Background TRPV4 mutations have been identified in Charcot–Marie–Tooth type 2 (CMT2), scapuloperoneal spinal muscular atrophy and distal hereditary motor neuropathy (dHMN).
Objective We aimed to screen the TRPV4 gene in 422 British patients with inherited neuropathy for potentially pathogenic mutations.
Methods We sequenced TRPV4 coding regions and splice junctions in 271 patients with CMT2 and 151 patients with dHMN. Mutations were clinically and genetically characterised and screened in ≥345 matched controls.
Results 13 missense and nonsense variants were identified, of which five were novel and absent from controls (G20R, E218K, N302Y, Y567X and T701I). N302Y and T701I mutations were present in typical CMT2 cases and are potentially pathogenic based on in silico analyses. G20R was detected in a patient with dHMN and her asymptomatic father and is possibly pathogenic with variable expressivity. The Y567X variant segregated with disease in a family with severe CMT2 but also with a MFN2 mutation reported to cause a mild CMT2 phenotype. Although Y567X caused nonsense mediated mRNA decay, the amount of TRPV4 protein on western blotting of patient lymphoblasts was no different to control. Y567X is therefore unlikely to be pathogenic. E218K is unlikely to be pathogenic based on segregation.
Conclusions In this comprehensive analysis of the TRPV4 gene, we identified mutations in <1% of patients with CMT2/dHMN. We found that TRPV4 likely harbours many missense and nonsense non-pathogenic variants that should be analysed in detail to prove pathogenicity before results are given to patients.
Statistics from Altmetric.com
The hereditary neuropathies are a clinically and genetically heterogeneous group of disorders characterised by degeneration of peripheral nerves leading to motor and/or sensory abnormalities.1 Since early 2010, heterozygous mutations in the vanilloid transient receptor potential cation channel 4 gene (TRPV4) have been identified in families with hereditary motor neuropathies, including congenital distal spinal muscular atrophy (also known as distal hereditary motor neuropathy (dHMN)) and scapuloperoneal spinal muscular atrophy and in families with Charcot–Marie–Tooth type 2 (CMT2).2–7 TRPV4 mutations are also known to cause a range of autosomal dominant skeletal dysplasias.8–12
So far, 14 different TRPV4 mutations have been identified in individuals with neuropathy. Initial studies suggested that mutations causing neuropathy clustered within the ankyrin repeat domain (ARD) and affected arginine residues (R232C, R269C, R269H, R315W, R316C and R316H).2–7 ,13 Interestingly, these six ARD mutations are predicted to lie within finger loops connecting the helices of the ARD. This domain is thought to be involved in protein–protein interactions, including oligomerisation of TRPV4 channels.14 Subsequent studies, however, suggested a more complicated picture. Three mutations previously reported to cause skeletal dysplasias were discovered in patients with both neuropathic and skeletal phenotypes (E278K, V601I and P799R).7 ,15 Furthermore, novel TRPV4 mutations have also been identified in families with an overlap between neuropathy and skeletal dysplasia. S542Y was present in a family with axonal neuropathy and short stature,3 A217S in a patient with skeletal dysplasia and peripheral neuropathy15 and three mutations (G78W, K276E and T740I) in four cases with metatropic dysplasia and a neurological phenotype presenting as fetal akinesia.16
Given the previously reported association of TRPV4 missense mutations with CMT2 and HMN, we screened TRPV4 coding exons and splice junctions for mutations in 279 unrelated CMT2 and 173 unrelated HMN patients from our inherited neuropathy patient cohort. The size of our cohort allowed us to provide an accurate estimate of the frequency of TRPV4 mutations among different disease categories.
We selected patients from our inherited neuropathy cohort with either CMT2 (n=279) or dHMN (n=173) who had not been found to have a mutation in the common genes causing CMT2 and dHMN or where a variant had been identified that did not explain the phenotype. Most patients had been screened for mutations in MFN2, MPZ and, where appropriate, BSCL2, GJB1, HSPB1 and HSPB8. We run an inherited neuropathy clinic as part of our peripheral nerve service in the National Hospital for Neurology and Neurosurgery (NHNN). We also run a diagnostic laboratory for some of the common CMT genes and, in addition, are sent many DNA samples from patients with CMT for research testing of the less common CMT genes from throughout the UK. One hundred and thirty-nine (50%) of the CMT2 patients and 118 (68%) of the dHMN patients were seen in the NHNN and the remainder were external DNA samples sent to us for diagnostic and research testing. The demographic breakdown of patients was nine from the Far East, 11 from Pakistan, 14 from India, 20 non-UK European and 368 British cases. For patients seen in the inherited neuropathy clinic, detailed information regarding phenotype was available. Clinical diagnosis was based on symptoms, signs, family history (including assessment of family members when possible) and neurophysiology. For external patients, more limited information was available and diagnosis was based on the information received. In order to distinguish CMT from HMN, both clinical and neurophysiological evidence of sensory involvement was taken into account. The majority of patients presented with length dependent muscle weakness and wasting, which was slowly progressive. This study was approved by the research ethics committee of the NHNN (99/N103). All patients gave written informed consent to undergo genetic testing.
Control individuals were from the Wellcome Trust 1958 British birth cohort series (we have access to 1000 samples). An additional 96 Pakistani controls from the CEPH human diversity panel were used for patients of Pakistani origin. The 1958 British birth cohort is based on all people born in Britain in 1 week in 1958. Between 2002 and 2004, a subset of the original cohort were followed-up, resulting in extraction of 8018 DNAs and the creation of 7526 immortalised cell lines. The CEPH human diversity panel comprises DNA from 1050 individuals from 52 world populations.
PCR and sequencing
Genomic DNA extracted from total blood samples was obtained from patients. Coding regions and exon–intron boundaries for all 15 coding exons of TRPV4 were amplified by PCR using 13 primer pairs (see supplementary table 1, available online only). Direct sequencing was performed using Big Dye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, USA) and capillary electrophoresis on an ABI3730xl Genetic Analyser (Applied Biosystems). Novel sequence variants were confirmed by sequencing of a second independent PCR. RNA was extracted from patient total blood samples using a PAX gene blood RNA kit (Qiagen/Preanalytix, Hombrechtikon, Switzerland), and first strand cDNA synthesis was performed using a SuperScript II RT kit (Invitrogen, Gaithersburg, MD, USA) using random primers according to the manufacturer's instructions. PCR was then carried out on cDNA using FastStart PCR master mix (Roche, Mannheim, Germany) and primers designed to amplify exon 11 from cDNA (forward GAA GAA ATG CCC TGG AGT GA, reverse GCA GTT GGT CTG GTC CTC AT). The reactions were run at 95°C for 10 min followed by 40 cycles of 95°C for 30 s, 58°C for 30 s and 72°C for 45 s. Finally, samples were held at 72°C for 10 min. The PCR products were then sequenced as described above.
Analysis of sequencing data
Sequence data were analysed using Sequencher v4.9 (Gene Codes Corporation, Ann Arbor, Michigan, USA). All sequences were compared with the ENST00000261740 TRPV4 transcript (http://www.ensembl.org/index.html), equivalent NCBI RefSeq transcript: NM_021625.4. The likelihood that a mutation is functionally damaging was assessed using five of the most commonly used web based programs: SIFT (http://blocks.fhcrc.org/sift/SIFT.html), PolyPhen (http://genetics.bwh.harvard.edu/pph/), PolyPhen2 (http://genetics.bwh.harvard.edu/pph2/), SNPs&GO (http://snps-and-go.biocomp.unibo.it/snps-and-go/) and SNAP (http://rostlab.org/services/snap/).
Cell culture and western blotting
Lymphoblasts from the proband with the TRPV4 Y567X mutation and an unrelated control were extracted from peripheral blood, immortalised with EB virus and then cultured as suspension cells in RPMI 1640 media containing L-glutamine and supplemented with 15% fetal calf serum. Lymphoblasts were lysed (50 mM Tris HCl, 1% Triton X100, 0.5% deoxycholate, 0.1% sodium dodecyl sulfate, 100 mM NaCl and 1 mM EDTA) for 30 min on ice prior to centrifugation at 10 000 g(av) for 10 min at 4⁰C. Equal amounts of patient and control protein were then electrophoresed on NuPAGE 4–12% Bis-Tris Gels (Invitrogen) and then transferred to a nitrocellulose membrane (Whatman, Stanford, ME, US). TRPV4 was detected using an anti-TRPV4 antibody raised against an intracellular epitope of rat TRPV4 corresponding to amino acid residues 853-871 (1:1000, Alomone Labs, Jerusalem, Israel) and horseradish peroxidise labelled antirabbit IgG secondary antibody (1:5000; Santa Cruz, California, USA)) followed by detection using enhanced chemiluminescence (Fisher Scientific, Rockford, IL, USA). Equal protein loading was confirmed using β-actin.
We sequenced TRPV4 coding regions and splice junctions in 271 patients with CMT2 and 151 patients with dHMN. Among these patients, we detected 23 sequence variants within the coding region (supplementary table 2, available online only), 11 of which were non-synonymous missense and two of which were nonsense variants. Five were novel heterozygous missense or nonsense variants which were absent from controls (table 1). These variants were spread throughout the protein, with two in the ARD (E218K and N302Y), two in transmembrane domains (Y567X and T701I) and one outside known domains (G20R) (figure 1).
The G20R variant was detected in a white UK female patient who presented at 13 years with equinus foot deformities and progressive distal weakness and wasting below the knees. By her mid-thirties she had developed mild proximal weakness in her lower limbs but was still ambulant. Reflexes were brisk but her ankle reflexes were absent. Neurophysiology confirmed normal sensory action potentials in her lower limbs but absent motor action potentials and distal denervation consistent with dHMN. Sequencing of her parents' DNA demonstrated that her father also carried the G20R variant (figure 2A). Her father was asymptomatic with normal muscle strength but EMG of the tibialis anterior showed occasional higher amplitude giant potentials (5–7 mV), possibly indicative of chronic denervation and reinnervation, suggesting that the father may have a milder affected phenotype, but not enough information to exclude an acquired cause such as a radiculopathy. G20R may therefore result in variable expressivity of disease. However, despite conservation of G20 in primates, rodents and other placental mammals (supplementary figure 1, available online only), four of five online prediction programs suggest that this variant is likely to be a neutral polymorphism and it is not located in any known functional domains within the protein.
E218K was found in a patient initially referred to us with a diagnosis of CMT2 but when we subsequently reviewed him it was clear he had a more complex phenotype with a demyelinating neuropathy. The variant was also present in two unaffected siblings (figure 2B) and is predicted to be benign by three out of five web based prediction algorithms.
The N302Y mutation was detected in a white UK female patient who developed symptoms of CMT2 in her forties followed by slowly progressive weakness of the lower limbs, first distally then proximally, and had bilateral foot drop by age 50 years. On examination at age 68 years, there was severe length dependent wasting in the upper and lower limbs, with weakness in the intrinsic hand muscles and proximal and distal weakness in the lower limbs. Vibration sensation was reduced distally. Her father, who is deceased, was reported to have been affected but no DNA was available. Her 35-year-old son was asymptomatic but has not been examined. Segregation analysis is therefore not possible for this patient. N302 is conserved in primates, rodents and other placental mammals but not birds, reptiles or fish (supplementary figure 1, available online only). However, only one of five online programs designed to predict whether or not an amino acid change has a damaging effect on protein function predicted N302Y to be damaging.
The TRPV4 Y567X variant was identified in an affected white UK male patient from a large family with a severe form of CMT2. A MFN2 (M376V) mutation that had been reported to cause a mild form of CMT217 had been identified in the proband and his father. Sequencing of his affected father and unaffected mother and brother demonstrated that the Y567X variant also segregates with disease in the family (figure 2C). Sequencing of the patient's cDNA revealed that RNA carrying the mutant TRPV4 allele was absent from the patient and his father's blood (figure 3), suggesting that the RNA carrying the premature stop codon is degraded. However, western blotting demonstrated that protein expression of TRPV4 in patient lymphoblasts was comparable with control lymphoblasts (figure 4), suggesting that haploinsufficiency of TRPV4 is not causing neuropathy in this family.
The T701I was present in a Pakistani female with onset of a progressive length dependent motor and sensory neuropathy in her fifties. She was areflexic with weakness in the intrinsic hand muscles and at the ankles with reduced vibration sense distally. She was unable to stand on her heels or toes and walked with foot drop. Nerve conduction studies demonstrated an axonal sensorimotor neuropathy. There were no other affected family members and both parents are deceased; however, an unaffected sister did not carry the T701I variant (figure 2D). Given the patient's ethnicity, we also excluded this variant from 91 Pakistani controls as well as 350 Caucasian controls. T701 is highly conserved in primates, rodents, placental mammals, marsupials, birds, reptiles and fish (supplementary figure 1, available online only). Furthermore, all five web based algorithms used in this study predicted T701I to be damaging to protein function.
Sequencing of 271 patients with CMT2 and 151 patients with dHMN revealed 11 missense and two nonsense variants, five of which were heterozygous missense or nonsense changes that were absent from controls (table 1). Three of these variants (G20R, N302Y and T701I) are considered potentially pathogenic. The G20R mutation was detected in a patient with dHMN and was also present in her asymptomatic father. However, her father showed possible evidence of chronic denervation and reinnervation, suggesting that this variant might be pathogenic with variable expressivity. Variable expressivity in terms of severity of disease and age at onset has been observed between and within families carrying known pathogenic TRPV4 mutations.2–7 ,13 The N302Y and T701I mutations were both identified in patients with CMT2. N302Y is located in the ARD of the protein, a region already known to harbour mutations causative of CMT2. T701I maps within a transmembrane domain and close to known skeletal dysplasia mutations, although no skeletal abnormalities were reported for this patient. T701 is highly conserved and predicted damaging to protein function by five out of five web based algorithms used in this study, in common with 19 out of 31 known disease causing TRPV4 mutations but only one out of 19 known TRPV4 SNPs. Although not enough family members were available for segregation analyses, N302Y and T701I are potentially pathogenic mutations requiring further functional validation.
The other two variants (E218K and Y567X) are considered unlikely to be pathogenic. The E218K variant was excluded as a pathogenic variant based on lack of segregation in the patient's family. The Y567X variant segregated with CMT2 in the proband's family (figure 2C) but the truncated copy is not present in patient cDNA, and TRPV4 protein expression was similar in patient lymphoblasts compared with controls. The MFN2 mutation in this family was previously reported as causing a mild CMT2 phenotype.17 It is possible that this mutation simply causes a more severe phenotype in this family or the combined effect of the TRPV4 variant and the MFN2 mutation causes a more severe form of CMT2. The western blotting data and the fact that a second stop variant, W733X, was identified in a patient with dHMN not segregating with disease and present in a control suggests that these loss of function mutations are not sufficient to cause disease. A gain of function pathological mechanism might render programs such as SIFT and PolyPhen less able to predict disease causing mutations.18 Interestingly, a recent paper demonstrated that loss of function heterozygous missense mutations in TRPV4 cause arthropathy of the hands and feet but no such phenotype was seen in our patients with stop mutations.19 However, the loss of function missense changes disrupt TRPV4 tetramer formation whereas the Y567X causes nonsense mediated decay of mutant transcripts which do not therefore interfere with complex formation.
Interestingly, E218K, Y567X and W733X were not the only novel but apparently non-pathogenic missense and nonsense changes to be found in this study. We also detected P638A in a control individual and I715V in a patient initially referred to as having CMT2 but excluded from the study when nerve conduction studies and a review of the phenotype showed no neuropathy. These data suggest that caution is needed before concluding that missense or nonsense TRPV4 mutations are pathogenic by virtue of their absence from controls. In some cases, researchers should also be wary about what they exclude as non-pathogenic given reports of non-segregation of known TRPV4 mutations with neuropathy in patient families.7 ,13 Detailed electrophysiological studies and MRI confirmed that three asymptomatic carriers of the R269C mutation are unaffected, suggesting that TRPV4 mutations can be non-penetrant.20 This makes it difficult to categorise TRPV4 variants as non-pathogenic based on segregation analyses and suggests an important role for in silico and functional studies and extensive screening of controls.
TRPV4 is a non-selective cation channel which is modestly permeable to calcium.21–23 It is activated in response to a variety of physical and chemical stimuli, including innocuous heat, mechanical stimuli such as cell swelling or shear stress, extracellular hypotonicity, and a number of endogenous and synthetic agonists.24–27 Its broad expression profile suggests that TRPV4 participates in a range of physiological processes in different cell types. Of particular relevance for neuropathies, TRPV4 is expressed in the brain and in peripheral neurons with cell bodies located in trigeminal and dorsal root ganglia (DRG), and appears to be involved in pain sensation,28–31 modulation of synaptic transmission,32 cytoskeletal organisation in DRG neurons33 and astrocyte survival of oxidative stress.34 A recent study also demonstrated that TRPV4 mediates neurotrophic factor derived neurite growth in PC12 cells and DRG neurons.35 Functional characterisations of neuropathy causing TRPV4 mutations to date suggest mutations have a toxic gain of function effect on the protein.4–6 ,12 ,36
Initial investigation of the biological mechanism by which TRPV4 mutations cause neuropathy gave conflicting results. Two studies suggested gain of function, with cells expressing R269C, R269H and R316C mutants demonstrating increased channel activity under resting conditions and enhanced responses to TRPV4 specific agonist 4α-phorbol 12,13-didecanoate (4αPDD), arachidonic acid, moderate heat and hypotonicity compared with cells expressing only wild-type TRPV4.4 ,6 This increased channel activity could not be accounted for by increased surface expression as wild-type and mutant proteins were present in approximately equal amounts.4 ,6 In contrast, another study of R269H, R315W and R316C mutants in HeLa cells demonstrated cytoplasmic aggregates of mutant proteins and reduced plasma membrane expression compared with wild-type.2 Furthermore, mutant proteins showed decreased responses to 4αPDD, hypotonicity and no response to moderate heat. It was postulated that this could be due to differences in the cells used, as gain of function had been demonstrated in HEK293 cells and Xenopus laevis oocytes, not HeLa cells. However, two more recent studies support the notion that TRPV4 neuropathy associated mutations (R232C, R316H, R269H, R315W and R316C) result in gain of function using a mixture of HEK293, HeLa and Neuro2a cells.5 ,36 Single channel recordings comparing the R269H mutant with wild-type demonstrated that single channel conductance was unlikely to account for the enhanced activity in mutants but showed that the R269H has a higher open probability.36 Increased calcium flux has been linked to cytotoxicity and indeed the R232C, R269H, R315W, R316C and R316H have all been associated with increased cell death in HEK293, HeLa, DRG neurons and/or Neuro2a cells.5 ,6 ,36 A possible mechanism for these biological effects could be that mutations interfere with the binding of inhibitors of TRPV4 activity. However, PACSIN3 and calmodulin binding was unaffected by R269C and R269H mutations.6 Skeletal dysplasia mutations, including V620I, have also been shown to result in gain of function of the TRPV4 channel.12 These functional data, taken together, suggest that TRPV4 mutations may cause neuropathy through increased channel activity resulting in increased cell death of neurons. Our data are not incompatible with a toxic gain of function mechanism for TRPV4 mutations but functional work would be required to demonstrate gain of function of our putative pathogenic mutations.
In summary, we have conducted the largest screening study to date on TRPV4 in patients with inherited axonal neuropathies and identified five novel variants which were absent from controls. However, only three of these variants (G20R, N302Y and T701I) may be pathogenic and further functional analyses will be required to establish pathogenicity. The frequency of TRPV4 disease causing mutations in patients with CMT2 and dHMN is therefore unlikely to exceed 1%. This is consistent with the two other cohort screenings that have been reported: one proband in 98 individuals with undifferentiated CMT2 across the two studies carried a TRPV4 mutation (∼1%).5 ,7 However, nine of 100 (9%) patients with unusual CMT2 with additional features such as vocal cord and diaphragmatic paresis carried TRPV4 mutations, suggesting routine TRPV4 screening could be restricted to patients with such features. Moreover, this study demonstrates that TRPV4 is a highly polymorphic gene in which rare missense and nonsense variants do not necessarily cause disease. Our results also suggest that haploinsufficiency is not sufficient to cause disease. These findings have important implications for the functional analysis of predicted TRPV4 mutations and suggest that any defects that are identified in patients need to be thoroughly investigated with segregation, in silico and most importantly control screening.
We are grateful to the patients and families for their essential help with this work.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data Supplement 1 - Online Materials
Funding MMR and HH are grateful to the Medical Research Council (MRC) and the Muscular Dystrophy Campaign, and SM and MMR are grateful to the NINDS/ORD (1U54NS065712-01) for their support. SW is funded by an Alzheimer's Research UK fellowship. We would also like to thank The Wellcome Trust for financial support to HH. This work was undertaken at University College London Hospitals/University College London, which received a proportion of funding from the Department of Health's National Institute for Health Research Biomedical Research Centres funding scheme.
Competing interests RQ was a principal investigator for a drug trial sponsored by PTC therapeutics. She has received research grants from the Muscular Dystrophy Campaign, Action Research and AGSD, and has received lecture fees from Genzyme.
Ethics approval Ethics approval was provided by UCLH/UCL Institute of Neurology.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.