Article Text

PDF

Survival transcriptome in the coenzyme Q10 deficiency syndrome is acquired by epigenetic modifications: a modelling study for human coenzyme Q10 deficiencies
  1. Daniel J M Fernández-Ayala1,2,
  2. Ignacio Guerra1,2,
  3. Sandra Jiménez-Gancedo1,2,
  4. Maria V Cascajo1,2,
  5. Angela Gavilán1,2,
  6. Salvatore DiMauro3,
  7. Michio Hirano3,
  8. Paz Briones2,4,
  9. Rafael Artuch2,5,
  10. Rafael De Cabo6,
  11. Leonardo Salviati7,
  12. Plácido Navas1,2
  1. 1Centro Andaluz de Biología del Desarrollo (CABD-CSIC), Universidad Pablo Olavide, Seville, Spain
  2. 2CIBERER, Instituto de Salud Carlos III, Seville, Spain
  3. 3Department of Neurology, Columbia University Medical Center, New York,  USA
  4. 4Instituto de Bioquímica Clínica, Corporació Sanitaria Clínic, Barcelona, Spain
  5. 5Department of Clinical Biochemistry, Hospital Sant Joan de Déu, Barcelona, Spain
  6. 6Laboratory of Experimental Gerontology, National Institute on Aging, NIH, Baltimore, USA
  7. 7Clinical Genetics Unit, Department of Woman and Child Health, University of Padova, Padova, Italy
  1. Correspondence to Dr Plácido Navas and Leonardo Salviati, pnavas{at}upo.es and Leonardo.salviati{at}unipd.it

Abstract

Objectives Coenzyme Q10 (CoQ10) deficiency syndrome is a rare condition that causes mitochondrial dysfunction and includes a variety of clinical presentations as encephalomyopathy, ataxia and renal failure. First, we sought to set up what all have in common, and then investigate why CoQ10 supplementation reverses the bioenergetics alterations in cultured cells but not all the cellular phenotypes.

Design Modelling study This work models the transcriptome of human CoQ10 deficiency syndrome in primary fibroblast from patients and study the genetic response to CoQ10 treatment in these cells.

Setting Four hospitals and medical centres from Spain, Italy and the USA, and two research laboratories from Spain and the USA.

Participants Primary cells were collected from patients in the above centres.

Measurements We characterised by microarray analysis the expression profile of fibroblasts from seven CoQ10-deficient patients (three had primary deficiency and four had a secondary form) and aged-matched controls, before and after CoQ10 supplementation. Results were validated by Q-RT-PCR. The profile of DNA (CpG) methylation was evaluated for a subset of gene with displayed altered expression.

Results CoQ10-deficient fibroblasts (independently from the aetiology) showed a common transcriptomic profile that promotes cell survival by activating cell cycle and growth, cell stress responses and inhibiting cell death and immune responses. Energy production was supported mainly by glycolysis while CoQ10 supplementation restored oxidative phosphorylation. Expression of genes involved in cell death pathways was partially restored by treatment, while genes involved in differentiation, cell cycle and growth were not affected. Stably demethylated genes were unaffected by treatment whereas we observed restored gene expression in either non-methylated genes or those with an unchanged methylation pattern.

Conclusions CoQ10 deficiency induces a specific transcriptomic profile that promotes cell survival, which is only partially rescued by CoQ10 supplementation.

Statistics from Altmetric.com

Article summary

Article focus

  • To analyse the common gene expression profile in primary cell cultures of dermal fibroblasts from patients suffering any of the clinical presentation of the human syndrome of coenzyme Q10 (CoQ10) deficiency (primary or secondary CoQ10 deficiency).

  • To determine why CoQ10 treatment, the current therapy for all forms of CoQ10 deficiency, restored respiration but not all the clinical phenotypes.

  • To investigate the stable genetic cause responsible for the survival adaptation to mitochondrial dysfunction owing to CoQ10 deficiency.

Key messages

  • The mitochondrial dysfunction owing to CoQ10 deficiency induces a stable survival adaptation of somatic cells in patients at early or postnatal development by epigenetic modifications of chromatin. Deficient cells unable to maintain this survival state during differentiation would die contributing to the pathological phenotype.

  • Supplementation with CoQ10 restores respiration through enhanced sugar rather than lipid metabolism; partially restores stress response, immunity, cell death and apoptotic pathways; and does not affect cell cycle, cell growth, and differentiation and development pathways.

  • Survival transcriptome in the CoQ10 deficiency syndrome is acquired by epigenetic modifications of DNA: DNA-demethylated genes corresponded to unaffected genes by CoQ10 treatment, whereas those with unchanged DNA-methylation pattern corresponded to genes with responsive expression to CoQ10 supplementation. These results would approach to explain the incomplete recovery of clinical symptoms after CoQ10 treatment, at least in some patients.

Strengths and limitations of this study

  • Human CoQ10 deficiencies are considered rare diseases with low prevalence, which limits the sample size.

  • The genetic heterogeneity of this disease is owing to mutations in any of the 11 genes directly involved in the synthesis of CoQ10 inside mitochondria, or other mutations altering somehow the mitochondria and its metabolism, affecting their inner CoQ10 synthesis as a side effect, will course with CoQ10 deficiency.

  • Among this genetic heterogeneity, all cells showed a common transcriptomic profile that justified their pathological phenotype, responded equally to CoQ10 treatment and presented the same DNA methylation pattern.

Introduction

Coenzyme Q10 (CoQ10) is a small electron carrier which is an essential cofactor for several mitochondrial biochemical pathways such as oxidative phosphorylation, β-oxidation and pyrimidine nucleotide biosynthesis. CoQ10 biosynthesis depends on a multienzyme complex1 that involves at least 11 proteins encoded by COQ genes. Mutations in any of these genes cause primary CoQ10 deficiencies, which are clinically heterogeneous mitochondrial diseases.2 Clinical presentations include encephalomyopathy with lipid storage myopathy and myoglobinuria,3 ataxia and cerebellar atrophy,4 severe infantile encephalomyopathy with renal failure,5 isolated myopathy,6 and nephrotic syndrome.7 Secondary CoQ10 deficiency has also been associated with diverse mitochondrial diseases.8–13 In all of these conditions, CoQ10 supplementation partially improves symptoms14 ,15 and usually induces a return to normal growth and respiration in CoQ10-deficient fibroblasts.8 ,16 ,17 Adaptation of somatic cells to CoQ10 deficiency may affect both onset and course of the disease. We document common transcriptomic profile alterations in somatic cells of CoQ-deficient patients, their response to CoQ10 supplementation, and the relationship with the DNA methylation status of specific genes.

Materials and methods

Cells

Primary skin fibroblasts from CoQ10-deficient patients and from aged-matched controls, at similar culture passage, were cultured at 37°C using Dulbecco's Modified Eagle Medium (DMEM) 1 g/l glucose, l-glutamine and pyruvate (Invitrogen, Prat de Llobregat, Barcelona) supplemented with an antibiotic/antimycotic solution (Sigma Chemical Co, St Louis, Missouri) and 20% fetal bovine serum (FBS, Linus). When required, CoQ10 prediluted in FBS was added to the plates at a final concentration of 30 µM (CoQ10, Synthetic Minimum 98%, high-performance liquid chromatography, Sigma). We studied five patients with primary CoQ10 deficiency: two siblings harboured a homozygous p.Y297C mutation in the COQ2 gene,5 other with a pathogenic mutation (c.483G>C) in the COQ4 gene (this paper), and another one with haploinsufficiency of COQ4.18 Patients with secondary CoQ10 deficiency included: a mitochondrial encephalopathy, lactic acidosis and stroke-like episodes patient harbouring the m.3243A>G in the mitochondrial tRNALeu(UUR) with 43% heteroplasmy level,8 a patient with mtDNA depletion syndrome12 and a third patient with ataxia of unknown origin.4 Table 1 summarises the clinical phenotype and biochemical studies of these patients.

Table 1

Clinical phenotype and biochemical studies performed in patients with coenzyme Q10 deficiency

Transcriptome analysis

RNA extraction, probe synthesis and hybridisation with two independent expression arrays (GeneChip Human Genome U133 Plus 2.0 and GeneChip Human Gene 1.0 ST, Affymetrix) were used as described.19 Gene expression was validated by the MyiQ Single Color Real Time PCR Detection System (Biorad). See supplementary methods for full description.

Data had been deposited with the NCBI-GEO database, at http://www.ncbi.nlm.nih.gov/geo/, accession number GSE33941 (this SuperSeries is composed of two subset Series, see online supplementary table S7 for an explanation).

Statistical analyses were performed comparing each signal of patient's fibroblasts RNA with the corresponding signal of control RNA by two different approaches. The main statistical analysis for both GeneChip Human Genome U133 Plus 2.0 Array and GeneChip Human Gene 1.0 ST Array was achieved as previously described,19 which selects the most significant genes commonly and equally regulated in all samples using very stringent parameters. In a few special cases, other unselected but regulated genes were studied because of their role in specific processes and pathways. They were equally described in table 2. The second statistical analysis approach for the Gene Ontology (GO) study was performed as previously described20 and analyses the most altered biological processes and pathways using a lower stringency analysis, which permits to select the hundred most altered GOs in different functional categories (see online supplementary table S4) and the hundred more distorted pathways (see online supplementary table S5) that had been regulated in CoQ10-deficient cells. GO regulated in both independent analysis of primary and secondary CoQ10 deficiencies (see online supplementary table 3), and those regulated by CoQ10 supplementation (see online supplementary table S9) were studied using the GORILLA software (Gene Ontology enrichment analysis and visualisation tool), at http://cbl-gorilla.cs.technion.ac.il/.21 Full description of statistical analysis be found in the supplementary material.

Table 2

Differentially expressed genes in coenzyme Q10 deficiency

Epigenetic analysis

DNA (CpG) methylation analysis was performed using a base-specific cleavage reaction with bisulfite combined with mass spectrometric analysis (MassCLEAVE). For the statistical analysis, the CpGs’ methylation degree for each gene was analysed with the MultiExperiment Viewer software developed by Saeed.22 See supplementary methods for full description.

Results

Transcriptome analysis

We studied skin fibroblasts from four patients with primary CoQ10 deficiency and three patients with secondary CoQ10 deficiency (table 1). We analysed the transcriptomic profiles and compared them with those of cells from age-matched control individuals, and evaluated the modifications induced by supplementation with 30 μM CoQ10 for 1 week to allow recovery of ATP levels.8 ,16 ,17 A very stringent analysis selected the most significant genes displaying a common and equally altered expression in all samples (summarised in table 2 and shown with full details in online supplementary table S1). Other genes unselected by this analysis, but still abnormally expressed were also included in the study because of their role in specific processes and pathways, such as NADH mobilisation, cell cycle and immunity (see online supplementary table S1) and energetic metabolism (see online supplementary table S2). GO classification of these genes showed similar profiles when comparing independently primary-deficient and secondary-deficient fibroblasts (see online supplementary table S3). A lower stringency analysis showing the most altered biological processes and pathways selected 100 most altered GO in different functional categories (see online supplementary table S4) and 100 more distorted pathways (see online supplementary table S5) in CoQ10-deficient cells. See supplementary data for description of statistical analyses.

CoQ10 treatment modified the specific transcriptomic profile displayed by CoQ10-deficient fibroblasts (see online supplementary tables S6 and S7). We classified genes into five groups according to the consequence of CoQ10 treatment on gene expression (see online supplementary table S8 for a graphical view). About 54% of probes with altered expression were unaffected by CoQ10 supplementation. Only 36% of probes showed partial or complete normalisation of expression and 2% showed inverse regulation (figure 1). Approximately 5% of probes were specifically altered after treatment in both deficient and non-deficient cells and 3% showed small or non-specific changes (these were not considered for further analysis). After statistical analysis, we obtained 70 altered GO with a significant p value (<0.001) and an enrichment value that represents the most altered GO within each group (see online supplementary table S9).

Figure 1

Cluster of genes differentially expressed in coenzyme Q10 (CoQ10)-deficiency and after CoQ10 supplementation. Four arrays of two representative fibroblasts from patients with Q deficiency were plotted with two arrays of control fibroblasts and nine arrays of five patient's fibroblasts with CoQ10 deficiency treated with 30 µM CoQ10. Activated genes were coloured in red and repressed ones in green. Between parentheses—group classification of genes after CoQ10 supplementation (see online supplementary Table S3).

Data have been deposited with the NCBI-GEO database, at http://www.ncbi.nlm.nih.gov/geo/, accession number GSE33941 (see online supplementary table S10 for an explanation). The functional description of each gene was updated from the GeneCard of The Human Gene Compendium (Weizmann Institute of Science), http://www.genecards.org/. See supplementary data for a full description of genes, biological process and pathways regulated in CoQ10 deficiency.

CoQ10-deficient fibroblasts readapt the energetic metabolism and CoQ10-treatment restores

In CoQ10-deficient fibroblasts, mitochondrial functions, including respiratory chain and tricarboxylic acid (TCA) cycle, were repressed, whereas 9 of 10 steps in glycolysis and pyruvate metabolism were activated, including lactate and pyruvate dehydrogenases (see online supplementary tables S2 and S4). Accordingly, genes involved in the negative regulation of glycolysis were downregulated, whereas those involved in its activation were upregulated (see online supplementary table S2). Furthermore, genes involved in cytosolic NADH oxidation (cytochrome b5 and several oxidoreductases) were slightly repressed (table 2). The expression of genes involved in cholesterol and fatty acid metabolism was downregulated (table 2), as well all the GO related with lipid metabolism (see online supplementary table S5).

CoQ10 supplementation normalised the expression (either partially or completely) of genes involved in the glycolytic pathway and activated the expression of repressed respiratory chain genes, whereas the TCA cycle remained unaffected (see online supplementary table S2). Most of the repressed enzymes of lipid metabolism and fatty acid β-oxidation remained downregulated (table 2), whereas several other pathways, such as monocarboxylic acid transport and the insulin response, were normalised (see online supplementary table S9). These results are in agreement with the recovery of aerobic metabolism observed in CoQ10-deficient fibroblasts after CoQ10 supplementation.8 ,16 ,17

CoQ10 deficiency induces specific adaptations of cells to promote survival

The major novel finding of transcriptome profiling in CoQ10-deficient fibroblasts was the altered expression of genes concerned with cell cycle and development and with resistance to stress and cell death (table 2). This suggests both a remodelling of differentiation and growth maintenance and an increase of cell survival mechanisms. Specifically, genes involved in cell cycle activation and maintenance were upregulated, and genes involved in cell cycle regulation increased or decreased their expression depending of their activating or repressing roles. This proliferative response was also enhanced by the repression of cellular attachment factors and by the activation of extracellular matrix proteins that reduce cell attachment and favour cell division. In parallel, GO clusters favouring cell cycle and cell division were activated, and those inhibiting cell growth were repressed (see online supplementary table S4). The differentiation of these cells was compromised because many required factors, transducers, antigens and structural proteins appeared downregulated, whereas repressors of differentiation during development were overexpressed (table 2). See supplementary material for a full description of genes, biological processes and related pathways.

Cell cycle activation was supported by the upregulation of CDK6 (table 2), a cyclin-dependent kinase that induces entry into the S-phase, and by a robust repression (more than ninefold) of p21/CDKN1A, an inhibitor of cyclin-dependent kinase that blocks cell cycle at the G1/S check point to stimulate cell differentiation. Moreover, subsequent pathways inactivated by p2123 were enhanced in CoQ10-deficient cells (see online supplementary table S5), as well as both transcription factors E2F7 and E2F8 (table 2), which push the progression of the cell cycle, activate cell survival and inhibit apoptosis.24

Cell survival in CoQ10-deficient cells was improved by the induction of DNA-repairing mechanisms, and by the establishment of pathways that regulate Jun kinases and activate NAD(P)H-CoQ oxidoreductase, which are involved in stress responses (table 2 and see online supplementary table S4). Components of apoptosis and cell death pathways were systematically repressed (table 2), including tumour suppressor genes, antigens, intracellular mediators and effectors of cell death. Also, cell surface receptors and modulators that inhibit apoptosis were greatly activated.

Interestingly, CoQ10 treatment did not alter the newly acquired resistance to cell death in CoQ10-deficient fibroblasts, kept cell growth activated, and allowed a higher degree of differentiation (tables 2 and see online supplementary table S8). However, genes controlling stress resistance pathways and cortical cytoskeleton were completely restored, as indicated by the shifts in gene expression listed in table 2. However, treated fibroblasts kept the DNA repair mechanism activated.

Signalling-related genes and pathways were differentially affected by CoQ10 deficiency, but most of immunity-related genes showed a general downregulation (table 2). Pathways and biological processes involved in immunity regulation were restored by CoQ10 supplementation (table 2 and see online supplementary table S5).

Stable DNA methylation profile is responsible for the specific gene expression profile in CoQ10 deficiency

CoQ10 supplementation modified the expression of 43% of genes that were abnormally expressed in CoQ10-deficient fibroblasts (see online supplementary table S8). In the majority of these cases, expression levels were restored to those of control fibroblasts (20%), but few showed inverse regulation (2%) and others were specifically altered after CoQ10 treatment in both deficient and non-deficient cells (5%). The remaining 16% corresponded to partially restored genes, which slightly alter their expression level without changing the CoQ10-deficient pattern. These genes along with the unaffected (54%) constitute 72% of regulated genes in CoQ10 deficiency, which were not significantly altered after CoQ10 supplementation.

To explain this differential response to respiratory dysfunction, we analysed the DNA-methylation profile of 20 among the most altered genes listed in table 2. These genes encompass the main biological processes and pathways affected by CoQ10 deficiency (table 3). Upregulated genes, which were unaffected by CoQ10 supplementation, had less-defined DNA methylation sites in their promoter regions.

Table 3

Epigenetic modifications in CoQ10 deficiency owing to DNA (CpG) methylation/demethylation

Genes with partial restoration of their expression after CoQ10 supplementation showed precise methylation and demethylation profiles that may explain their altered expression during CoQ10 deficiency. The methylation degree of these genes changed after treatment, and may be responsible for the modulation of expression (table 3). The patterns of methylation of activated and repressed genes in CoQ10 deficiency that could be normalised by CoQ10 supplementation, were either unaffected or only slightly affected by the treatment, and we did not detect new methylation sites after CoQ10 supplementation.

However, a few genes showed significant differences in the methylation degree after the treatment, which correspond to the partially restored genes that maintain the specific expression pattern of untreated CoQ10 deficient cells at a lower level.

Finally, reviewing the biological processes and molecular functions of regulated genes in CoQ10 deficiency, the main adaptation for cell survival activated genes by DNA demethylation, which increased the expression of genes involved in cell cycle activation, apoptosis inhibition, and cell stress resistance, meanwhile the undifferentiated state could be owing to gene repression by DNA methylation, which decreased the expression of genes involved in cell differentiation. CoQ10 treatment did not alter the methylation degree of these genes and subsequently the expression level was maintained.

Discussion

CoQ10-deficient fibroblasts readapt the energetic metabolism and CoQ10-treatment restores

CoQ10 is an essential component of the mitochondrial respiratory chain,1 therefore dysfunctional mitochondria are a common finding in both primary CoQ10 deficiencies3–7 and secondary forms.8–13 Although each form presents a specific clinical phenotype, all these conditions display a substantial reduction of cellular CoQ10 content and deficit in the mitochondrial enzymatic activities of respiratory chain (table 1). Accordingly to these results, we have shown here that fibroblasts from patients with CoQ10 deficiency have reorganised their genetic resources to cope with this mitochondrial dysfunction. Consistent with the role of CoQ10 in bioenergetics, the lack of CoQ10 would force the cell to support it mainly by glycolysis, whereas both mitochondrial lipid metabolism and respiratory chain were repressed (see online supplementary tables S2 and S4). These findings, together with the mild repression of cytosolic enzymes that oxidise NADH (cytochrome b5 and its oxidoreductases listed in table 2) could indicate that NADH is mainly used for biosynthetic purposes rather than for energy production.

Supplementation with CoQ10, the current therapy for all forms of CoQ10 deficiency, restored respiration through enhanced sugar utilisation, but did not stimulate lipid metabolism. The expression of genes involved in the glycolytic pathway was partially or completely normalised after CoQ10 treatment, whereas the repressed genes involved in the respiratory chain were activated. The TCA cycle remained unaffected (see online supplementary table S2). These results are in agreement with the recovery of aerobic metabolism observed in CoQ10-deficient fibroblasts after CoQ10 supplementation.8 ,16 ,17

CoQ10 deficiency induces a stable survival adaptation of cells

CoQ10-deficient fibroblasts adapted several physiological processes to acquire a cellular-resistance state for survival under the conditions of mitochondrial dysfunction induced by CoQ10 deficiency. The new genetic pattern increases cell survival by activating cell cycle and growth, maintaining an undifferentiated phenotype, upregulating stress-induced proteins and inhibiting apoptosis and cell death pathways. These results recapitulate a survival network that can be observed in nutritional stress such as when cells are grown in galactose-enriched media.25

The survival adaptation shown by CoQ10-deficient cells included a global resistance mechanism that is observed also during the initial phase of tumorigenesis. In fact, the CoQ10-deficient expression profile was very similar to that described during myeloid cell transformation26 and breast tumours.27 Moreover, some of the regulated genes in CoQ10 deficiency (listed in table 2 as italicised letter) are used as biomarkers in several types of cancer,28 like KRT34, the cell cycle-related POSTN, MCAM, EFEMP1 and VCAN, and the apoptotic and cell resistance-related CYP1B1, XAF1 and TNFRSF10D. Although these biomarkers behaved in CoQ10 deficiency (increased or decreased) as described by Yoo et al,28 there is no sign of tumour formation reported in the patients so far. In addition, cellular senescence, a defining feature of premalignant tumours,29 is characterised by a gene expression pattern similar to that of CoQ10-deficient fibroblasts (see online supplementary table S5).

Supplementation with CoQ10 enhanced both stress response and immunity pathways. Although the pathway of cell death was partially restored, cell cycle and growth, and the mechanisms to prevent differentiation and development were not. These results indicate that the mitochondrial dysfunction owing to CoQ10 deficiency induces a stable survival adaptation of somatic cells in patients at early or postnatal development, and we speculate that cells unable to institute, or to maintain, this survival mechanism during differentiation will die, contributing to the pathological phenotype.

A stable DNA methylation profile is responsible of specific gene expression in CoQ10 deficiency

The cellular adaptation to CoQ10 deficiency-enhanced DNA demethylation of genes that regulate cell cycle activation, apoptosis inhibition and cell stress resistance as part of an adaptation survival mechanism. Comparable results were observed in several models of epigenetic regulation by demethylation (see online supplementary table S5), whereas DNA methylation inhibits activation of genes related to tumorogenesis and apoptosis.30

Pathways unaffected by CoQ10 treatment corresponded to stably demethylated genes, whereas those that responded to CoQ10 supplementation were controlled by genes with unchanged methylation patterns.

We did not find changes in the methylation degree of all genes affected by CoQ10 deficiency, suggesting that other modalities of gene regulation are responsible, including epigenetic mechanisms such as histone modifications by methylation and acetylation, or even DNA methylation in CpG islands other than those studied here. Interestingly, it has been reported that CoQ10 regulates lipid metabolism in mice liver without any effect on the DNA methylation profile,31 indicating that supplemented CoQ10 by itself may not alter the DNA methylation pattern that cells acquired during the survival adaptation to CoQ10 deficiency.

Mechanisms unaffected by therapy corresponded to stably DNA demethylated genes, which were responsible for the acquisition of the undifferentiated state for survival and resistance that cells obtain during the adaptation to CoQ10 deficiency, whereas the responsive to CoQ10 supplementation were controlled by genes with unchanged methylation patterns and correspond mainly to metabolic genes and those related with the restoration of mitochondrial function.

We propose that these epigenetic changes may be established as early as during the fetal life32 in order to cope with CoQ10 deficiency; these cells then maintain this adaptive response throughout their life. We speculate that cells unable to maintain this survival mechanism during differentiation would die contributing to the pathological phenotype.

Our model has some limits: we treated cells only for 1 week and in principle we cannot rule out that prolonged exposure to CoQ10 could restore also some of the other unaffected pathways. Alternatively, incomplete recovery of the gene expression profiles could be explained by the fact that exogenous CoQ10 can rescue the bioenergetic defect, but not all other functions of CoQ10 in these cells, as it has been observed in other organisms.33

Acknowledgments

We appreciate the generous cooperation of the families of the patients.

References

View Abstract
  • Supplementary Data

    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:

Footnotes

  • Senior authorship is shared by LS and PN.

  • Contributors DJMFA and PN designed the experiments and drafted the manuscript; DJMFA, IG, SJG, MVC and AG performed the experiments and performed the statistical analyses; PB, RA and LS provided cells and patient's clinical information; DJMFA, PB, RA, PN, SDM, MH, RDC and LS analysed the data and edited the manuscript. All authors read and approved the final manuscript.

  • Funding This work was supported by the following funders: Spanish Ministerio de Sanidad (FIS) grant numbers PI11/00078 and PI11/02350; National Institutes of Health (NIH, USA) grant number 1R01HD057543; Junta de Andalucía (Spanish Regional Government) grant number CTS-3988; Telethon (Italy) grant number GGP09207; and from a grant from Fondazione CARIPARO.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No additional data are available.

Request permissions

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.