Neurobiology of Disease 94 (2016) 237–244

Contents lists available at ScienceDirect

Neurobiology of Disease

journal homepage: www.elsevier.com/locate/ynbdi

Pathogenic mechanisms underlying X-linked Charcot-Marie-Tooth neuropathy (CMTX6) in patients with a kinase 3mutation

Gonzalo Perez-Siles a,c,⁎,CarolynLya, Adrienne Grant a, Alexander P. Drew a, Eppie M. Yiu d,e,f, Monique M. Ryan d,e,f, David T. Chuang g, Shih-Chia Tso g, Garth A. Nicholson a,b,c,MarinaL.Kennersona,b,c,⁎ a Northcott Neuroscience Laboratory, ANZAC Research Institute, University of Sydney, Concord, NSW, Australia b Molecular Medicine Laboratory, Concord Hospital, Concord, NSW, Australia c Sydney Medical School, University of Sydney, Sydney, NSW, Australia d Department of Neurology, Royal Children's Hospital, Flemington Road, Parkville, VIC, Australia e Neuroscience Research, Murdoch Childrens Research Institute, Melbourne, VIC, Australia f Department of Pediatrics, The University of Melbourne, VIC, Australia g Department of Biochemistry, University of Texas Southwestern Medical Center, Dallas, TX, USA article info abstract

Article history: Charcot-Marie-Tooth disease (CMT) is the most common inherited peripheral neuropathy. An X-linked form of Received 11 March 2016 CMT (CMTX6) is caused by a missense mutation (R158H) in the pyruvate dehydrogenase kinase isoenzyme 3 Revised 22 June 2016 (PDK3) gene. PDK3 is one of 4 isoenzymes that negatively regulate the activity of the pyruvate dehydrogenase Accepted 3 July 2016 complex (PDC) by reversible phosphorylation of its first catalytic component pyruvate dehydrogenase (designat- Available online 5 July 2016 ed as E1). Mitochondrial PDC catalyses the oxidative decarboxylation of pyruvate to acetyl CoA and links glycol- ysis to the energy-producing Krebs cycle. We have previously shown the R158H mutation confers PDK3 enzyme Keywords: fi R158H X-linked Charcot-Marie-Tooth neuropathy hyperactivity. In this study we demonstrate that the increased PDK3 activity in patient broblasts (PDK3 ) Pyruvate dehydrogenase kinase 3 leads to the attenuation of PDC through hyper-phosphorylation of E1 at selected serine residues. This hyper- Pyruvate dehydrogenase complex phosphorylation can be reversed by treating the PDK3R158H fibroblasts with the PDK inhibitor dichloroacetate Mitochondria (DCA). In the patient cells, down-regulation of PDC leads to increased lactate, decreased ATP and alteration of Patient fibroblasts the mitochondrial network. Our findings highlight the potential to develop specific drug targeting of the mutant Dichloroacetic acid PDK3 as a therapeutic approach to treating CMTX6. © 2016 Published by Elsevier Inc.

1. Introduction disease and the cellular processes of the peripheral nerve degeneration. However, there are still no effective treatment therapies for CMT. Hereditary motor and sensory disorders of the peripheral nerve The identification of mutations in the pyruvate dehydrogenase ki- form one of the most common groups of human genetic diseases, collec- nase isoenzyme 3 (PDK3) gene as a cause of an X-linked dominant tively called Charcot–Marie–Tooth (CMT) neuropathy. CMT is a clinical- form of CMT (CMTX6) (Kennerson et al., 2013)(Kennerson et al., ly and genetically heterogeneous disorder affecting 1 in 2500 people 2016) has added to the growing list of CMT genes related to the biology (Fowler et al., 1997). Clinical features include progressive weakness of mitochondria (Pareyson et al., 2015) suggesting that pathways lead- and atrophy of distal muscles, high arched feet (pes cavus) and loss of ing to mitochondrial deficits may be a common theme in inherited axo- deep tendon reflexes. Mutations in N80 genes cause CMT and related nal neuropathies. The nuclear-encoded pyruvate dehydrogenase disorders. The diversity of the cellular and molecular function of pro- complex (PDC) is located in the mitochondrial matrix and catalyses teins implicated in CMT is providing insight to the pathophysiology of the conversion of pyruvate to acetyl CoA, a key regulatory step of the en- ergy-producing Krebs cycle. The mammalian PDC is a 9.5 million-Dalton protein machine comprising multiple copies of pyruvate dehydroge- nase (E1), dihydrolipoyl transacetylase (E2), dihydrolipoamide dehy- ⁎ Corresponding authors at: Northcott Neuroscience Laboratory, ANZAC Research drogenase (E3), and the E3-binding protein (E3BP) (Reed, 2001). PDC Institute, University of Sydney, Concord, NSW, Australia. is regulated through reversible phosphorylation of the E1 subunit of E-mail addresses: [email protected] (G. Perez-Siles), [email protected] (M.L. Kennerson). PDC by the four PDK isoenzymes (PDK1 to PDK4) that act to inactivate Available online on ScienceDirect (www.sciencedirect.com). the PDC (Korotchkina and Patel, 1995) and dephosphorylation by

http://dx.doi.org/10.1016/j.nbd.2016.07.001 0969-9961/© 2016 Published by Elsevier Inc. 238 G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244 pyruvate dehydrogenase phosphatases (PDPs) restores PDC activity detection at 450 nm. Briefly, 1 × 106 fibroblasts were trypsinized and re- (Huang et al., 1998). suspended in PBS containing proteinase inhibitors (cOmplete, Mini Pro- Previously we have shown that the R158H mutation confers PDK3 tease Inhibitor, Roche). After protein quantification, the concentration hyperactivity and binds with stronger affinity than its wild-type coun- was adjusted to 10 mg/ml using PBS. Intact functional PDC was solubi- terpart to the inner-lipoyl (L2) domain of the E2 component of PDC lized by adding a detergent (1:9 volumes) provided in the kit. Using (Kennerson et al., 2013). In this study, we further explore the down- the 1XBuffer, sample concentration was adjusted to 1 mg/ml and pro- stream consequences of the R158H mutation, using primary fibroblasts tein (200 μg per well) was assayed in triplicate and incubated at RT (PDK3R158H) cultured from a CMTX6 patient. We demonstrate the im- for 3 h. Wells were washed twice with 1× Stabilizer reagent and the pact of the PDK3 mutation on the phosphorylation status of the E1 sub- Assay solution (200 μl) was added to the wells. Using an EnSpire Multi- unit, its consequences on the PDC activity as well as the subsequent mode Plate Reader (PerkinElmer) the assay absorbance at 450 nm was biochemical outcomes and the impact on mitochondrial morphology measured at RT for 30 min at 30 s intervals. in the PDK3R158H cells. 2.5. Lactate production assay 2. Methods Prior to the experiment (72 h), 1 × 105 cells per well were seeded in 2.1. Fibroblasts culture a P6 well culture plate. Lactate production was evaluated using the Lac- tate Colorimetric Assay Kit II (BioVision). On the day of the experiment, Patients participating in this study provided written consent accord- media containing 10% (v/v) FBS was replaced with DMEM and after a ing to protocols approved by the Sydney Local Health District Human 4 h incubation media (1 ml) was collected and kept on ice. A dilution Ethics Review Committee, Concord Repatriation General Hospital, Syd- (1 in 10) in Assay Buffer (50 μl) plus 50 μl of Reaction Mix was added fi ney, Australia (reference number: HREC/11/CRGH/105). Primary bro- to the ELISA plate. The reaction was incubated for 30 min at RT. The ab- blasts were cultured from patient skin biopsies and maintained with sorbance at 450 nm was measured with an EnSpire Multimode Plate fi broblast medium: DMEM (Gibco, Life technologies) sup- Reader (PerkinElmer). Levels of lactate were corrected per mg of pro- plemented with 10% (v/v) fetal bovine serum (SAFC Biosciences), 1% tein. Protein concentration was determined using the Pierce BCA Pro- (v/v) Penicillin Streptomycin (Gibco, Life technologies) and 1% (v/v) tein Assay Kit (ThermoScientific). L-glutamine (Gibco, Life technologies) and maintained at 37 °C in hu- midified air and 5% CO . 2 2.6. ATP assay 2.2. Real-time (RT) quantitative PCR Prior to the experiment (24 h), 2 × 104 cells per well were plated in P96 well plates. Cellular ATP was measured using an ATPlite assay kit RNA was extracted from fibroblasts using the RNeasy mini kit (PerkinElmer, Massachusetts, UK). Briefly, mammalian cell lysis solu- (Qiagen). Reverse transcribed template was prepared using the High- tion (50 μl) was added to 100 μl cell suspension per well and the plate Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quanti- incubated for 10 min. Substrate solution (50 μl) was added to the cell ly- tative RT-PCR was performed using the PDK3-specificTaqManGeneEx- sate and incubated in the absence of light for 10 min. Luminescence was pression Assay Hs00178440_m1 (Applied Biosystems) and run on a measured on an EnSpire Multimode Plate Reader (PerkinElmer). Data is StepOnePlus™ real-time PCR System (Applied Biosystems). The com- represented as the nmoles of ATP detected for the 2 × 104 cells assayed. parative 2−ΔΔCt method (Schmittgen and Livak, 2008) was used to mea- The average of the signal was determined for the 8 replicates plated for sure the relative quantitation of PDK3 expression in affected and control each cell line. individuals. 18S (Hs99999901_s1; Applied Biosystems) was used as the housekeeping gene. 2.7. Assessment of mitochondrial morphology 2.3. Immunofluorescence Prior to the experiment (48 h) 3 × 104 cells per well were plated on Control and patient fibroblasts were incubated with 200 nM cover slips and placed in P24 well plates. The mitochondrial network in fi MitoTracker Red CMX Red (Invitrogen) to visualize mitochondrial control and patient broblasts was stained with 200 nM MitoTracker fi structures. Cells were fixed with 4% formaldehyde, permeabilized in Red CMX Red for 30 min. Cells were xed and mounted as previously fi phosphate-buffered saline (PBS) containing 0.3% (v/v) Triton X-100 described. N = 30 individual well de ned cells were analyzed per sam- and blocked in 3% (w/v) bovine serum albumin (BSA). Cells were incu- ple. Using the image processing package Fiji (Schindelin et al., 2012), fi bated with affinity-purified anti-pSer293 (phosphorylation site 1), anti- images were binarised by conversion to 8 bit image types. Nonspeci c fl “ ” pSer300 (site 2) or anti-pSer232 (site 3) (Millipore) at 500 ng/ml in PBS noise of the uorescent signal was reduced by using the despeckle containing 3% (w/v) BSA overnight at 4 °C, followed by Alexa Fluor 488 function and the mitochondrial structures highlighted using the convo- fi goat anti-rabbit-conjugated (Invitrogen) at 1:500 dilution for lution lter. A threshold was then applied to the images to obtain an ap- “ ” 2 h at RT. Nuclei were stained with 300 nM 4,6-diamidino-2- propriate signal-to-noise ratio. Using the analyzed particles option, phenylindole (DAPI, Molecular Probes) and mounted using Prolong morphological characteristics for each mitochondrion per cell were an- Gold antifade reagent (Invitrogen). When required, cells were treated alyzed for area, perimeter, as well as major and minor axes. These indi- with the PDK inhibitor dichloroacetic acid DCA (Sigma-Aldrich) at vidual parameters were used to calculate the aspect ratio (ratio 5 mM for 2 h before staining. Cells were visualized using a Leica SPE-II between the major and minor axes of the ellipse equivalent to the ob- fi confocal microscope and images acquired at 63× magnification. ject) which de nes the length of mitochondria, as well as the form fac- tor (perimeter2 / (4PIarea)) which reflects the degree of branching and 2.4. Pyruvate dehydrogenase complex (PDC) activity assay the number of mitochondria per cell (Mortiboys et al., 2008).

PDC activity in control and patient fibroblasts was measured using 2.8. Statistical analysis the PDC Enzyme Activity Microplate Assay Kit (MitoSciences). In this assay, PDC is immunocaptured within the microplate and the activity For the statistical analysis, 3 independent experiments under the determined by following the reduction of NAD+ to NADH. This reduc- same conditions were performed and a Student's t-test used to assess tion is coupled to a reporter dye that yields a colored product for the significance of the results. The data are expressed as mean ± SEM. G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244 239

The following statistical thresholds have been applied throughout the PDK3R158H fibroblasts at the Ser293 (site1)(4.4foldchange)and manuscript: *p b 0.05; **p b 0.01; ***p b 0.001. Ser300 (site 2) (2.9 fold change) when compared with controls (Fig. 2B). No significant change was observed for phosphorylation levels at 3. Results position Ser232.

3.1. PDK3 is expressed in human fibroblasts and the R158H mutation does 3.3. Biochemical consequences of E1 hyper-phosphorylation in patient not change levels of the PDK3 gene expression PDK3R158H fibroblasts Peripheral tissues such as fibroblasts have been shown to be a useful As the levels of phosphorylation of E1 determine the activity of PDC, model of neurological diseases, replicating pathophysiological abnor- we postulated the increased phosphorylation at Ser293 and Ser300 ob- malities present in the affected tissues. Specifically, studies using pa- served in the PDK3R158H fibroblasts would reduce the activity of the tients fibroblasts have provided relevant functional information about PDC and affect the conversion of pyruvate to acetyl CoA in the mito- the molecular mechanism underlying numerous forms of CMT (Noack chondria (Fig. 3A). To test this hypothesis we first measured the PDC ac- et al., 2012)(Irobi et al., 2012)(Casasnovas et al., 2010)(Perez-Siles et tivity in whole cell extracts from PDK3R158H and 3 unrelated control al., 2016). fibroblast cell lines, using a commercially available PDC Enzyme Activity PDK3 is expressed in testis, kidney, lungs, brain (Bowker-Kinley et Microplate Assay Kit. The results demonstrated the activity of PDC is sig- al., 1998), spinal cord and skeletal muscle (Kennerson et al., 2013). To nificantly reduced in the CMTX6 patient cells when compared to con- determine if PDK3 is expressed in human fibroblasts and if the R158H trols (Fig. 3B). mutation alters PDK3 mRNA expression, we performed quantitative Lactic acid is the product of anaerobic of glucose and is real time PCR analysis on reverse transcribed template from a generated by the reduction of pyruvate by lactate dehydrogenase PDK3R158H patient and control subjects (Fig. 1). Our data confirmed (LDH). The reduced PDC activity observed in PDK3R158H fibroblasts that both mutant and wild type PDK3 is expressed in human fibroblasts. may lead to the accumulation of mitochondrial pyruvate being convert- Levels of PDK3 mRNA in the patient cells is within the range of expres- ed into lactate. The cytotoxic effects of accumulated lactate in neuronal sion found in the three age-matched controls, indicating the R158H mu- cells has been reported (Staub et al., 1993) and there is increasing evi- tation does not affect the level of expression of the PDK3 gene. dence and established mechanisms by which lactic acidosis might spe- cifically affect the highly specialized neuromuscular junction (NMJ) 3.2. The E1 component of PDC is hyper-phosphorylated at selected serine structure (Vadakkadath Meethal and Atwood, 2012). To test if PDC de- residues in patient PDK3R158H fibroblasts ficiency in the CMTX6 patient cells leads to increased lactic acid, we test- ed for the levels of lactate produced in PDK3R158H and control We previously showed the R158H mutation confers enzyme hyper- fibroblasts. Fig. 3C shows lactate levels were significantly higher in the activity and binds with stronger affinity than the wild-type to the inner- patient cells than in the controls. lipoyl (L2) domain in the E2 chain of the PDC (Kennerson et al., 2013). In the mitochondria, oxidative decarboxylation of pyruvate pro- We proposed that these combined biochemical effects on PDK3 function duces acetyl-CoA, linking glycolysis to the energy-producing Krebs could impact the PDC phosphorylation state in the patient cells. The (TCA) cycle. The high energy demands of motor neurons, which rely function of this enzyme is tightly regulated by reversible phosphoryla- heavily on axonal transport mechanisms within their long axonal com- tion at three known serine sites at the E1 subunit: Ser293 (site 1, full- partments, make this cell type highly dependent on maintaining suffi- length sequence beginning with the initiation Met), Ser300 (site 2) and cient levels of energy production. ATP depletion has been previously Ser232 (site 3) (Yeaman et al., 1978)(Teague et al., 1979)(Sale and described in other cell models of motor neuron disease (MND) and Randle, 1981). All PDK isoenzymes phosphorylate site 1 and site 2 CMT (Tamiya et al., 2014)(Strickland et al., 2014)(Xu et al., 2012) sug- with different rates, whereas only PDK1 phosphorylates site 3 gesting compromised energy levels may be common in this group of (Korotchkina and Patel, 2001). To determine the phosphorylation levels disorders. To further investigate if energy deficits could be observed in of the PDK3R158H fibroblasts compared to control fibroblasts we per- the patient fibroblasts, we tested the production of ATP in the formed immunofluorescence analysis using phospho-specificantibod- PDK3R158H and control fibroblasts. We confirmed the PDK3 R158H fibro- ies corresponding to the three E1 serine residues. blasts produce significantly lower levels of ATP compared to control We observed increased phosphorylation in the PDK3R158H cells cells (Fig. 3D). when compared to control fibroblasts for two of the serine residues, Ser293 and Ser300 (Fig. 2A). Measurement of the corrected total cell fluo- rescence (CTCF) confirmed the visual confocal microscopy observations 3.4. PDK3R158H fibroblasts show an altered mitochondrial network with statistically significant increases of phosphorylation in the Metabolic dysfunction and decreased ATP production can result in mitochondrial morphologic defects. This phenomenon underlies several neurodegenerative disorders including Parkinson disease (Grunewald et al., 2010), Alzheimer disease (Wang et al., 2008) and amyotrophic lat- eral sclerosis (ALS) disease progression (Borthwick et al., 1999) (Bowling et al., 1993)(Carri et al., 1997)(Fujita et al., 1996)(Menzies et al., 2002). We performed a detailed characterization of mitochondrial morphology in PDK3R158H and control fibroblasts. Mitochondria were measured for the length, degree of branching and the number of mito- chondria per cell line (Fig. 4). Our studies confirmed significant reduced numbers of mitochondria in the mutant cells when compared with the 3 control cell lines. A signif- icantly lower aspect ratio in the PDK3R158H cells was observed when Fig. 1. Relative PDK3 gene expression (mRNA) levels. Controls and patient PDK3 mRNA comparing control 1 and 3 to the patient cells, and a trend (did not expression was calculated using the comparative 2−ΔΔCt method with 18S as the reach significance) of decreased aspect ratio was observed when com- housekeeping gene. PDK3 mRNA expression in control 1 is used as baseline. Data was obtained from 3 independent experiments and is represented as the mean ± standard paring control 2 to the patient. This suggests the length of the mitochon- error. drial network may also be affected. No differences were found in the 240 G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244

Fig. 2. (A) Patient E1 subunit is hyper-phosphorylated at Ser293 and Ser300. Patient and control fibroblasts were stained with E1-PSer232 (site 3), -PSer293 (site 1), and -PSer300 (site 2) specific (green). Nuclei were stained with DAPI (blue). Images were captured at 63× magnification. Scale bar is 20 μm. (B) Corrected total cell fluorescence (CTCF) determined for each treatment condition represented as the mean value obtained from 50 cells for 3 independent experiments. **p b 0.01; ***p b 0.001

degree of mitochondrial branching between the patient and control of the E1 phosphorylation is due to the specific inhibition of the PDK3 cells (form factor). isoform.

3.5. E1 hyper-phosphorylation can be reduced by treating PDK3R158H fibro- 4. Discussion blasts with a pan-PDK inhibitor In the present study, using primary fibroblasts we describe the path- Given the biochemical consequences and mitochondrial abnormali- ogenic consequences underlying X-linked Charcot-Marie-Tooth neu- ties shown in the PDK3R158H fibroblasts, E1 hyper-phosphorylation may ropathy (CMTX6) in patients with a R158H mutation in the PDK3 represent a pharmacological target for therapeutic intervention in the gene. Previous biochemical analyses using recombinant forms of CMTX6 patients. human PDK3 had shown the R158H mutation confers hyperactivity in As a proof of principle we have tested the effect of the pan-PDK in- PDK3 and binds with stronger affinity to the PDC than wild type hibitor dichloroacetate (DCA) on the levels of phosphorylation at PDK3. However, the mutation effects on biochemical consequences of Ser293 and Ser300 in the PDK3R158H fibroblasts. PDK3 hyperactivity and organelle morphology remained unexplored. The sensitivity of the PDK isoforms to inhibition by DCA varies nota- We have shown the R158H mutation leads to an attenuation of PDC ac- bly (Tso et al., 2014)(Kato et al., 2007)(Baker et al., 2000) and apparent tivity in PDK3R158H patient fibroblasts, resulting in numerous down- Ki values of the different isoforms for DCA varies 40-fold, from 0.2 mM stream abnormalities including increased lactate, decreased ATP and for PDK2, 0.5 mM for PDK4 and 1 mM for PDK1, up to 8 mM for PDK3 mitochondrial structural abnormalities. (Bowker-Kinley et al., 1998). Patient cells were treated with increasing By using phospho-specific antibodies against each of the three serine concentrations of DCA (0–25 mM) prior to the staining with the sites known to be phosphorylated at the E1 subunit, we determined that phospho-specific antibodies. Only incubations above 5 mM are able to reduced activity of PDC in the PDK3R158H patient fibroblasts is caused by revert E1 hyper-phosphorylation (Fig. 5A). Quantification of the levels specific modification of Ser293 and Ser300. These results support previ- of phosphorylation at Ser293 and Ser300 in the PDK3R158H fibroblasts ous biochemical studies (Korotchkina and Patel, 2001)thatdemonstrat- confirms that treatments with lower doses of DCA (0.2 mM, 0.5 mM, ed PDK3 specifically phosphorylates these two serine sites (sites 1 and 1 mM) did not have a significant impact on the hyperphosphorylation 2) but does not modify Ser232, (site 3); the latter residue is known to shown at the patient cells (Fig. 5B). This result demonstrates that be exclusively modified by PDK1. hyperphosphorylation can be reversed in the patient fibroblasts. The Our experiments demonstrate the PDK3R158H mutation causes a dose dependent response to DCA has indirectly shown the reversibility marked reduction in the capacity of the patient-derived cells to produce G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244 241

Fig. 3. Biochemical consequences of E1 hyper-phosphorylation in patient's PDK3R158H fibroblasts. (A) Schematic representation of a mitochondria. Pyruvate dehydrogenase complex (PDC) contributes to transforming pyruvate into acetyl-CoA (AcCoA), which feeds the Krebs Cycle (TCA) rendering the production of ATP molecules. In the cytosol, accumulation of pyruvate promotes the production of lactate via lactate dehydrogenase (LDH) activity. (B) PDC activity was measured in whole cell extracts from control and patient fibroblasts (400 μg) by measuring the absorbance at 450 nm at 30 s intervals for 30 min. PDC activity is shown as the increase in milliOD per minute per mg protein calculated from the linear range of the curves (600–1500 s). (C) Lactate was measured in control and patient fibroblasts 4 h after replacing the media with fresh DMEM. Data is represented as nmoles of lactate produced per mg protein per min. (D) ATP production in control and patient fibroblasts was determined from lysates obtained from 2 × 104 cells. Data is represented as the total nmoles of ATP produced. Biochemical experiments were repeated 3 times and data represented as the mean ± standard error. *p b 0.05.

ATP. The long of the peripheral nerves, which are affected in CMT, (Pitceathly et al., 2012). The maintenance of a balanced mitochondrial require particularly high levels of ATP and therefore make the motor network is achieved through highly regulated mitochondrial dynamics, neurons especially vulnerable to energy deficits. Other CMT gene muta- including ATP-dependent mitochondrial transport along cytoskeletal tions reporting mitochondrial bioenergetic deficits of reduced ATP tracks, regulated fission and fusion events and the establishment of con- levels include COX6A1 (Tamiya et al., 2014), DHTKD1 (Xu et al., 2012), tact sites between mitochondria and a specialized subdomain of the en- VCP (Gonzalez et al., 2014)(Bartolome et al., 2013)andMT-ATP6 doplasmic reticulum (ER) termed mitochondria-associated membranes

Fig. 4. Morphology of the mitochondrial network in control and R158H patient fibroblasts. (A) Representative binarized images of control and patient fibroblasts. Mitotracker Red was used to visualize the mitochondrial network. Scale bar is 20 μm. Boxplots showing the distributions of number of mitochondria (B), aspect ratio (C) and form factor (D). Analysis of the data obtained from 30 individual cells per sample revealed statistically significant differences in mitochondrial length and number of mitochondria between the control and patient fibroblasts. *p b 0.05. 242 G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244

Fig. 5. P E1 hyper-phosphorylation at Ser293 and Ser300 in the patient cells is reduced at high doses of DCA. (A) Representative images of fibroblasts stained using the E1-PSer293 and - PSer300 specific antibodies (green) after 2 h treatment with increasing concentrations of DCA (0–10 mM). Cells were stained with the nuclear marker DAPI (blue). Images were captured at 63× magnification. Scale bar is 20 μm. (B) Mean fluorescence within the region of interest (defined by staining the cells with 200 nM MitoTracker Red) was calculated for PSer293 and PSer300 after DCA treatments (0–25 mM) as the mean ± standard error obtained for 50 cells from 2 independent experiments. *p b 0.05; ***p b 0.001.

(MAMs) (Krols et al., 2016). These processes ensure optimal mitochon- Nehrke, 2010), and recent investigations reported increased lactate as drial metabolism and, in neurons, the correct distribution of mitochon- a metabolic signature in motor neuron dysfunction (Dodge et al., 2013). dria to dendrites and synapses. Genes affecting mitochondrial dynamics The hyperactivity of the PDK3R158H kinase, leading to the inhibition in CMT include MFN2 (Zuchner et al., 2004), INF2 (Boyer et al., 2011) of an integral component of the cell metabolism and bioenergetics, (Kijima et al., 2005), GDAP1 (Baxter et al., 2002; Cuesta et al., 2002), make the PDC and associated pathways an ideal pharmacological target KIF1Bbeta (Zhao et al., 2001)andDYNC1H1 (Weedon et al., 2011). for the development of treatment therapies. We have shown that pa- The data presented for PDK3R158H fibroblasts supports several func- tient cells treated with high doses of DCA, an inhibitor of all PDK isoen- tional studies using both patient-derived primary cells and animal zymes, reduces the hyper-phosphorylation at Ser293 and Ser300 in the models, in which CMT genes involved in mitochondrial dynamics have PDK3R158H fibroblasts, proving that E1 hyperphosphorylation in the pa- been genetically modified. These studies have repeatedly shown that tient cells is a reversible process. mitochondrial fragmentation is a recurrent event in the etiology of Although DCA has been used in humans for decades in the treatment CMT (Johnson and Nehrke, 2010)(Strickland et al., 2014)(Saporta et of lactic acidosis and inherited mitochondrial diseases (Whitehouse and al., 2015). Research in Caenorhabditis elegans led to the hypothesis that Randle, 1973)(Stacpoole et al., 2006) and recent studies have shown a lactic acidosis following mitochondrial fragmentation may play a role therapeutic potential of systemic DCA administration in an ALS in both the tissue specificity and progression of CMT2A (Johnson and SOD1G93A mouse model (Miquel et al., 2012) the clinical benefits of G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244 243

DCA are controversial since other investigations have reported worsen- Bowling, A.C., Schulz, J.B., Brown Jr., R.H., Beal, M.F., 1993. Superoxide dismutase activity, oxidative damage, and mitochondrial energy metabolism in familial and sporadic ing of peripheral neuropathy and tumor growth after high doses of the amyotrophic lateral sclerosis. J. Neurochem. 61 (6), 2322–2325. pyruvate analog (Kaufmann et al., 2006). Studies using rat Schwann Boyer, O., Nevo, F., Plaisier, E., Funalot, B., Gribouval, O., Benoit, G., Huynh Cong, E., cells and dorsal root ganglia neurons have proved that DCA causes a Arrondel, C., Tete, M.J., Montjean, R., Richard, L., Karras, A., Pouteil-Noble, C., Balafrej, L., Bonnardeaux, A., Canaud, G., Charasse, C., Dantal, J., Deschenes, G., dose- and exposure dependent decrease of myelination (Felitsyn et al., Deteix, P., Dubourg, O., Petiot, P., Pouthier, D., Leguern, E., Guiochon-Mantel, A., 2007) that may account for its clinical peripheral neuropathic effects. Broutin, I., Gubler, M.C., Saunier, S., Ronco, P., Vallat, J.M., Alonso, M.A., Antignac, C., Although effectiveness of DCA on improving the metabolic function of Mollet, G., 2011. INF2 mutations in Charcot-Marie-Tooth disease with glomerulopa- – skin fibroblasts from patients with PDC deficiency has been shown thy. N. Engl. J. Med. 365 (25), 2377 2388. Carri, M.T., Ferri, A., Battistoni, A., Famhy, L., Gabbianelli, R., Poccia, F., Rotilio, G., 1997. Ex- (Fouque et al., 2003), our experimental approach measuring total ATP pression of a Cu,Zn superoxide dismutase typical of familial amyotrophic lateral scle- intracellular levels failed to detect statistically significant improvement rosis induces mitochondrial alteration and increase of cytosolic Ca2+ concentration – on the capacity of the PDK3R158H fibroblasts to produce ATP (Supple- in transfected neuroblastoma SH-SY5Y cells. FEBS Lett. 414 (2), 365 368. Casasnovas, C., Banchs, I., Cassereau, J., Gueguen, N., Chevrollier, A., Martinez-Matos, J.A., mentary Fig. S1). Moreover, the high doses needed to reverse the Bonneau, D., Volpini, V., 2010. Phenotypic spectrum of MFN2 mutations in the Span- hyperphosphorylation in the CMTX6 patient cells (Fig. 5) suggests any ish population. J. Med. Genet. 47 (4), 249–256. therapeutic intervention using DCA to inhibit the mutant PDK3 would Cuesta, A., Pedrola, L., Sevilla, T., Garcia-Planells, J., Chumillas, M.J., Mayordomo, F., LeGuern, E., Marin, I., Vilchez, J.J., Palau, F., 2002. The gene encoding -in- impact on the activity of the other PDK3 isoforms. duced differentiation-associated protein 1 is mutated in axonal Charcot-Marie- The existence of multiple PDK isoenzymes makes the development Tooth type 4A disease. Nat. Genet. 30 (1), 22–25. of isoform-specific drugs highly desirable. Increasing the specificity of Dodge, J.C., Treleaven, C.M., Fidler, J.A., Tamsett, T.J., Bao, C., Searles, M., Taksir, T.V., Misra, K., Sidman, R.L., Cheng, S.H., Shihabuddin, L.S., 2013. Metabolic signatures of amyotro- the compound would help to minimize undesirable secondary effects. phic lateral sclerosis reveal insights into disease pathogenesis. Proc. Natl. Acad. Sci. U. The employment of structure-guided design for the development of S. A. 110 (26), 10812–10817. an inhibitor highly specific for the PDK2 isoform has been recently Felitsyn, N., Stacpoole, P.W., Notterpek, L., 2007. Dichloroacetate causes reversible demy- achieved (Tso et al., 2014). This supports the feasibility of designing elination in vitro: potential mechanism for its neuropathic effect. J. Neurochem. 100 (2), 429–436. highly selective kinase inhibitors for PDK3 by using structural informa- Fouque, F., Brivet, M., Boutron, A., Vequaud, C., Marsac, C., Zabot, M.T., Benelli, C., 2003. tion available for this isoform (Kato et al., 2005). Differential effect of DCA treatment on the pyruvate dehydrogenase complex in pa- fi – Given that CMT neuropathy specifically affects peripheral nerves, tients with severe PDHC de ciency. Pediatr. Res. 53 (5), 793 799. Fowler Jr., W.M., Abresch, R.T., Koch, T.R., Brewer, M.L., Bowden, R.K., Wanlass, R.L., 1997. Em- the need for a suitable model to test potential drugs aimed to ameliorate ployment profiles in neuromuscular diseases. Am. J. Phys. Med. Rehabil. 76 (1), 26–37. the progression of the disease in CMTX6 is essential. The development Fujita, K., Yamauchi, M., Shibayama, K., Ando, M., Honda, M., Nagata, Y., 1996. Decreased of patient iPSC-derived motor neurons would therefore represent an ex- cytochrome c oxidase activity but unchanged superoxide dismutase and glutathione R158H peroxidase activities in the spinal cords of patients with amyotrophic lateral sclerosis. cellent platform for testing potential drugs targeting the PDK3 mu- J. Neurosci. Res. 45 (3), 276–281. tant while maintaining the disease-associated genetic background of Gonzalez, M.A., Feely, S.M., Speziani, F., Strickland, A.V., Danzi, M., Bacon, C., Lee, Y., Chou, the patient (Saporta et al., 2015). T.F., Blanton, S.H., Weihl, C.C., Zuchner, S., Shy, M.E., 2014. A novel mutation in VCP – fi fi causes Charcot-Marie-Tooth type 2 disease. Brain 137 (Pt 11), 2897 2902. In summary, our ndings using patient broblasts have revealed Grunewald, A., Voges, L., Rakovic, A., Kasten, M., Vandebona, H., Hemmelmann, C., pathogenic pathways underlying CMTX6 and highlight an important Lohmann,K.,Orolicki,S.,Ramirez,A.,Schapira,A.H.,Pramstaller,P.P.,Sue,C.M., causative link between peripheral nerve degeneration and essential bio- Klein, C., 2010. Mutant parkin impairs mitochondrial function and morphology in fi R158H fi human broblasts. PLoS One 5 (9), e12962. energetic pathways. The evidence that the PDK3 broblasts show a Huang, B., Gudi, R., Wu, P., Harris, R.A., Hamilton, J., Popov, K.M., 1998. Isoenzymes of py- marked hyper-phosphorylation at selected serine residues of the E1 ruvate dehydrogenase phosphatase. DNA-derived amino acid sequences, expression, component in PDC and the observation that this modification can be re- and regulation. J. Biol. Chem. 273 (28), 17680–17688. Irobi, J., Holmgren, A., De Winter, V., Asselbergh, B., Gettemans, J., Adriaensen, D., duced by using PDK inhibitors highlights this pathway as a potential tar- Ceuterick-de Groote, C., Van Coster, R., De Jonghe, P., Timmerman, V., 2012. Mutant get for therapeutic intervention and treatment of CMTX6. HSPB8 causes protein aggregates and a reduced mitochondrial membrane potential Supplementary data to this article can be found online at http://dx. in dermal fibroblasts from distal hereditary motor neuropathy patients. – doi.org/10.1016/j.nbd.2016.07.001. Neuromuscul. Disord. 22 (8), 699 711. Johnson, D., Nehrke, K., 2010. Mitochondrial fragmentation leads to intracellular acidifica- tion in Caenorhabditis elegans and mammalian cells. Mol. Biol. Cell 21 (13), 2191–2201. Acknowledgments Kato, M., Chuang, J.L., Tso, S.C., Wynn, R.M., Chuang, D.T., 2005. Crystal structure of pyru- vate dehydrogenase kinase 3 bound to lipoyl domain 2 of human pyruvate dehydro- The authors thank the families. This research was supported by Na- genase complex. EMBO J. 24 (10), 1763–1774. Kato, M., Li, J., Chuang, J.L., Chuang, D.T., 2007. Distinct structural mechanisms for inhibi- tional Health and Medical Research Council Project Grants tion of pyruvate dehydrogenase kinase isoforms by AZD7545, dichloroacetate, and (APP1046680 and APP1007705) and USA Muscular Dystrophy Associa- radicicol. Structure 15 (8), 992–1004. tion Project Grant (MDA158509) awarded to M.L.K. and G.A.N. as well as Kaufmann, P., Engelstad, K., Wei, Y., Jhung, S., Sano, M.C., Shungu, D.C., Millar, W.S., Hong, the National Institutes of Health (DK62306) and the Welch Foundation X., Gooch, C.L., Mao, X., Pascual, J.M., Hirano, M., Stacpoole, P.W., DiMauro, S., De Vivo, D.C., 2006. Dichloroacetate causes toxic neuropathy in MELAS: a randomized, con- (I-1286) grants awarded to D.T.C. trolled clinical trial. Neurology 66 (3), 324–330. Kennerson, M.L., Yiu, E.M., Chuang, D.T., Kidambi, A., Tso, S.C., Ly, C., Chaudhry, R., Drew, A.P., Rance, G., Delatycki, M.B., Zuchner, S., Ryan, M.M., Nicholson, G.A., 2013. Anew References locus for X-linked dominant Charcot-Marie-Tooth disease (CMTX6) is caused by mu- tations in the pyruvate dehydrogenase kinase isoenzyme 3 (PDK3) gene. Hum. Mol. Baker, J.C., Yan, X., Peng, T., Kasten, S., Roche, T.E., 2000. Marked differences between two Genet. 22 (7), 1404–1416. isoforms of human pyruvate dehydrogenase kinase. J. Biol. Chem. 275 (21), Kennerson, M.L., Kim, E.J., Siddell, A., Kidambi, A., Kim, S.M., Hong, Y.B., Hwang, S.H., 15773–15781. Chung, K.W., Choi, B.O., 2016. X-linked Charcot-Marie-Tooth disease type 6 Bartolome, F., Wu, H.C., Burchell, V.S., Preza, E., Wray, S., Mahoney, C.J., Fox, N.C., Calvo, A., (CMTX6) patients with a p.R158H mutation in the pyruvate dehydrogenase kinase Canosa,A.,Moglia,C.,Mandrioli,J.,Chio,A.,Orrell,R.W.,Houlden,H.,Hardy,J., isoenzyme 3 gene. J. Peripher. Nerv. Syst. 21 (1), 45–51. Abramov, A.Y., Plun-Favreau, H., 2013. Pathogenic VCP mutations induce mitochon- Kijima,K.,Numakura,C.,Izumino,H.,Umetsu,K.,Nezu,A.,Shiiki,T.,Ogawa,M.,Ishizaki,Y., drial uncoupling and reduced ATP levels. Neuron 78 (1), 57–64. Kitamura, T., Shozawa, Y., Hayasaka, K., 2005. Mitochondrial GTPase mitofusin 2 muta- Baxter, R.V., Ben Othmane, K., Rochelle, J.M., Stajich, J.E., Hulette, C., Dew-Knight, S., tion in Charcot-Marie-Tooth neuropathy type 2A. Hum. Genet. 116 (1–2), 23–27. Hentati, F., Ben Hamida, M., Bel, S., Stenger, J.E., Gilbert, J.R., Pericak-Vance, M.A., Korotchkina, L.G., Patel, M.S., 1995. Mutagenesis studies of the phosphorylation sites of Vance, J.M., 2002. Ganglioside-induced differentiation-associated protein-1 is mutant recombinant human pyruvate dehydrogenase. Site-specific regulation. J. Biol. Chem. in Charcot-Marie-Tooth disease type 4A/8q21. Nat. Genet. 30 (1), 21–22. 270 (24), 14297–14304. Borthwick, G.M., Johnson, M.A., Ince, P.G., Shaw, P.J., Turnbull, D.M., 1999. Mitochondrial Korotchkina, L.G., Patel, M.S., 2001. Site specificity of four pyruvate dehydrogenase kinase enzyme activity in amyotrophic lateral sclerosis: implications for the role of mito- isoenzymes toward the three phosphorylation sites of human pyruvate dehydroge- chondria in neuronal cell death. Ann. Neurol. 46 (5), 787–790. nase. J. Biol. Chem. 276 (40), 37223–37229. Bowker-Kinley, M.M., Davis, W.I., Wu, P., Harris, R.A., Popov, K.M., 1998. Evidence for ex- Krols, M., van Isterdael, G., Asselbergh, B., Kremer, A., Lippens, S., Timmerman, V., istence of tissue-specific regulation of the mammalian pyruvate dehydrogenase com- Janssens, S., 2016. Mitochondria-associated Membranes as Hubs for Neurodegenera- plex. Biochem. J. 329 (Pt 1), 191–196. tion. Neuropathology, Acta. 244 G. Perez-Siles et al. / Neurobiology of Disease 94 (2016) 237–244

Menzies, F.M., Cookson, M.R., Taylor, R.W., Turnbull, D.M., Chrzanowska-Lightowlers, D.W., Valenstein, E., 2006. Controlled clinical trial of dichloroacetate for treatment Z.M., Dong, L., Figlewicz, D.A., Shaw, P.J., 2002. Mitochondrial dysfunction in a cell cul- of congenital lactic acidosis in children. Pediatrics 117 (5), 1519–1531. ture model of familial amyotrophic lateral sclerosis. Brain 125 (Pt 7), 1522–1533. Staub, F., Mackert, B., Kempski, O., Peters, J., Baethmann, A., 1993. Swelling and death of Miquel, E., Cassina, A., Martinez-Palma, L., Bolatto, C., Trias, E., Gandelman, M., Radi, R., neuronal cells by lactic acid. J. Neurol. Sci. 119 (1), 79–84. Barbeito, L., Cassina, P., 2012. Modulation of astrocytic mitochondrial function by Strickland, A.V., Rebelo, A.P., Zhang, F., Price, J., Bolon, B., Silva, J.P., Wen, R., Zuchner, S., dichloroacetate improves survival and motor performance in inherited amyotrophic 2014. Characterization of the mitofusin 2 R94W mutation in a knock-in mouse lateral sclerosis. PLoS One 7 (4), e34776. model. J. Peripher. Nerv. Syst. 19 (2), 152–164. Mortiboys, H., Thomas, K.J., Koopman, W.J., Klaffke, S., Abou-Sleiman, P., Olpin, S., Wood, Tamiya, G., Makino, S., Hayashi, M., Abe, A., Numakura, C., Ueki, M., Tanaka, A., Ito, C., N.W., Willems, P.H., Smeitink, J.A., Cookson, M.R., Bandmann, O., 2008. Mitochondrial Toshimori, K., Ogawa, N., Terashima, T., Maegawa, H., Yanagisawa, D., Tooyama, I., function and morphology are impaired in parkin-mutant fibroblasts. Ann. Neurol. 64 Tada, M., Onodera, O., Hayasaka, K., 2014. A mutation of COX6A1 causes a recessive (5), 555–565. axonal or mixed form of Charcot-Marie-Tooth disease. Am. J. Hum. Genet. 95 (3), Noack, R., Frede, S., Albrecht, P., Henke, N., Pfeiffer, A., Knoll, K., Dehmel, T., Horste, G.M.Z., 294–300. Stettner, M., Kieseier, B.C., Summer, H., Golz, S., Kochanski, A., Wiedau-Pazos, M., Teague, W.M., Pettit, F.H., Yeaman, S.J., Reed, L.J., 1979. Function of phosphorylation sites Arnold, S., Lewerenz, J., Methner, A., 2012. Charcot-Marie-Tooth disease CMT4A: on pyruvate dehydrogenase. Biochem. Biophys. Res. Commun. 87 (1), 244–252. GDAP1 increases cellular glutathione and the mitochondrial membrane potential. Tso, S.C., Qi, X., Gui, W.J., Wu, C.Y., Chuang, J.L., Wernstedt-Asterholm, I., Morlock, L.K., Hum. Mol. Genet. 21 (1), 150–162. Owens, K.R., Scherer, P.E., Williams, N.S., Tambar, U.K., Wynn, R.M., Chuang, D.T., Pareyson, D., Saveri, P., Sagnelli, A., Piscosquito, G., 2015. Mitochondrial dynamics and 2014. Structure-guided development of specific pyruvate dehydrogenase kinase in- inherited peripheral nerve diseases. Neurosci. Lett. 596, 66–77. hibitors targeting the ATP-binding pocket. J. Biol. Chem. 289 (7), 4432–4443. Perez-Siles, G., Grant, A., Ellis, M., Ly, C., Kidambi, A., Khalil, M., Llanos, R.M., Fontaine, S., Vadakkadath Meethal, S., Atwood, C.S., 2012. Lactate dyscrasia: a novel explanation for Strickland, A.V., Zuchner, S., Bermeo, S., Neist, E., Brennan-Speranza, T.C., Takata, amyotrophic lateral sclerosis. Neurobiol. Aging 33 (3), 569–581. R.I., Speck-Martins, C.E., Mercer, J.F., Nicholson, G.A., Kennerson, M.L., 2016. Charac- Wang, X., Su, B., Fujioka, H., Zhu, X., 2008. Dynamin-like protein 1 reduction underlies mi- terizing the molecular phenotype of an Atp7a conditional knock in mouse model tochondrial morphology and distribution abnormalities in fibroblasts from sporadic for X-linked distal hereditary motor neuropathy (dHMNX). Metallomics http://dx. Alzheimer's disease patients. Am. J. Pathol. 173 (2), 470–482. doi.org/10.1039/C6MT00082G. Weedon, M.N., Hastings, R., Caswell, R., Xie, W., Paszkiewicz, K., Antoniadi, T., Williams, Pitceathly, R.D., Murphy, S.M., Cottenie, E., Chalasani, A., Sweeney, M.G., Woodward, C., M., King, C., Greenhalgh, L., Newbury-Ecob, R., Ellard, S., 2011. Exome sequencing Mudanohwo, E.E., Hargreaves, I., Heales, S., Land, J., Holton, J.L., Houlden, H., Blake, identifies a DYNC1H1 mutation in a large pedigree with dominant axonal Charcot- J., Champion, M., Flinter, F., Robb, S.A., Page, R., Rose, M., Palace, J., Crowe, C., Marie-Tooth disease. Am. J. Hum. Genet. 89 (2), 308–312. Longman, C., Lunn, M.P., Rahman, S., Reilly, M.M., Hanna, M.G., 2012. Genetic dysfunc- Whitehouse, S., Randle, P.J., 1973. Activation of pyruvate dehydrogenase in perfused rat tion of MT-ATP6 causes axonal Charcot-Marie-Tooth disease. Neurology 79 (11), heart by dichloroacetate (short communication). Biochem. J. 134 (2), 651–653. 1145–1154. Xu, W.Y., Gu, M.M., Sun, L.H., Guo, W.T., Zhu, H.B., Ma, J.F., Yuan, W.T., Kuang, Y., Ji, B.J., Wu, Reed, L.J., 2001. A trail of research from lipoic acid to alpha-keto acid dehydrogenase com- X.L., Chen, Y., Zhang, H.X., Sun, F.T., Huang, W., Huang, L., Chen, S.D., Wang, Z.G., 2012. plexes. J. Biol. Chem. 276 (42), 38329–38336. A nonsense mutation in DHTKD1 causes Charcot-Marie-Tooth disease type 2 in a Sale, G.J., Randle, P.J., 1981. Occupancy of sites of phosphorylation in inactive rat heart py- large Chinese pedigree. Am. J. Hum. Genet. 91 (6), 1088–1094. ruvate dehydrogenase phosphate in vivo. Biochem. J. 193 (3), 935–946. Yeaman, S.J., Hutcheson, E.T., Roche, T.E., Pettit, F.H., Brown, J.R., Reed, L.J., Watson, D.C., Saporta, M.A., Dang, V., Volfson, D., Zou, B., Xie, X.S., Adebola, A., Liem, R.K., Shy, M., Dimos, Dixon, G.H., 1978. Sites of phosphorylation on pyruvate dehydrogenase from bovine J.T., 2015. Axonal Charcot-Marie-Tooth disease patient-derived motor neurons dem- kidney and heart. Biochemistry 17 (12), 2364–2370. onstrate disease-specific phenotypes including abnormal electrophysiological prop- Zhao, C., Takita, J., Tanaka, Y., Setou, M., Nakagawa, T., Takeda, S., Yang, H.W., Terada, S., erties. Exp. Neurol. 263, 190–199. Nakata, T., Takei, Y., Saito, M., Tsuji, S., Hayashi, Y., Hirokawa, N., 2001. Charcot- Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, Marie-Tooth disease type 2A caused by mutation in a microtubule motor KIF1Bbeta. S., Rueden, C., Saalfeld, S., Schmid, B., Tinevez, J.Y., White, D.J., Hartenstein, V., Cell 105 (5), 587–597. Eliceiri, K., Tomancak, P., Cardona, A., 2012. Fiji: an open-source platform for biolog- Zuchner, S., Mersiyanova, I.V., Muglia, M., Bissar-Tadmouri, N., Rochelle, J., Dadali, E.L., ical-image analysis. Nat. Methods 9 (7), 676–682. Zappia, M., Nelis, E., Patitucci, A., Senderek, J., Parman, Y., Evgrafov, O., Jonghe, P.D., Schmittgen, T.D., Livak, K.J., 2008. Analyzing real-time PCR data by the comparative C(T) Takahashi, Y., Tsuji, S., Pericak-Vance, M.A., Quattrone, A., Battaloglu, E., Polyakov, method. Nat. Protoc. 3 (6), 1101–1108. A.V., Timmerman, V., Schroder, J.M., Vance, J.M., 2004. Mutations in the mitochondrial Stacpoole, P.W., Kerr, D.S., Barnes, C., Bunch, S.T., Carney, P.R., Fennell, E.M., Felitsyn, N.M., GTPase mitofusin 2 cause Charcot-Marie-Tooth neuropathy type 2A. Nat. Genet. 36 Gilmore, R.L., Greer, M., Henderson, G.N., Hutson, A.D., Neiberger, R.E., O'Brien, R.G., (5), 449–451. Perkins, L.A., Quisling, R.G., Shroads, A.L., Shuster, J.J., Silverstein, J.H., Theriaque,