 6C and F) cell lines. To confirm if decreased intensity levels of exportin-5 protein under stressful conditions were really a decrease in its concentration or simply a decrease in its intensity due to consolidation in SGs after stress treatment, we carried out subcellular fractionation. Remarkably, under control conditions, exportin-5 was only detected in whole cell and cytoplasmic fractions but not in nuclear fractions by immunoblotting. Strikingly, after stress treatment, exportin-5 was detected in nuclear fraction as well (Fig. 6D and G).

Overall, these findings indicate stress-mediated redistribution of XPO5 and its sequestration into PrP-positive foci.

We then examined the colocalization between PrP–XPO5 by performing immunofluorescence using antibodies specific to each protein. In control cells, XPO5 was localized mainly in the perinuclear region in HeLa cells, whereas PrP (SAF32) protein was detected in both, cytoplasmic and nuclear regions. Following stress treatment, XPO5 translocated into distinct granules localized around the nucleus (Fig. 6H). The cytoplasmic exportin-5 and PrP proteins colocalized in the granules after stress treatment (yellow dots in the enlarged image), although some of the XPO5, especially nuclear XPO5, remained in distinct regions, as shown by the green signal (Fig. 6H). Thus, these yellow dots show the partial co-localization of exportin-5 and PrP in the granules.