Hereditary spastic paraplegia (HSP) comprises a group of clinically and genetically heterogeneous diseases that affect the upper motor neurons and their long axonal projections. The hallmark of hereditary spastic paraplegia (HSP) is progressive degeneration of the corticospinal tracts, often in a length-dependent manner, and progressive spasticity as the major clinical feature; neuropathological studies have identified axon degeneration of the corticospinal tracts and fasciculus gracilis fibers [1]. The key diagnostic findings are lower limb weakness, increased muscle tone, hyperreflexia, extensor plantar responses, and gait spasticity [2].

The burden inherent to corticospinal neurons with their long axonal processes spanning from motor cortex to distant segments of the spinal cord is likely to involve various cellular vulnerabilities that could account for the broad genetic heterogeneity of HSPs. Variant forms of HSP arise from mutations in genes directly implicated in membrane traffic, mitochondrial function, endoplasmic reticulum shaping/distribution, myelination, lipid/cholesterol metabolism, bone morphogenetic protein signaling, and endolysosomal function, as well as endosome and microtubule dynamics [3, 4]. Typically, the longest fibers — those connecting to the lower spinal cord segments — are the earliest to experience changes in their terminal regions that predate changes in cell bodies, leading to suggestions that impact on axonal traffic is a common mechanism of HSP. In this regard, HSP may be viewed as a central nervous system counterpart of the axonal form of Charcot-Marie-Tooth (CMT2) neuropathy, a clinically and genetically heterogeneous group of disorders of the peripheral nervous system marked by length-dependent distal axon degeneration of motor and sensory nerves [5].

We report on a novel association of an autosomal dominant HSP phenotype with a heterozygous DNM2 mutation. Dynamins are highly conserved large GTPases that play a critical role in clathrin-mediated endocytosis. They participate in converting the nascent invaginated clathrin-coated pit into a fully formed vesicle and in detaching the vesicle from the plasma membrane (membrane fission) [6]. In developing neurons, both endocytosis and exocytosis are critical for delivery of nutrients and building materials. Clathrin-mediated endocytosis plays a particularly important and specialized role at neuronal synapses [7]. Dynamin isoforms dynamin 1 and dynamin 3 are expressed in neurons, while dynamin 2 is found ubiquitously, including the CNS [8]. Dynamin 2 is involved in several other cellular processes, such as regulation of neuronal morphology, axonal growth and formation of growth cones, centrosome cohesion, actin- and microtubular dynamics [9].

Dynamin 2 is a 100 kDa multidomain protein composed of a highly conserved catalytic N-terminal GTPase domain (GTPase), a middle domain (MD) driving dynamin oligomerization, a pleckstrin homology domain (PH) critical for the interaction with membrane phosphoinositides, a GTPase effector domain (GED) that activates GTPase upon assembly of dynamin oligomers into higher order structures, and a less conserved C-terminal proline/arginine rich domain (PRD) which directs dynamin to endocytic sites and is a major site for interacting with other proteins [10, 11] (Fig. 1).

Distinct mutations in dynamin 2 have previously been associated with other phenotypes including two forms of Charcot-Marie-Tooth disease: axonal CMT2M (MIM# 606482) originally reported by Züchner et al. [12] and intermediate form CMTDIB (MIM# 606482) reported by Kennerson et al. [13]; centronuclear myopathy ADCNM (MIM# 160150) originally reported by Bitoun et al. [14], and lethal congenital contractures syndrome type 5 LCCS5 (MIM#615368). The majority of CMT-associated DNM2 mutations is located in the N-terminal part of the pleckstrin homology domain (Fig. 1), while a distinct set of mutations, mainly those found in the middle, PH, and PH/GTPase-effector boundary, is linked to autosomal-dominant centronuclear myopathy. The mechanisms by which various DNM2 mutations affect discrete tissues and lead to distinct phenotypes remain unknown.

The Siberian family under study had a history compatible with autosomal dominant transmission of a young-adult onset slowly progressive bilateral leg spasticity corresponding to typical clinical manifestations of HSP: each studied patient had lower extremity hyperreflexia, muscle weakness, and spastic gait. Some patients also showed wasting of lower leg muscles, pes cavus, decreased vibratory sense in the ankles, and urinary urgency as late features of illness. There was no optic neuropathy, retinopathy, extrapyramidal symptoms, dementia, ataxia, ichthyosis, or deafness described as part of the “complicated” subtypes of HSP. Alternative diagnoses were excluded based on the results of clinical studies and routine laboratory investigations. Nerve conduction studies in patients with advanced disease revealed motor and sensory axonal neuropathy in distal lower extremities, a feature reported in other subtypes of HSP [31].

Exome sequencing methodology was used for genetic analysis of this HSP family. There are well known limitations to exome sequencing as a method of detection of the causative mutation. Structural genomic alterations, changes in regions not captured by this technology such as translocations, changes in regulatory regions or introns, may be responsible for the clinical phenotypes. But subsequent analysis of tissue expression, structural and functional characteristics of the candidate genetic variants, and segregation within the pedigree as applied in this study are helpful in the process of detection of the disease-associated variant.

The identification of the DNM2 p.R719W mutation as the cause of HSP in this family brought attention to the fact that mutations in this same gene are responsible for other phenotypes, including two forms of CMT peripheral neuropathy. DNM2-associated autosomal dominant CMT2 neuropathy is characterized by progressive distal muscle weakness and atrophy, predominantly in the lower extremities, decreased or absent tendon reflexes, steppage gait, foot drop, pes cavus [32, 33]. These clinical features are in contrast with the presentation observed in patients with DNM2-associated HSP described here, although some overlap with CMT2 neuropathy exists, especially late in the disease. There has been active discussion on the subject of similarity between some forms of CMT neuropathy and HSP based on the identification of shared mechanisms involving central and peripheral motor neurons (for review, see [5]). Interestingly, mutations in KIF5A, a gene encoding the neuronal kinesin heavy chain implicated in anterograde axonal transport, have been associated with “pure” HSP, although a fraction of family members shows electrophysiological evidence of sensory-motor peripheral axonal neuropathy [34, 35]. Our findings give some weight to the view that HSP and CMT caused by DNM2 mutations, although clinically and anatomically separate syndromes, may have some overlap due to shared pathological vulnerabilities in their extended axonal processes.

Another DNM2-associated disease, autosomal-dominant Centronuclear myopathy, is a slowly progressive congenital myopathy characterized by delayed motor milestones, generalized muscle weakness, ptosis, and ophthalmoplegia [14]. It includes a wide spectrum of phenotypes ranging from severe neonatal to mild late-onset familial forms [36, 37]. CNM is characterized pathologically by myofiber atrophy, abnormal nuclear centralization and internalization, and type 1 muscle fiber predominance [38]. Some patients with CNM show clinical overlap of myopathy with mild axonal involvement in peripheral nerves [39].

It remains to be determined why specific DNM2 mutations result in involvement of skeletal muscle, peripheral nerves, peripheral or corticospinal neurons. Existing hypotheses concentrate on different roles of dynamin 2 domains involved in CMT or CNM. CMT mutations are located within the pleckstrin homology [12] and proline-rich domains [33], whereas the CNM mutations are found within the PH domain. The significance of involvement of specific dynamin domains is further underlined by the evidence we obtained in this study that the HSP dynamin 2 mutation, p.R719W, is the only known mutation in the bundle-signaling element of dynamin, which is structurally and most likely functionally different from the stalk and PH domains in which CMT and CNM mutations are located. A mutation in this region may cause a conformational change to the helical element and affect dynamin assembly. Moreover, the mutation results in fewer hydrogen bonds between the bundle-signaling element and the stalk. A weaker connection potentially uncouples the power stroke of dynamin from the rest of the molecule, a step that is predicted to be essential for endocytosis [30].

Modeling in cell cultures of all known DNM2 mutations causing either disease show their destructive effect on clathrin-mediated endocytosis considered a major mechanism for uptaking molecules into eukaryotic cells regulated by dynamin [40, 41]. Clathrin-coated pits capture cargo molecules via adaptors, the pits invaginate and pinch off to form clathrin-coated vesicles that carry the cargo into the cell. Data presented here show that the HSP-causing dynamin 2 p.R719W mutation induces a marked decrease of clathrin-mediated endocytosis similarly to other disease-causing dynamin 2 mutations, suggesting that impairment of the membrane trafficking process contributes to the pathogenesis of all DNM2-associated diseases. Prominent punctate staining in mutant cells also suggests that the HSP DNM2 mutation affects cellular morphology by mechanisms similar to previously characterized CMT and CNM mutations [12, 28].

The superfamily of dynamin GTPases comprises a group of proteins that are involved in diverse cellular functions and are often closely associated with biological membranes; they function in vesicle scission, as well as in fusion and fission of organelles [9]. The large GTPase atlastin belongs to the dynamin superfamily and shares the domain architecture with dynamin 2. Atlastin is involved in ER fusion, vesicular trafficking, axon formation and elongation. Mutations in atlastin isoform 1 (atlastin 1) have been identified in patients with early-onset HSP (SPG3). Curiously, mutations in atlastin 1 have also been associated with hereditary sensory neuropathy type 1D (HSN1D; MIM# 613708), a condition characterized by distal axonal sensory deficits leading to late distal skin ulceration and amputation, with some patients showing upper motor involvement [42]. The majority of atlastin 1 mutations are located in the GTPase domain, although some have been identified in the stalk, the BSE, as well as in the transmembrane domains [43] suggesting that membrane association that is common to both representatives of the dynamin family, dynamin 2 and atlastin 1, may be critical in the pathophysiology of HSP.

Lack of DNM2 mutations in other Siberian HSP families suggests that more than a single gene is responsible for this phenotype; HSP is an extremely heterogeneous disorder with mutations in more than 60 genes associated with this syndrome.

Patients in the 4-generation Siberian family suffered from hereditary spastic paraplegia and mild peripheral axonopathy. Our molecular, structural and functional analyses and segregation studies allowed to identify DNM2 c.2155C > T, p.R719W mutation as a cause of this disease. The mutation is located in a unique position of the multidomain dynamin 2 protein and its pathogenic effects may depend on structural relationships this mutation disrupts. Awareness of the distinct, but partly overlapping clinical phenotypes associated with mutations in DNM2, will facilitate identification of these disorders in additional families and direct future research toward better understanding of the role dynamin 2 and related networks have in health and disease.