Adult-onset autosomal dominant spastic paraplegia linked to a GTPase-effector domain mutation of dynamin 2

Chronic neurological disorders are highly prevalent in the Sakha (Yakut) Republic, Russian Federation. In the process of systematic ascertainment of patients with Viliuisk encephalomyelitis and related disorders [15], a multigenerational HSP family was identified. This 4-generational Siberian family included 9 patients. The oldest family member (I:2, Fig. 2) known to be affected by history had stiff gait and progressive muscle weakness of the lower extremities for at least 10 years before his death at 63. His 3 sons from two marriages (II:1, II:4, and II:6) developed the “family disease”. They were repeatedly examined over the following three decades and diagnosed with hereditary spastic paraplegia. In the third generation, three sons of patient II:4, also from separate marriages, and a daughter of patient II:6 developed the same disease (III:2, III:3, III:5, and III:7). Finally, a great-grandson (IV:1) was the youngest patient to be diagnosed. Pedigree was constructed based on cross interviews of patients and closest family members.

The study was approved by the Institutional Review Boards of the Institute of Health, North-Eastern Federal University, Yakutsk, and the Yakutsk Research Center, Siberian Branch of the Russian Academy of Medical Sciences. The statement confirmed in writing that the Clinical protocol complied with the Declaration of Helsinki. A written informed consent was obtained from each participant.

After obtaining informed consent, 5 affected and 5 unaffected family members underwent a neurological exam that included assessment of mental status, cranial nerves, muscle strength (MRC scale), coordination, tendon reflexes, muscle bulk, muscle tone, plantar responses, foot deformity, and gait. Evaluation of sensory impairment included clinical testing for pain and temperature sensation, vibration and position sense. Electrophysiological investigation performed in 3 patients (III:3, III:5, and IV:1) included motor and sensory nerve conduction velocities (NCV), compound muscle action potential (CMAP) amplitudes, distal motor latencies (DML), sensory NCV and sensory nerve action potential (SNAP) amplitudes recorded under standard conditions from the median, ulnar, peroneal, tibial, and sural nerves. Routine clinical MRI of the spinal cord was also obtained in three cases. Blood for DNA extraction was drawn from 9 family members.

WES was performed using genomic DNA extracted from peripheral white blood cells of patients II:6 and III:3. Exome capture utilized TruSeq Kit v1 (Illumina, Sand Diego, California) in accordance with manufacturer’s instructions. Library construction, sequence generation, sequence alignment to the reference genome (UCSC GRCh37/hg19), variant calling, and identification of potentially pathogenic variants were performed as described [16]. Variant analysis was performed using an autosomal dominant genetic model. Only variants satisfying stringent coverage and genotype quality criteria were included in further downstream variant analysis. WES generated variants were further validated by segregation analysis to 3 additional affected and 4 unaffected members of the Siberian family using standard Sanger sequencing of amplified DNA fragments.

The wild-type human DNM2 in pmCherry-N1 vector (Addgene plasmid 27689) was a kind gift from Dr. C. Merrifield [17]. The c.2155C > T, p.R719W mutation was introduced into the pmCherry-N1 DNM2 plasmid using the Quick-Change Site Directed Mutagenesis kit according to manufacturer’s protocol (Agilent Technologies, Santa Clara, CA). The integrity of all plasmids was confirmed by DNA sequencing.

HeLa cells were grown in Dulbecco's modified Eagle’s medium (Life Technologies, Grand Island, NY) supplemented with 10 % fetal bovine serum. 50-60 % confluent cells were transfected with plasmids containing wild type or mutant DNM2 by using transfection reagent HilyMax (Dojindo Molecular Technologies, Rockville, MD) according to the manufacturer’s protocol. Cultures were then incubated in growth medium. At 20 hrs post-transfection, the cells were serum starved for 1 h and treated with 25ug/ml Alexa-Flour 488 conjugated Transferrin (Life Technologies, Frederick, MD) for 15 min at 37 °C. Cells labeled with fluorescent transferrin were then washed and fixed with 4 % paraformaldehyde. Immunofluorescene analysis was performed as previously described [18]. Cells were visualized with a Zeiss LSM 510 confocal microscope. Transferrin immunofluorescent signal was measured by utilizing ImageJ (Image Processing and Analysis in Java, National Institutes of Health) software. Mean fluorescence was first measured in the background; the corrected total transferrin fluorescence signal was then compared between cells expressing WT and mutant DNM2 constructs. Statistical significance was evaluated using the Student’s t-test. P-value <0.05 was considered significant.

Molecular models of dynamin 2 wild type and p.R719W were generated by I-TASSER [19‐21], using dynamin 1 as a template [22]. The sequence homology between dynamin 1 and 2 is 78 % identical overall and 87 % in the GTPase domain. The position of R719 in dynamin 1 is R725. The molecular models were visualized and H-bonds calculated by Chimera [23]. The dynamin 1 tetramer image was generated from docked crystal structures in the 3D density map of K44A-dynamin 1 [24].

The pattern of disease inheritance in this family was autosomal dominant (Fig. 2). Clinical information obtained at evaluation of 5 affected family members (II:6, III:3, III:5, III:7, and IV:1) is summarized in Table 1. The illness had an insidious onset between the ages of 10 and 37 years (mean, 26 years) with impaired gait, muscle stiffness and weakness in the lower limbs. Further progression of illness in patients II:6 and III:5 led to severe bilateral lower limb muscle weakness requiring the use of cane and eventual wheel-chair dependency by the 28^th and 23^rd years of illness. Three patients (I:2, II:1, and III:2) died after an illness lasting for 23 to 32 years, and one (II:4) died in an accident.

At examination (Table 1), a single patient had mild developmental cognitive delay. No optic nerve atrophy, nystagmus, or signs of ataxia were found. Cranial nerves were intact, bulbar functions preserved until late in the illness, none showed dysphagia, and only a single patient (III:5) had moderate spastic dysarthria. Upper extremities muscle bulk, strength, muscle tone, reflexes, sensation and coordination were not affected. In the lower limbs, typical features of upper motor neuron involvement were present in all patients, including spasticity, reduced muscle strength, tightness of Achilles heels. Deep tendon reflexes were increased in all patients with clonus in 3 and bilateral Babinski sign in 4. Moderate muscle atrophy of lower leg muscles was noticed in advanced illness in patients II:6, III:3, III:5, and one patient had flexion contracture of leg muscles and chronic decubitus ulceration (II:6). All patients had profoundly spastic gait. Sphincter control abnormalities manifesting as urinary urgency were observed late in the illness in two patients (II:6 and III:5). Clinical testing for pain, temperature, and position sense did not reveal any deficits. Mild distal vibratory sense impairment was detected in patients II:6 (inconsistently), III:3, and III:5 late in the course of illness. Scoliosis was present in one patient and bilateral pes cavus in all examined patients. MRI of the spinal cord performed in patients III:3, III:5, and IV:1 showed no evidence for cord compression, cord atrophy or intrinsic spinal cord abnormalities. There was no evidence of multisystem involvement as determined by routine blood count and routine blood chemistry. In summary, the clinical course in five affected individuals over a multi-decade observation period was overwhelmingly consistent with upper motor neuron dysfunction leading to progressive spasticity. Only late in the course of illness symptoms suggestive of a mild peripheral involvement in the form of mild sensory changes and distal muscle atrophy became apparent. No chronic neurological diseases were found among other members of the extended family or other inhabitants of this village.

Motor and sensory nerve conduction studies were performed in patients III:3, III:5, and IV:1 (Table 2). Patients III:3 and III:5 were in advanced phases of illness at the 17^th and 32^nd year from disease onset, patient IV:1 was the youngest in this family at the age of 17 years, in the 7^th year after the disease onset. In the n.medianus, conduction studies revealed mostly preserved distal motor latency (DML), motor nerve conduction velocities (NCV) and compound muscle action potentials (CMAP). In the lower limbs, stimulation of n.peroneus profundus and n.tibialis posterior showed reduced NCV and CMAP, most significantly in patients with advanced disease. Sensory nerve action potential (SNAP) amplitudes in n.suralis were also severely affected in patients with advanced disease. The electrodiagnostic study revealed an axonal impairment in motor and sensory peripheral nerves confined to the distal lower extremities .

Prior to exome sequencing, coding regions of common dominant HSP genes had been analyzed by Sanger sequencing, including SPAST (spastin) linked to autosomal dominant SPG4 (MIM# 18601) and ATL1 (atlastin) causing SPG3A (MIM# 182600), genes that account for 37-46 % and 6-11 % of dominant HSP cases [25]. No pathogenic changes were identified.

Exome sequencing was performed in two affected individuals, II:6 and III:3 (Fig. 2). A number of filtering steps were used to prioritize sequence variants (Table 3), beginning with the requirement that the variant had to be shared by two studied affected members of the Siberian family in heterozygous state, assuming a dominant model. Variants in non-coding regions and synonymous SNPs were excluded. Candidates were further restricted by excluding variants reported to have occurred in adult cohort of over 900 subjects participating in the ClinSeq project (ClinSeq www.genome.gov/20519355 ) or the 1000 genome project. Variant population frequency was additionally examined using the ExAC data set. Missense variants were further prioritized by the degree of functional disruption predicted by PolyPhen-II ( www.genetics.bwh.harvard.edu/pph2/ ), SIFT [26], MutationTaster ( www.mutationtaster.org/ ), and CADD (Combined Annotation–Dependent Depletion) - a single-annotation method used to prioritize functional, deleterious and pathogenic variants across many functional categories, effect sizes, and genetic architectures [27]. Exome sequencing did not identify any predicted or known pathogenic changes in previously identified HSP-associated genes. Selected variants were then assessed based on encoded protein expression and function. Considering that our patients had an adult onset neurodegenerative disorder with high penetrance, we excluded genes with no expression in the CNS or genes involved in early embryonic development. Based on analysis of whole exome sequencing data, 4 heterozygous variants were identified under autosomal dominant model. Table 4 provides information regarding the encoded proteins function, expression, and the known disease associations. These four strongest candidates were tested for segregation within the Siberian family (Table 5). The results of this analysis indicated that DNM2 p.R719W variant is the only one that segregated perfectly.

The identified DNM2 missense substitution is located at NM_001005360:c.2155C > T; Chr19(GRCh37):g.10939808C > T in exon 19 of the DNM2 gene, which replaces Arginine (R) with Tryptophan (W) (p.R719W) in the encoded dynamin 2 (Fig. 3a). The region is highly conserved at the amino acid level, up to the worm (Fig. 3b). The p.R719W mutation is in the GTPase effector (GED) domain (Fig. 1). The variant is predicted to be damaging or deleterious by all methods employed in the study. The variant was seen only once in an individual of South Asian descent in the entire Exome Aggregation Consortium data set that comprises exomes of 60,706 unrelated individuals sequenced as part of various disease-specific and population genetic studies (www.​exac.​broadinstitute.​org).

Validation by Sanger sequencing and further segregation to other family members shows that only the affected members II:6, III:2, III:3, III:7, and IV:1 are heterozygous for the c.2155C > T, p.R719W change whereas unaffected individuals III:1, IV:2, IV:3, and IV:4 are homozygous reference (Table 5). We conclude that the DNM2 c.2155C > T, p.R719W variant is the underlying molecular basis for the HSP phenotype in the Siberian family.

To investigate the effects of the disease-related dynamin 2 p.R719W mutant, we evaluated clathrin-mediated endocytosis in HeLa cells transiently transfected with the wild type and mutant dynamin 2 following incubation with fluorescently labeled transferrin. We observed a prominent punctate staining in cells transfected with p.R719W DNM2 (Fig. 4), a phenomenon morphologically similar to that reported for other DNM2 mutations causing CMT and CNM [28]. There was a significant decrease in transferrin uptake in cells expressing mutant dynamin 2. This decrease was evident in all transfected cells and was especially significant (more than 50 %) in cells marked by arrows. The granules are localized most prominently in the perinuclear area, likely in the endosomal compartment. The results suggest that inhibited endocytosis is part of the pathophysiological mechanisms leading to HSP in this family.

The HSP mutation p.R719W is located in the three-helix bundle of dynamin called the bundle-signaling element (BSE) (Fig. 5a). It is intriguing that p.R719W is the only disease-associated mutation found in this region. Both CMT and CNM mutations are located in the stalk (Middle/GED) and PH domains of the protein (Fig. 5a). The BSE undergoes a large conformational change upon GTP hydrolysis that has been described as the power stroke of dynamin [29, 30]. A mutation in this region suggests a defect in propagating the BSE conformational change to the rest of the molecule and assembled helical polymer. In addition, the mutation is located near a hinge between the BSE and the stalk and as indicated by fewer H-bonds (1 H-bond in W719 compared to 3 H-bonds in R719), p.R719W may disrupt the connection between these domains (Fig. 5b). The potential disruption of R719W in the assembled protein is further illustrated by examining the dynamin tetramer (Fig. 5c, asterisks).