Cellular analysis of SOD1 protein-aggregation propensity and toxicity: a case of ALS with slow progression harboring homozygous SOD1-D92G mutation

Mutations within Superoxide dismutase 1 (SOD1) cause amyotrophic lateral sclerosis (ALS), accounting for approximately 20% of familial cases. The pathological feature is a loss of motor neurons with enhanced formation of intracellular misfolded SOD1. Homozygous SOD1-D90A in familial ALS has been reported to show slow disease progression. Here, we reported a rare case of a slowly progressive ALS patient harboring a novel SOD1 homozygous mutation D92G (homD92G). The neuronal cell line overexpressing SOD1-D92G showed a lower ratio of the insoluble/soluble fraction of SOD1 with fine aggregates of the misfolded SOD1 and lower cellular toxicity than those overexpressing SOD1-G93A, a mutation that generally causes rapid disease progression. Next, we analyzed spinal motor neurons derived from induced pluripotent stem cells (iPSC) of a healthy control subject and ALS patients carrying SOD1-homD92G or heterozygous SOD1-L144FVX mutation. Lower levels of misfolded SOD1 and cell loss were observed in the motor neurons differentiated from patient-derived iPSCs carrying SOD1-homD92G than in those carrying SOD1-L144FVX. Taken together, SOD1-homD92G has a lower propensity to aggregate and induce cellular toxicity than SOD1-G93A or SOD1-L144FVX, and these cellular phenotypes could be associated with the clinical course of slowly progressive ALS.

www.nature.com/scientificreports/ So far, more than 200 mutations of SOD1 have been described, and most of them are heterozygous mutations, the pathological mechanism of which is probably related to gain of a toxic function rather than loss of function. Previous studies have pointed out that part of the toxicity of mutant SOD1 may be involved in its aggregation. FALS patients carrying heterozygous G93A mutation for SOD1 (SOD1-G93A) exhibit rapid progression of motor symptoms and die of pneumonia or respiratory insufficiency approximately 3 years after onset of symptoms 9 . In vitro SOD1-G93A protein has been reported to have a higher propensity to aggregate than wild-type SOD1 (SOD1-WT), leading to formation of cytoplasmic inclusions and cellular toxicity 10,11 . A FALS patient carrying SOD1-L144FVX has been reported with severe motor symptoms, resulting in death only 10 months after onset 12 . We have demonstrated that induced pluripotent stem cells (iPSC) of the patient carrying SOD1-L144FVX showed misfolded SOD1 13 .
A rare case of ALS showing slow progression that harbored a novel SOD1 homozygous D92G mutation (homD92G) is presented. Protein-aggregation propensity and cellular toxicity were examined using cell lines and motor neurons derived from a patient carrying SOD1-homD92G.

Results
Identification of a novel SOD1 homozygous mutation of D92G in an ALS patient with slow progression. A 54-year old woman noticed difficulty standing up from a chair. She felt difficulty grabbing objects with her right hand at the age of 60 years, and then the same symptom gradually progressed to the left hand. Her temperature sensation was diminished at the age of 62 years. These symptoms gradually worsened, and she recently could not walk without a cane. She visited our hospital at 71 years of age.
General cognitive function was preserved. Neurological findings showed muscle atrophy of both thenar muscles and thighs with fasciculations. Manual muscle testing (MMT) showed proximal dominant muscle weakness (MMT 2-3/V). Gowers sign was present. Deep tendon reflexes were increased on both side limbs. Babinski's sign was present on the right side. Temperature sensation was diminished in the lower limbs.
Magnetic resonance imaging (MRI) of the brain and spinal cord showed no specific abnormality (Fig. 1a,b), and MRI of the lower limbs demonstrated atrophy and high intensity on T1-weighted imaging (WI) in the calf (Fig. 1c). Electromyography showed acute and chronic neurogenic changes of the 1st dorsal interosseous, T7 paraspinal, tibialis anterior, and gastrocnemius muscles. The nerve conduction study showed decreased compound muscle action potentials (CMAPs) in the median and peroneal nerves and a mild decrease of the sensory nerve action potential (SNAP) in the sural nerve, but no evidence of conduction block or demyelination. Taken together, she was clinically diagnosed at the age of 71 years as having probable ALS according to the  www.nature.com/scientificreports/ updated Awaji criteria 14,15 , though the progression of motor symptoms was slow compared to that of typical ALS patients. Her family history showed repeated consanguineous marriages, indicating autosomal recessive inheritance (Fig. 1d). DNA sequencing showed homozygous D92G mutation in SOD1 gene (homD92G) (Fig. 1e). We performed whole exome sequencing using the patient's sample to examine the responsible genes of FALS (Supplementary Table 1). The SOD1-D92G was the only mutation with clinical significance determined by ClinVar (https:// www. ncbi. nlm. nih. gov/). Her parents had passed away without motor dysfunction. It is assumed that they could have been subclinical carriers with heterozygous SOD1-D92G mutation.
In silico analysis of SOD1-D92G. To examine the structure-related protein propensity of SOD1-D92G, several in silico analyses were performed in comparisons with known FALS mutations, SOD1-L84F, N86S, D90A, G93A, and L126S. SOD1 mutations showing a rapid progression or severe phenotype, SOD1-L84F, N86S, G93A, and L126S showed predictive values of "probably damaging" or "deleterious" protein propensity on 3-4/4 software analyses (Table 1). On the other hand, SOD1 mutations showing a slow progression phenotype, SOD1-D90A and D92G, showed predictive values of "deleterious" protein propensity on 0-1/4 software analyses (Table 1). In addition, SOD1-D92G showed "deleterious" with a PROVEAN score of − 2.872 (cutoff = − 2.5), although the score was near the cutoff. Taken together, the severity of the phenotype in FALS with SOD1 mutation may be correlated with the predicted value of protein propensity on in silico analysis.
Low aggregation propensity and minor cell toxicity of the Neuro2a cell line overexpressing SOD1-D92G. The aggregation property and cellular toxicity of SOD1-D92G were assessed in mouse neuroblastoma Neuro2a cells transfected with pSOD1-D92G-EGFP vector. As a positive control, pSOD1-G93A-EGFP, which mutation has shown aggregation phenotypes in in vitro cell lines, as well as in the motor neurons of patients and rodent models, was used 11,17 . No significant difference in transfection efficiencies was found among pSOD1-WT-EGFP, D92G-EGFP, and G93A-EGFP ( Supplementary Fig. 1). Confocal microscopy showed large aggregates in cells expressing SOD1-G93A-EGFP (G93A), whereas fine aggregates were seen in cells expressing SOD1-D92G-EGFP (D92G) (Fig. 2d-i). Both aggregates were immunostained by monoclonal antibody to specifically detect misfolded SOD1 (A5C3) 18 . Cells expressing SOD1-WT-EGFP (WT) did not show misfolded aggregates (Fig. 2a-c). Low magnification images of G93A showed a dense round formation (EGFP-aggregate) in the cytoplasm (Fig. 2l), whereas most D92G and WT showed disperse formation (Fig. 2j,k). The ratio of the number of EGFP-aggregate positive cell relative to whole EGFP-expressing cells was significantly less in D92G than in G93A, whereas the aggregation propensity of D92G was similar to that of WT (Fig. 2m). Next, immunoblotting analysis was performed. The soluble fraction of G93A was the most reduced among G93A, D92G, and WT, and the differences between them were significant ( Fig. 3a,b). In contrast, the insoluble fraction of G93A was slightly increased among them (Fig. 3a,c). The relative ratio of the insoluble to the soluble fraction of G93A was the most increased. There were significant differences in the relative ratio between two groups among WT, D92G and G93A (Fig. 3d). The condition of D92G may be different from that of WT.
Finally, cellular toxicity was examined in each subtype. The LDH assay showed that the toxicity of G93A was significantly higher than of D92G and WT (p < 0.05), whereas there was no significant difference between D92G and WT (Fig. 3e). Therefore, D92G has a lower propensity to aggregate and lower toxicity than G93A.
Minor misfolded SOD1 and neurodegenerative phenotype of the motor neurons derived from patient iPSCs carrying SOD1-homD92G. Aggregation propensity and neurodegenerative phenotype were evaluated in an in vitro human cellular model using patient-derived motor neurons carrying SOD1-homD92G. We have established one iPSC line carrying D92G. We used one FALS-derived iPSC carrying het- www.nature.com/scientificreports/  = 4). Data are shown as means ± SD. One-way ANOVA (followed by Tukey's test) was used for statistical analysis. *p < 0.05, ****p < 0.0001. OD optical density. www.nature.com/scientificreports/ erozygous SOD1-L144FVX 13 and healthy control-derived iPSCs 201B7 19 . The extent of misfolded SOD1 was examined using an ELISA assay detecting misfolded SOD1. The level of misfolded SOD1 was lower in motor neurons carrying SOD1-homD92G than in motor neurons carrying SOD1-L144FVX (Fig. 4a). Finally, the cellular vulnerability of patient-derived motor neurons was analyzed by a survival assay calculating cell viability expressed as a ratio of the number of neurons on Day 14 to the number of neurons on Day 7. The ratio of the number of survived motor neurons carrying SOD1-homD92G at Day 14 relative to those at Day 7 was significantly higher than that of L144FVX (Fig. 4b,c and Supplementary  www.nature.com/scientificreports/ accompanied by the loss of SOD1 activity 24,25 , suggesting that loss of SOD1 function could be associated with the severe clinical course of ALS. The present study is the first report of a case of FALS carrying homSOD1-D92G that showed onset in middle age and very slow progression. In a similar case, homozygous SOD1-D90A in FALS was reported to show slow progression in a Swedish/Finnish population 26 . The present in silico analyses showed that both SOD1-D90A and D92G had less "pathogenic" protein propensity than L84F, N86S, and L126S, of which homozygous mutations are linked with a severe clinical course. This suggests that protein structure-related properties in the various mutations of SOD1 could explain the different propensities to aggregate and the clinical prognosis of SOD1-related FALS. Further studies are needed to elucidate the mechanism how these SOD1 mutations could determine the clinical course of each form of ALS. To investigate the causal relationship between protein propensity from a mutation of SOD1 gene and a neurodegenerative phenotype characterized by loss of motor neurons, protein aggregation, solubility and cellular vulnerability were examined using Neuro2a cells and patient-derived motor neurons. We found that SOD1-D92G is slightly more prone to aggregate than WT, but less prone to aggregate and less cytotoxic than G93A. There are several limitations in this study. First, detailed pathological assessment using postmortem human samples of motor neurons was not performed. Further research is needed to determine whether mild motor neuron loss and aggregation could contribute to slower disease progression. A major barrier in the study of neurological disorders including ALS is that it is difficult to perform a highly invasive biopsy. It is more difficult to obtain the central nervous system (CNS) tissue, which is vulnerable to postmortem degeneration, suitable for biochemistry and molecular biology from autopsy. To overcome these barriers, we generated iPSC-derived motor neuron which has been reported to recapitulate the post-mortem CNS pathology in neurodegenerative disease 27 . Second, the mechanism of how homozygosity of SOD1-D92G could exert more toxicity than heterozygosity could not be elucidated. A deeper analysis (e.g. all-atom Molecular Dynamics simulation) of how the homD92G mutation affected the SOD1 dimer structure and how it could alter potential interactions with other proteins seems necessary to further comprehend the slow-progressing phenotype of this novel gene variant. Since there were no ALS cases in the presumed heterozygous parental generation, it was thought that the SOD1-D92G acquired ALS disease propensity only in homozygotes. A gene-correction study using patient-derived iPSCs might show the more precise cellular phenotypes of heterozygous and homozygous SOD1-D92G.
This study showed a relationship between the mutant SOD1 aggregation propensity and FALS prognosis in silico and a relationship between the SOD1 aggregation propensity and cell fragility in vitro. Future therapeutic approaches to regulate the aggregation propensity of SOD1 may alter the progression and prognosis of SOD1related disease.

Materials and methods
Ethics statements and consent to participate. This study was approved by the Kyoto University Graduate School of Medicine Ethical Committee (G1178, R91). Written, informed consent was obtained from the participants prior to inclusion in the study. Samples from the participants were identified by numbers, not by names. All methods were performed in accordance with the relevant guidelines and regulations. MRI data acquisition. MRI scans were performed using the 3 T Magnetom Prisma or Avanto system for the brain or muscles, respectively (Siemens, Erlangen, Germany). Data analysis was performed with Centricity PACS 4.0 (GE Healthcare, Chicago, IL, USA). DNA sequencing. DNA was extracted from a blood sample, and previously reported primers were used 28 .
The exons of SOD1 were amplified and Sanger sequencing was performed on 3730xl DNA Analyzer (Thermo Fisher Scientific, Waltham, MA, USA) with Thermo Fisher ScientificTerminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific). A novel homD92G mutation in the SOD1 gene was identified (Fig. 1).

Whole exome sequencing.
For whole-exome sequencing, genomic DNA was extracted, and 1 μg was sheared and used for the construction of a paired-end sequencing library as described in the protocol provided by Illumina. Enrichment of exonic sequences was then performed for each library using the SureSelect Human All Exon V6 (Agilent Technologies Inc., Santa Clara, CA, USA) following the manufacturer's instructions. Libraries for whole-exome sequencing were sequenced with a NovaSeq 6000 (Illumina Inc., San Diego, CA, USA).
In silico analysis. The effects of the newly detected SOD1-homD92G and previously reported mutations of www.nature.com/scientificreports/ Warrington, UK) reagent according to the manufacturer's protocol. After overnight incubation, the medium was changed with new growth medium. At 48 h after transfection, the cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS). After 10-min incubation with PBS/0.1% Tween, the samples were incubated overnight at 4 °C with primary antibodies against misfolded SOD1 (A5C3, 1:200, MEDIMABS, Montreal, Canada). The samples were subsequently incubated with donkey-derived secondary antibodies (Alexa Fluor 594, 1:1000, Thermo Fisher Scientific) for 1 h at room temperature, and they were then covered with antifade mounting medium, VECTASHIELD ® with DAPI, (Vector Laboratories, Burlingame, CA, USA). High magnification images were acquired with an FV-1000 confocal laser scanning microscope (Olympus, Tokyo, Japan). SOD1-EGFP aggregates were observed with a BZ-X710 fluorescence microscope (KEYENCE, Osaka, Japan). A 'SOD1-EGFP-aggregate' was defined as a dense round formation more than 10 µm in diameter and counted in a double-blind manner. An 'aggregate-positive cell' was defined as a cell having an 'EGFP-aggregate' . The ratio of the number of 'aggregate-positive cells' to that of whole EGFP-expressing cells was calculated.
The Neuro2a cells were transfected with the plasmid carrying SOD1-EGFP (WT, D92G, or G93A) using LTX (Thermo Fisher Scientific) reagent according to the manufacturer's protocol. After overnight incubation, the medium was changed with new growth medium. At 48 h after transfection, the cells in 6-well dishes were washed twice with PBS and then scraped in 300 µl of PBS. The cells were sonicated (Cosmo Bio, Bioruptor, Tokyo, Japan) for 5 min and shaken for 15 min at 4 °C, and then the lysates were centrifuged at 12,000g for 10 min at 4 °C. The resulting pellet was resolved in 20 µl of 2% SDS as the PBS-insoluble fraction, and the supernatant was designated the PBS-soluble fraction. These lysates were incubated in 2 × SDS sample buffer and boiled at 95 °C for 5 min. One-fourth of the insoluble fraction and 1/60 of the soluble fraction were processed for the detection of SOD1. These proteins (PBS-soluble 5 µl, PBS-insoluble 5 µl) were loaded onto SuperSep™ Ace 10-20% (Wako, Tokyo, Japan). The proteins were then transferred onto polyvinylidene difluoride membranes and blocked with 5% skim milk in Tris-buffered saline Tween-20 (TBST) for 30 min. The membranes were incubated with an anti-SOD1 primary antibody (Enzo Life Science abi-sod-100F, 1:1000) at 4 °C overnight. Next, the membranes were incubated for 1 h at room temperature with a horseradish peroxidase secondary antibody (Santa Cruz #sc-2005, 1:10,000), and the protein bands were visualized using ECL Western Blotting Substrate (Thermo Fisher Scientific). Chemiluminescent signals were detected using an Amersham Imager 600 imager (GE Healthcare, Chicago, IL, USA). To evaluate the protein levels, these bands were analyzed with ImageJ ver. 1.50i (https:// imagej. nih. gov/ ij/). Total protein was measured by Coomassie Brilliant Blue staining (CBB Stain One, Nacalai Tesque, Kyoto, Japan). The insoluble fraction ratio was calculated as I/S, where S is the band intensity of the soluble fraction and I of the insoluble fraction.
Cell viability assay. Neuro2a cells were plated onto 24-well plastic plates. Cell viability was evaluated with the LDH assay kit (Dojindo, Tokyo, Japan) 48 h after transfection. The LDH assay was performed according to the manufacturer's protocol.
Generation of motor neurons. Motor neurons were generated from iPSCs as previously described 13 .
Motor neuron survival assay. The iPSCs derived motor neurons were cultured on iMatrix-coated 96-well plates (BD Bioscience, San Jose, CA) for 7 days and 14 days. The number of surviving motor neurons stained with the βIII-tubulin antibody was quantified by IN Cell Analyzer 6000 and IN Cell Developer toolbox software 1.9, and the ratio of surviving neurons on day 14 to those on day 7 was shown as a percent, as previously described 13 .
After washing and blocking with TBS-T containing 1% BSA, 200 μg protein/100 µl of samples were added, and incubation was carried out for 2 h at room temperature. Recombinant mutant SOD1 protein (G93A) was used to obtain a standard curve. For detection, the plates were incubated with 3 μg/ml anti-SOD1 antibody (ENZO), www.nature.com/scientificreports/ followed by sheep anti-rabbit IgG F(ab)'2 fragment linked to horseradish peroxidase (1:3000; GE Healthcare). After incubation with tetramethylbenzidine solution (BD Bioscience) at room temperature for 30 min, absorbance at 450 nm was measured by VersaMax (Molecular Device, Sunnyvale, CA, USA).