Wildtype σ1 receptor and the receptor agonist improve ALS-associated mutation-induced insolubility and toxicity

Abstract

Genetic mutations related to ALS, a progressive neurological disease, have been discovered in the gene encoding σ-1 receptor (σ1R). We previously reported that σ1RE102Q elicits toxicity in cells. The σ1R forms oligomeric states that are regulated by ligands. Nevertheless, little is known about the effect of ALS-related mutations on oligomer formation. Here, we transfected NSC-34 cells, a motor neuronal cell line, and HEK293T cells with σ1R-mCherry (mCh), σ1RE102Q-mCh, or nontagged forms to investigate detergent solubility and subcellular distribution using immunocytochemistry and fluorescence recovery after photobleaching. The oligomeric state was determined using crosslinking procedure. σ1Rs were soluble to detergents, whereas the mutants accumulated in the insoluble fraction. Within the soluble fraction, peak distribution of mutants appeared in higher sucrose density fractions. Mutants formed intracellular aggregates that were co-stained with p62, ubiquitin, and phosphorylated pancreatic eukaryotic translation initiation factor-2-α kinase in NSC-34 cells but not in HEK293T cells. The aggregates had significantly lower recovery in fluorescence recovery after photobleaching. Acute treatment with σ1R agonist SA4503 failed to improve recovery, whereas prolonged treatment for 48 h significantly decreased σ1RE102Q-mCh insolubility and inhibited apoptosis. Whereas σ1R-mCh formed monomers and dimers, σ1RE102Q-mCh also formed trimers and tetramers. SA4503 reduced accumulation of the four types in the insoluble fraction and increased monomers in the soluble fraction. The σ1RE102Q insolubility was diminished by σ1R-mCh co-expression. These results suggest that the agonist and WT σ1R modify the detergent insolubility, toxicity, and oligomeric state of σ1RE102Q, which may lead to promising new treatments for σ1R-related ALS.

The σ1 receptor (σ1R) was initially identified as the binding site (σ) of SKF-10047, a benzomorphine-related compound, and has since been classified as a nonopioid receptor according to its pharmacological properties (1–3). Recently, σ1R has been revealed to be a type II transmembrane protein localized in the endoplasmic reticulum (ER) membrane with its C-terminal chaperone domain toward the ER lumen (4, 5). σ1R is unique in its involvement in diverse cellular functions like calcium transport, stress response, lipid metabolism, regulation of neuronal activity, and RNA transcription by binding to various proteins in ER, plasma membrane, and nuclear envelope (6, 7). A number of physiological and pharmacological studies have been conducted to reveal that σ1R acts as the binding target not only for SKF-10047 but also for intrinsic molecules, several drugs, and other compounds (8–11).

Recent investigations have uncovered the genetic mutations of the human SIGMAR1 gene in motor neuron diseases like ALS, distal hereditary motor neuropathy, and Silver-like syndrome (12–20). These mutations cause single amino acid substitutions or various deletions in the C-terminal chaperone domain; however, it remains unclear how these mutations affect motor neurons and lead to the onset of these disorders. We previously reported how the variation (E102Q) identified in juvenile ALS showed abnormal localization in neuroblastoma Neuro2A cells (21). The transient expression of this mutation (σ1RE102Q) caused deficits in mitochondrial calcium transport and function, leading to a decrease in proteasomal activity and cell death. Abnormal protein aggregation is the prominent pathological feature in neurodegenerative diseases (22, 23). TAR DNA binding protein-43 and other proteins of which genetic mutations have been identified in familial ALS form subcellular inclusions containing ubiquitin and p62 proteins in motor neurons (23). This suggests that the deficit in the clearance of toxic proteins and aggregates by the ubiquitin-proteasome system and autophagy is associated with the pathology of ALS. Mutations in genes involved in the ubiquitin-proteasome system and autophagy are related to ALS and frontotemporal dementia with ALS (24–28).

σ1R localizes preferentially to the mitochondria-associated ER membranes (MAMs) or ER-associated lipid droplets (ER-LDs), which are highly resistant to detergents such as Triton X-114 in various types of cultured cells (29–31). Hayashi et al. (31) proposed that ER-LDs are detergent-resistant membranes (DRMs) that contain MAMs and are enriched in cholesterol and sphingolipids tethering σ1R to these microdomains. They also reported σ1R to be localized to DRMs, which were separated in the light fraction of sucrose density gradient centrifugation (i.e. third to fifth of 12 or 13 fractions) and that the agonist (+)-pentazocine caused σ1R disassociation from the ER-LDs (29, 32); however, it remains uncertain how ALS-related mutations influence the characteristics of σ1R such as detergent solubility/insolubility and subcellular dynamics in motor neurons.

Size exclusion chromatography and FRET analyses revealed that σ1R forms oligomeric states ranging from dimers to octamers and even higher oligomeric forms (33–35). Radiographic structural analysis confirmed the formation and trimeric structure of this receptor (4). Further studies elucidated how σ1R agonists and antagonists affect the oligomeric states (7, 36). When inactivated, σ1R forms a complex with the ER chaperone protein immunoglobulin heavy chain-binding protein/glucose-regulated protein (BiP/GRP)78. Binding of agonists causes σ1R to dissociate from BiP/GRP78, thereby enabling its translocation to intracellular domains where it can interact with protein substrates as a chaperone protein. On the other hand, whereas agonists stabilize monomers and dimers and thus increase activity of the receptor, antagonists can stabilize higher-order oligomers and consequently negatively affect its activity (7, 36).

Here, we found that the σ1R mutant fused with the red fluorescent protein (RFP) mCherry (σ1RE102Q-mCh) showed abnormal insolubility to detergents, subcellular distribution, and aggregations. These features were distinct from conventional MAMs or ER-LDs. Treatment with σ1R agonist SA4503 abrogated the insolubility and toxicity of the ALS mutant. Whereas σ1R-mCh formed exclusively monomers and dimers, σ1RE102Q-mCh also formed trimers and tetramers in NSC-34 cells, which could be partly reversed by SA4503. Co-expression of σ1R-mCh was found to inhibit σ1RE102Q fractionation in the insoluble fraction. These findings indicate that the agonist and WT σ1R can rescue the aberrant characteristics of this ALS-related σ1R mutant.

Results

ALS-related mutant shows abnormal insolubility to detergents

We transfected mouse motor neuron-like hybrid NSC-34 cells with WT σ1R and the ALS-related mutant E102Q, which were C-terminally fused with the RFP mCherry (σ1R-mCh and σ1RE102Q-mCh). Successful expression was confirmed using an anti-RFP antibody (Fig. 1A, top). Cells were lysed in buffer containing 1% NP-40 and separated by ultracentrifugation to soluble supernatants and insoluble pellets. Insoluble pellets were mixed with SDS lysis buffer and sonicated. Under these conditions, BiP/GRP78, the ER stress sensor inositol requiring (IRE)1α, and voltage-dependent anion channel, which were reported to interact with σ1R (30, 37, 38), were fractionated exclusively in the soluble fraction (Fig. 1B). σ1R-mCh was also fractionated in the soluble fraction; in contrast, σ1RE102Q-mCh was mainly separated in the insoluble fraction (Fig. 1C). This abnormality was also evident by a significantly higher ratio of protein level in the insoluble fraction relative to its soluble counterpart (Fig. 1D). Similar results were obtained using other detergents such as 0.5% Triton X-100 and 20 mg/ml CHAPS (Fig. S1, A–C), which is consistent with previous reports for WT σ1R (29, 31). Consistently, σ1RE102Q-mCh showed similar insolubility to 1% Nonidet P-40 (NP-40) in HEK293T cells (Fig. S1, D and E). These results demonstrate that σ1RE102Q-mCh shows abnormal properties regarding detergent insolubility.

Figure 1.

ALS-related mutation increases detergent insolubility of σ₁R. A, representative images of expression of WT σ₁R and ALS-related mutant (σ₁R^E102Q) C-terminally fused with mCherry (mCh) in total fraction. B, representative images of σ₁R-mCh, σ₁R^E102Q-mCh, and marker proteins for cellular organelles in each fraction. C, quantitative analysis for relative levels of σ₁R. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using two-way ANOVA followed by Bonferroni post hoc test. n = 6. F (1, 20) = 79.6 (WT/E102Q), 49.4 (soluble/insoluble) and 344.6 (interaction). **p < 0.01 versus soluble. ##p < 0.01 versus σ₁R-mCh. D, quantitative analysis for the ratio of insoluble to soluble fraction. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using unpaired t test. n = 6. t (10) = 13.0. **p < 0.01 versus σ₁R-mCh.

ALS-related mutant shows that altered subcellular distribution and p62- and ubiquitin-positive structures which are correlated with ER stress induction in NSC-34 cells

To evaluate the subcellular distribution of σ1RE102Q-mCh, we further separated both fractions by sucrose density gradient centrifugation (Fig. 2A). Western blotting analysis following the collection of 12 fractions from the top showed a wide range of distribution of σ1R-mCh, which partially overlapped with that of BiP/GRP78 (Fig. 2A, fractions 2–6). The distribution peak of σ1RE102Q-mCh within the soluble fraction changed slightly into the latter fractions (Fig. 2B, σ1R-mCh: fractions 5 and 6, σ1RE102Q-mCh: fractions 7 and 8). Within the insoluble fraction, σ1RE102Q-mCh was most abundant in the bottom portion together with p62 (Fig. 2, A and B). These results were distinct from the distribution of σ1R in conventional ER-LDs or DRMs, which are separated in the light fractions (29, 32). Furthermore, we performed the same fractionation with nontagged forms to exclude the possibility that the fluorescent tag influences the distributions and mutation-induced alteration (Fig. 2, C and D). As a result, transfected σ1R in the soluble fraction was preferentially separated in the front half of 12 fractions. This result was similar to endogenous σ1R observed in mock-transfected cells. However, σ1RE102Q was additionally fractionated in the 12th fraction with a slight decrease in mobility in SDS-PAGE (Fig. 2C, third panel, arrow, σ1RE102Q; arrowhead, endogenous σ1R) when PAGE was performed with longer time exceptionally to distinguish them. Simultaneously, σ1RE102Q in the insoluble fraction was exclusively collected in the 12th fraction (Fig. 2, C and D).

Figure 2.

E102Q mutation changes σ₁R distribution in sucrose gradient density fractionation. A, representative images of WT σ₁R-mCh in the soluble fraction and σ₁R^E102Q-mCh mutant in the soluble and insoluble fraction followed by sucrose gradient fractionation. Twelve fractions were collected from the top. B, quantification of relative abundance of σ₁R-mCh in the 12 fractions. C, representative images of nontagged σ₁R in the soluble fraction and E102Q mutant in the soluble and insoluble fraction followed by sucrose gradient fractionation. Twelve fractions were collected from the top. Third panel, arrow and arrowhead indicate transfected mutant and endogenous (endo.) σ₁R, respectively. D, quantification of relative abundance of σ₁R in the twelve fractions. The quantification of σ₁R^E102Q in soluble fractions was simultaneously performed with lower bands of endogenous σ₁R.

We further performed immunocytochemistry to analyze the cellular distribution of σ1RE102Q-mCh in NSC-34 cells (Fig. 3). When cells were labeled with an antibody against σ1R, the appearance of red fluorescence of mCherry (magenta) and detected signal (green) were slightly different in the signal:noise ratio; nonetheless, the localization patterns were indeed similar (Fig. 3, A and B). σ1RE102Q-mCh showed abnormal aggregate-like structures as we reported previously using Neuro2A cells (Fig. 3B) (21). BiP/GRP78 was found to be partially colocalized with both σ1R-mCh and σ1RE102Q-mCh, suggesting that a portion of these proteins do interact (Fig. 3, C and D). The immunostained signal for p62 did not colocalize with σ1R-mCh (Fig. 3E) but colocalized with the aggregate-like structures of σ1RE102Q-mCh (Fig. 3F, inset and arrows). We observed an aberrant accumulation of ubiquitin, which colocalized with σ1RE102Q-mCh aggregates (Fig. 3, G and H). As previously reported, σ1RE102Q-mCh aggregates were co-stained especially with phosphorylated pancreatic eukaryotic translation initiation factor-2-α kinase (pPERK), an ER stress sensor in NSC-34 cells (Fig. 3, I and J) (39). The aggregates of σ1RE102Q-mCh did not show colocalization with inositol 1,4,5-triphosphate receptor type 2 (IP3R2), a major subtype of IP3Rs expressed in motor neurons (Fig. 3, K and L) (40). The localization of translocase of outer mitochondrial membrane (TOM)20, a protein in the mitochondrial outer membrane, showed partial colocalization with tubular structures of σ1R-mCh and σ1RE102Q-mCh (Fig. 3, M and N). In contrast, TOM20 did not colocalize with aggregate-like structures of σ1RE102Q-mCh (Fig. 3N, inset). The analyzed images were also processed and converted into binary images to visualize MAMs (Fig. S2). It showed similar localizations of σ1R-mCh and σ1RE102Q-mCh in these structures.

Figure 3.

Subcellular localization of WT and ALS-related mutant of σ₁R-mCh. NSC-34 cells transfected with σ₁R-mCh and σ₁R^E102Q-mCh were immunostained with anti-σ₁R (ab53852) (A and B), BiP/GRP78 (C and D), p62 (E and F), ubiquitin (MAB1510) (G and H), phosphorylated PERK (Thr-980) (pPERK) (I and J), ITPR2 (IP₃R2) (K, L), and TOM20 (M, N) antibodies. White squares indicate the regions of highly magnified insets (A–N). Arrows indicate the large aggregate-like structures, which show an internally immunoreactive signal with the anti-p62 antibody. Scale bars, 10 µm (original) and 1 µm (insets).

Acute SA4503 treatment does not affect intracellular dynamics of σ1RE102Q-mCh tubular structures and aggregates

Next, we performed fluorescence recovery after photobleaching (FRAP) analysis to evaluate potential changes in the intracellular dynamics of σ1RE102Q-mCh tubular and aggregate-like structures (Fig. 4). FRAP analysis enables the assessment of molecular mobility in the cells using fluorescence. The rigid aggregate-like structures elicit the decline of protein penetration toward aggregates and therefore lead to slow fluorescent recovery after bleaching (41, 42). The fluorescence of the σ1R-mCh tubular structures recovered just after bleaching within 39 s (0.38 ± 0.08 in recovery) (Fig. 4, A, C, and E and Movie S1); however, the fluorescence of the σ1RE102Q-mCh aggregate-like structures did not recover (0.09 ± 0.05 in recovery) (Fig. 4, B, D, and E and Movie S2). We calculated the normalized intensity in bleached regions (Fig. 4, A and B, black circles) in control and SA4503-treated cells (Fig. 4, C and D). The recovery derived from the normalized intensity was significantly lower in σ1RE102Q-mCh aggregate-like structures than in both σ1R-mCh and σ1RE102Q-mCh tubular structures (Fig. 4E). This suggests that the abnormal structures were indeed intracellular aggregates. SA4503 treatment for 1 h decreased the recovery of σ1R-mCh and σ1RE102Q-mCh tubular structures slightly, but there was no statistically significant difference (p = 0.13 for σ1R-mCh, p = 0.10 for σ1RE102Q-mCh).

Figure 4.

Fluorescence recovery after photobleaching analysis for subcellular dynamics of σ₁R-mCh and σ₁R^E102Q-mCh. Representative images of tubular structure of σ₁R-mCh (A) and aggregate structure of σ₁R^E102Q-mCh (B). Ten and 30 images were taken before and after photobleaching, respectively. The magnified images were expressed in pseudo-color following the fluorescent intensity. Black circles indicate the photobleached position. C and D, quantitative analysis. Each line graph shows mean ± S.D. (error bars). Black arrows indicate the time point of bleaching. E, recovery was calculated as the changes in the normalized intensity after bleaching and the final image. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using two-way ANOVA followed by Bonferroni post hoc test. n = 9–13. F (2, 63) = 79.8 (WT-tubular/E102Q tubular/E102Q aggregates) and 2.20 (interaction). F (1, 63) = 4.97 (control/SA4503). a, p < 0.01 versus WT-tubular control. b, p < 0.01 versus E102Q-tubular control. c, p < 0.01 versus WT-tubular SA4503. d, p < 0.01 versus E102Q-tubular SA4503. N.S., not significant. Scale bars, 5 µm (original) and 1 µm (insets).

Prolonged SA4503 treatment decreases σ1RE102Q-mCh insolubility and cell apoptosis

Based on our observations that a short period of treatment did not have any effect on intracellular dynamics, we next treated cells with SA4503 for 48 h after transfection. SA4503 treatment decreased the ratio of σ1RE102Q-mCh protein levels in insoluble versus soluble fractions significantly (Fig. 5A, top panel, and Fig. 5B). We next evaluated the concentration dependence by treatment at various conditions (0, 0.2, 1, 5, and 20 μm) (Fig. 5, C and D). As a result, the ratio of σ1RE102Q-mCh protein levels gradually decreased: there was statistically significant difference at 1, 5, and 20 μm (p < 0.01). Because abnormal aggregates were generally considered to be degraded by autophagy, we analyzed the expression levels of marker proteins reflecting autophagic flux in total fractions (Fig. S3). The ratio of microtubule-associated protein 1 light chain 3 type II/I (LC3-II/I) decreased slightly in SA4503-treated cells; however, expression of p62, which is degraded by autophagy, was not altered by SA4503 treatment. Hence, we concluded that binding of SA4503 to σ1RE102Q-mCh changed its solubility directly but did not affect the autophagic machinery or cause the degradation of aggregates. Cells transfected with σ1RE102Q-mCh, but not with σ1R-mCh, were immunostained with an antibody against activated caspase-3 (Fig. 5, E and F). Remarkably, the percentage of cells with nuclei featuring a hazy outer boundary in a DNA diffusion assay increased significantly in σ1RE102Q-transfected cells (Fig. 5, G and H). These results suggest that the ALS-associated mutant of σ1R induced apoptotic cell death, which could be rescued by treatment with the σ1R agonist SA4503 for 48 h.

Figure 5.

Pharmacological effect of SA4503 on σ₁R^E102Q-mCh insolubility and toxicity. A, representative images of σ₁R, σ₁R^E102Q-mCh after SA4503 treatment for 2 days in soluble and insoluble fractions. B, quantitative analysis for the ratio of insoluble to soluble fraction. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using two-way ANOVA followed by Bonferroni post hoc test. n = 4. F (1, 12) = 73.1 (WT/E102Q), 7.81 (control/SA4503) and 7.18 (interaction). **p < 0.01 versus σ₁R-mCh Cont or SA4503. ##p < 0.01 versus σ₁R^E102Q-mCh control. C, representative images of σ₁R^E102Q-mCh after various concentrations of SA4503 treatment in soluble and insoluble fractions. D, quantitative analysis for the ratio of insoluble to soluble fraction. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using one-way ANOVA followed by Bonferroni post hoc test. n = 6. F (4, 25) = 22.17. **p < 0.01 versus control. E and F, cells were immunostained with anti-activated caspase-3 antibody (C8487). σ₁R, σ₁R^E102Q-mCh (mCh) are shown in magenta, and activated caspase-3 (aCas3) is shown in green. Nucleus was stained by 4′,6-diamidino-2-phenylindole and shown in blue. Scale bars, 10 µm. G, cells transfected with nontagged σ₁R and σ₁R^E102Q and treated with SA4503 (20 µm) were subjected to DNA diffusion assay and stained with ethidium bromide (red). White arrows and arrowheads show apoptotic cells with hazy outer boundary. Arrowheads indicate the regions of highly magnified insets. Scale bars, 50 µm (original) and 10 µm (insets). H, quantitative analysis for percentage of apoptotic cells. Each bar shows mean ± S.D. (error bars), and each plot indicates the percentage of apoptotic cells in 10 regions randomly selected. Statistical significance was tested using two-way ANOVA followed by Bonferroni post hoc test. n = 10. F (2, 54) = 86.4 (mock, WT, E102Q) and 8.12 (interaction). F (1, 54) = 23.1 (control/SA4503). **p < 0.01 versus mock control or SA4503. ##p < 0.01 versus σ₁R^E102Q control. N.S., not significant; Cont, control.

Nontagged mutant showed similar aberrant aggregates in NSC-34 cells, but tagged mutant failed in HEK293T cells

We further evaluated the fluorescent study of σ1RE102Q with nontagged form in NSC-34 cells and with tagged form in HEK293T cells (Fig. 6). Nontagged WT σ1R localized in diffuse pattern in the cytosol (Fig. 6, A, C, E, and G); however, σ1RE102Q caused aberrant aggregates which were co-stained with p62, ubiquitin, and pPERK (Fig. 6, B, D, and F). Cells expressing σ1RE102Q were stained with activated caspase-3 (Fig. 6H). σ1R-mCh showed similar patterns in HEK293T cells (Fig. 6, I, K, M, and O), whereas σ1RE102Q-mCh showed large aggregates (Fig. 6, J, L, N, and P). These aggregates were stained with ubiquitin but not with p62 or pPERK antibodies. Nevertheless, cells highly expressing σ1RE102Q-mCh localized in abnormal structure were stained with activated caspase-3 (Fig. 6P).

Figure 6.

Subcellular localization of WT and ALS-related mutant of σ₁R in NSC-34 cells and mCh-tagged forms in HEK293T cells. NSC-34 cells transfected with σ₁R and σ₁R^E102Q and HEK293T cells transfected with mCh-tagged forms were immunostained with anti-σ₁R (sc-137075) (A–H), p62 (A, B, I, and J), ubiquitin (AB1690) (C, D, K, and L), phosphorylated PERK (Thr-980) (pPERK) (E, F, M, and N), and activated caspase-3 (aCas3) (9661) (G, H, O, and P) antibodies. White squares indicate the regions of highly magnified insets. Scale bars, 10 µm (original) and 1 µm (insets).

SA4503 reduces σ1RE102Q-mCh monomer-tetramer formation in the insoluble fraction but increases monomer formation in the soluble fraction

We hypothesized that SA4503 stabilizes and thus increases monomers/dimers, causing the observed changes in abnormal insolubility and aggregations of σ1RE102Q-mCh (36). To test this, we subjected cells to a crosslinking procedure with the uncleavable crosslinker disuccinimidyl suberate (DSS) following transfection and treatment. We analyzed the oligomeric state in soluble and insoluble fractions by Western blotting (Fig. 7). σ1R-mCh primarily formed monomers and dimers in the soluble fraction (Fig. 7A), whereas σ1RE102Q-mCh was not only found to form monomers/dimers but also trimers/tetramers in both the soluble and insoluble fractions (Fig. 7, A and B). Quantitative analysis showed that SA4503 treatment increased σ1RE102Q-mCh monomer levels in the soluble fraction and decreased monomer/dimer/trimer/tetramer levels in the insoluble fraction (Fig. 7, C and D). The level of σ1RE102Q-mCh dimers in the soluble fraction was slightly higher in cells treated with SA4503, although the difference was not statistically significant (Fig. 7C, p = 0.15). There was a smear pattern but no apparent specific band with a higher molecular weight of σ1RE102Q-mCh in the insoluble fraction (Fig. S4).

Figure 7.

Effect of E102Q mutation and SA4503 on oligomeric state of σ₁R. The cells were processed by crosslinking with DSS, followed by subcellular fractionation. Representative images of the oligomeric state of σ₁R-mCh and σ₁R^E102Q-mCh are shown in soluble (A) and insoluble (B) fractions. Blue arrowhead shows monomers of ∼55 kDa. Black arrowheads show dimers (bottom), trimers (middle), and tetramers (top). Quantitative analysis of relative levels of σ₁R^E102Q-mCh in soluble (C) and insoluble (D) fractions. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using unpaired t test. n = 5. t (8) = 4.57 (monomers), 1.61 (dimers), 1.62 (trimers), and 0.59 (tetramers) of soluble, and 6.37 (monomers), 4.85 (dimers), 4.18 (trimers), and 3.61 (tetramers) of insoluble fractions. **p < 0.01 versus control. N.S., not significant; Cont, control.

Co-expression of σ1R-mCh diminishes σ1RE102Q insolubility in NSC-34 cells

We further analyzed whether WT σ1R could influence σ1RE102Q-mCh insolubility (Fig. 8). We transfected NSC-34 cells with σ1R C-terminally fused with GFP (σ1R-GFP) together with σ1R-mCh or σ1RE102Q-mCh and then performed co-immunoprecipitation using an antibody against GFP (Fig. 8A). As a result, σ1R-mCh and σ1RE102Q-mCh were immunoprecipitated with σ1R-GFP (upper panel, lanes 7 and 8) but not with GFP (upper panel, lanes 3 and 4). This indicates that the ALS-associated mutant can form chimeric oligomers with WT σ1R. Co-expression of σ1R-mCh reduced σ1RE102Q fractionation in the insoluble fraction and the ratio (insoluble/soluble) significantly (Fig. 8, B and C). In the regular time-course of PAGE, endogenous σ1R and transfected σ1RE102Q could not be separated. This seems to cause the apparent soluble σ1R level to appear higher and the ratio to appear lower compared with the mCh-tagged form (Fig. 1, B and D and Fig. 8, B and C). These results indicate that WT σ1R could bind to and abrogate the insolubility of the ALS-related σ1R mutant and its aggregates.

Figure 8.

Effect of co-expression of σ₁R-mCh on σ₁R^E102Q insolubility. A, cell lysates were co-immunoprecipitated with an anti-GFP antibody, and representative images after detection using anti-RFP (mCh) and GFP antibodies are shown for immunoprecipitated samples (Co-IP) and total lysates (Input). Blue arrowheads show mCh- and GFP-tagged σ₁R, and black arrowheads show mCh and GFP. Asterisk indicates the immunoglobulin heavy chain used for co-immunoprecipitation. B, representative images of σ₁R and σ₁R^E102Q transfected with σ₁R-mCh in soluble and insoluble fractions. C, quantitative analysis for the ratio of insoluble to soluble fractions. Each bar shows mean ± S.D. (error bars). Statistical significance was tested using two-way ANOVA followed by Bonferroni post hoc test. n = 4. F (1, 12) = 26.6 (WT/E102Q), 18.6 (mCh/σ₁R-mCh) and 11.6 (interaction). **p < 0.01 versus σ₁R control (mCh). ##p < 0.01 versus σ₁R^E102Q control (mCh).

Discussion

In the present study we have demonstrated that the ALS-related mutation of σ1R, σ1RE102Q, affected the solubility and oligomer formation of this receptor, which induced toxicity in NSC-34 cells. The smear pattern of σ1RE102Q-mCh with a higher molecular weight, which appeared in the insoluble fractions (Fig. S4), suggests a correlation between the aggregation and its insolubility (Fig. 9). The aggregates of σ1RE102Q-mCh colocalized with pPERK but not with IP3R2. This suggests that σ1RE102Q-mCh aggregation leads to a dysregulation of calcium homeostasis and ER stress, eventually resulting in cell death as previously reported (21, 39). In spinal motor neurons of ALS model mice, ER stress and unfolded protein response were induced in degenerating neurons but not in preserved neurons in the period of onset (43), suggesting the progressive role in the disease. Co-expression of σ1R-mCh inhibited σ1RE102Q insolubility, which was possibly caused by direct interaction. Nontagged σ1RE102Q showed similar p62-, ubiquitin-, and pPERK-positive aggregates in NSC-34 cells (Fig. 6, A–H). However, in HEK293T cells, σ1RE102Q-mCh provoked ubiquitin-positive and p62- and pPERK-negative aggregates, although an extremely highly expressed cell was stained with anti-activated caspase-3 antibody (Fig. 6, I–P). These results might suggest the vulnerability of motor neurons to toxicity from ER stress and the usefulness of NSC-34 cells as a motor neuron model.

Figure 9.

Schematic diagram of the hypothesized action of SA4503/σ₁R expression on σ₁R^E102Q abnormal characteristics. Upper, σ₁R is present exclusively in monomers/dimers that are detergent-soluble in NSC-34 cells. Lower, σ₁R^E102Q forms not only monomers/dimers but also trimers/tetramers and aggregates, which are detergent-insoluble. Aggregation of σ₁R^E102Q correlates with the induction of ER stress and might result in cell death. σ₁R agonist SA4503 and WT co-expression rescued these mutation-induced abnormalities.

Hayashi et al. (29, 31) reported that σ1R localizes preferentially in ER-LDs, which are resistant to Triton X-114 and show a ring-like structure in neuroblastoma NG-108 cells. They also reported that overexpression of a tagged σ1R targets enlarged ER-LDs (29, 32). The appearance of σ1RE102Q-mCh aggregates was similar to these large ER-LDs (Fig. 3), with the following differences: 1) σ1RE102Q-mCh localized in a tubular rather than a ring-like structure in NSC-34 cells; 2) whereas ER-LDs tend to be fractionated in light fractions in sucrose density gradient centrifugation, the mutation changed the distribution peak of tagged form from the fifth and sixth to the seventh and eighth in the soluble fraction, mainly to the bottom in the insoluble fraction (Fig. 2, A and B). This result was identical in nontagged σ1R regarding the altered direction (Fig. 2, C and D); and 3) in contrast to ER-LDs, which contain enriched MAMs, both σ1R-mCh and σ1RE102Q-mCh showed similar colocalization with TOM20 (Fig. 3, M and N and Fig. S2). We therefore conclude that the fractionation of σ1RE102Q-mCh in the insoluble fraction did not indicate a localization in ER-LDs or MAMs in NSC-34 cells. Indeed, sucrose density gradient centrifugation confirmed that σ1R-mCh and σ1RE102Q-mCh appeared only in the bottom fractions (Fig. S5). A common feature between our study and that of Hayashi et al. (32) is that (+)-pentazocine dissociated σ1R from ER-LDs and that SA4503 changed σ1RE102Q-mCh fractionation from the insoluble to the soluble fraction (Fig. 5B). Furthermore, a glycosylphosphatidylinositol-anchored protein (alkaline phosphatase) was reported to become insoluble to detergents when binding with cholesterol/sphingolipids (44, 45). These molecules might alter the insoluble/soluble property of σ1R, although we at least failed to confirm σ1RE102Q localization in ER-LDs (Fig. S5). Further studies will be required to clarify any discrepancies and to fully understand σ1R's role in this context.

Our FRAP analysis demonstrated that recovery of σ1RE102Q-mCh in tubular structures was identical to that of σ1R-mCh, which was significantly faster than that of σ1RE102Q-mCh in aggregates (Fig. 4). Treatment with SA4503 for 1 h failed to influence the recovery. The intracellular dynamics of σ1R, including the effects of the mutation and agonist/antagonist, were already investigated by two other groups recently (39, 46). Dreser et al. (39) reported that a yellow fluorescent protein-tagged version of σ1RE102Q (σ1RE102Q-YFP) shows slower recovery compared with WT in COS-7 cells, which is consistent with our observations. Wong et al. (46) found that σ1R-YFP in reticular structures exhibits a significantly lower recovery than BiP-mCh in mouse embryonic fibroblasts. The σ1R agonist SKF-10047 (100 μm, 1 h) increased the recovery of σ1R-YFP in reticular structures but not σ1RE102Q-YFP, which had a faster recovery. The speed of recovery of both σ1R- and σ1RE102Q-YFP in puncta structures was slow and similar to that of σ1R-YFP in reticular structures. Although we cannot explain the discrepancy between these reports and our results, σ1R may have different characteristics according to different cell types because the localization in mouse embryonic fibroblasts showed much more diffuse structures, which were not identical to the tubular structures seen in NSC-34 cells (Fig. 4) (46).

We hypothesized that monomers and dimers of σ1RE102Q-mCh would be soluble to detergents, whereas higher-order oligomers such as tetramers would not be. However, interestingly, σ1RE102Q-mCh monomers, dimers, trimers, and tetramers were fractionated in both soluble and insoluble fractions (Fig. 7). This suggests that the oligomeric state is not sufficient to determine the vulnerability of σ1RE102Q to detergents. SA4503 treatment decreased the formation of these four oligomeric states in the insoluble fraction but increased the accumulation of monomers in the soluble fraction. The effects of agonists and antagonists on the oligomeric state of σ1R were recently investigated by another group (47–50), who transfected HEK293 cells with nontagged or tagged forms of σ1R and performed normal SDS-PAGE or PAGE with mild detergents (perfluorooctanoic acid or sodium lauroyl sarcosinate). They found that most of the agonists decrease multimer formation with a molecular weight of over 130 kDa, whereas antagonists increase such formations. Their results also showed that σ1RE102Q reduces multimer or high-order oligomers over dimers (49, 50). In our study, σ1R-mCh formed exclusively monomers and dimers, whereas σ1RE102Q-mCh was also observed in trimers and tetramers (Fig. 7, A and B). Although we cannot explain this discrepancy, it suggests that these cell lines have no appropriate properties including protein expressions or lipid composition that might require σ1R oligomerization. It is not fully demonstrated how the oligomer forms and is regulated in the cells/during extraction. Considering this, our procedure (DSS + SDS-PAGE) might be useful to assess the intracellular behavior of the receptor. The smear pattern of σ1RE102Q-mCh with higher molecular weight especially in the insoluble fraction (Fig. S4) suggests a correlation of the insolubility and aggregate formation. Furthermore, in the spinal cord, σ1R is reportedly enriched in the post-synaptic terminal in the cell soma, axons, and dendrites of motor neurons (51, 52). However, it is not determined that NSC-34 cells can form the synapse between cells. Further studies are required for assessing the influence of synaptic formation on σ1R mutant features.

We performed co-immunoprecipitation and revealed that σ1R exerted homomeric interactions in NSC-34 cells (Fig. 8A). This result supports previous reports using size exclusion chromatography and FRET analysis (4, 33–35). We also revealed that σ1RE102Q preserves the interaction with WT σ1R. The co-expression of σ1R-mCh with the mutant inhibited the aberrant insolubility (Fig. 8, B and C). The inheritance of most genetic mutations associated with familial ALS identified so far are autosomal or X-linked dominant (24, 25, 53–59), suggesting a high toxicity of those mutations. The σ1RE102Q mutation causes juvenile ALS, thus also evidencing high toxicity (12). However, the inheritance of this variation is autosomal recessive, with the homozygous mutation affecting motor neurons severely, whereas the heterozygous mutation has no effect. Our finding that WT σ1R-mCh abolished the abnormal insolubility of σ1RE102Q may explain this discrepancy.

In conclusion, the σ1RE102Q mutation caused abnormal insolubility to detergents, which correlated with aggregation, although these were distinct from the conventional ER-LDs or MAMs. Treatment with the σ1R agonist SA4503 and WT σ1R co-expression could abrogate the aberrant oligomer formation caused by the mutation. In the future, pharmacological treatment with SA4503 and genetic therapy using WT σ1R expression may be promising for the treatment of σ1R mutation-related ALS.

Experimental procedures

Materials

The σ1R agonist SA4503, 1-[2-(3,4-dimethoxyphenyl)ethyl]-4-(3-phenylpropyl) piperazine, was synthesized by Dr. Chen Zhang (Zhejiang University) (60, 61). Primary antibodies used were as follows: mouse anti-RFP mAb (M165), rabbit anti-p62 (SQSTM1) polyclonal antibody (pAb) (PM045), and rabbit anti-LC3 pAb (PM036) (Medical & Biological Laboratories); rabbit anti-σ1-receptor pAb (ab53852), rabbit anti-Histone H3 pAb (ab1791), and rabbit anti-ITPR2 (IP3R2) pAb (ab136412) (Abcam); mouse anti-β-tubulin mAb (T0198), rabbit anti-caspase 3, active pAb (C8487), mouse anti-ubiquitin mAb (MAB1510), and rabbit anti-ubiquitin pAb (AB1690) (Merck); rabbit anti-IRE1α mAb (3294), rabbit anti–voltage-dependent anion channel pAb (4866), rabbit anti-phospho PERK (Thr-980) mAb (3179), and rabbit anti-cleaved caspase-3 (Asp-175) (9661) pAb (Cell Signaling Technology); rabbit anti-GFP pAb (632592) (Takara Bio); mouse anti-σ1 receptor mAb (sc-137075), and rabbit anti-TOM20 pAb (sc-11415) (Santa Cruz Biotechnology); and mouse anti-BiP/GRP78 mAb (610978) (BD Biosciences). All other materials were purchased from FUJIFILM Wako Pure Chemical (Osaka, Japan) unless otherwise specified.

Cell culture, transfection, and treatment

NSC-34 cells, a mouse motor neuron-like cell line, were a kind gift from Dr. Yasuhiro Kosuge (Nihon University) (62). HEK293T cells (CRL-3216) were purchased from American Type Culture Collection (63). The cells were maintained in DMEM supplemented with 10% FBS (Biowest, Nuaillé, France) and penicillin/streptomycin (Thermo Fisher Scientific) in a 5% CO₂ incubator at 37 °C. The cells were transfected with plasmids encoding WT or ALS-related mutant (E102Q) of σ1R C-terminally fused with mCherry or GFP, nontagged forms or control (pcDNA3.1, mCherry or GFP) for 6 h using Lipofectamine LTX Reagent (Thermo Fisher Scientific) (21). After transfection, the cells were cultured in complete DMEM and incubated with 0, 0.2, 1, 5, and 20 μm SA4503 for 48 h, then used for further experiments. For FRAP, SA4503 was added only 1 h before analysis. SA4503 was dissolved in sterilized double distilled water (ddw) at 20 mm stock solutions and stored at –20 °C. For analysis of concentration dependence, stock solutions were sequentially diluted with sterilized ddw before use.

Detergent-soluble/insoluble fractionation

Detergent-soluble/insoluble fractionation was performed according to previous reports with slight modifications (31, 64). The cells were washed with PBS, harvested in lysis buffer containing 1% NP-40 (20 mm HEPES-NaOH, pH 7.4, 100 mm KCl, 1.5 mm MgCl₂, 250 mm sucrose, 1 mm EDTA, and 1 mm EGTA), and gently agitated at 4 °C for 0.5 h. The lysates were centrifuged at 10,000 × g for 10 min, and the supernatant was collected in a new tube. The samples were further centrifuged at 100,000 × g for 1 h, and the supernatant was collected as a soluble fraction. The pellets were washed with homogenizing buffer followed by adding half the volume of SDS lysis buffer (50 mm Tris-HCl, pH 8.0, 150 mm NaCl, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, and 1% glycerol) and sonication as with the insoluble fraction. For further investigation of the insoluble features of σ1RE102Q, the cells were homogenized in lysis buffer containing 0.5% Triton X-100 or 20 mg/ml CHAPS (10 mm Tris-HCl, pH 7.4, 150 mm NaCl, and 5 mm EDTA) as previously reported (31). Following agitation at 4 °C for 0.5 h, the cells were centrifuged at 12,000 × g for 20 min. The supernatants were further separated into soluble and insoluble fractions by ultracentrifugation as described above. For total fractions, the cells were harvested in SDS lysis buffer, sonicated, and centrifuged at 10,000 × g for 10 min before the supernatant was collected. Protease/protein phosphatase inhibitors (50 μg/ml trypsin inhibitor, 0.1 mm leupeptin, 1 mm DTT, 75 μm pepstatin A, and 25 nm calyculin A) were added to the lysis buffer. For Western blotting, each fraction was boiled at 100 °C for 3 min following the addition of 6× Laemmli SDS sample buffer.

Sucrose density gradient centrifugation

Sucrose density gradient centrifugation was performed according to a previous study (30). Insoluble fractions were adjusted to 250 mm sucrose, which was identical to the concentration of the soluble fraction. Both fractions were laid on top of 40–80% sucrose density gradients (5 volumes) and centrifuged at 100,000 × g for 17 h. Twelve fractions were collected from the top, and the proteins were precipitated by adding an equal volume of 40% TCA. After centrifugation, the pellets were washed with ethanol, followed by adding SDS lysis buffer and sonication. Proteins were boiled with 6× Laemmli SDS sample buffer for Western blotting.

Another condition of sucrose density gradient centrifugation was used to investigate the localization of σ1R-mCh and σ1RE102Q-mCh in ER-LDs or MAMs (29). Briefly, transfected cells were extracted with lysis buffer containing 0.5% Triton X-100, and the lysates were separated by 0–35% sucrose density gradient using centrifugation at 141,000 × g for 24 h. Eleven fractions were collected from the top, and the proteins were precipitated by TCA and denatured with SDS sample buffer.

Crosslinking procedure

The cells were washed twice and harvested in PBS. The cells were added with DSS crosslinker (final 2 mm; Thermo Fisher Scientific) and incubated for 0.5 h at room temperature. The reaction was stopped by adding Tris-HCl (pH 7.5, final 50 mm) and incubation at room temperature for 15 min. The cells were subjected to detergent-soluble/insoluble fractionation as described above.

Western blotting analysis

Western blotting analysis was performed as previously reported (21). Briefly, an equal number of samples were applied to SDS-PAGE with 5–12/15% separated polyacrylamide gels or 2–15% gradient gels (MULTIGEL II mini, Cosmo Bio, Tokyo, Japan), followed by transfer to polyvinylidene fluoride membranes (Merck). The membranes were agitated in 5% skim milk in TBS with Tween 20 for 1 h, then shaken with primary antibody (1:1,000 dilution except for the antibody against β-tubulin (1:2,000)) as described above at 4 °C overnight. After three washes with TBS with Tween 20, the membranes were shaken with goat anti-rabbit or mouse IgG antibody conjugated with horseradish peroxidase (1:5,000 dilution; Southern Biotech, Birmingham, Alabama, USA). Chemiluminescent signal was detected by imaging analyzer (LAS4000 Mini; FUJIFILM, Tokyo, Japan). Densitometry analysis was visualized using Multi Gauge Software (FUJIFILM).

Immunocytochemistry

The cells were cultured on poly-l-lysine–coated cover glass in a 12-well multiplate. Immunocytochemistry was performed at room temperature unless otherwise specified. After fixation with 4% paraformaldehyde in PBS for 10 min, the cells were permeabilized with 0.1% Triton X-100 in PBS for 10 min. The cells were incubated with 1% BSA in PBS for 0.5 h for blocking, then probed with the primary antibody in BSA solution (1:200 or 1:500) at 4 °C overnight. After three washes, the cells were incubated with a secondary antibody of Alexa Flour 488-conjugated donkey anti-rabbit, mouse IgG antibody, or Alexa Flour 594-conjugated donkey anti-mouse IgG antibody (1:500; Thermo Fisher Scientific). After washing, the nucleus was stained with 4′,6-diamidino-2-phenylindole (Merck), and the cells were immersed in Vectashield (Vector Laboratories, Burlingame, CA, USA). Fluorescent images were obtained by confocal microscopy (TCS SP8, Leica Microsystems, Wetzlar, Germany). For immunostaining proteins in ER using antibodies against σ1R, BiP/GRP78, pPERK, and IP3R2, antigen retrieval was performed by incubating cells in 6 m urea in 0.1 m Tris-HCl, pH 9.5, at 80 °C for 10 min after fixation (65). The fluorescent images of mCherry and TOM20 were automatically adjusted in intensity, the background signal was subtracted, and the threshold was determined and converted into binary images using Image J software to visualize the MAMs (66). Merged images with these binary images show the colocalization as MAMs.

Fluorescence recovery after photobleaching

FRAP was performed for analyzing the molecular dynamics within cells using LAS X Software (Leica Microsystems). The cells were cultured in a poly-lysine–coated glass-bottom dish (Matsunami Glass Ind., Osaka, Japan) for 2 days after transfection. The cells were stimulated with SA4503 (20 μm) for 1 h, then washed with and incubated in Hank's Balanced Salt Solution, and red fluorescence of mCherry was observed in live imaging using the confocal microscope. Imaging settings were as follows: 512 × 512 pixels, 63× magnification lens, 3× zoom, and scan speed 400 Hz. In this setting, 40 images were taken in ∼1.3 s/image with photobleaching just after taking the 10th image. The fluorescent intensity of regions of interest was measured using Image J software, and the normalized intensity was calculated compared with the intensity in the first image. The difference between intensity in the final and the 11th images was calculated as the index of recovery.

Immunoprecipitation

Immunoprecipitation was performed as previously reported (67). Briefly, the anti-GFP antibody was mixed with Protein A-Sepharose beads (GE Healthcare) and gently agitated at 4 °C for 0.5 h. Cell lysates were added to the beads and rotated at 4 °C for 2 h. The beads were washed with homogenizing buffer and denatured for Western blotting.

DNA diffusion assay

Cell apoptosis was determined by DNA diffusion assay (68). All procedures were performed in the dark as much as possible to avoid DNA damage and at room temperature unless otherwise specified. The cells were transfected with nontagged σ1R and σ1RE102Q or control vector (pcDNA3.1) to avoid the fluorescence of mCherry and collected in PBS 2 days after transfection. Then, the cells were mixed with 0.5% low melting temperature agarose (Cambrex, East Rutherford, NJ, USA) in PBS at 37 °C. The mixture lay on 0.5% normal melting temperature agarose (Nippon Gene, Tokyo, Japan)–coated slide glass and covered with a cover glass. The gels were kept at 4 °C until getting solid for 3 min. The cover glass was removed, and the top layer of 2% low melting temperature agarose was made in the same procedure. The gels were immersed in chilled lysis solution (10 mm Tris-NaOH, pH 10, 2.5 m NaCl, 1% Triton X-100, and 2 mm EDTA) and kept at 4 °C for 1 h. Then, the gels were immersed in alkaline solution (0.3 N NaOH, pH > 13.5, and 0.2% DMSO) for 10 min to denature the genomic DNA and then dipped into precipitation solution (1 m CH3COONH4 in ethanol) for 15 min. After washing with 75% ethanol, the gels were air-dried and incubated in prestaining solution (10 mm NaH₂PO₄, 5% DMSO, and 5% sucrose) for 5 min and stained with 20 μg/ml ethidium bromide (Nacalai Tesque) for 5 min. The gels were washed with ddw, immersed in Vectashield, and observed using the confocal microscope. The cells with a round appearance with clear borders were counted as nonapoptotic cells, and the cells with a halo of granular DNA with a hazy outer boundary were counted as apoptotic cells (68). Over 100 cells in 10 randomly selected regions were analyzed, and the percentages of apoptotic cells were calculated. Counting and analysis was performed in a blinded manner.

Statistical analysis

Data were expressed as mean ± S.D. with individual plots. Unpaired t test for two groups and one-way or two-way analysis of variance (ANOVA) followed by Bonferroni post hoc test for more than two groups were performed using GraphPad Prism version 8.4.2 (GraphPad Software, San Diego, CA, USA). t values in t test and F values in ANOVA are described in the Fig. legends. p < 0.05 was considered statistically significant.

Data availability

Data are to be shared upon request.