Mouse models of mitochondrial DNA defects and their relevance for human disease

Abstract

Qualitative and quantitative changes in mitochondrial DNA (mtDNA) have been shown to be common causes of inherited neurodegenerative and muscular diseases, and have also been implicated in ageing. These diseases can be caused by primary mtDNA mutations, or by defects in nuclear-encoded mtDNA maintenance proteins that cause secondary mtDNA mutagenesis or instability. Furthermore, it has been proposed that mtDNA copy number affects cellular tolerance to environmental stress. However, the mechanisms that regulate mtDNA copy number and the tissue-specific consequences of mtDNA mutations are largely unknown. As post-mitotic tissues differ greatly from proliferating cultured cells in their need for mtDNA maintenance, and as most mitochondrial diseases affect post-mitotic cell types, the mouse is an important model in which to study mtDNA defects. Here, we review recently developed mouse models, and their contribution to our knowledge of mtDNA maintenance and its role in disease.

Glossary

8-OHdG 8-hydroxydeoxyguanosine

ANT1 adenine nucleotide translocator; heart, muscle and brain-specific isoform

ATP adenosine triphosphate

cDNA complementary DNA

COI mitochondrially encoded cytochrome c oxidase 1

dNTP deoxyribonucleotide triphosphate

dTTP deoxythymidine triphosphate

HSP40 heat-shock protein 40 mitochondrial chaperone

ND6 mitochondrially encoded NADH dehydrogenase 6

p53R2 ribonucleotide reductase subunit

POLG mitochondrial DNA polymerase γ

RNaseH1 ribonuclease H1

RRM2B ribonucleotide reductase subunit M2B

TFAM mitochondrial transcription factor A

TK2 thymidine kinase 2

tRNA transfer RNA

Introduction

Mitochondrial dysfunction is a major cause of inherited metabolic diseases, and mitochondrial DNA (mtDNA) mutations have an estimated prevalence of 1:200 and carrier frequencies of more than 1:100 for nuclear-encoded gene mutations in Western populations (Elliott et al, 2008; Hakonen et al, 2005; Hudson & Chinnery, 2006). Furthermore, mitochondrial dysfunction has an important role in age-related pathologies such as cancer, type II diabetes and neurodegeneration, and possibly even in normal ageing. This exceptionally wide clinical spectrum can be attributed to the range of mitochondrial functions, which extend from oxidative ATP production and synthesis of iron–sulphur clusters, haem, amino acids, steroid hormones and neurotransmitters, to the regulation of cytoplasmic calcium levels and key events in apoptosis (Chan, 2006). However, most of the known mitochondrial disorders are caused primarily by a dysfunctional respiratory chain (DiMauro & Schon, 2003; Shoubridge, 2001; Zeviani & DiDonato, 2004). Thirteen key catalytic subunits of oxidative phosphorylation (OXPHOS) complexes are encoded by the mitochondrial genome (Anderson et al, 1981), whereas the rest of the approximately 80 OXPHOS proteins, as well as all other mitochondrial proteins, are encoded by nuclear genes. The existence of mtDNA is completely controlled by the nucleus, which encodes all of the proteins that are required for mtDNA maintenance (Fig 1). Mutations in these nuclear genes cause autosomal inherited secondary mtDNA instability.

Figure 1.

Essential mitochondrial DNA maintenance proteins. All mtDNA maintenance proteins are encoded by nuclear genes, which are transcriptionally regulated by factors such as NRF1. Some proteins are involved directly in mtDNA replication and some provide nucleotides for DNA synthesis. The specific function of some of these proteins is still unclear but their absence induces mtDNA defects. ANT1, adenine nucleotide translocator 1; DGUOK, deoxyguanosine kinase; HSP40, heat-shock protein 40 mitochondrial chaperone; MPV17, mitochondrial inner membrane protein; mtDNA, mitochondrial DNA; NRF1, nuclear respiratory factor 1; OXPHOS, oxidative phosphorylation; p53R2, ribonucleotide reductase subunit; SSBP, single-stranded DNA-binding protein; TFAM, mitochondrial transcription factor A; TK2, thymidine kinase 2; TOPO, topoisomerase; TP, thymidine phosphorylase.

Several key questions have yet to be answered in order to understand mtDNA-associated pathology (Sidebar A). If OXPHOS deficiency leads to insufficient ATP synthesis, why are there tens of different mtDNA diseases instead of just one? Mutations in different mitochondrial tRNA genes cause specific syndromes: tRNA^Leu mutations are involved in hearing loss, diabetes, cardiomyopathy and early strokes (Finsterer, 2007; Kobayashi et al, 1991); tRNA^Lys mutations lead to myoclonic epilepsy (Shoffner et al, 1990); and tRNA^Ser mutations typically cause deafness (Reid et al, 1994). This pathogenic variability cannot be explained on the basis of ATP deficiency. How does the level of heteroplasmy affect the disease outcome? Primary mtDNA mutations cause maternally inherited disorders, the severity of which is loosely correlated with the amount of mutant mtDNA—that is, the level of heteroplasmy. However, low levels of one mutation cause diabetes, whereas high levels of the same mutation cause stroke, indicating that subtle-to-severe dysfunction must be interpreted differently by different sets of cells. Does developmental and somatic mtDNA segregation affect disease outcomes? During oogenesis, the amount of mtDNA undergoes a bottleneck and then rapidly expands (Jenuth et al, 1996; Cree et al, 2008). Are some mutations actively selected away during development? Does the heteroplasmy level change somatically? Understanding this process would be the cornerstone of preimplantation diagnosis. In addition, what are the physiological consequences of mtDNA disorders? Which cells are susceptible at which stage of development? Why do families with the same nuclear mutation show wide variation in their disease severity? Can ‘nuclear background' affect the individual capacity for stress response or can mtDNA copy number explain variation in disease manifestations? What proteins are required for mtDNA maintenance? Many such proteins have been found because, when they are mutated, they lead to mitochondrial disease.

Sidebar A | In need of answers.

1. Why are mitochondrial DNA (mtDNA) diseases so phenotypically variable?

2. How does the level of mtDNA heteroplasmy affect the phenotypic outcome?

3. What are the mechanisms of mtDNA segregation?

4. What are the physiological consequences of mtDNA disorders?

5. Which set of proteins is required for mtDNA maintenance?

These fundamental questions about mtDNA biology remain mostly unsolved and will only be answered by using appropriate animal models. Therefore, the genetic manipulation of mtDNA in mice has become a crucial tool for understanding mitochondrial pathology (Table 1).

Table 1.

Manipulation of mitochondrial DNA or proteins involved in its maintenance in mice

Mouse model	References
Transmitochondrial models
Mito-mice	Inoue et al, 2000; Nakada et al, 2004
Chloramphenicol resistance	Marchington et al, 1999; Sligh et al, 2000
COI T6589C + ND6 13885insCdelT	Fan et al, 2008
NZB+BALB	Jenuth et al, 1996; Battersby et al, 2003
Restriction endonuclease targeted to mitochondria
PstI, muscle-specific	Srivastava & Moraes, 2005
Manipulation of mtDNA by modifying nuclear disease genes
ANT1−/−	Graham et al, 1997
TP−/−,UP−/−	Haraguchi et al, 2002
Twinkle dup353–365 transgenic, Deletor	Tyynismaa et al, 2005
POLGA−/−	Hance et al, 2005
POLGA D181A heart-specific transgenic	Zhang et al, 2000
POLGA, D257A knock-in, Mutator	Trifunovic et al, 2004; Kujoth et al, 2005
POLGA D181A brain-specific transgenic	Kasahara et al, 2006
POLGA Y955C heart-specific transgenic	Lewis et al, 2007
TFAM−/−	Larsson et al, 1998
TFAM−/− heart	Wang et al, 1999; Li et al, 2000
TFAM−/− pancreatic β-cell	Silva et al, 2000
TFAM−/− neocortex, neuronal	Sorensen et al, 2001
TFAM−/− skeletal muscle	Wredenberg et al, 2002
TFAM−/− dopamine neurons	Ekstrand et al, 2007
RNaseH1−/−	Cerritelli et al, 2003
RRM2B−/−	Kimura et al, 2003; Bourdon et al, 2007
TK2−/−	Akman et al, 2008
TK2 H126N knock-in	Zhou et al, 2008
NRF1−/−	Huo & Scarpulla, 2001
MPV17−/−	Weiher et al, 1990; Viscomi et al, 2008
HSP40−/−	Hayashi et al, 2006
Mice with increased mtDNA copy number
TFAM overexpression	Ekstrand et al, 2004
TFAM overexpression	Ikeuchi et al, 2005
TK2 overexpression	Hosseini et al, 2007
Twinkle overexpression	Tyynismaa et al, 2004

See supplementary Tables 1, 2, 3 online for further details. ANT1, adenine nucleotide translocator 1; BALB, inbred mouse strain; COI, mitochondrially encoded cytochrome c oxidase 1; del, deletion; dup, duplication; HSP40, heat-shock protein 40 mitochondrial chaperone; ins, insertion; MPV17, mitochondrial inner membrane protein; mtDNA, mitochondrial DNA; ND6, mitochondrially encoded NADH dehydrogenase 6; NRF1, nuclear respiratory factor 1; NZB, inbred mouse strain; POLGA, DNA polymerase-γ α subunit; PstI, restriction endonuclease; RNaseH1, ribonuclease H1; RRM2B, ribonucleotide reductase subunit M2B; TFAM, mitochondrial transcription factor A; TK2, thymidine kinase 2; TP, thymidine phosphorylase; UP, uridine phosphorylase.

Direct manipulation of mtDNA

The generation of heteroplasmic mice with pathological mtDNA mutations would be an important tool for understanding mtDNA segregation and the tissue specificity of disease manifestations. However, engineering mtDNA has proved challenging owing to the multicopy nature of the mitochondrial genome: hundreds to thousands of mtDNAs exist within one cell. Furthermore, the transfection of plasmids or modified mtDNA into animal mitochondria has not been successful (McGregor et al, 2001). Supplementary Table 1 online summarizes the indirect approaches that have been used to introduce primary mtDNA mutations into mice.

Ageing-related somatic mtDNA mutagenesis was used to generate heteroplasmic mice: synaptosomes with mutant mtDNA were isolated and fused to cells lacking mtDNA, and enucleated cytoplasts were fused to fertilized pronuclear-state embryos (Inoue et al, 2000). The transmitochondrial ‘Mito-mice' carried a heteroplasmic single mtDNA deletion in high amounts, and presented respiratory-chain deficiency in the heart, skeletal muscle and kidneys. The mice developed severe anaemia, myopathy, cardiomyopathy, deafness and renal failure, and died by 6 months of age (Inoue et al, 2000; Nakada et al, 2004). Surprisingly, the deleted mtDNA was transmitted through the maternal line to the offspring, although single mtDNA deletions are not usually inherited. The transmission of the defect through the germline might have been possible through mtDNA duplications. The inherited nature of the trait allowed the maintenance of the mouse line and emphasized the importance of Mito-mice in studying mitochondrial disease pathogenesis. In humans, early-onset Pearson syndrome is caused by a single heteroplasmic mtDNA deletion that leads to anaemia, pancreatic and renal insufficiency, and mitochondrial myopathy, closely mimicking the phenotype of Mito-mice.

Heteroplasmic mice have also been generated by cytoplasmic fusion between two normal inbred mouse strains, generating mice with variable degrees of heteroplasmy (Jenuth et al, 1996). These mice did not show pathology, but there was consistent directional selection for one mtDNA genotype in some tissues (Battersby et al, 2003). Tissue-specific single factors mediated this selection (Battersby et al, 2005); the identification of the responsible proteins will be important to understand the basis of tissue specificity and somatic segregation. A similar strategy was used to introduce a chloramphenicol-resistance mutation into mtDNA (Marchington et al, 1999; Sligh et al, 2000), and to generate mice that carried two heteroplasmic point mutations in the ND6 and COI genes of mtDNA, which led to mitochondrial myopathy and cardiomyopathy (Fan et al, 2008). These results hold promise for obtaining models of mtDNA point mutations that are relevant in human diseases such as those that involve tRNA mutations.

Engineering nuclear genes to target mtDNA

The manipulation of nuclear mtDNA maintenance genes has been revealed as a successful way in which to affect mtDNA stability indirectly (supplementary Table 1 online). The mtDNA replication machinery—especially the DNA polymerase POLG (Kaguni, 2004; Graziewicz et al, 2006) and the replicative helicase Twinkle (Spelbrink et al, 2001; Korhonen et al, 2004)—has been an effective target to induce mtDNA mutagenesis (Fig 2).

Figure 2.

Mouse models of mitochondrial DNA and its maintenance. (A) Main tissue manifestations of mtDNA disease models. (B) Crucial tissue manifestations of mice lacking mtDNA maintenance proteins. Mice in which mtDNA replication proteins are absent die during embryonic development, whereas those without proteins involved in maintaining the nucleotide pools present postnatal symptoms. (C) Various tissue manifestations of prematurely aged Mutator mice. ΔmtDNA, Mito-mice with single mtDNA deletion; ANT1, adenine nucleotide translocator; HSP40, heat-shock protein 40 mitochondrial chaperone; MPV17, mitochondrial inner membrane protein; mtDNA, mitochondrial DNA; NRF1, nuclear respiratory factor 1; p53R2, ribonucleotide reductase subunit; POLG, DNA polymerase γ; PstI, restriction endonuclease; RNaseH, ribonuclease H1; TFAM, mitochondrial transcription factor A; TK2, thymidine kinase 2.

Dominant mutations in Twinkle predispose individuals to human adult-onset progressive external ophthalmoplegia, which is associated with multiple mtDNA deletions (Suomalainen et al, 1997; Spelbrink et al, 2001). This disease is closely replicated in transgenic mice that ubiquitously express a mutant murine Twinkle cDNA—which carries a mutation that is homologous to a dominant patient mutation—in a 1:1 ratio with respect to endogenous Twinkle. This mouse suffers from late-onset (1 year of age) progressive respiratory-chain deficiency in the skeletal muscle, as well as in distinct neuronal populations such as cerebellar Purkinje cells and hippocampal neurons, with the accumulation of multiple mtDNA deletions (Tyynismaa et al, 2005), and was hence named ‘Deletor'. This phenotype could be replicated in several transgenic lines, whereas the overexpression of wild-type Twinkle caused no pathology (Tyynismaa et al, 2004). The lifespan and reproductive capabilities of the Deletor mouse were normal, and the distinct pathology makes it an attractive target for long-term therapeutic trials to treat adult-onset mitochondrial myopathy and neurodegeneration.

Human mutations of the mitochondrial replicative polymerase POLG have been shown to be a major cause of mitochondrial disease, and more than 2% of the population carries pathogenic POLG variants (Hakonen et al, 2005; Hudson & Chinnery, 2006). Knock-in inactivation of the POLG proof-reading exonuclease activity in mice results in efficient mtDNA mutagenesis (Trifunovic et al, 2004; Kujoth et al, 2005). These ‘Mutator' mice show signs of premature ageing, including weight loss, reduced subcutaneous fat content, hair loss, kyphosis, osteoporosis, anaemia, reduced fertility and an enlarged heart, and their median lifespan is shortened to 48 weeks (Fig 2C; Trifunovic et al, 2004; Kujoth et al, 2005). Mutator mice were the first experimental demonstration that extensive mtDNA mutagenesis can cause premature ageing-like symptoms, and will be important in revealing the contribution of mitochondria to various degenerative phenotypes. Studies on the Mutators have shown that coding-region mutations, which are potentially severely defective in function, are selected against during oocyte development, and an over-representation of tRNA mutations, which have a potentially milder functional effect, is maintained through the selection (Stewart et al, 2008). This intriguing finding could explain the abundance of tRNA mutations in human pathology. Furthermore, the Mutator might provide an excellent tool for creating mtDNA mutation-specific mouse lines by breeding the animals with wild-type strains, thereby maintaining and ‘diluting out' the original mtDNA mutations in the maternal line.

Overexpression of mutant POLG in mice caused cardiomyopathy when expressed in the heart and mood disorder-like behaviour with the neuronal promoter (Zhang et al, 2000; Kasahara et al, 2006; Lewis et al, 2007). However, these studies lacked the wild-type POLG overexpression control or reported only one transgenic line, leaving the contribution of insertional mutagenesis or protein overload to the phenotypes still to be excluded.

Eliminating mutant mtDNA

A potentially successful treatment strategy could shift the heteroplasmy to favour wild-type over mutant mtDNA. As a proof-of-principle for the enzymatic modification of mtDNA in vivo, PstI restriction endonuclease—which cuts mtDNA twice inducing double-stranded DNA breaks (DSBs)—was targeted to mouse skeletal muscle mitochondria (Srivastava & Moraes, 2005). Surprisingly, this led to mtDNA depletion and multiple large-scale mtDNA deletions, indicating that the 5′ ends of the DSBs had often been end-joined to the partly single-stranded D-loop, thereby creating large mtDNA deletions. The homologous D-loop region is also a common deletion breakpoint in humans and in the Deletor mouse (Zeviani et al, 1989; Tyynismaa et al, 2005). These similarities indicate that the formation of multiple mtDNA deletions might involve DSB repair. Between the ages of 6 and 7 months, the PstI mice developed a severe mitochondrial myopathy, but the PstI transgene was not transmitted to their offspring. Nonetheless, this experiment shows that mtDNA modifying enzymes can be introduced into animal tissues, indicating that point mutation-specific DNA-processing enzymes could be used for the sequence-specific degradation of mutant mtDNA.

No cell can live for long without mtDNA

MtDNA is completely dependent on nuclear genes: if an essential maintenance protein is defective, mtDNA is lost. The elucidation of the molecular background of mitochondrial DNA depletion syndrome (MDS)—a prevalent cause of respiratory-chain deficiency in childhood (Sarzi et al, 2007)—has uncovered several new proteins that are required for mtDNA maintenance.

The absence of proteins that directly regulate mtDNA copy number in mice leads to mtDNA loss and is not tolerated beyond mouse embryonic day 8.5 (Fig 2B), at which point mtDNA has probably been diluted too much from the ovum origin to be able to maintain life. At this developmental stage, cardiac function is required to enable oxygen supply because the embryo can no longer survive by oxygen diffusion. Essential mtDNA maintenance proteins—such as RNaseH1 and HSP40 (supplementary Table 2 online; Cerritelli et al, 2003; Hayashi et al, 2006)—have been found by studying knockout mice that die around embryonic day 8.5 and lack mtDNA.

TFAM is a histone-like packaging protein of mtDNA that is essential for mtDNA transcription and replication (Fisher & Clayton, 1988; Kanki et al, 2004; Kaufman et al, 2007). mtDNA levels closely follow TFAM levels, and TFAM knockout therefore leads to the progressive loss of mtDNA, severe OXPHOS defects and the death of the targeted cell type (supplementary Table 2 online; Larsson et al, 1998; Wang et al, 1999; Li et al, 2000; Silva et al, 2000; Sorensen et al, 2001; Wredenberg et al, 2002; Ekstrand et al, 2007). Conditional knockout studies have shown that different cell types vary considerably in their tolerance to reduced mtDNA levels. Tissue-specific TFAM knockouts are good models of respiratory-chain dysfunction at stages in which mtDNA is still present in reduced amounts.

Nucleosides and nucleotides for mtDNA maintenance

Molecular genetics studies of MDS have revealed an expanding group of proteins involved in mitochondrial nucleoside pool regulation (supplementary Table 2 online). These disorders are typically tissue specific, although this phenomenon is not currently understood. MtDNA maintenance requires both the mitochondrial dNTP salvage pathway and cytoplasmic de novo dNTP synthesis. TK2 is a mitochondrial deoxyribonucleoside kinase that is part of the salvage pathway. TK2 mutations manifest as early childhood-onset MDS with progressive myopathy and/or encephalopathy (Saada et al, 2001; Götz et al, 2008). Mice lacking functional TK2 showed rapid and progressive mtDNA depletion in all studied tissues, although only after postnatal day 12 (Akman et al, 2008; Zhou et al, 2008). The most severely affected tissue in the TK2 null mice was the brain rather than skeletal muscle, as is the case in human MDS. Defects in the de novo dNTP synthesis pathway also lead to a severe phenotype: mutations affecting the p53R2 subunit of the cytoplasmic ribonucleotide reductase, RRM2B, underlie a severe form of MDS that is associated with infantile myopathy, encephalopathy and nephropathy, which is fatal before 4 months of age (Bourdon et al, 2007). The RRM2B knockout mouse reproduces this phenotype relatively well, as it develops normally until weaning, but then experiences growth retardation and multiple organ failure leading to death 12–14 weeks after birth (Kimura et al, 2003). Lack of ANT1, which is the main regulator of adenine pools in post-mitotic tissues, does not compromise mouse lifespan or fertility, but leads to severe exercise intolerance, marked proliferation of mitochondria in the skeletal muscle and mtDNA rearrangements in the enlarged heart (Graham et al, 1997). The involvement of ANT1 in mtDNA maintenance has been verified in humans through the identification of dominant ANT1 mutations in familial progressive external ophthalmoplegia (Kaukonen et al, 2000), and recessive mutations in adult-onset myopathy and cardiomyopathy, both of which are associated with multiple mtDNA deletions (Palmieri et al, 2005).

On the basis of the available models, the proteins that are involved in mitochondrial dNTP pool regulation in various species seem to be conserved and, if inactivated, lead to mtDNA depletion. However, the tissue-specific need for these proteins seems to differ between species. The dNTP-knockout models vary from other mtDNA maintenance protein knockouts in that all of the former undergo normal development, despite the great need for DNA replication and dNTPs during embryonal life, and disease develops only after weaning, resembling the situation in humans. This points to an alternative, perhaps maternal, source of mitochondrial dNTPs during embryonal development, which is a possibility that deserves further study.

Is an increase in mtDNA copy number beneficial?

The copy number of mtDNA is tightly regulated; however, overexpression of human TFAM or murine Twinkle in mice has led to mtDNA increase (supplementary Table 3 online; Ekstrand et al, 2004; Tyynismaa et al, 2004; Ikeuchi et al, 2005). Twinkle helicase could be rate limiting for replication initiation and might affect mtDNA copy number through increased synthesis, whereas TFAM could stabilize mtDNA and increase its half life. Elevated levels of mtDNA seem not to be deleterious in either model; by contrast, mice overexpressing TFAM were found to have increased survival after myocardial infarction (Ikeuchi et al, 2005) as well as reversed age-dependent memory impairment (Hayashi et al, 2008), implying a beneficial effect of increased mtDNA levels. TK2 overexpression—that is, 300-fold increased TK2 activity—in the heart of transgenic mice led to a 40% increase in mtDNA copy number (Hosseini et al, 2007), indicating that heavy upregulation of the dNTP salvage pathway/dTTP pool can also affect the amount of mtDNA. Mice with increased mtDNA provide an intriguing model in which to study the cross-talk between energy metabolism and the environment, and to identify conditions in which having more mtDNA could be advantageous.

Need for extensive pathogenesis studies

The growing number of mtDNA mouse models has indicated an essential role of mtDNA maintenance proteins for viability. Knockouts of such proteins inevitably kill the targeted cell, and hence their physiological relevance for understanding mitochondrial dysfunction is low. Knockouts of dNTP pool regulators have provided important information about early-onset disorders, reflecting the marked consequences of loss-of-activity mutations for early postnatal life; perhaps surprisingly, none of those proteins seems to be essential for development. As mitochondrial proteins are usually well conserved among various species, mouse strains carrying homologous patient mutations can often be obtained. This strategy—either through knock-in techniques, transgenesis or the introduction of primary mtDNA mutations—has been relatively successful in replicating human diseases, as exemplified by the Mito-mice and the Deletor mice. However, even these true disease models have not been fully utilized to study pathogenesis or therapeutic-intervention strategies. Surprisingly, most models studied so far have not shown increased production of reactive oxygen species (Silva & Larsson, 2002; Kujoth et al, 2005; Trifunovic et al, 2005), which is generally thought to be a crucial mediator of mitochondrial pathology, thereby implying that an increase in reactive oxygen species is not a general consequence of respiratory-chain dysfunction. The exception is the POLGA Y955C heart-specific transgenic model (Lewis et al, 2007), which did show elevated 8-OHdG in mtDNA.

Conclusions

Much remains to be elucidated about mtDNA, its maintenance and its effect on disease; however, the existing mouse models will be useful for studies of pathogenesis and, in some cases, for testing potential treatments for mitochondrial disorders. However, more models carrying patient mutations would need to be generated in order to understand the phenotypic variation, tissue specificity and physiological consequences of mitochondrial diseases. The effect of ‘nuclear background' on phenotypic variability within one disorder could be studied by back-crossing disease models to various pure inbred backgrounds; if variation was found, the responsible genes could then be mapped. Intriguingly, mitochondrial dysfunction is increasingly associated with common disease entities such as neurodegeneration and metabolic syndrome, as well as ageing, which emphasizes the role of the mtDNA-mouse models in understanding the role of mitochondrial dysfunction in common pathologies.

Supplementary information is available at EMBO reports online (http://www.emboreports.org).

Supplementary Material

Supplementary information

Henna Tyynismaa (left) & Anu Suomalainen