SPG7 mutations in amyotrophic lateral sclerosis: a genetic link to hereditary spastic paraplegia

The study was approved by the Ethics Board of Hannover Medical School (ID # 6269). Written informed consent was obtained from all patients. The study cohort consisted of 214 European patients (126 males, 88 females) diagnosed with ALS [6 familial ALS (fALS), 208 sporadic ALS (sALS) cases] based and recruited at the ALS/MND Clinic of the Department of Neurology of Hannover Medical School, Germany. Patients were subdivided into one of eight clinical subtypes [7]: (1) upper motor neuron (UMN) dominant ALS: predominant UMN involvement and clinical/neurophysiological signs of lower motor neuron (LMN) affection during follow-up; (2) bulbar phenotype (B): bulbar onset and predominant but not exclusive bulbar affection throughout the disease course; (3) flail arm syndrome: progressive, predominantly proximal, symmetric weakness and wasting in the upper limbs with little or no functional impairment of the bulbar muscles or legs 12 months after symptom onset; (4) flail leg syndrome: isolated weakness and wasting in the lower limbs for 12–24 months after symptom onset; (5) respiratory phenotype: predominant respiratory impairment at onset and only mild spinal or bulbar signs for 12 months; (6) progressive muscular atrophy (PMA): progressive LMN involvement and no clinical UMN signs; (7) LMN dominant ALS: predominant LMN symptoms with UMN signs at some point in the disease; (8) classic (Charcot’s) ALS: LMN and UMN signs in more than two regions within a period of 12 months after symptom onset. Disease progression was evaluated using the revised version of the ALS functional rating score (ALSFRS-R). At first presentation, the progression rate was calculated using the following formula: 48 − total ALSFRS-R/symptom duration in months [8]. Neurocognitive or behavioral impairments were assessed using the Edinburgh Cognitive and Behavioural ALS Screen (ECAS) [9] or the frontal assessment battery (FAB) [10] in 32 patients, and based only on clinical impression in 182 patients. Clinical cerebellar dysfunction was diagnosed if symptoms of cerebellar ataxia including gait ataxia and intention tremor in extremities not affected by weakness or oculomotor disturbances (saccadic pursuit, nystagmus) were present.

Genomic DNA was extracted from whole blood using the QIAamp DNA Blood Kit (Qiagen, Hilden, Germany). WES was performed on DNA from 23 ALS patients using Agilent SureSelect Human All Exon v4 Target Enrichment System on an Illumina HiSeq 2000 by Oxford Gene Technology, Begbroke, UK, as described previously [11]. All samples were sequenced to a mean target coverage of > 50×. WES data were analyzed using our in-house workflow based on INGENUITY Variant Analysis (QIAGEN Bioinformatics, Redwood City, CA, USA). To screen all exons (1–17) and flanking intronic sequences of SPG7 (NM_003119.2) for variants in 191 further ALS patients and to verify selected SPG7 variants identified by WES, conventional chain termination protocols were used. Primers and amplification conditions were adapted from a previously published study [12]. Minor allele frequencies (MAF) of genetic variants in non-Finnish Europeans were extracted from the ExAC Browser Beta database (https://​gnomad.​broadinstitute.​org/​). Variant pathogenicity was predicted using SIFT according to Alamut Visual software, version 2.11 (Interactive Biosoftware, Rouen, France), PolyPhen-2 (http://​genetics.​bwh.​harvard.​edu/​pph2/​), and the American College of Medical Genetics (ACMG) criteria [13]. Potential effects on gene splicing were evaluated using Alamut Visual software. The Human Gene Mutation Database Professional 2018.1, the ALS Online Genetics Database ALSoD (https://​alsod.​ac.​uk), and PubMed (https://​pubmed.​ncbi.​nlm.​nih.​gov) were interrogated for previously reported SPG7 variants.

Total RNA was extracted from whole blood of patient MD018 using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized on total RNA using the SuperScript III First-Strand Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, cDNA was sequenced using conventional chain termination protocols.

Predicted amino acid alterations were introduced into the crystal structure of paraplegin (PDB: 2qz4) [14], and the mutated structures were energy minimized using Macromodel of the Schrödinger Suite and the OPLS3 force field [15, 16]. A structural model of the paraplegin hexamer was constructed by alignment on the hexameric FtsH complex structure (PDB: 2dhr).

All patients were examined by nerve conduction studies (NCS) and electromyography (EMG) to detect sensory/motor neuropathy and the extent of LMN affection. Multisequence brain MRI studies of seven SPG7 mutation carriers done during diagnostic workup were reassessed by two neuroradiologists (MPW, PR) with regard to imaging findings associated with neurodegenerative diseases, such as corticospinal tract hyperintensities and global or regional cortical atrophies, particularly corpus callosum (CC) thinning and cerebellar atrophy [17‐19]. The total midsagittal CC area was evaluated on T1- or T2-weighted sagittal MRI images, manually estimated using ROI analysis, and compared to anatomical references (normal range 580–1040 mm²) [20].

To detect rare ALS-associated variants, WES was performed on DNA from whole blood of 23 unrelated ALS cases, and variants with a MAF of < 1% in genes associated with ALS according to ALSoD (n = 126) that were classified as pathogenic or likely pathogenic using ACMG guidelines were retrieved from the datasets. This filtering strategy identified three variants, each of which was carried by one patient, in two genes, SPG7 and LIPC, whereby SPG7 was recurrently affected (Supplementary Table 1). Both rare SPG7 variants, the missense c.1529C > T p.(A510V) variant in patient TALS002-01 and the splice site c.1552 + 1G > T variant in patient MD018, were heterozygous as confirmed by Sanger sequencing (Fig. 1a, d), and had previously been described in ALS or HSP patients (Table 1). To test the consequence of the splice site variant c.1552 + 1G > T, we amplified a part of SPG7 comprising exons 10–12 on cDNA from whole blood of patient MD018. By cDNA sequencing, an aberrant SPG7 transcript lacking exon 11 was identified (Fig. 1d). The altered transcript encoded by the c.1552 + 1G > T variant was predicted to result in an aberrant protein after amino acid 483 and a premature truncation after amino acid 556.

To study the SPG7 variant frequency in a larger ALS cohort, we performed mutational analysis of all 17 SPG7 exons and adjacent splice site regions (± 20 base pairs) on whole blood DNA of 191 further ALS patients. We detected seven patients carrying four different rare heterozygous missense variants in SPG7 predicted to be deleterious according to SIFT or PolyPhen-2 and affecting highly or very highly conserved amino acid residues (Fig. 1a, b and Table 1). All identified SPG7 nucleotide changes were summarized in Supplementary Table 2.

Taken together, in 9 of 214 (4.2%) ALS patients we identified five different rare SPG7 variants predicted to be deleterious, four missense and one essential splice site variant, all of which were heterozygous (Fig. 1a, d). None of the SPG7 variant carriers harbored additional non-synonymous SPG7 variants, excluding compound heterozygosity for pathogenic SPG7 variants. All identified SPG7 variants were previously described in HSP patients, and two of the missense variants were reported in ALS cases. Two SPG7 variants were inherited from mothers (aged 75 and 76 years) not affected by ALS to date suggesting reduced penetrance.

The SPG7 gene encodes paraplegin, a mitochondrial metalloprotease containing an AAA+ (ATPases associated with diverse cellular activities) domain that couples ATP hydrolysis to protein remodeling flanked by an FtsH-extracellular and a peptidase domain, which is thought to form a hexameric complex. All five different rare SPG7 variants identified here in ALS patients clustered in exons 7–13 that code for the paraplegin AAA+ domain, i.e. amino acid residues 305–565 (Fig. 1c, d) [14]. Variant p.(G349S) is located in the N-terminal P-loop of the AAA+ domain. Glycine at position 349 is part of the Walker A motif with its consensus sequence GxxGxGK(T/S), involved in the nucleotide binding process and ATP hydrolysis. Variant p.(R400W) is part of the α3-helix in the P-loop domain. The exchange of the positively charged arginine by the bulky hydrophobic tryptophan might destabilize the N-terminal domain and the adjacent five-stranded β-sheet. As this variant maps to the interface between the AAA+ monomers (see Supplementary Fig. 1) and changes the surface properties, it may also have a destabilizing effect on the supracomplex formation. Variants p.(R486Q) and p.(A510V) map to the α-helical bundle domain within a radius of 3 Å. The slightly more bulky side chain of p.(A510V) might interfere with the interactions that stabilize α-helices 5 and 6, disturbing the opening and closing of the α-helical bundle domain, which is thought to be important for oligomerization to the supracomplex. Additionally, p.(A510V) could affect the nearby Sensor 2 motif that is assumed to play a role in coupling the hydrolysis state to the oligomerization state [14]. The close vicinity of p.(R486Q) and p.(A510V) indicates a similar effect of both variants on the stability and function of the α-helical bundle domain, Sensor 2, and the supracomplex formation. The splice site variant c.1552 + 1G > T leads to a transcript lacking exon 11 predicted to cause a frameshift and a premature stop codon [18] that would alter the entire α-helical bundle domain of the  AAA+ domain and completely delete the peptidase domain (Fig. 1d) resulting in a loss of function of the aberrant protein.

A summary of clinical, electrophysiological, and neuroradiological features of the ALS patients carrying rare deleterious SPG7 variants (n = 9) is given in Table 2. In all SPG7 variant carriers, acute and chronic denervation signs were detected by EMG (Table 2). NCS showed a predominantly axonal motor neuropathy in all SPG7 variant carriers attributable to loss of motor axons in ALS, and additionally an axonal sensory neuropathy in three SPG7 variant carriers (VALS125, MD075, MD018). The latter three and three other SPG7 variant carriers (MD087, VALS008, VALS093) presented with pallhypaesthesia. No apparent underlying causes for the sensory impairment in these six patients such as long-term diabetes mellitus, chronic alcohol abuse, or chemotherapy were known. Brain MRI studies were available for seven patients with SPG7 variants (Table 2). The corticospinal tract T2 hyperintensity frequently described in ALS was not detected in any of the SPG7 variant carriers. MRI analysis showed varying degrees and patterns of brain atrophy consistent with a pattern suggestive of FTD in three cases carrying SPG7 variants (VALS008, VALS125, VALS093) (Fig. 2). A clinical diagnosis of FTD was confirmed 1 year after ALS symptom onset for patient VALS125, who only reached 10 of 18 points in the FAB screen showing cognitive impairment in conceptualization, motor programming and executive control of action as well as inhibitory control. Patient VALS093 also displayed clinical features of FTD, i.e. verbal impairment with respect to speech and comprehension with normal memory function. No neuropsychological testing was performed in patient VALS008, who had a rapid ALSFRS-R disease progression score of 1.76. These three patients carrying SPG7 variants also showed the most prominent CC thinning (< 500 mm²). The FTD-specific cognitive impairments disturbed executive function and reduced verbal fluency were identified by ECAS in the SPG7 variant carrier MD087. Cranial MRI revealed vermis atrophy in SPG7 variant carriers MD087 and VALS008 (Fig. 2a), and two others (TALS012-01, VALS020) showed a cerebellar atrophy pattern, clinically manifesting as intention tremor in patients VALS008 and VALS020. Cerebellar dysfunction was also clinically observed in SPG7 variant carriers VALS125 (saccadic pursuit) with no correlating cerebellar atrophy on brain MRI at the time of investigation, TALS002-01 (ataxic gait) and MD018 (saccadic pursuit, nystagmus) with no available brain MRI. Patient VALS125, harboring the p.(R486Q) variant, and patients VALS093 and TALS002-01, both carrying the p.(A510V) variant, were diagnosed with flail arm syndrome.

When comparing the characteristics in ALS patients from our cohort with (n = 9) and without (n = 205) rare deleterious SPG7 variants (Supplementary Table 3), clinical cerebellar dysfunction was more frequent in SGP7 variant carriers (5/9: 56%) compared to non-SPG7 variant carriers (16/205: 8%; P < 0.001, two-sided Fisher’s exact test). Furthermore, SPG7 variant carriers were overrepresented in two ALS subgroups, i.e. flail arm and flail leg syndrome, whereby flail arm syndrome was significantly more frequent in SPG7 variant carriers (3/9: 33%) versus non-SPG7 variant carriers (17/205: 8%; P = 0.04, two-sided Fisher’s exact test). However, no significant difference was detected when comparing SPG7 variant carriers with clinical or radiological signs of FTD (4/9: 44%) and non-SPG7 variant carriers exhibiting cognitive impairment consistent with FTD as assessed by ECAS (9/30: 30%; P = 0.337, two-sided Fisher’s exact test). Mean disease duration in SPG7 variant carriers was longer, i.e. 4.9 years, than in non-carriers, i.e. 3.7 years. However, the difference in disease duration (comparing 9 SPG7 variant carriers versus 205 non-carriers; P = 0.20, T test) and in survival (comparing 6 SPG7 variant carriers versus 88 non-carriers; P = 0.749, log-rank test) did not reach statistical significance (Supplementary Table 3).