SPG7 and SPAST HSP patients involved in this study were examined by neurologists CMS and KRK. Details of the patient clinical characteristics and causative gene mutations are listed in Table 1. Control and patient samples used were matched for age and gender. Mean age range of SPG7 patients was 60 years, and those of SPAST patients and healthy controls were 51 and 60 years, respectively. Olfactory neurosphere-derived cells (ONS cells) were prepared from olfactory mucosa biopsies of participants (Féron et al., 1998, 2013) with their informed and written consent.

All patients except patient 3 have been previously reported (refer to Table 1). Patient 3 was identified using targeted next-generation sequencing, as described elsewhere (Kumar et al., 2013). In brief, patients with an apparent autosomal recessive or sporadic mode of inheritance were screened for nine autosomal recessive HSP-associated genes using a PCR-based library preparation and Roche Junior 454 next-generation sequencing, after exclusion of SPAST mutations. Amplicons that did not achieve 30 reads were resequenced with a conventional Sanger sequencing method. All pathogenic or potentially pathogenic variants were confirmed by Sanger sequencing on an ABI Prism 3130xl Genetic Analyzer platform (Applied Biosystems).

ONS cells were derived from olfactory mucosa biopsies using our published techniques (Matigian et al., 2010; Féron et al., 2013). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM)/F12 (Gibco) with 10% fetal bovine serum at 37°C and 5% CO₂. Please refer to our previous publication (Féron et al., 2013) for a detailed protocol of generating olfactory neurosphere-derived cells from olfactory mucosal biopsies.

Cells were treated with 100 nM MitoTracker green for 30 min at 37°C to label active mitochondria (Park et al., 2014). Images of mitochondria-labeled cells were captured using a 63× objective on a confocal microscope. Using ImageJ image analysis software, raw images were binarized, and mitochondrial morphological characteristics of the aspect ratio (AR), indicating mitochondrial length, form factor (FF), and mitochondrial branching, were calculated as described before (Mortiboys et al., 2008). AR is the ratio of the major to minor axis of the mitochondria, and FF is calculated as (perimeter of mitochondria²)/(4π ^∗ area of mitochondria) (Mortiboys et al., 2008). Thirty cells were imaged and analyzed per cell line per technical replicate experiment.

Mitochondrial mass and membrane potential were assessed by labeling mitochondria with 100 nM MitoTracker green for 30 min (as above) and 50 nM non-quenching TMRM for 30 min at 37°C, respectively. The cells were washed twice with PBS before measuring fluorescence using a BD LSRFortessa^™. Fluorescence was measured for 10,000 events per sample, and the data were analyzed using the BD FACS Diva^™ software (Gautier et al., 2012). Mitochondrial membrane potential was normalized to MitoTracker green fluorescence, a relative estimate of the mitochondrial mass per cell.

ONS cells (10,000) were plated in each well of a black-walled 96-well plate and cultured for 24 h before being assessed. The cells were stained with 5 μM MitoSOX Red for 15 min (as an indication of mitochondrial oxidative stress) (Teves et al., 2018) or 5 μM CM-H₂DCFDA for 15 min (a measure of general cellular oxidative stress) (Wali et al., 2016) at 37°C. The cells were washed twice with PBS before fluorescence was measured using a VICTOR Nivo Multimode Microplate Reader. MitoSOX and CM-H₂DCFDA fluorescence levels were normalized to mitochondrial mass, using dual MitoTracker green staining.

ONS cells (10,000 per well) were plated in a white-walled 96-well plate for 24 h before being assessed. The assay was performed as per the manufacturer’s instructions and as described before (Wali et al., 2016). Luminescence was measured using the VICTOR Nivo Multimode Microplate Reader. The ATPlite luminescence values were normalized to mitochondrial mass, using dual MitoTracker green staining.

A Seahorse XF24-3 Bioanalyzer (Agilent Technologies) was used to measure mitochondrial oxygen consumption as described previously (Koentjoro et al., 2017). On day 1, 70,000 ONS cells per well were plated in a 24-well XF24 V7 cell culture microplate and cultured in DMEM/F12 (Gibco) with 10% fetal bovine serum at 37°C and 5% CO₂ for 24 h. On day 2, the DMEM/F12 media was replaced with Seahorse XF Base media supplemented with 2 mM L-glutamine, and the pH was adjusted to 7.35 ± 0.05 (basal media). Cells were incubated in the basal media for 1 h at 37°C in a non-CO₂ incubator. The plate was then assembled with a hydrated sensor cartridge and placed in the Seahorse flux analyzer. The oxygen consumption rate (OCR) was measured at baseline and after subsequent stepwise injection of the mitochondrial respiratory chain inhibitors, oligomycin (1 μM), carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP) (1 μM) and rotenone/antimycin A (0.5 μM). The OCR readings were normalized to cell number measured by CyQUANT assay in a sister plate. Multiple parameters of mitochondrial respiration were calculated based on the response of the cells to the respiratory chain inhibitors: (a) basal respiration, basal OCR - non-mitochondrial respiration rate; (b) ATP production, difference in OCR before and after oligomycin treatment; (c) maximal respiration, OCR measurement after FCCP treatment - OCR measurement after oligomycin treatment; and (d) spare respiratory capacity, maximal respiration - basal respiration.

ONS cells (5,000) were seeded into each well of a 96-well plate. On day 0 (4 h post seeding), day 1 (24 h post seeding), and day 2 (48 h post seeding), the relative cell numbers were assessed using the CyQUANT^® NF Cell Proliferation Assay Kit, following the manufacturer’s instructions.

Three technical replicates were performed for all experiments: three control cell lines, three SPG7 patient lines, and three SPAST patient cell lines. One-way analysis of variance with Tukey’s multiple comparisons test was performed to compare control, SPG7, and SPAST sample results. Statistical analysis and graphing were performed with GraphPad (Prism 8).

All patients with mutations in SPG7 had complex HSP phenotypes (clinical features detailed in Table 1). Complicating features in SPG7 patients included ophthalmoplegia (patient 1), ptosis (patient 1), ataxia (patients 2 and 3), and mitochondrial cytopathy on muscle biopsy (patient 1). In contrast, no SPAST patients (patients 4, 5, and 6) had typical mitochondrial disorder-related clinical signs (such as ocular myopathy, present in a range of mitochondrial myopathy syndromes; Pfeffer and Chinnery, 2013), and only one of three (patient 4) had additional features of dysarthria and upper limb weakness indicative of complicated HSP.

Six HSP patients, three with SPG7 mutations and three with SPAST mutations, were recruited to the study. These patients carried a variety of mutations (Table 1). SPG7 patients were patient 1 [c.1454_1462del (het); c.1529C > T (het)], patient 2 [c.1529C > T (het); c.1727C > G (het)], and patient 3 [c.1449 + 1G > A (novel het); c.1529C > T (het)]. SPAST patients were patient 4 (c.1392A > T), patient 5 (c.583C > G), and patient 6 (c.1096G > A). All patients except patient 3 have been reported previously (Abrahamsen et al., 2013; Kumar et al., 2013). SPG7 patient 3 was a 59-year-old man with a severe HSP phenotype complicated by upper limb cerebellar ataxia and cerebellar atrophy on MRI brain (Figures 1A,B). He was found to have the known common mutation NM_003119.3:c.1529C > T (p.Ala510Val) (Brugman et al., 2008; Bonn et al., 2010) and a novel canonical splice variant NM_003119.3:c.1449 + 1G > A, classified as pathogenic by the ACMG criteria (Richards et al., 2015) (PVS1, PM2, and PP3). Both variants were also present in a similarly affected sister (Figure 1C).

All three SPG7 patients had a common heterozygous variant (c.1529C > T; p.A510V) in addition to different compounding heterozygous variants. All coding variants were predicted to be pathogenic by several prediction algorithms (e.g., MutationTaster, PolyPhen 2, SIFT, and PROVEAN), likely due to high evolutionary conservation of affected amino acids and the considerable physicochemical or likely structural changes imposed. Assessment of possible splicing changes for the non-coding variant using Human Splicing Finder 3.1 provided possible aberrant splicing outcomes (see below).

The pathogenicity of the c.1529C > T variant (p.A510V) common to all SPG7 patients in this study has been extensively studied (Bonn et al., 2010). It affects a highly conserved and completely buried alanine residue at the beginning of the structural AAA+ Lid domain (Figure 1D). Because of the constrained environment around the alanine residue, the introduction of a bulkier valine side chain (89 vs. 117 Da) would be predicted to destabilize the hydrophobic core of the Lid domain. Inspecting the crystal structure of the paraplegin AAA+ ATPase domain (amino acids 305–565; PDB entry 2qz4; Karlberg et al., 2009), of the three possible conformations of the valine side chain, two are predicted to clash with Arg486, and one could potentially be accommodated. This may afford expression of the mutant p.510V protein, although it would likely be dysfunctional for ATPase activity.

The second variant in patient 1 was a deletion variant (c.1454_1462del; p.485_487delRRE) removing three charged amino acids from the highly conserved alpha helix bundle of the AAA+ Lid domain (Figure 1D). The deletion would remove an entire alpha helix turn and three charged residues necessary for structural bonds to other helices in the domain (Karlberg et al., 2009) and is predicted to cause severe structural destabilization.

The second variant in patient 2 was a missense mutation (c.1727C > G; p.S576W) located within the magnesium metal ion coordination and peptidase active site of the M41 peptidase domain (Figure 1D). The substitution of tryptophan for serine is a very large physicochemical change, a mass change of 105 to 204 Da and a hydrophilic serine residue common in active sites vs. tryptophan, which has a bulky hydrophobic aromatic side chain. This would likely cause structural and functional disruption to the peptidase active site. Based on previous mutagenesis studies of this functional site, if the variant protein were expressed, it would likely form dysfunctional complexes (Shanmughapriya et al., 2015). In this situation, peptidase activity could be compensated for by AFG3L2, as demonstrated in yeast (Bonn et al., 2010).

The second variant in patient 3 was a novel canonical donor splice site disrupting mutation (c.1449 + 1G > A) that would impact the polypeptide chain from the end of the AAA+ ATPase domain at the start of the AAA+ domain linker region (Figure 1D). A number of possibilities for the consequent splicing defect include (1) skipping of exon 10 and a frameshift leading to termination after six amino acids; (2) intronic read-through into intron 10 before an alternate donor splice site is encountered 90 nucleotides downstream, resulting in an in-frame inclusion of 30 amino acids that would alter the length of the AAA+ ATPase domain linker; and (3) intronic read-through into intron 10 before encountering a termination codon 96 nucleotides downstream. Transcript analysis would be required to resolve the resulting splicing mechanism, although the variant protein product would almost certainly not be expressed.

Paraplegin expression in patient and control cells was evaluated by Western blot analysis and by flow cytometry (Figures 2A–D). Western blot analysis showed that SPG7 patient cells had significantly higher expression levels of paraplegin protein, compared to control cells (p < 0.01) (Figures 2A,C). SPAST patient cells had a trend of higher expression levels of paraplegin when compared to control cells, but this difference was not statistically significant (p = 0.06) (Figure 2C). To validate this finding, we evaluated paraplegin expression using flow cytometry. SPG7 patient cells had significantly higher mean fluorescence intensity, compared to control cells (p < 0.01) and SPAST patient cells (p < 0.01) (Figures 2B,D). Both techniques showed that paraplegin expression was significantly increased in SPG7 patient cells compared to control and SPAST patient cells.

Mitochondria in SPG7 patient cells were highly fragmented and short (low AR) with a low degree of branching and interconnectivity (low FF; Figures 2E–G), showing typical signs of a dysfunctional mitochondrial network. Conversely, mitochondria in control cells and SPAST patient cells were longer and interconnected (p < 0.05; Figures 2E–G), indicative of a healthy mitochondrial network.

Consistent with their aberrant mitochondrial morphology, SPG7 patient cells also displayed lower mitochondrial mass (p < 0.05) and lower mitochondrial membrane potential (p < 0.05) (Figures 2H,I).

To investigate if oxidative phosphorylation was impaired in HSP patient cells, we measured mitochondrial OCR using a Seahorse Bioanalyzer (Figure 3A). Compared to control and SPAST patient cells, SPG7 patient cells had reduced basal respiration (Figure 3B, control vs. SPG7: p < 0.05; SPG7 vs. SPAST: p = 0.09), significantly lower levels of oxygen consumption attributed to ATP production (Figure 3C, control vs. SPG7: p < 0.05; SPG7 vs. SPAST: p < 0.05), reduced maximal respiration (Figure 3D, control vs. SPG7: p < 0.001; SPG7 vs. SPAST: p < 0.01), and reduced spare respiratory capacity (Figure 3E, control vs. SPG7: p < 0.01; SPG7 vs. SPAST: p < 0.05). Consistent with this, SPG7 patient cells had lower total ATP content compared to both control and SPAST patient cells (p < 0.05) (Figure 3F). SPAST patient cells were not significantly different from control cells on any of these measures (Figures 3B–F).

MitoSOX fluorescence, a mitochondria-specific superoxide indicator, was significantly higher in SPG7 patient cells compared to controls (p < 0.05) and SPAST patient cells (p < 0.01) (Figure 3G). CM-H₂DCFDA fluorescence, a general reactive oxygen species indicator, was significantly higher in SPG7 patient cells compared to control cells (p < 0.05) and SPAST patient cells (trend, p = 0.08) (Figure 3H).

MELAS and MERRF mitochondrial disease patient cells have increased oxidative stress and reduced cellular proliferation (James et al., 1996). To test if SPG7 patient cells were similarly affected, we evaluated cellular proliferation in SPG7, SPAST, and control cells. SPG7 patient cells had reduced cellular proliferation compared to both control and SPAST patient cells (p < 0.05) (Figure 3I).