[Skip to Navigation]
Sign In
Case Report/Case Series
December 2013

COX10 Mutations Resulting in Complex Multisystem Mitochondrial Disease That Remains Stable Into Adulthood

Author Affiliations
  • 1Medical Research Council Centre for Neuromuscular Diseases, University College London Institute of Neurology and National Hospital for Neurology and Neurosurgery, London, England
  • 2Department of Clinical Neuroscience, University College London Institute of Neurology, London, England
  • 3Mitochondrial Research Group, Clinical and Molecular Genetics Unit, University College London Institute of Child Health, London, England
  • 4Centre de Génétique Moléculaire, UPR3404, Centre national de la recherche scientifique, Gif-sur-Yvette, France
  • 5Division of Mathematical Biology, National Institute for Medical Research, London, England
  • 6Dubowitz Neuromuscular Centre, University College London Institute of Child Health and Great Ormond Street Hospital for Children NHS Foundation Trust, London, England
  • 7Neurometabolic Unit, National Hospital for Neurology and Neurosurgery, London, England
  • 8Neurogenetics Unit, National Hospital for Neurology and Neurosurgery, London, England
  • 9The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, England
  • 10Department of Clinical Neurophysiology, University College London Institute of Neurology and National Hospital for Neurology and Neurosurgery, London, England
  • 11Department of Clinical Neurophysiology, Norfolk and Norwich University Hospital, Norwich, England
  • 12Division of Neuropathology, University College London Institute of Neurology, London, England
  • 13Department of Molecular Neuroscience, University College London Institute of Neurology, London, England
JAMA Neurol. 2013;70(12):1556-1561. doi:10.1001/jamaneurol.2013.3242

Importance  Isolated cytochrome-c oxidase (COX) deficiency is one of the most frequent respiratory chain defects seen in human mitochondrial disease. Typically, patients present with severe neonatal multisystem disease and have an early fatal outcome. We describe an adult patient with isolated COX deficiency associated with a relatively mild clinical phenotype comprising myopathy; demyelinating neuropathy; premature ovarian failure; short stature; hearing loss; pigmentary maculopathy; and renal tubular dysfunction.

Observations  Whole-exome sequencing detected 1 known pathogenic and 1 novel COX10 mutation: c.1007A>T; p.Asp336Val, previously associated with fatal infantile COX deficiency, and c.1015C>T; p.Arg339Trp. Muscle COX holoenzyme and subassemblies were undetectable on immunoblots of blue-native gels, whereas denaturing gels and immunocytochemistry showed reduced core subunit MTCO1. Heme absorption spectra revealed low heme aa3 compatible with heme A:farnesyltransferase deficiency due to COX10 dysfunction. Both mutations demonstrated respiratory deficiency in yeast, confirming pathogenicity. A COX10 protein model was used to predict the structural consequences of the novel Arg339Trp and all previously reported substitutions.

Conclusions and Relevance  These findings establish that COX10 mutations cause adult mitochondrial disease. Nuclear modifiers, epigenetic phenomenon, and/or environmental factors may influence the disease phenotype caused by reduced COX activity and contribute to the variable clinical severity related to COX10 dysfunction.

Cytochrome-c oxidase (COX) is the final enzyme (complex IV) of the mitochondrial respiratory chain, catalyzing electron transfer from ferrocytochrome c to oxygen. Three subunits (MTCO1-3) form the enzyme’s catalytic core and are encoded by mitochondrial DNA (mtDNA); the remaining 11 subunits (including the subunit formerly considered a constituent of complex I1,2) are nuclear encoded. Cytochrome-c oxidase deficiency has been linked to mutations in all 3 mtDNA-encoded subunits, 3 nuclear-encoded subunits (COX4I2, COX6B1, and COX7B), and 11 nuclear-encoded factors needed for assembly of the enzyme complex (SURF1, SCO1, SCO2, COX10, COX15, LRPPRC, TACO1, FASTKD2, C2orf64/COA5, C12orf62/COX14, and FAM36A/COX20). Affected patients typically present with fatal neonatal/childhood-onset disease.3,4

We now present an adult patient harboring compound heterozygous missense mutations in COX10 associated with a complex multisystem clinical phenotype and isolated COX deficiency. In contrast to most patients with severe COX deficiency, disease progression was limited, and the patient is clinically stable at 37 years of age and working as a physician. We aimed to elucidate the molecular mechanisms underlying this unusual and comparatively mild disease phenotype.

Report of a Case

A 37-year-old European woman was born at term to nonconsanguineous parents. She presented at 12 months of age with poor weight gain with no obvious cause. At 10 years of age, she had a short stature and mild exercise intolerance. By 21 years of age, proximal muscle weakness and fatigue were prominent, and at 25 years of age, renal Fanconi syndrome and metabolic acidosis were detected. She subsequently developed premature ovarian failure. There is no family history of neuromuscular disease.

Currently, exercise tolerance is limited to 180 m (200 yd) with a walking aid. She has a short stature (height, 1.3 m, 25.4 cm [4 ft, 10 in]) and a proximal myopathic gait, with upper and lower limb proximal muscle weakness (Medical Research Council grade 4/5) and absent tendon reflexes, but no other signs of neuropathy. She has an asymptomatic pigmentary maculopathy and mild right-sided sensorineural hearing loss.

A laboratory evaluation revealed an elevated serum lactate level at 34.2 mg/dL (normal level, <15.3 mg/dL [to convert to millimoles per liter, multiply by 0.111]). The results of magnetic resonance imaging of her brain were normal. The results of nerve conduction studies and electromyography were compatible with a demyelinating peripheral neuropathy and proximal myopathy. A sural nerve biopsy showed depletion of myelinated fibers, and the surviving myelinated fibers had inappropriately thin myelin sheaths (Figure 1A). Electron microscopy confirmed frequent demyelination (Figure 1B and C) and membrane-bound filament-containing inclusions associated with the Golgi complex cisternae of myelinated fiber Schwann cells (Figure 1D and E). Teased fibers demonstrated frequent segmental demyelination. Larger mitochondria were noted in a few Schwann cells, but the cristae were of normal density and architecture. A muscle biopsy revealed global reduction in COX histochemical staining (Figure 1F). A spectrophotometric assay demonstrated severely decreased COX activity in skeletal muscle tissue (COX to citrate synthase ratio of 0.004; for controls, the ratios were in the range of 0.014-0.034). However, polarographic studies using tetramethylphenylenediamine ascorbate as a substrate showed normal oxygen uptake of intact muscle mitochondria. The result of genetic analysis of the entire mitochondrial genome was normal.

Figure 1.  Results of Sural Nerve Biopsy, Electron Microscopy, and Muscle Biopsy
Results of Sural Nerve Biopsy, Electron Microscopy, and Muscle Biopsy

A, Sural nerve biopsy specimen. High-magnification view of one fascicle (semithin section of resin stained with methylene blue-azure basic fuchsin) shows moderate depletion of myelinated fibers with several large diameter axonal profiles displaying inappropriately thin myelin sheaths for their diameter, suggesting demyelination (A [arrowheads]). Electron microscopy demonstrates 2 examples of demyelinating axons (B and C) and tightly packed inclusions within a myelinated fiber Schwann cell cytoplasm (D [asterisk]). E, Inspection at higher magnification shows membrane-bound filamentous inclusions (asterisks) closely associated with the Golgi cisternae (arrowheads). F, Muscle biopsy specimen shows decreased COX staining in the majority of fibers. Ragged-red fibers were not present. Scale bars represent 100 μm.

Whole-exome sequencing, performed using the HiSeq platform (IIIumina Inc) (eTable 1 in Supplement for exome metrics and eTable 2 in Supplement for bioinformatic filtering pathway used to identify candidate genes for COX deficiency), revealed compound heterozygous missense mutations in COX10: the known pathogenic, previously reported c.1007A>T; p.Asp336Val mutation,5 and the novel c.1015C>T; p.Arg339Trp mutation (absent from dbSNP132, the 1000 Genomes Project, the UK10K rare disease cohort [823 exomes at the time of the analysis], and the National Heart, Lung, and Blood Institute Exome Sequencing Project database). ExomeDepth (available at http://cran.rproject.org/) was used to exclude copy number variants from the exome data.6 Dosage analysis of chromosome 17p11.2 using multiplex ligation-dependent probe amplification was performed because of the patient’s demyelinating neuropathy. The multiplex ligation-dependent probe amplification kit includes probes to COX10. No deletions or duplications were detected. Segregation analysis confirmed that the clinically unaffected parents were heterozygous carriers for each mutation, and evolutionary conservation studies demonstrated invariance of Asp336 and Arg339 across species.

The structural consequences of both mutations on COX assembly were investigated using blue-native polyacrylamide gel electrophoresis of mitochondrial membrane proteins extracted from muscle. Immunoblot analysis revealed undetectable COX holoenzyme with no abnormal subassemblies (Figure 2A). Western blot of sodium dodecyl sulfate–denaturing polyacrylamide gels and immunocytochemistry demonstrated reduced steady state levels of MTCO1 in muscle and fibroblasts, respectively (Figure 2B and C). COX10 protein levels were normal in muscle. Muscle heme absorption spectra showed reduction in heme aa3 levels with loss of the heme aa3 γ band at 445 nm, confirming impaired incorporation of heme aa3 into MTCO1 due to dysfunctional COX10 (Figure 2D). The functional significance of the mutations was demonstrated by introducing Asp336Val and Arg339Trp into the yeast Saccharomyces cerevisiae. Growth on a nonfermentable carbon source confirmed that each mutation caused respiratory deficiency individually, which was even more pronounced when the mutations coexisted (Figure 3A and B).

Figure 2.  Immunoblot Analysis, Immunocytochemical Micrographs, and Muscle Heme Absorption Spectra
Immunoblot Analysis, Immunocytochemical Micrographs, and Muscle Heme Absorption Spectra

Immunoblot analysis of blue-native (A) and sodium dodecyl sulfate (B) polyacrylamide gels loaded with mitochondrial membrane proteins extracted from muscle. A, The cytochrome-c oxidase (COX) holoenzyme was undetectable in the patient (P), unlike the controls (C1-3). B, The COX10 protein steady state levels were normal in the patient compared with the controls, indicating that the COX10 missense mutations do not impair COX10 protein stability, whereas MTCO1 was undetectable in the patient, unlike the controls. C, Micrographs of a control’s and the patient’s fibroblasts are shown immunocytochemically stained for MTCO1 (green fluorescence) and MitoTracker Red (Life Technologies; red fluorescence), and counterstained with the DNA fluorochrome 4′,6-diamidino-2-phenylindole (blue fluorescence); the patient’s cells show a general decrease in MTCO1 levels compared with the control’s cells. D, Air-oxidized vs sodium dithionite–reduced heme spectra of the patient’s muscle mitochondrial proteins are shown. There is loss of the heme aa3 γ band at 445 nm (D [arrowhead]) in the patient compared with the control. MW indicates molecular weight.

Figure 3.  Growth of Yeast on a Nonfermentable Carbon Source and Hypothetical COX10 Protein Model
Growth of Yeast on a Nonfermentable Carbon Source and Hypothetical COX10 Protein Model

Growth of COX10 p.Asp336Val, COX10 p.Arg339Trp and COX10 Asp336Val plus Arg339Trp yeast strains and a control on glucose (A) and glycerol/ethanol (B) as carbon source. Both COX10 mutations cause respiratory deficiency, as demonstrated by the impaired growth on the nonfermentable carbon source glycerol/ethanol. When both mutations coexisted, the respiratory deficiency appeared to be even more pronounced (A and B). C, A topology diagram of the COX10 protein shows locations of both known and the herein reported mutated COX10 residues highlighted in red; the transmembrane segments are labeled with Roman numerals (I-IX). A structural model of COX10 shows the locations of residues 336, 339, and 225 (D) and residues 196, 204, 225, 336, and 339 looking up from the mitochondrial matrix at the putative face of the bundle (E). COOH indicates the C-terminus (also known as the carboxyl terminus) of the protein; and H2N, the N-terminus (also known as the amino terminus) of the protein.

Because the crystal structure for COX10 is undetermined, we mapped the reported novel Arg339Trp, in addition to the previously reported COX10 amino acid substitutions Thr196Lys, Asp204Lys, Pro225Leu, Asp336Gly, and Asp336Val, to a hypothetical COX10 protein model predicted using computational analysis (Figure 3C-E). The resulting structural consequences were subsequently analyzed to generate hypotheses for potential mechanisms underlying the pathogenicity of these mutations.


The COX10 protein is required to farnesylate protoheme (heme B) to heme O in the heme A biosynthetic pathway.7 Two heme A moieties (a and a3) are essential prosthetic groups of the MTCO1 subunit and are involved in electron transfer.8 To date, 4 infants with COX10 mutations have been reported (Table).5,9,10 None survived beyond 2 years. Although our adult patient shares certain clinical features exhibited by these children, specifically renal tubulopathy and sensorineural hearing loss, her phenotype is considerably milder. We speculate that the difference in clinical severity relates to a nuclear modifier or epigenetic phenomenon that partially compensates for the underlying biochemical defect. It is also possible that our patient has never been exposed to a physiological stress sufficient to trigger a severe metabolic decompensation.

Table.  Clinical and Molecular Data of Published Cases With Genetically Confirmed COX10 Mutations
Clinical and Molecular Data of Published Cases With Genetically Confirmed COX10 Mutations

To study the effect of both mutations on COX assembly, we used blue-native polyacrylamide gel electrophoresis, which showed undetectable COX holoenzyme and subassemblies in muscle. Lack of COX assembly intermediates is consistent with previous reports of infantile COX10 dysfunction and reflects the requirement for COX10 at an early stage of COX assembly.5 Insertion of heme A is thought to occur prior to folding of MTCO1 during its association with subunits COX4 and COX5A. Consequently, COX10 mutations do not cause the accumulation of COX subassemblies typical of SCO1 and SURF1 defects.11 The absence of assembly intermediates is probably related to impaired stability and downregulation of MTCO1, which requires an association with COX4 and COX5A and the insertion of heme A to prevent its rapid degradation by mitochondrial proteases.11,12 Yeast studies provided further support for the pathogenicity of each mutation; both variants caused respiratory deficiency in isolation, and this was further exaggerated when the mutations coexisted.

Our protein model indicates that helices 7 and 9 of COX10 contain 2 highly conserved histidine residues (360 and 418) in close proximity, suggesting that the heme-binding site might be located in this C-terminal region. Substitutions Thr196Lys and Asp204Lys, implicated in severe infantile mitochondrial disease, affect hydrophilic residues in helix 2 that interact with other helices in the bundle. The larger lysine side chain would disrupt these interactions and destabilize the overall protein structure, thereby explaining the deleterious effects of both these variants. The reported Pro225Leu substitution is in a long loop region predicted to localize to the mitochondrial matrix between helices 2 and 3. Thus, this mutation may cause dynamic changes by increasing local flexibility and disrupting interactions with partner proteins. Mutations to residues 336 (Asp to Gly and Val) and 339 (Arg to Trp) are found in a loop region that, based on the positive-inside rule of von Heijne and Gavel,13 would be expected to be intraorganellar. These residues may participate in functional interactions taking place in the matrix. Despite the limitations of this computational model, it is tempting to hypothesize that the Asp336Gly substitution (compound heterozygous with Asp336Val in a pediatric case with death at 4 months) might generate more significant changes to the COX10 protein structure than the Arg339Trp substitution, thereby contributing toward the milder clinical phenotype observed.

The normality of the patient’s oxygen polarographic studies, a technique that has not previously been reported in patients with COX10 mutations, was surprising given the severe histochemical and spectrophotometric reduction in COX activity. We suggest that this finding reflects the capacity of the remaining assembled COX to partially compensate for the underlying biochemical defect. Polarography measures respiratory chain function at physiological levels when electron flux through complex IV is at a controlled/low rate. This scenario mimics the cellular environment in which only suboptimal substrate concentrations are usually available. In contrast, a spectrophotometric assay of respiratory chain function records the optimal enzyme activity under conditions in which there is maximum electron flux. The residual COX activity in our patient could be sufficient to maintain electron transport and proton translocation activities under nonstrenuous conditions, explaining the normal polarographic results. However, during times of physiological stress, it is not possible to upregulate COX activity to meet the higher energy demands required owing to impaired COX10 function. This is reflected by the reduced COX activity during a spectrophotometric assay.

Finally, we speculate whether the filamentous inclusions closely linked to the Golgi complex in the Schwann cells of myelinated axons may result from protein misfolding as a result of an endoplasmic reticulum-Golgi stress response caused by COX deficiency, possibly leading to impaired myelination.14 In conclusion, we have provided some evidence to explain the variable clinical severity seen with COX10 mutations. Furthermore, COX10 genetic analysis should be added to the diagnostic armamentarium for adult mitochondrial disease with COX deficiency. It remains essential that additional unrelated cases harboring the reported mutations are identified to unequivocally prove their pathogenicity and fully understand the extreme variation in clinical severity associated with COX10 mutations reported herein.

Back to top
Article Information

Accepted for Publication: April 23, 2013.

Corresponding Author: Michael G. Hanna, FRCP, Medical Research Council Centre for Neuromuscular Diseases, University College London Institute of Neurology and National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, England (m.hanna@ucl.ac.uk).

Published Online: October 7, 2013. doi:10.1001/jamaneurol.2013.3242.

Author Contributions: Drs Taanman and Rahman contributed equally to the study.

Drs Pitceathly and Hanna had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pitceathly, Taanman, Rahman, Reilly, Hanna.

Acquisition of data: Pitceathly, Taanman, Rahman, Meunier, Sadowski, Cirak, Hargreaves, Land, Nanji, Polke, Woodward, Sweeney, Solanki, Foley, Hurles, Stalker, Blake, Holton, Phadke, Muntoni.

Analysis and interpretation of data: Pitceathly, Taanman, Rahman, Meunier, Sadowski, Cirak, Hargreaves, Land, Nanji, Polke, Woodward, Sweeney, Solanki, Foley, Hurles, Stalker, Blake, Holton, Phadke, Muntoni.

Drafting of the manuscript: Pitceathly.

Critical revision of the manuscript for important intellectual content: Taanman, Rahman, Meunier, Sadowski, Cirak, Hargreaves, Land, Nanji, Polke, Woodward, Sweeney, Solanki, Foley, Hurles, Stalker, Blake, Holton, Phadke, Muntoni, Reilly, Hanna.

Study supervision: Taanman, Rahman, Reilly, Hanna.

Conflict of Interest Disclosures: None reported.

Funding/Support: Funding for the UK10K Consortium was provided by the Wellcome Trust under award WT091310. Dr Pitceathly is funded by the Medical Research Council (grant G0800674). Drs Taanman, Rahman, and Muntoni are supported by the Great Ormond Street Hospital Children’s Charity. Drs Reilly and Hanna are supported by Medical Research Council Centre for Neuromuscular Diseases grant G0601943. Dr Reilly received grant 1U54NS065712-01 from the Medical Research Council and the National Institute of Neurological Disorders and Stroke/Office of Rare Diseases. Drs Holton and Hanna are supported by the Myositis Support Group, and Dr Holton is also supported by the Reta Lila Weston Institute for Neurological Studies. Dr Sadowski is funded by Medical Research Council grant U117581331. Dr Foley is a Muscular Dystrophy Campaign Fellow, and Dr Cirak is supported by the EU FP7 Bio-NMD grant to Dr Muntoni. This study was supported by the UK NHS Specialised Service for Rare Mitochondrial Diseases of Adults and Children and the National Institute for Health Research University College London Hospitals/University College London Comprehensive Biomedical Research Centre, and 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.

Role of the Sponsor: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: The UK10K Consortium members are Saeed Al-Turki, Carl Anderson, Dinu Antony, Inês Barroso, Phil Beales, Jamie Bentham, Shoumo Bhattacharya, Keren Carss, Krishna Chatterjee, Sebahattin Cirak, Catherine Cosgrove, Petr Danecek, Richard Durbin, David Fitzpatrick, Jamie Floyd, A. Reghan Foley, Chris Franklin, Marta Futema, Steve E. Humphries, Matthew E. Hurles, Chris Joyce, Shane McCarthy, Hannah M. Mitchison, Dawn Muddyman, Franceso Muntoni, Stephen O’Rahilly, Alexandros Onoufriadis, Felicity Payne, Vincent Plagnol, Lucy Raymond, David B. Savage, Peter Scambler, Miriam Schmidts, Nadia Schoenmakers, Robert Semple, Eva Serra, Jim Stalker, Margriet van Kogelenberg, Parthiban Vijayarangakannan, Klaudia Walter, Ros Whittall, and Kathy Williamson.

Additional Contributions: We are extremely grateful to the patient and her family for their participation in this study, the UK10K Consortium for making the research possible, and Jean M. Jacobs, PhD, for contributing to the electron microscopy morphometric analysis.

Balsa  E, Marco  R, Perales-Clemente  E,  et al.  NDUFA4 is a subunit of complex IV of the mammalian electron transport chain.  Cell Metab. 2012;16(3):378-386. doi:10.1016/j.cmet.2012.07.015.PubMedGoogle ScholarCrossref
Pitceathly  RDS, Rahman  S, Wedatilake  Y,  et al; UK10K Consortium.  NDUFA4 mutations underlie dysfunction of a cytochrome c oxidase subunit linked to human neurological disease [published correction appears in Cell Rep. 2013;4(2):402].  Cell Rep. 2013;3(6):1795–1805. PubMedGoogle ScholarCrossref
Soto  IC, Fontanesi  F, Liu  J, Barrientos  A.  Biogenesis and assembly of eukaryotic cytochrome c oxidase catalytic core.  Biochim Biophys Acta. 2012;1817(6):883-897.PubMedGoogle ScholarCrossref
Szklarczyk  R, Wanschers  BFJ, Nijtmans  LG,  et al.  A mutation in the FAM36A gene, the human ortholog of COX20, impairs cytochrome c oxidase assembly and is associated with ataxia and muscle hypotonia.  Hum Mol Genet. 2013;22(4):656-667.PubMedGoogle ScholarCrossref
Antonicka  H, Leary  SC, Guercin  GH,  et al.  Mutations in COX10 result in a defect in mitochondrial heme A biosynthesis and account for multiple, early-onset clinical phenotypes associated with isolated COX deficiency.  Hum Mol Genet. 2003;12(20):2693-2702.PubMedGoogle ScholarCrossref
Plagnol  V, Curtis  J, Epstein  M,  et al.  A robust model for read count data in exome sequencing experiments and implications for copy number variant calling.  Bioinformatics. 2012;28(21):2747-2754.PubMedGoogle ScholarCrossref
Moraes  CT, Diaz  F, Barrientos  A.  Defects in the biosynthesis of mitochondrial heme c and heme a in yeast and mammals.  Biochim Biophys Acta. 2004;1659(2-3):153-159.PubMedGoogle ScholarCrossref
Taanman  JW.  Human cytochrome c oxidase: structure, function, and deficiency.  J Bioenerg Biomembr. 1997;29(2):151-163.PubMedGoogle ScholarCrossref
Valnot  I, von Kleist-Retzow  JC, Barrientos  A,  et al.  A mutation in the human heme A:farnesyltransferase gene (COX10 ) causes cytochrome c oxidase deficiency.  Hum Mol Genet. 2000;9(8):1245-1249.PubMedGoogle ScholarCrossref
Coenen  MJH, van den Heuvel  LP, Ugalde  C,  et al.  Cytochrome c oxidase biogenesis in a patient with a mutation in COX10 gene.  Ann Neurol. 2004;56(4):560-564.PubMedGoogle ScholarCrossref
Williams  SL, Valnot  I, Rustin  P, Taanman  J-W.  Cytochrome c oxidase subassemblies in fibroblast cultures from patients carrying mutations in COX10, SCO1, or SURF1.  J Biol Chem. 2004;279(9):7462-7469.PubMedGoogle ScholarCrossref
Arnold  I, Langer  T.  Membrane protein degradation by AAA proteases in mitochondria.  Biochim Biophys Acta. 2002;1592(1):89-96.PubMedGoogle ScholarCrossref
von Heijne  G, Gavel  Y.  Topogenic signals in integral membrane proteins.  Eur J Biochem. 1988;174(4):671-678.PubMedGoogle ScholarCrossref
Bravo  R, Vicencio  JM, Parra  V,  et al.  Increased ER-mitochondrial coupling promotes mitochondrial respiration and bioenergetics during early phases of ER stress.  J Cell Sci. 2011;124(pt 13):2143-2152.PubMedGoogle ScholarCrossref