HSPB1

HSPB1 codes for HSP27, one of 10 small heat-shock proteins (sHSP) (figure 2). Mutations in the HSPB1 gene were first reported in 2004 in Russian, English, Belgian, Croatian and Austrian families with autosomal dominant CMT2 and dHMN.14 Fifteen different autosomal dominant mutations have now been reported in families with both CMT2 and dHMN.2 14–23 The majority of these mutations lie in the α-crystallin domain, a highly conserved region implicated in oligomer formation. An autosomal recessive mutation in HSPB1 has been reported with a similar age of onset and clinical phenotype.17 The average age of onset is 30 years but can be as late as 6319 and as young as 4 years of age.22 The phenotype is characterised by progressive distal weakness in the legs with mild or subclinical sensory involvement. The upper limbs are involved with time, as would be expected with a length-dependent neuropathy. The rate of progression is slow with a minority of patients requiring a wheelchair in the seventh decade.17 In a Korean family of 12 patients, MRI of the lower legs revealed an unusual pattern of fatty atrophy whereby the anterolateral compartments were relatively spared. This correlates with our own experience whereby patients display disproportionate weakness of the ankle plantar flexors. Brisk reflexes are often noted, and spastic paraplegia and mild cerebellar signs have also been reported in patients with dHMN owing to HSPB1 mutations.23 24

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Photograph of a patient with a mutation in heat-shock protein B1 demonstrating distal wasting of (A) the anterior and (B) the posterior aspects of the legs.

Mutations in the C-terminal domain of HSP27 seem to confer a more severe phenotype with ages of onset as young as 4 and 7.20

The small heat-shock proteins are a group of ubiquitously expressed, stress-induced chaperone proteins classified according to their molecular weight. Unlike HSP70 and HSP90, they lack an ATPase and are unable to independently refold misfolded proteins. Instead, through the formation of oligomers, they are able to maintain misfolded proteins in a refolding competent state.

Much of the research into the pathomechanism of mutant HSP27 has focused on aggregate formation and its interaction with neurofilament light (NFL) and medium chains (NFM). Expression of the p.Ser135Phe HSP27 mutant and NFL in cell lines results in the formation of NFL/HSP27 aggregates,14 25 and expression of the p.Phe182Leu mutant in primary mouse cortical neurons disrupts the incorporation of NFM into the cytoskeletal network.26 Transgenic mice expressing either the p.Ser135Phe or the p.Pro182Leu mutations develop a distal motor neuropathy associated with a reduction in acetylated α-tubulin and impaired axonal transport of mitochondria. Treatment with a histone deacetylase 6 (HDAC6) inhibitor increased the proportion of acetylated α-tubulin and reversed the phenotype suggesting a role for HSP27 in microtubule dynamics.27 Whether the aggregation of neurofilaments seen in previous experiments is simply a result of protein overexpression remains to be seen.

HSPB8

Mutations in HSPB8 were first identified as a cause of dHMN in four separate families from the Czech Republic, Belgium, England and Bulgaria,28 and subsequently in a Chinese family with CMT2.29 Only two missense mutations, both at position Lys 141 within the α-crystallin domain, have been reported that correspond to the Arg116 and 120 substitutions in αA and αβ-crystallins that cause autosomal-dominant congenital cataracts and desmin-related myopathy.30 The phenotype of dHMN/CMT2 owing to HSPB8 mutations is indistinguishable from that owing to mutations in HSPB1.31 The screening of large cohorts of patients with dHMN and CMT2 has failed to identify new HSPB8 mutation carriers highlighting their low prevalence in these populations.17 29

Like HSPB1, HSPB8 is a member of the small heat-shock family of proteins and codes for the protein HSP22.32 It is ubiquitously expressed and possesses intrinsic chaperone and apoptotic regulatory activity.30 It has been proposed that mutant HSP22 results in disease by sequestering proteins with essential housekeeping functions. In support of this, immunoprecipitation studies reveal that mutant HSP22 pulls down HSP27 and in greater quantities than wild type.28 In addition, expression of mutant HSP22 in simian fibroblast cell lines leads to the formation of aggregates containing HSP27 and increased cell death.33 The interaction with HSP27 appears to be particularly strong in comparison with the other small heat-shock proteins, HSP20 and αβ crystallin.34 The formation of protein aggregates in primary neuronal cells expressing mutant HSP22 appears to be specific to motor neurons, suggesting that the sequestration of proteins in vitro may be representative of events in vivo.35

It has also been proposed that mutant HSP22 results in disease through loss of its normal chaperone function. In support of this, overexpression of mutant HSP22 leads to an increase in the accumulation of mutant Huntington protein (Htt43Q) and ALS-causing G93A-SOD1 oligomers in immortal cell lines compared with wild-type HSP22.36 37 Similarly, Drosophila expressing mutant HSP22 were significantly less effective than wild type at reducing aggregation of mutated ataxin 3 and mutant HSP27 protein.38 As part of its chaperoning function, HSP22 directs misfolded proteins to the autophagosome, inserting them into a multiheteromeric complex containing BAG3, Hsc70 and CHIP.39 The significance of this complex in neuropathy is illustrated by the discovery of mutations in BAG3 as a cause of aggressive, childhood-onset myopathy and peripheral neuropathy.40

Finally, the RNA helicase Ddx20, a component of the survival motor neuron (SMN) complex, was identified as an interacting partner of WT HSP22.41 Ddx20 binds to the SMN protein, which is mutated in SMA, linking HSP22 to potential deficits in RNA metabolism.

Despite advances in our understanding of the interactions of mutant HSP22 and HSP27, the pathological events by which the abnormally interacting mutants lead to motor neuron death remain unclear.

HSPB3

Mutations in HSPB3 were discovered in two sisters with a length-dependent motor neuropathy using a candidate gene approach. HSPB3 codes for the small heat-shock protein HSPL27 that has a divergent protein structure compared with HSP27 and HSP22.42 Unfortunately, DNA was not examined from the affected mother, and functional studies are awaited in order to confirm the pathogenicity of the mutation.43

GARS

Glycyl-tRNA synthetase (GARS) is one of 37 nuclear encoded amino acyl tRNA synthetases that function to attach amino acids onto their respective tRNA for protein translation.44 Since the discovery of GARS as a cause of dHMNV and CMT2D, mutations in three other amino acyl tRNAs have been identified as causes of intermediate CMT (DI-CMTC) (YARS, tyrosyl-tRNA synthetase),45 CMT2N (AARS, alanyl-tRNA synthetase)46 and autosomal recessive CMT2 (RI-CMTB) (KARS, lysyl-tRNA synthetase).47

In 2003, four different GARS mutations were discovered in five families with upper-limb predominant distal motor neuropathy (dHMNV or CMT2D).10 Six additional familial and sporadic cases of dHMN owing to GARS mutations have been reported.48–51 In an extended phenotype/genotype study of the original five families, 75% of affected family members presented in the second decade of life, with the majority functioning independently 40 years after disease onset.52 In one patient, weakness began in the lower limbs, demonstrating that GARS mutations can present identically to dHMN owing to mutations in HSPB1 and HSPB8.53 In all other affected individuals, the disease presented in the hands. Reduced penetrance was reported in two carriers who were asymptomatic at 49 years of age. Conversely, a congenital form of dHMN owing to a GARS mutation was reported in a boy who developed floppy feet at 6 months of age and required a walking frame at 18 months.50

Screening for GARS mutations in selected patients with both upper-limb and lower-limb dHMN reveals a detection rate of approximately 1 in 100.2 48 51

GARS is a ubiquitously expressed protein that is essential for translation in all cells. Why mutations in this protein should result in a motor-neuron-specific phenotype is unclear. However, in vitro cell-based and murine models suggest that GARS and also YARS mutations result in disease through a toxic gain of function.54

Initial studies on GARS mutations attempted to assess whether the mutations were associated with a reduction in the ability of GARS to charge tRNA molecules with glycine. In vitro studies using capture of tritiated glycine as a measure of amino-acylation and radioactive inorganic phosphate as a marker for charging activity did not demonstrate a reduction in activity across all the mutations tested.55 56 Mice heterozygous for the hypomorphic GARS allele XM256 have a normal lifespan and no neuromuscular junction pathology, despite having a 50% reduction in GARS mRNA levels.57 These experiments would suggest that GARS mutations do not result in disease through loss of amino-acylation. It is not clear, however, whether GARS mutations have a neuron-specific effect on glycine charging.

Immunohistochemistry of normal human tissue reveals GARS containing puncta in spinal cord, nerve root and peripheral nerve sections suggesting that the GARS protein may be transported to the distal axon.58 It has been hypothesised that axonal transport of mutant GARS may be impaired, leading to distal axonal degeneration.54 In support of this hypothesis, transfection of mutant GARS failed to localise to sprouting neurities in N2a cells in culture.55

The Nmf249 GARS mouse model, a spontaneous CC-to-AAATA mutation resulting in an in-frame change (equivalent to the p.Pro234LysTyr in the human protein), results in a severe phenotype with sensory and motor deficits and a reduced lifespan of 6–8 weeks.57 A second mutant GARS mouse, C201R (equivalent to human p.Cys157Arg) has a less severe phenotype with a 50% reduction in grip strength by 1 month of age.59 Both mouse models demonstrate normal amino-acylation activity in the heterozygote state. Compound heterozygotes for both mutations with the hypomorphic allele GarsXM256 are not viable. This is unexpected, given that both mutants retain normal amino-acylation activity, and may imply a non-canonical function of GARS protein affected by disease causing mutations.54 Non-canonical functions have been described in other amino-acyl tRNAs including the role of KARS as a transcription factor.60

The pathomechanisms of mutant GARS remain enigmatic. Impaired axonal transport of GARS, mitochondrial dysfunction (GARS has a mitochondrial isoform) and loss of a non-canonical function have been suggested as the most likely mechanisms.54

BSCL2

The Berardinelli–Seip Congenital Lipodystrophy type 2 (BSCL2) gene was first identified as a cause of Congenital Generalised Lipodystrophy Type 2.61 This is a rare autosomal recessive condition characterised by the absence of functional adipocytes. In 2004, Windpassinger and colleagues identified two autosomal dominant mutations (p.Asn88Ser,p. Ser90Leu) in BSCL2 in a series of new and previously reported families with upper-limb onset dHMN,62 Silver syndrome63 64 and hereditary spastic paraplegia (HSP).11 The same two mutations have since been reported in several geographically diverse and unrelated families.65–72

A detailed phenotype study from 14 unrelated Austrian families with the p.Asn88Ser mutation divided the clinical phenotype into six subgroups; (1) asymptomatic carriers (4.4%); (2) mild disease—for example, wasting of the thenar muscles (20%); (3) exclusive or predominant wasting of the hands (dHMNV) (31%); (4) classical Silver syndrome (14%); (5) classical length-dependent dHMN beginning in the legs (dHMN II) (20%); (6) classical HSP (10.6%). Sensory symptoms were absent in most carriers, although median sensory nerve action potentials were significantly reduced compared with controls.65

The BSCL2 gene encodes the endoplasmic reticulum (ER) resident glyco protein, Seipin.73 Much of the work on the molecular pathogenesis of BSCL2 mutations in dHMN has focused on a toxic gain of function resulting from activation of the unfolded protein response (UPR).73 The UPR is a highly conserved intracellular signalling pathway initiated in response to ER stress arising from an increase in protein misfolding.74 75 The UPR enables the cell to reduce the unfolded protein load in the ER by halting further protein translation and promoting protein refolding or degradation. If these measures fail to relieve ER stress, the UPR initiates pro-apoptotic pathways.

Both the p.Asn88Ser and p.Ser90Leu missense mutations in BSCL2 disrupt N-glycosylation sites,11 important for the linkage of N-linked oligosaccharides to seipen protein and its correct folding. Expression of the p.Asn88Ser and p.Ser90Leu mutant proteins in vitro leads to the formation of misfolded protein that localises in the ER.11 73 76 Cells expressing the two mutant proteins upregulated BIP and CHOP (markers of the UPR) and resulted in increased apoptotic cell death.73 BIP expression was also increased in the cerebral cortex of transgenic mice overexpressing the human p.Asn88Ser mutant protein.77 Taken together, these findings suggest that mutant seipin leads to a motor neuropathy as a result of its incorrect folding and activation of the UPR.

IGHMBP2

Spinal muscular atrophy with respiratory distress type 1 (SMARD1), also classified as dHMN type VI, is an autosomal recessive condition characterised by severe length-dependent weakness with diaphragmatic palsy beginning before the age of 13 months.78 The prominent distal weakness differentiates it from other forms of SMA. Homozygous and compound heterozygous mutations in the gene immunoglobulin μ Binding Protein 2 (IGHMBP2) were identified as a cause of SMARD1 and are also responsible for the phenotype of the NMD mouse, which develops limb weakness and respiratory insufficiency.79 The prognosis is poor, although there have been case reports of patients alive at 12 and 20 years of age.80 81 In a series of 13 patients with infantile onset respiratory distress, two were heterozygous and 11 homozygous or compound heterozygous for mutations in IGHMBP2.78 Sural nerve biopsies demonstrated a reduction in the number of large diameter axons but with no evidence of demyelination or axonal degeneration. Unusually, motor-nerve conduction velocities were slow, which, in view of the pathological findings, was postulated to be due to the failure to develop large-diameter fast-conducting axons.78

Based on sequence homology, IGHMBP2 has been classified as a member of the UPF1-like group within the helicase superfamily 1.82 In a series of in vitro experiments using purified IGHMBP2, Guenther et al demonstrated that IGHMBP2 is an ATP dependent 5′–3′ helicase and unwinds RNA and DNA duplexes.82 The majority of mutations in IGHMBP2 lie within or in close proximity to the helicase domain, and all bar one were found to impair helicase activity, implying that reduced protein translation may be a final common pathway in the pathogenesis of SMARD1.82

DCTN1

Heterozygous point mutations in the p150 subunit of dynactin 1 (DCTN1) are often listed as a cause of dHMN. Only one family has been reported with a lower motor-neuron-disease phenotype.13 The disease presented in early adulthood with vocal-cord paralysis, progressive facial and arm weakness, with leg weakness developing later. The clinical picture is therefore not strictly a length-dependent neuropathy, presenting with bulbar features and more suggestive of ALS/SMA.

DCTN1 is required for retrograde axonal transport of vesicles and organelles along microtubules, the binding to which is impaired in mutant constructs.13 Missense mutations in DCTN1 have been reported in several ALS families with an average age of onset of 50 and disease duration of 4–8 years. In one family, two mutation carriers developed ALS, and two developed a fronto temporal dementia.83 84 Unaffected carriers were identified in two families, suggesting either reduced penetrance or non pathogenicity.83 Mutations in DCTN1 have also been identified as a cause of Perry syndrome.85

Hereditary forms of ALS are being discovered at an increasing rate. The atypical forms of dHMN, ALS, SMA and PLS are not mutually exclusive, and there is a degree of overlap. Genes that have been implicated in familial ALS include Cu/Zn superoxide dismutase, TDP-43, Fused in Sarcoma (FUS), Alsin, SETX, Spatacsin, VAPB, Angiogenin, FIG4, optineurin and valosin-containing protein (VCP) mutations.86–88 While mutations in these genes affect peripheral motor neurons, they do not as a general rule lead to a distal motor neuropathy. FIG4 is intriguing, however, as homozygous and compound heterozygous mutations result in CMT4J, a demyelinating sensory and motor neuropathy.

ATP7A

ATP7A is a copper-transporting ATPase mutated in the severe infantile-onset developmental disorder, Menke's disease and its milder variant, occipital horn syndrome.89–92 Menke's disease is an X-linked disorder characterised by infantile onset cerebral and cerebellar neurodegeneration, failure to thrive, kinky hair and connective-tissue abnormalities.92 93

In 2010, mutations in the ATP7A gene were discovered in two previously reported North American and Brazilian families with X-linked distal hereditary motor neuropathy.94 Affected individuals developed an axonal distal motor neuropathy in the lower limbs with minimal sensory symptoms. The age of onset ranged from 2 to 61 (mean 25 years), and serum copper concentrations were within the normal range. The two reported missense mutations have not been reported in Menke's disease.

ATP7A is responsible for the metallation of copper enzymes in the trans-Golgi network and also for preventing toxic copper accumulation.95 In order to provide this dual function, ATP7A is trafficked from the trans-Golgi network to the plasma membrane in response to copper. Transfection of the two recently described ATP7A mutants in cell lines resulted in a greater accumulation of the mutant protein at the plasma membrane compared with wild type, implying impaired endocytic recycling to the Golgi network. This in turn may lead to osmotic imbalance or a deficiency of copper-dependent enzymes leading to neuropathy.94 95

Transport of copper across the blood–brain barrier is mediated by ATP7A and is impaired in Menke's disease leading to hypocupraemia. While copper replacement has been shown to be effective in patients with Menke's disease, there is no evidence to support its use in treating or preventing neuropathy in patients with dHMN owing to ATP7A mutations.96–100 Nevertheless, the response of acquired sensory motor neuropathy owing to copper deficiency to exogenous replacement suggests that this therapeutic avenue is one that should be explored in greater detail.

SETX

Heterozygous mutations in the gene senataxin have been identified as a cause of both dHMN with pyramidal features and juvenile ALS type 4.5 101 In two families from Belgium and Austria, affected individuals developed slowly progressive distal muscle weakness with brisk reflexes and mild sensory symptoms.5 In a large US kindred, distal lower-limb weakness was present in 90%, brisk reflexes in 84% and sensory abnormalities in 10%. Four per cent of carriers developed facial weakness.101 Postmortem studies in two patients revealed loss of motor neurons and the presence of ubiquitin and phosphorylated neurofilament containing spheroids in the lumbar cord. A novel heterozygous mutation was identified in a 42-year-old Chinese gentleman who developed clinical ALS with bulbar involvement and died of respiratory failure 2 years after diagnosis.102 Interestingly, homozygous and compound heterozygous mutations in SETX give rise to Ataxia with Oculomotor Apraxia type 2 (AOA2).103 This condition is characterised by cerebellar atrophy, sensory motor axonal neuropathy and occasional oculomotor apraxia. Three patients with heterozygous SETX mutations presenting with ALS, dHMN and an ALS/ataxia overlap syndrome have recently been reported.104

While the exact function of SETX is unknown, an analysis of the SETX protein sequence using domain-prediction software predicted a super family 1 DNA/RNA helicase at the C-terminal domain.105 This sequence shows 42% homology to the DNA/RNA helicase IGHMBP2 mutated in SMARD1. Immunoprecipitation studies have demonstrated that SETX interacts with pre-mRNA processing factors,106 RNA polymerase II and the RNA processing protein SMN that are essential for the biogenesis of small nuclear RNA (snRNA) ribonucleoproteins and pre-mRNA splicing.107 Knockout of SETX in cell lines using silencing RNA results in the abnormal termination of transcription and splicing of mRNA.106 While a loss of function is likely to explain the pathomechanisms of AOA2, an as-yet unknown toxic gain of function appears more probable for dHMN/ALS 4.

TRPV4

The transient receptor vallanoid 4 gene (TRPV4) encodes a non-selective cation channel that is moderately permeable to calcium ions.108 Dominant mutations in this gene were originally described as causative of the skeletal dysplasia, brachyolmia type 3.109 110 More recently, dominant mutations in TRPV4 have also been linked to congenital distal hereditary motor neuropathy, scapulo peroneal muscular atrophy (SPMA) and CMT2C.111–113

Scapuloperoneal SMA is a syndrome characterised by scapular winging, distal wasting in the lower limbs and vocal-cord paralysis.114 115 CMT2C describes an autosomal dominant, distal, predominantly motor axonal neuropathy with vocal-cord paralysis and diaphragmatic weakness.116–119 Congenital dHMN presents at birth with severe distal motor neuropathy and arthrogryphosis.120 121

In a large case series, Zimon et al identified mutations in seven families with distal weakness in the lower limbs and minimal sensory loss.6 In the majority of carriers, symptoms developed before 14 years of age, and vocal-cord palsy was present in 50%.6 Sensory neuronal hearing loss and urinary incontinence were present in two separate families, features that are also observed in TRPV4 knock-out mice.122 123 Facial asymmetry, tongue fasciculations and third and sixth cranial nerve palsies have been reported in affected carriers.124 125 Within families, there is often marked phenotypic heterogeneity. In one reported family, a daughter was wheelchair-bound by 7 years of age, while an unaffected 70-year-old relative was found to carry the same TRPV4 mutation.126

The TRPV4 protein has several distinct domains including a six repeat N-terminal ankyrin domain, a central ion pore and a C-terminal intracellular domain. Mutations in the first ankyrin domain have been shown to abolish binding of PACSIN3, a protein that strongly inhibits the basal activity of TRPV4.127

Mutations in TRPV4 are thought to result in skeletal dysplasias owing to increased calcium influx.109 In vitro studies of neuropathy causing mutations have been less conclusive, with three studies demonstrating increased calcium influx,112 113 125 and one reduced calcium influx.111 The discrepancy in these findings has been postulated to be due to the use of different non-neuronal cell lines (Hek293, HeLa) and experimental conditions.128 Using the neuronal-derived cell line, Neuro2a, Fecto et al demonstrated increased intracellular calcium, cytotoxicity and channel open time in TRPV4 mutant transfected cells suggesting that a toxic gain of function appears the most likely pathomechanism.129

Selective TRPV4 antagonists have been developed and provide a potential therapeutic strategy.125 While TRPV4 mutants appear to be associated with increased calcium influx, there is scepticism that this underlies the pathophysiology in view of the complex and diverse phenotype of such mutations.128 Phenotypic similarities with TRPV4 knockout mice suggest that the use of an antagonist should be approached with caution.

DYNC1H1

A heterozygous mutation in the cytoplasmic dynein heavy chain 1 (DYNC1H1) gene was recently identified in a family with a distal motor neuropathy using whole exome sequencing. The age of onset was in early childhood, and in a proportion of affected family members, proximal weakness was prominent. The disease was classified as CMT2, although several affected individuals had normal sensory action potentials and therefore could be considered to have dHMN.130

Dynein is a retrograde motor protein, and the reported mutation, p.His306Arg lies within the homodimerisation domain of cytoplasmic dynein heavy chain 1. A missense mutation in DYNC1H1 in the Loa mouse leads to axonal transport deficits in vivo.131

PLEKHG5

Homozygous mutations in the Pleckstrin homology domain-containing, family G member 5 gene (PLEKHG5) are often classified as a cause of dHMN.132 The mutation has only been described in one Malian family with a motor neuropathy characterised by proximal muscle weakness beginning by 3 years of age.133 The phenotype is therefore more analogous to SMA than dHMN and should be considered alongside other hereditary motor disorders with proximal involvement including those owing to mutations in vesicle-associated membrane protein B (VAPB).134

Other dHMN loci

Five other separate genetic loci have been reported for both autosomal dominant and recessive forms of dHMN (see table 1). Linkage to chromosome 11q13 has been reported in a large, consanguineous Lebanese family with a chronic form of dHMN. The locus is associated with both dHMN III and IV owing to the variable presence of diaphragmatic palsy. The age of onset varied from 6 months to 43 years of age. Respiratory involvement was seen in patients with a younger age of onset and may simply be a reflection of the severity of the disease.135

An autosomal recessive form of dHMN has been linked to chromosome 9p21.1–p12 in seven consanguineous families from the Jerash region of Jordan (dHMN-Jerash).7 The age of onset ranged from 6 to 10 years. Distal atrophy and weakness presented in the lower limbs and progressed to the upper limbs after 1 year. Reflexes were often brisk, and the Babinski sign positive.

Linkage to chromosome 2q14.2 has been reported in two large families with dHMN VII. Affected carriers developed vocal-cord paralysis and weakness in the lower limbs by 10 years of age.12

Autosomal dominant dHMN with pyramidal signs has been mapped to chromosomes 4q34.3–q35.2 and 7q34–q36 in Italian and Australian families respectively.8 9 The age of onset was slightly later (25–40 years) in the Italian compared with the Australian family (mean 10 years). Central motor conduction times were found to be prolonged in six patients with dHMN localised to 7q34–36, although cortical excitability studies were normal.136