Mutations causing mitochondrial disease: What is new and what challenges remain?

See allHide authors and affiliations

Science  25 Sep 2015:
Vol. 349, Issue 6255, pp. 1494-1499
DOI: 10.1126/science.aac7516


Mitochondrial diseases are among the most common and most complex of all inherited genetic diseases. The involvement of both the mitochondrial and nuclear genome presents unique challenges, but despite this there have been some remarkable advances in our knowledge of mitochondrial diseases over the past few years. A greater understanding of mitochondrial genetics has led to improved diagnosis as well as novel ways to prevent transmission of severe mitochondrial disease. These and other advances have had a major impact on patient care, but considerable challenges remain, particularly in the areas of therapies for those patients manifesting clinical symptoms associated with mitochondrial dysfunction and the tissue specificity seen in many mitochondrial disorders. This review highlights some important recent advances in mitochondrial disease but also stresses the areas where progress is essential.

Mitochondrial diseases are a common group of genetic human disorders, which we define here as those leading to a primary defect in mitochondrial oxidative phosphorylation (OXPHOS), the main source of cellular adenosine triphosphate (ATP). The mitochondrial electron transport chain, which is required for human life, is composed of four multisubunit complexes (CI to CIV) and two mobile electron carriers (ubiquinone and cytochrome c). This system produces a transmembrane proton gradient that is harnessed by the protein complex known as complex V (FoF1 ATP synthase) to synthesize ATP, a usable energy source for the cell. Freely moving respiratory complexes and mobile carriers coexist in the inner mitochondrial membrane with larger structures called respiratory supercomplexes (1).

OXPHOS proteins are uniquely under the dual genetic control of the mitochondrial and nuclear genomes. The circular mitochondrial genome (mtDNA) consists of only 16,569 base pairs (2) but is present in multiple copies in all cells. MtDNA encodes only 37 gene products, of which 13 are polypeptides that are structural OXPHOS subunits, plus 22 transfer RNAs (tRNAs) and two ribosomal RNAs (rRNAs) required for their synthesis. The remaining mitochondrial proteins including the majority of OXPHOS subunits, the assembly and ancillary factors of the OXPHOS complexes, and those involved in maintenance and expression of mtDNA, intraorganellar protein synthesis, and mitochondrial dynamics are nuclear-encoded, synthesized in the cytosol, and imported into mitochondria.

Clinical features and prevalence of mitochondrial disease

One of the great challenges of mitochondrial disease remains the marked variation in clinical features in patients, involving several different organs and leading to multisystem presentations. Mitochondrial diseases can arise throughout each decade of life. Patients presenting in childhood often have severe and progressive disease due to recessively inherited nuclear gene disorders (3). Clinical syndromes include Leigh syndrome and Alpers syndrome, with prominent involvement of the central nervous system; however, some patients may present with cardiac, skeletal muscle, or other organ involvement reflecting genetic heterogeneity. In adult-onset disease, mtDNA mutations predominate, although Mendelian disorders caused by autosomal dominant mutations of gene products, such as components of the mtDNA replication machinery [e.g., DNA polymerase gamma (POLG) and Twinkle helicase (PEO1)] that usually lead to severe autosomal recessive disease in children, may manifest later in life. As with pediatric presentations, there are commonly recognized clinical phenotypes in adult-onset mitochondrial disease including chronic progressive external ophthalmoplegia, subacute blindness associated with Leber hereditary optic neuropathy (LHON), MELAS (mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes), and MERRF (myoclonic epilepsy and ragged red fibers). However, many patients do not fit clearly defined syndromes, and this is perhaps especially true for those with the common m.3243A>G mutation (4), present in almost one-third of adult patients with mitochondrial disease, often delaying patient diagnosis.

Another key goal in mitochondrial disease is to understand the tissue specificity associated with specific mitochondrial genotypes. A good example of this is the frequent isolated involvement of the optic nerve in patients with primary LHON mutations (5). It is presumed that the retinal ganglion cells are particularly susceptible to the effects of these complex I mutations, but it is of interest that the same mutation that causes LHON can also lead to a devastating severe dystonia with no obvious eye involvement in other patients (6). This is also exemplified by the remarkable phenotypic differences observed in patients with mutations of mt-aminoacyl tRNA synthetases, particularly given these are ubiquitous enzymes (see below). Such selective vulnerability, however, is not unique to mitochondrial disease but is seen with many neurological disorders. Progress in this area is slow because of the difficulties in modeling human mitochondrial diseases. At present, modifying the mitochondrial genome in a specific way is difficult (see below), and many different animal models of nuclear genetic defects do not replicate the phenotype in humans (7, 8). Clearly, mtDNA heteroplasmy cannot explain the remarkable tissue variability in these disorders, but it is interesting to note that the mitochondrial transcriptome also harbors substantial heteroplasmy in unaffected individuals (9); little is currently known about what may cause such RNA heteroplasmy, but it is another way in which the composition of OXPHOS components could vary within an individual, tissue, or cell.

Extensive clinical and genetic heterogeneity means that the exact prevalence of mitochondrial disease is difficult to establish. Published studies looking at the prevalence in children suggest a minimum birth prevalence of 6.2/100,000, although the frequency is much higher in some consanguineous communities because of the high incidence of autosomal recessive disease (10). Adult mitochondrial disease, which includes pathogenic mutations of both the mitochondrial and nuclear genomes, is estimated to be approximately 1 in 4300 of the population affected or at risk of developing mitochondrial disease, with mtDNA mutations identified in more than 75% of total clinically affected adults (11). Although classification of mitochondrial diseases is complex, we present a simplified version in Table 1.

Table 1 A simplified genetic classification of mitochondrial diseases.

Shown are examples of mutations (mtDNA) or disease genes (nuclear disorders).

View this table:

Mitochondrial DNA genetic disorders

A major advance for families with mtDNA mutations has been the increase in reproductive options available and a realistic possibility that the transmission of these diseases can be prevented. Such reproductive techniques to prevent the transmission of mitochondrial disease are important because of the lack of curative treatment and the progressive nature of such diseases. For nuclear mutations that cause mitochondrial disease, similar options are available as for other nuclear genetic conditions, including prenatal and pre-implantation genetic diagnosis. For families with pathogenic mtDNA mutations, the challenges are greater, reflecting the different genetic mechanisms and unique maternal transmission of mtDNA (12). MtDNA mutations may be homoplasmic (with all mtDNA copies mutated), as observed in the majority of patients with LHON; more commonly they are heteroplasmic, with a mixture of mutated and wild-type mtDNA present within the individual and in cells. A cellular threshold level of mutated mtDNA is believed to be a requirement for a manifest biochemical defect (13) (almost certainly due to the absolute amount of wild-type mtDNA). The same is true for clinical defects, with the risk of developing severe disease correlated with the level of heteroplasmy.

Not all mtDNA mutations behave similarly in terms of the distribution of heteroplasmy, which is crucial for our understanding of disease pathogenesis, diagnosis, and inheritance. MtDNA is maternally inherited (12). For heteroplasmic mutations, this inheritance pattern is complicated by a genetic bottleneck due to relatively few mtDNA copies in individual cells of developing embryos and relaxed mtDNA replication (14, 15). For heteroplasmic mutations, such as m.8344A>G and m.3243A>G, the genetic bottleneck results in variable levels of mtDNA mutation in mature oocytes. However, single, sporadic large-scale mtDNA deletions can occur, although the risks of transmission to offspring are small.

It is estimated that the average number of births per year from women at risk of transmitting serious mtDNA disease is 152 in the United Kingdom and 778 in the United States (16). For these women, expert genetic counseling is important to inform them of the specific risks associated with their particular mtDNA mutation and to discuss their reproductive options (17). These include adoption or ovum donation, but there is a limitation of donor oocytes and many women want their own genetically related children. After counseling, some families prefer to conceive naturally, albeit with some risk of having an affected child. Some at-risk mothers may consider prenatal testing at the chorionic villus stage or amniocentesis to determine the level of mtDNA mutation in the developing fetus. Preimplantation genetic diagnosis, determined on the basis of analysis of either one or two blastomeres of a day-3 embryo, is increasingly used to reduce the risk of transmitting some mtDNA mutations and has become an option for some women with specific mtDNA mutations (18, 19). Furthermore, TALENS and other potential genetic editing techniques (see below) may be able to reduce the level of mtDNA mutation in individual oocytes (20). The latter is an attractive approach, but in cases with high levels of heteroplasmy, the total mtDNA copy number may be lowered to levels that prevent implantation, and these techniques would not be amenable for homoplasmic mtDNA mutations.

For women with high levels of heteroplasmic mutations or those with homoplasmic mutations, the present options are in reality limited to ovum donation or adoption, although mitochondrial donation may soon become a viable option. Mitochondrial donation is an in vitro fertilization (IVF) technique that involves the transfer of the nuclear genes from an oocyte or zygote at the metaphase II spindle, polar body, or pronuclear stage (2124) into an enucleated donor oocyte or zygote (Fig. 1). These techniques use an established method (25) successfully applied both to animal models (26) and to human oocytes with minimal carryover of mtDNA (22, 24, 27). These techniques could prevent the transmission of mtDNA disease, but they raise ethical issues (28) and may not be universally available. In the United Kingdom, after extensive ethical deliberations, public consultation, and independent scientific scrutiny, new regulations to allow mitochondrial donation were debated and approved in both Houses of Parliament in February 2015 and became legal in March 2015, and are regulated by the Human Fertilisation and Embryology Authority.

Fig. 1 Mitochondrial donation.

Mitochondrial donation involves the transfer of the nuclear chromosomal DNA from an oocyte or zygote from a woman with pathogenic mtDNA mutation (shown in pink) into an enucleated, recipient donor oocyte or zygote (blue). In these techniques the mitochondria (and mitochondrial DNA) are from the donor (shown in green). (A) Metaphase II spindle transfer involves removal of the spindle from the donor egg and transfer of the patient’s spindle into the donor oocyte followed by fertilization and development. (B) Pronuclear transfer occurs after fertilization and the pronuclei are transferred from the patient zygote to the donor zygote.

Nuclear genetic mitochondrial disease

One area of mitochondrial disease research that has seen important recent advances has been the application of next-generation sequencing technologies to the identification of genes linked with Mendelian mitochondrial disorders, and the subsequent characterization of their associated proteins. Mitochondrial disorders are particularly amenable to such strategies, as they are characterized by a complex and expanding spectrum of clinical phenotypes associated with marked genetic heterogeneity difficult to resolve with candidate gene approaches.

The mitochondrial proteome is well characterized (29). Therefore, specific biochemical signatures of OXPHOS dysfunction identified as part of a multidisciplinary diagnostic work-up (e.g., an isolated or multiple OXPHOS deficiency) and a mitochondrial localization can help prioritize and diagnose candidate gene variants when incorporated in the bioinformatic pipeline. Both exon capture/sequencing of all predicted mitochondrial genes (so-called “Mitoexome”) (30, 31) and whole-exome sequencing (WES) (32, 33) have been applied to cohorts of patients with severe, early-onset presentations of mitochondrial disease. This assumes an autosomal recessive mode of inheritance, although in some patient cohorts, particularly the late-onset adult mitochondrial presentations, this filtering strategy may not be as applicable. Where large, predominantly pediatric, cohorts have been studied, diagnostic yields range from ~20% to 60% (3033). In cases of small families and singleton cases, however, the availability of robust next-generation sequencing and variant filtering protocols can produce diagnoses (34), facilitating reliable prenatal screening and genetic counseling. More recently, WES has successfully been used to identify genetic defects in cohorts of adult patients (35).

In addition to molecular diagnostics, WES of patients with mitochondrial disease has provided important insight into mitochondrial pathophysiology and basic biochemical pathways. This has been particularly valuable in those patients with multiple OXPHOS disorders due to a generalized disorder of mitochondrial protein synthesis implicating a number of processes in intraorganellar translation that have been studied in detail. These include processing of mt-mRNA transcripts, assembly and function of the mitoribosome and translation machinery, posttranscriptional modification of mt-tRNA, and function of mt-aminoacyl tRNA synthetases (mt-ARSs) (Fig. 2).

Fig. 2 Nuclear gene defects causing multiple mitochondrial OXPHOS abnormalities.

Schematic showing mitochondrial genes and pathways, many involved in key component steps of mitochondrial translation, in which human mitochondrial diseases associated with combined OXPHOS activity deficiencies have been identified. Those shown in red indicate disease genes identified by next-generation, largely whole-exome sequencing studies.

Rescue of the mitochondrial phenotype after lentiviral delivery of a wild-type copy of the candidate gene is the gold-standard practice to assign pathogenicity. Defects involving mt-ARS enzymes highlight many of the peculiarities of mitochondrial disease presentations, with mutation of different genes in a common pathway leading to a wide-ranging pathology with central nervous system and other organs involved. For some genes, however, there are clear genotype-phenotype correlations, particularly in relation to neuroradiological features, such as pontocerebellar hypoplasia type 6 for RARS2 mutations (36) and leukoencephalopathy with thalamus and brainstem involvement for EARS2 mutations (37).

One of the most fascinating groups of disorders relates to nuclear-encoded mt-tRNA–modifying enzymes, given that >30 different modified mt-tRNA positions have been identified as necessary to facilitate faithful protein translation (38). Human defects arising from mutations in these mt-tRNA–modifying enzymes also show organ specificity. For example, the related proteins MTO1 (39) and GTPBP3 (40) are almost always associated with cardiomyopathy when mutated. The application of mitoribosome profiling to these disorders (41) shows promise in unraveling the molecular mechanisms, which may lead to novel treatment strategies (42).

By applying WES to mitochondrial patients, proteins previously thought not to be involved in mitochondrial function, nor with clear evidence of mitochondrial localization, have been identified as potentially causative of disease. For example, The FBXL4 protein was previously thought to localize exclusively in the nucleus but has been implicated as a cause of early-onset mitochondrial encephalopathy associated with severe loss of mtDNA copy number (43, 44).

Studies have also highlighted avenues for therapeutic intervention and the “repurposing” of some mitochondrial proteins relative to their function. ACAD9 encodes an enzyme of mitochondrial β-oxidation that also functions as a key chaperone of complex I biogenesis (45); patients with mutations in this gene and complex I deficiency can partially respond to riboflavin therapy. Also, the neurodegenerative disease genes SPG7 and AFGL32 have been identified as causing late-onset disorders of mtDNA maintenance in patients with complex neurological presentations, characterized by progressive external ophthalmoplegia and cerebellar ataxia (46, 47).

Although WES has been successful in identifying Mendelian mitochondrial disease, challenges remain. Late-onset, likely dominant mitochondrial disorders may prove difficult to solve by WES; a diagnostic application of whole-genome sequencing should facilitate the identification of copy number variations (CNVs) and deep intronic mutations missed by exome capture. However, debate is ongoing to determine when to undertake WES within a multidisciplinary diagnostic algorithm that incorporates functional testing (assessment of mitochondrial enzyme activities); in patient populations with high rates of parental consanguinity, WES should be a frontline test prior to a muscle biopsy. Clearly, in some scenarios these tests assist the interpretation of the sequence data.

Treatment of mitochondrial disease

Treating patients with established mitochondrial disease remains a major challenge. For the vast majority of patients, therapy is limited to either preventing or treating the complications of mitochondrial diseases, such as diabetes, cardiac involvement, and epilepsy. These limitations explain the efforts being made by many groups to develop effective treatments and to prevent the transmission of mitochondrial disease.

As mentioned above, there are some rare genetic defects that may respond to specific therapies. Specifically, patients unable to synthesize ubiquinone may respond to dietary ubiquinone supplementation, or patients with ACAD9 deficiency may benefit from treatment with riboflavin. Many compounds or approaches have been touted as putative therapeutics, and there have been numerous reports of cocktails of assorted nutrients, vitamins, and antioxidants. However, the number of properly controlled clinical trials has been limited (48). This is due in major part to the difficulties in finding sufficient patients with similar clinical presentations and/or genotypes to conduct controlled, meaningful clinical trials. This problem is being addressed by the development of patient cohorts and registries (4). Below, we highlight examples with potential therapeutic applications (Fig. 3).

Fig. 3 Potential therapeutic interventions in patients with mitochondrial disease.

A generic dysfunctional mitochondrion is shown (blue; middle). It can house heteroplasmic mtDNA (black, normal; red, mutated) or defective nuclear gene products. Therapeutic approaches include mitochondrial replacement (brown). Additional strategies include targeting alternative oxidases or single-subunit NADH dehydrogenases to mitochondria to bypass defective components of the oxidative phosphorylation machinery (67) and antioxidant ubiquinone derivatives (idebenone, EPI-473) (68, 69). Rescuing defects of mitochondrially encoded proteins by recoding for nuclear expression and mitochondrial targeting (allotopic expression) is controversial but undergoing trial (70).

Relieving metabolic stress and increasing mitochondrial biogenesis

Many OXPHOS disorders exhibit decreased cellular NAD+/NADH ratios (ratios of nicotinamide adenine dinucleotide to its reduced form), leading to a loss of important NAD+-linked enzymatic activities. This is exemplified by the sirtuins SIRT1 and SIRT3, which activate key transcription factors that in turn promote expression of gene products that regulate mitochondrial biogenesis. Treatment of mice deficient in the replicative helicase Twinkle (“deletor” mice) with the NAD+ precursor nicotinamide riboside (NR) led to increased NAD+ in skeletal muscle and brown adipose tissue (49). Consistent with the importance of retaining NAD+ levels under conditions of mitochondrial dysfunction, inhibition of the NAD+-consuming enzyme poly(adenosine diphosphate) ribose phosphorylase (PARP1) has also been shown to be efficacious with models of mitochondrial dysfunction (50), although human trials have not been reported. Increasing mitochondrial mass alone may be sufficient to help resolve mitochondrial dysfunction, because an increase in partially functioning mitochondria may be sufficient to restore OXPHOS activity. In particular, a key question has been how to induce PGC1 isoforms, the master regulators of mitochondrial proliferation. Several mitochondrial proliferators have been championed, including the pan-PPAR (peroxisome proliferator–activated receptor) agonist bezafibrate, the polyphenols resveratrol and epicatechin, and the ribonucleotide AICAR. The literature surrounding the efficacy of these putative proliferators in ameliorating mitochondrial dysfunction is complex and often contradictory (51), requiring further study. It should also be considered that mitochondrial biogenesis can be engineered by exercise programs and has been shown to be safe and beneficial for patients with mitochondrial disease (52).

Improving the efficiency of OXPHOS or the mitochondrial protein synthesis system

The majority of cytochrome c is loosely bound to the anionic lipid cardiolipin, which is found mostly in the outer leaflet of the inner mitochondrial membrane. A small proportion, however, binds through hydrophobic interaction, causing a conformational change that results in cytochrome c peroxidase activation and subsequent cardiolipin oxidation (53). This can result in a loss of membrane curvature and cristae formation, detachment of cytochrome c from the membrane, reduced electron transfer efficiency, and eventually apoptosis (54, 55). An unusual tetrapeptide derivative, SS-31 or Bendavia, has been shown to localize to mitochondria and bind selectively to cardiolipin in vitro, inhibiting this hydrophobic interaction and promoting an increase in OXPHOS efficiency in isolated organelles (56). This compound has also been shown to protect against loss of mitochondrial structure in a rat model of ischemic reperfusion injury, which is suggested to explain the more rapid recovery of ATP production following injury (57).

Many disorders are due to mutations in genes encoding mt-tRNAs, leading to a deficiency in mitochondrial protein synthesis and subsequent OXPHOS abnormality. Often, the mutation causes instability of the mt-tRNA. Mitochondrial leucyl tRNA synthetase can bind and protect RNA in the mitochondrial matrix of Saccharomyces cerevisiae. Overexpression of just 69 C-terminal residues of the human ortholog was sufficient to promote stabilization of pathogenic mt-tRNA species in cultured human cells, leading to an increase in mitochondrial protein synthesis and partial restoration of oxidative phosphorylation (58, 59). We propose that small-molecule screening to identify compounds that can stabilize mt-tRNAs is worth investigating.

Restoring mtDNA homeostasis or shifting mtDNA heteroplasmy

Mitochondrial neurogastrointestinal encephalomyopathy (MNGIE) is caused by a defect in thymidine phosphorylase that catabolizes thymidine and deoxyuridine. The subsequent buildup of these compounds eventually causes a deoxynucleotide triphosphate (dNTP) imbalance evidenced by increased matrix thymidine triphosphate (TTP) and decreased deoxycytidine triphosphate (dCTP) levels. By transducing the liver of an MNGIE mouse model with adeno-associated virus (AAV) expressing human thymidine phosphorylase, the levels of nucleoside pools in several, but not all tissues, were restored as nucleoside levels can equilibrate between cells and tissues (60). Although the MNGIE model has its limitations, these results are impressive and clinical trials have been suggested for those patients who may not be suitable for stem cell transplantation.

Another mitochondrial disorder associated with an imbalance in nucleotide metabolism is caused by a defect in the enzyme thymidine kinase 2 (TK2). This mitochondrial matrix protein phosphorylates deoxypyrimidine nucleosides to generate their respective monophosphates, thymidine monophosphate (TMP) and deoxycytidine monophosphate (dCMP). Patients with autosomal recessive mutations can present at any stage of life, but often present in early childhood with profound neuromuscular disease and mtDNA depletion. Encouraging results have been reported in a TK2-deficient mouse model, simply by orally administering the two deoxypyrimidine monophosphates early after birth (61). Partial restoration of mitochondrial TTP concentrations, mtDNA levels, and OXPHOS components was noted in various tissues, although this effect was markedly reduced in the brains of pups at a stage after the blood-brain barrier had been fully developed. Notably, an increase in life expectancy was measurable. The simplicity of this treatment is striking, and the potential for therapy in humans is clear.

In patients with heteroplasmic mtDNA mutations, the pathogenic mutation is normally recessive, with dysfunction only becoming apparent when >60% of mtDNA carried the mutation. Recent studies with cultured human cells have involved the design of TAL effector nucleases (TALENs) to specifically cleave mtDNA carrying defined pathogenic deletions and point mutations (62), or have used zinc finger nucleases to destroy the mtDNA molecules carrying either a single deletion or the pathogenic m.8993T>G point mutation (63). The natural next step for this approach would be to deliver AAV expressing such designer nucleases to heteroplasmic animal models with stable populations of the pathogenic mtDNA.

Could genome editing be used to target pathogenic mtDNA? There is a great deal of current interest in the CRISPR-Cas9–mediated genome editing process. Cas9 is an RNA-guided endonuclease and the RNA can be engineered to provide the sequence selectivity. Although numerous RNA species have been postulated to be imported into the human mitochondrial matrix, none of these species has a defined mitochondrial function. Irrespective of this, RNA vectors for mitochondrial import have been designed and published, along with details of a putative RNA import pathway (64, 65). Cas9-mediated cleavage of the mitochondrial genome could potentially be attained by targeting Cas9 protein and a designed RNA chimera to the mitochondrion, with error-prone nonhomologous end joining effectively producing a gene knockout. Further complications would have to be resolved if Cas9-mediated knock-in mutations were ever likely, as homologous recombination of mammalian mtDNA appears to be rare (66).


The past 5 years have been an exciting time in mitochondrial disease research. Until recently, the complexity of the genetics and the clinical variability made mitochondrial disease a particularly challenging area of medicine. However, remarkable progress has been made, including the establishment of large patient cohorts; application of next-generation sequencing techniques to previously undiagnosed patients; the development of animal models of mitochondrial disease (not covered in this review); ethical, scientific, and legislative aspects of mitochondrial donation; and the development of potential therapeutic strategies. The next few years are likely to be equally exciting, with the major challenges of pinpointing the genetic defects in those cases not identified by WES, understanding the mechanisms involved in tissue specificity, and predicting disease progression. Finally, and most important, the development of effective treatments for all patients with mitochondrial disease must become a reality.

References and Notes

  1. Acknowledgments: Supported by The Wellcome Trust Centre for Mitochondrial Research (G906919), Newcastle University Centre for Aging and Vitality [supported by the Biotechnology and Biological Sciences Research Council and Medical Research Council (G016354/1)], MRC Centre for Neuromuscular Disease (G000608-1), The MRC Centre for Translational Research in Neuromuscular Disease Mitochondrial Disease Patient Cohort (UK) (G0800674), The Lily Foundation, the UK NIHR Biomedical Research Centre in Age and Age Related Diseases award to the Newcastle upon Tyne Hospitals NHS Foundation Trust and UK NHS Specialist Commissioners “Rare Mitochondrial Disorders of Adults and Children” Service.
View Abstract

Stay Connected to Science

Navigate This Article