The Hereditary Spastic Paraplegias: Nine Genes and Counting

The hereditary spastic paraplegias (HSPs) are inherited neurologic disorders in which the primary symptom is insidiously progressive difficulty walking due to lower extremity weakness and spasticity. There have been great strides in our knowledge of this group of disabling disorders; 20 HSP loci and 9 HSP genes have been discovered. Insights into the molecular causes of HSPs are beginning to emerge. This review summarizes these advances in HSPs' genetics.

Hereditary spastic paraplegia (HSP) is a syndromic designation encompassing more than 30 disorders in which the predominant feature is spastic gait.¹ The HSP syndromes are classified clinically as uncomplicated (also known as pure or nonsyndromic) if symptoms are limited to progressive spastic weakness in the legs, often accompanied by urinary urgency and subtle dorsal column impairment. The HSP syndromes are classified as complicated or syndromic if the inherited disorder includes other neurologic abnormalities (eg, neuropathy, amyotrophy, ataxia, mental retardation, or thin corpus callosum) or systemic disturbances (eg, cataracts) for which alternate disorders have been excluded. Postmortem studies of uncomplicated HSP have shown axonal degeneration that is most marked in the distal terminals of corticospinal tracts and fasciculus gracilis fibers.²

Symptoms of HSP may begin at any age, from infancy to older than 60 years. If symptoms begin during the teenage years or later, then spastic gait disturbance usually progresses insidiously over many years. Canes, walkers, and wheelchairs may eventually be required. If symptoms begin in late infancy or early childhood, however, then there may be relatively little functional worsening, even over many years. Individuals with early-onset, apparently nonprogressive HSP may be distinguishable from those with spastic diplegic cerebral palsy only by family history (which may be absent in recessive, X-linked, or dominantly inherited HSP with variable age of symptom onset).

Autosomal dominant, autosomal recessive, and X-linked HSP are each genetically heterogeneous. Genetic mapping has identified 20 different HSP loci (Table 1) designated SPG (SPastic parapleGia) 1 through 21 in order of their discovery (SPG18 has been identified but not published). Thus far, 10 loci for autosomal dominant HSP, 7 loci for autosomal recessive HSP, and 3 loci for X-linked HSP have been published. The occurrence of HSP in families for which known HSP loci are excluded indicates the existence of additional loci (J.K.F. and S. Rainier, PhD, unpublished observations, 2003).

Different genetic types of uncomplicated HSP usually cannot be distinguished by clinical and neuroimaging parameters alone. This reflects both the clinical similarity between different types of HSP and the phenotypic variability within a given genetic type of HSP. There may be significant clinical variability both within a given family in which all subjects have the same HSP gene mutation; between families with the same genetic type of HSP; and between families with different genetic types of HSP. For example, some families with SPG4 HSP (due to spastin gene mutations described in the next section) include individuals with child hood-onset symptoms and individuals whose symptoms begin after age 30 years.

Mutations in the SPG4 gene (spastin protein) are responsible for approximately 40% of autosomal dominant HSP cases.⁴ Hereditary spastic paraplegia due to SPG4 gene mutation is the single most common form of autosomal dominant HSP, and possibly the single most common form of any type of HSP. At this stage, we know very little about the function of spastin, the encoded protein of SPG4, and the mechanisms by which spastin abnormalities lead to axonal degeneration in HSP. It is widely held that most, if not all, SPG4 mutations are pathogenic because of haploinsufficiency (ie, decreased abundance of functionally normal spastin) rather than a dominant negative mechanism.^5-10 This conclusion is based on (1) the fact that many mutations reduce the abundance of full-length, sequence-normal spastin, such as mutations that cause premature translation termination, insertions, or deletions that lead to nonsense transcripts and mutations that cause aberrant messenger RNA splicing, and (2) recent studies in which expression of SPG4 bearing nonsense or frameshift mutations resulted in the absence of immunologically detectable spastin.¹¹

SPG4 is expressed widely and undergoes alternate splicing variable inclusion of exon 4. Thus far, none of the more than 80 reported HSP-specific SPG4 mutations have occurred in exon 4. In addition to variable splicing, there is also evidence that tissue-specific posttranslational modification results in 75- and 80-kDa spastin isoforms.¹¹ Charvin et al¹¹ showed spastin was present in neurons, particularly in the brain, and also in anterior horn motor neurons, but not within glia. Spastin contains an ATPase associated with diverse cellular activities (AAA) domain.⁵ It also contains a nuclear localization signal,⁵ and very recent immunofluorescent studies indicate spastin is a nuclear protein. The intracellular location of spastin is controversial because previous studies indicated that spastin was distributed within the cytoplasm.¹²

Emerging evidence suggests that spastin may interact with microtubules. Azim et al¹³ showed that antitubulin antibodies could precipitate spastin in vitro. More recently, Errico et al¹¹ showed that a spastin fusion protein colocalized with microtubules in Cos-7 and HeLa cells transfected with wild-type and mutant SPG4 expression vectors. Findings that spastin may interact directly with microtubules support the hypothesis that disturbances in axonal cytoskeleton or transport underlie some forms of HSP.¹⁴

Autosomal dominant HSP linked to the chromosome 14q SPG3A locus represents approximately 10% of dominantly inherited HSP cases¹⁵ and is particularly prevalent among those autosomal dominant HSP kindreds in which each affected subject developed symptoms in childhood.¹⁶ Zhao et al¹⁷ identified mutations in a novel gene (SPG3A) as the cause of this form of HSP. At this stage, insight into the possible function of SPG3A comes only from analysis of the sequence of its encoded protein, atlastin. Atlastin does not contain an AAA motif and is not homologous to spastin or other proteins implicated in HSP. In contrast, atlastin contains conserved motifs for GTPase binding and hydrolysis¹⁷ and is structurally homologous to guanylate binding protein 1. The functional importance of atlastin's GTPase motif is indicated by a recently identified HSP mutation that disrupted a conserved GTPase domain.¹⁸

Guanylate binding protein 1, to which atlastin shows homology, is a member of the dynamin family of large GTPases. Dynamins play essential roles in a wide variety of vesicle trafficking events.^19-24 Dynamins play essential roles in the action of many neurotrophic factors and are critical elements in the rapid and efficient process of recycling of synaptic vesicles. Dynamins are also involved in the maintenance and distribution of mitochondria²⁵ and, through their association with actin and microtubules, have been implicated in maintenance of the cytoskeleton.²⁶ The important and diverse functions of dynamins raise many interesting possibilities by which atlastin mutations could cause axonal degeneration. These possibilities include defective synaptic vesicle recycling leading to abnormal synaptic structure and impaired neurotransmission, impaired activation of selected neurotrophic factors, and impaired mitochondrial distribution.

Kinesin heavy chain (KIF5A) is a molecular motor that participates in the intracellular movement of organelles and macromolecules along microtubules in both anterograde and retrograde directions. KIF5A gene mutation was recently identified in affected subjects with HSP linked to the SPG10 locus.²⁷ Subjects with KIF5A mutation exhibited either uncomplicated HSP or HSP associated with distal muscle atrophy.²⁷ The HSP-specific KIF5A mutation disrupted an invariant asparagine residue that, when mutated in orthologous kinesin heavy chain motor proteins, prevented stimulation of the motor ATPase by microtubule binding.²⁷ Finding KIF5A mutations in SPG10 HSP suggests that degeneration of distal axons in this and possibly of other forms of HSP may be related to disturbance of axonal transport.²⁸

Mutation in the mitochondrial protein heat shock protein 60, also known as chaperonin 60, causes SPG13-linked autosomal dominant HSP. SPG13 HSP is an autosomal dominant form of uncomplicated HSP mapped to chromosome 2q24-34. Recently, Hansen et al²⁹ identified a mutation in heat shock protein 60 in affected subjects from an SPG13-linked HSP kindred. The mechanism by which heat shock protein 60 mutations cause HSP are not yet known. It is intriguing however, that 2 HSP genes, heat shot protein 60 or chaperonin 60 and SPG7 or paraplegin, encode mitochondrial proteins.

NIPA1 gene mutations cause autosomal dominant HSP linked to chromosome 15q (SPG6 HSP). SPG6 HSP is a prototypical example of adolescent or adult-onset, slowly progressive, uncomplicated HSP. SPG6 and SPG8 are perhaps the most severe forms of dominantly inherited, uncomplicated HSP.³⁰ Chai et al³¹ identified nonimprinted Prader-Willi/Angleman (NIPA) locus genes as candidate's for SPG6. Rainier et al³² identified a disease-specific mutation in a novel gene (NIPA1) in an SPG6-linked HSP kindred and in an unrelated kindred that was too small for linkage analysis. Precisely the same NIPA1 gene mutation (T45R) was discovered in 2 unrelated kindreds.³² The function of NIPA1 is unknown. It is widely expressed, particularly in the central nervous system. The presence of 9 alternating hydrophobic-hydrophilic domains suggests that NIPA1 encodes a membrane protein.³² This feature makes NIPA1 unique among HSP proteins. The NIPA1 mutation T45R appears to act through a "dominant negative" gain of function. This prediction is based on the observation that subjects who are missing one NIPA1 gene entirely (eg, subjects with Prader-Willi and Angleman syndromes who have deletions involving this region of chromosome 15q) do not develop HSP.

SPG7 encodes a mitochondrial protein (paraplegin). De Michele et al³³ discovered disease-specific mutations in SPG7 as the cause of chromosome 16q–linked autosomal recessive HSP. This is a rare form of autosomal recessive HSP with only a few reported families affected.^33,34 Within these families, some individuals had pure HSP, and others had HSP complicated by dysarthria, dysphagia, optic disc pallor, axonal neuropathy, and evidence of vascular lesions, cerebellar atrophy, or cerebral atrophy on cranial magnetic resonance imaging. Paraplegin is highly homologous to the yeast mitochondrial ATPases AFG3, RCA1, and YME1, which have both proteolytic and chaperone-like activities at the inner mitochondrial membrane. Paraplegin is also localized to mitochondria.³³ Muscle biopsy specimens from some, but not all, HSP patients with SPG7 gene mutations showed ragged-red and cytochrome-oxidase–negative fibers and abnormal mitochondrial structure typical of mitochondriopathy.³³

SPG7 knockout mice exhibit signs of HSP. Ferreirinha et al³⁵ reported preliminary studies in homozygous SPG7/paraplegin knockout mice. These animals displayed impaired performance on a rotorod apparatus that started at age 6 months and worsened with age. Histological analysis of the spinal cord showed axonal swelling, particularly in the lateral columns of the lumbar spinal cord, consistent with a retrograde axonopathy. The changes were progressive with signs of axonal degeneration becoming prominent at age 12 months.

Although 2 HSP genes (SPG7/paraplegin and SPG13/chaperonin 60) are mitochondrial proteins, it does not appear that all types of HSP are due to mitochondrial dysfunction. Biochemical and histological evidence of mitochondrial abnormalities has been sought but has not been observed in SPG4 (spastin), SPG3A (atlastin), SPG8, and SPG6 autosomal dominant HSP.^5,36-38

Spartin mutations cause HSP associated with distal muscle wasting (SPG20 Troyer syndrome).³⁹ This gene is designated "spartin" (spastin-related autosomal recessive Troyer protein) because its amino-terminal region is similar to that of spastin, mutations in which cause SPG4 HSP. Spartin is also homologous to proteins involved in endosome morphology and membrane trafficking.

Proteolipid protein (PLP) is an intrinsic myelin protein. PLP gene duplications and point mutations cause Pelizaeus-Merzbacher disease, an infantile-onset, progressive leukodystrophy, and are responsible for dysmyelination in jimpy mice.^40,41 In addition, some individuals with X-linked HSP linked to the SPG2 locus on Xq22 also have PLP mutations.^42,43 Although some of these patients have clinical features consistent with uncomplicated HSP, others have abnormal-appearing white matter on brain magnetic resonance imaging scans or evidence of peripheral neuropathy. Leukodystrophy is not a feature of other forms of uncomplicated HSP.

Some of the phenotypic variation of PLP mutation syndromes (infantile-onset leukodystrophy vs childhood-onset slowly progressive spastic paraparesis) can be attributed to different mutations in the PLP gene. However, both syndromes have occurred in the same family in individuals who share the same PLP gene mutation (F. Cambi, MD, PhD, J.K.F., unpublished observation, 2001). This observation indicates that, in some cases, the neurologic consequences of PLP gene mutation are influenced by modifying factors that presumably include additional genes and possibly environmental factors.

Neuronal cell adhesion molecule L1 (L1CAM) gene mutations cause a variety of X-linked neurologic disorders, including complicated spastic paraplegia, hydrocephalus, and mental retardation aphasia, shuffling gait, and adducted thumbs (MASA) syndrome.⁴⁴ Although there is correlation between some of these syndromes and specific L1CAM mutations,⁴⁵ X-linked hydrocephalus, MASA syndrome, and X-linked spastic paraplegia have occurred in kindreds in which affected individuals had the same L1CAM mutation.⁴⁶ L1CAM is an integral membrane glycoprotein and a member of the immunoglobulin superfamily of cell adhesion molecules that mediate cell-to-cell attachment. L1CAM is found primarily in the nervous system, and its functions include guidance of neurite outgrowth during development, neuronal cell migration, and neuronal cell survival.^47,48 Dahme et al⁴⁹ used gene targeting to create transgenic mice in which the L1CAM gene was disrupted. These animals exhibited weak hind limbs and reduced size of corticospinal tracts.

The very recent discovery of many HSP genes is rapidly shaping new concepts of the pathophysiologic mechanisms of HSP. Whereas the uniform clinical appearance of uncomplicated HSPs initially suggested that a common biochemical disturbance underlies most types of HSP, this appears to not be the case. Rather, it appears that very long central nervous system axons (ie, corticospinal tracts and dorsal column fibers) are particularly vulnerable to a number of distinct biochemical disturbances and that the highly similar clinical features of genetically diverse types of uncomplicated HSP reflect the limited repertoire of symptoms from corticospinal tract and, to a lesser extent, dorsal column fiber disturbance.

Based on the diversity of the HSP genes discovered, a biochemical classification of HSP is emerging. For example, one can consider (1) HSP due to mitochondrial abnormality (including SPG13 due to chaperonin 60 mutation and SPG7 due to paraplegin mutation); (2) HSP due to axonal transport abnormality (including SPG10 due to kinesin heavy chain mutation and possibly SPG3A due to atlastin mutation and SPG4 due to spastin mutation; (3) HSP due to primary myelin disturbance (SPG2 due to PLP mutation); and (4) HSP due to embryonic development of corticospinal tract neurons (SPG1 due to L1CAM mutation). Whether these disparate primary biochemical disturbances converge into one or more common pathways remains to be determined. Each HSP gene discovery permits direct exploration of the molecular mechanisms that underlie HSP, insights that bring us one step closer to developing real treatment.

Corresponding author and reprints: John K. Fink, MD, 5214 Cancer Geriatrics Center Bldg, Box 0940, 1500 E Medical Center Dr, Ann Arbor, MI 48109 (e-mail: jkfink@umich.edu).

Accepted for publication March 25, 2003.

This study was supported by grants R01NS33645, R01NS36177 and R01NS38713 from the National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Md, and grants from the Veterans Affairs Merit Review, Washington, DC, and the National Organization for Rare Disorders, Danbury, Conn.

I gratefully acknowledge the expert secretarial assistance of Lynette Girbach and the participation of subjects with HSP and their families, without whom our investigations of HSP would not be possible.