**3. Models to study OXPHOS diseases**

#### **3.1 Cybrids and Rho<sup>0</sup> cells**

34 Bioenergetics

autosomal recessive syndrome showed mtDNA deletions, depletion, or both. The MNGIE locus was mapped to chromosome 22q13.32-qter, a region that contains the thymidine phosphorylase (TP) gene (gene symbol ECGF1). Studies on patients showed that TP activity was markedly decreased. Ichizo Nishino and collaborators found various homozygous as well as compound heterozygous ECGF1 mutations in the genomic DNA of MNGIE patients (Nishino et al., 1999). The precise mechanism by which TP deficiency leads to mtDNA rearrangements have still to be explained, but imbalance of the mitochondrial nucleotide pool is likely to have a role. Autosomal dominant Progressive External Ophtalmogia (adPEO) is an adult-onset mitochondrial disorder that is characterized by progressive external ophthalmoplegia and variable additional features, including exercise intolerance, ataxia, depression, hypogonadism, hearing deficit, peripheral neuropathy and cataract (Zeviani et al., 1990). Some patients carry mtDNA deletions, although the disease is inherited in an autosomal fashion. Of the two autosomal loci for this disorder, the 4q-adPEO locus includes the gene for the heart and skeletal muscle isoform of the ANT1. Kaukonen and collaborators (Kaukonen et al., 2000) identified two heterozygous missense mutations in

Enzyme complex I and IV deficiencies are by far the most frequently observed abnormalities of the OXPHOS system. In sharp contrast to isolated complex I deficiencies, no mutations have been found as yet in the ten nuclear genes that encode the structural proteins of complex IV (Adams et al., 1997). The discovery of mutations in a nuclear assembly gene that is associated with COX deficiency resulted from chromosomal transfer experiments. This approach identified mutations in the SURF1 gene in patients with COX-deficient Leigh syndrome (Tiranti et al., 1998; Zhu et al., 1998). SURF1 is part of a cluster of unrelated housekeeping genes and is the only gene of this cluster that is known or believed to be involved in COX assembly (Tiranti et al., 1999). Nuclear gene defects that are associated with isolated complex III or complex V deficiencies have not yet been discovered. In recent years, four inherited neurodegenerative diseases, Friedreich ataxia, hereditary spastic paraplegia, human DDP syndrom (deafness/dystonia peptide) and dominant optic atrophy (OPA1) have also been shown to be mitochondrial disorders that are caused by nuclear DNA mutations in the genes for frataxin, paraplegin, DDP and OPA1, respectively. Mitochondria obtained from heart biopsies of Friedreich ataxia patients disclosed specific defects in the citric-acid cycle enzyme aconitase, and complex I–III activities (Rotig et al., 1997). The causative Friedreich ataxia protein, frataxin, has an essential role in mitochondrial iron homeostasis, and Friedreich ataxia can therefore be considered as an OXPHOS homeostasis defect. Muscle biopsies from the autosomal recessive form of patients with hereditary spastic paraplegia revealed histochemical signs of a mitochondrial disorder, namely RRFs, COX-negative fibers and succinate dehydrogenase-positive hyperintense fibers (Casari et al., 1998). Linkage and subsequent mutation analysis revealed large deletions in a gene dubbed paraplegin (Casari et al., 1998). Owing to the homology with a yeast mitochondrial ATPase with both proteolytic and chaperone-like activities, it has been suggested that this form of hereditary spastic paraplegia could be a neurodegenerative disorder due to OXPHOS deficiency, attributing a putative function in the assembly or import of respiratory chain subunits or cofactors to paraplegin (Di Donato, 2000). The DDP syndrome, an X-linked recessive disorder also known as the Mohr–Tranebjaerg syndrome, is associated with a novel defect in mitochondrial protein import (Koehler et al., 1999). The defective gene is homologous to the yeast protein Tim8, which belongs to a family of

this gene in several families and in one sporadic patient with adPEO.

Cybrids, or "cytoplasmic hybrids," are cultured cells manipulated to contain introduced mitochondrial DNA (mtDNA). Cybrids have been a central tool to unravel effects of mtDNA mutations in OXPHOS diseases. In this way, the nuclear genetic complement is held constant so that observed effects on OXPHOS can be linked to the introduced mtDNA. The cybrids are produced by first treating mitochondrial donor cells with cytochalasin B to weaken the cytoskeleton, before subjecting the cells to a centrifugal force, either as attached cells or in suspension. The dense nuclei are extruded, leaving plasma membrane-bound "cytoplasts" containing cell cytoplasm and organelles, including mitochondria. These cytoplasts are then fused with a nuclear donor cell line. The first mammalian cultured cell phenotype identified to segregate with mtDNA was in human (HeLa) cells, where mtDNA imparted resistance to the antibiotic chloramphenicol (Spolsky & Eisenstadt, 1972). Several other mtDNA‐linked drug‐resistant phenotypes were identified in mammalian cells in the 1970s and 1980s, including resistance to the complex III inhibitors antimycin and myxothiazol (Howell & Gilbert, 1988) and to the complex I inhibitor rotenone (Bai & Attardi, 1998). The development of robust DNA‐sequencing methods leads to the identification of single‐base substitutions in the 16S rRNA gene of the mtDNA of independently derived yeast, mouse, and human chloramphenicol cell lines (Blanc & Dujon, 1980; Kearsey & Craig, 1981). These pioneering studies were in turn followed by identification of the first cytochrome b mutants (Howell & Gilbert, 1988) and more recently ND5, ND6, and COI mutants.

The second cellular OXPHOS model corresponds to the isolation of a human cell line without mtDNA (called Rho0 cells). Employing an approach first used in yeast (Slonimski et al., 1968), cells were incubated with low levels of the drug ethidium bromide, which intercalates DNA. Low levels of the drug selectively inhibits the gamma‐DNA polymerase responsible for mtDNA replication, and with ongoing cell division, the mtDNAs are ''diluted'' to the point where clones can be isolated without detectable organelle genomes. King and Attardi (King & Attardi, 1989) also discovered the absolute requirement for pyruvate gained by these cells and confirmed the previous observation from Paul Desjardins and collaborators (Desjardins et al., 1985) that mtDNA‐less cells also required added uridine for growth. This allowed a selection regime to be used after cytoplast– Rho0 cell fusion so that unfused Rho0 cells could be eliminated and cybrids selected with the use of an appropriate nuclear drug‐resistant marker (King & Attardi, 1989). Apart from their value in cybrid experiments, such Rho0 cells represent a unique research tool by themselves. They are a surprising reminder that OXPHOS is dispensable, at least for some differentiated mammalian cell types. In this instance, ATP production is 100% from glycolysis, so the cells acidify culture media very rapidly by producing large quantities of lactate. They retain functional mitochondria (except lacking OXPHOS), which show a transmembrane potential

Mitochondrial Calcium Signalling: Role in Oxidative Phosphorylation Diseases 37

mitochondrial biogenesis and embryonic development, and the conditional knockouts have indicated that the OXPHOS system is crucial for normal heart function and insulin secretion. Five mouse models that were specifically designed to mimic isolated complex I deficiency in humans involve the NDUFS4 gene. This gene constitutes a mutational hotspot in humans. Four models are KO or conditioned KO for NDUFS4, the fifth one corresponds to a point mutation in NDUFS4. The whole-body and neuron-targeted NDUFS4 KO mice displayed small size and displayed weight loss. This was accompanied by ataxia, blindness, hearing loss, loss of motor skills and death from a fatal encephalomyopathy. The Purkinje cell specific KO mice only manifested mild behavioral and neuropathological abnormalities. Homozygote point mutation NDUFS4 mice were not viable, demonstrating that the presence of mutated NDUFS4 protein leads to a much more severe phenotype than

Two mouse models for Friedreich ataxia have also been created (Puccio et al., 2001). Like the ANT1- and SOD2-deficient mice, these mice can also be considered as secondary OXPHOSdeficient mice. The frataxin-deficient mammals showed time-dependent iron accumulation and will allow the detailed study of the mechanism of frataxin involvement in iron

Finally, Jun-Ichi Hayashi's group, using a completely different approach, generated mice that carry large-scale mtDNA deletions (Inoue et al., 2000). Synaptosomes from mouse brains with naturally occurring somatic mtDNA mutations were fused with Rho0 cells. Each fusion event introduced a variable number of mutant and wild-type mtDNAs, which then repopulate the Rho0 cell, creating a cybrid cell line. Enucleated cybrid cells were fused to donor embryos and implanted in pseudo pregnant females. In this way, they generate heteroplasmic founder female animals in which mtDNA deletion transmission was obtained

Calcium (Ca2+) is one of the most common second messengers in intracellular signalling networks. Periodic fluctuations in cytosolic calcium concentration ([Ca2+]cyt) is driven by electrical activation of voltage-gated Ca2+ channels (VGCC) or by agonist stimulation of plasma membrane receptors and the subsequent formation of Ca2+-mobilizing second messengers, such as inositol 1,4,5-trisphosphate (IP3). IP3 binds to its receptor the IP3R (inositol 1,4,5-trisphosphate) on the endoplasmic reticulum (ER) membrane leading to Ca2+ release from the ER to the cytosol. In excitable cells, Ca2+ release from the ER occurs also through ryanodine receptors (RyR) that function as Ca2+-activated Ca2+ channels which

The frequency, amplitude and/or duration of cytosolic [Ca2+]cyt spikes can be detected and decoded by downstream Ca2+-sensitive proteins providing a versatile pathway for extracellular stimuli to exert control over a wide range of metabolic pathways (Berridge et

Complex buffering systems that include multiple Ca2+-buffering proteins, ATP-dependent Ca2+ pumps (SERCA (sarco-endoplasmic Reticulum Ca2+ ATPase) accumulating Ca2+ from the cytosol to the ER, and PMCA (Plasma membrane Ca2+ ATPase) extruding Ca2+ from cytosol to the extracellular space), and the sodium-Ca2+ exchanger (Na+/Ca2+), work together to restore [Ca2+] back to resting levels. Mitochondria also play an important role in shaping Ca2+ signals by utilizing potent mitochondrial Ca2+ uptake mechanisms. Ca2+

complete absence of NDUFS4 (reviewed in (Roestenberg et al., 2011)).

**4. Calcium signalling and mitochondrial OXPHOS physiology** 

further amplify Ca2+ signals originating from other sources.

metabolism and iron–sulphur biogenesis.

for three generations (Inoue et al., 2000).

al., 2000).

(probably from the electrogenic exchange of ATP for ADP) and can import the hundreds of other proteins needed for non‐OXPHOS functions. The pioneering cybrid work using the selectable markers was limited in the sense that endogenous mtDNAs were also present; that is, the cybrids were heteroplasmic. The Rho0 cell approach allowed creation of homoplasmic or heteroplasmic cells, depending on the mtDNA donor cell(s) used.

## **3.2 Human fibroblasts**

The use of individual patient's cells in tissue cultures enables the study of specific defects. With respect to cell type, myoblasts are most likely to express the phenotype observed in muscle, but it is generally not feasible to derive myoblasts for each diagnostic muscle biopsy, because most of the muscle tissue is used up for enzymatic, pathological and molecular workup. Moreover, myoblasts are not representative of some liver-specific phenotypes. An alternative to myoblasts, are fibroblasts, which are much easily obtained during a muscle biopsy or after (Robinson, 1996). Fibroblasts cultures are in general, the most obtainable and renewable source of cells for both diagnosis and research. The major drawback with fibroblasts in culture is that they sometimes fail to maintain the diseased phenotype. This is especially true for fibroblast cultures derived from tissue specific forms of mitochondrial diseases. Nevertheless, many patients do express mitochondrial dysfunction in primary fibroblasts albeit the defect is sometimes unmasked only under stressful growth conditions in culture media, devoid of glucose or serum (Iuso et al., 2006; Robinson, 1996; Taanman et al., 2003). Therefore, patient's fibroblast harboring nuclear encoded mutations can be a suitable tool to study OXPHOS diseases and a platform for the search for treatments by small molecules, using individual approaches tailored to a specific defect.

#### **3.3 Mouse models**

Despite some obvious limitations, our ability to mimic human disease in animal models is undoubtedly one of the most important technological breakthroughs in modern genetics.

Since the first knockout mice with impaired OXPHOS were generated in 1995 (reviewed in (Smeitink et al., 2001) and (Larsson & Rustin, 2001)), eight others have been described.

Classical knockout (KO) technology has been achieved for the manganese superoxide dismutase gene (SOD2) and the ANT1. These mice can be considered as secondary OXPHOS-deficient mice because the genes are only indirectly related to the OXPHOS system. SOD2 is an oxygen radical scavenger in the mitochondrial matrix, which acts as a first line of defense against the superoxide that is produced as a by-product of OXPHOS (Li et al., 1995). To gain further insight into the effects of the ANT1 mutation in particular, study the regulation of nuclear and mitochondrial genes in the skeletal muscle of mice KO of ANT1 (Murdock et al., 1999) revealed upregulation of 17 genes that fall into four categories: nuclear and mitochondrial genes that encode OXPHOS components; mitochondrial tRNA and ribosomal RNA genes; genes involved in intermediary metabolism; and an eclectic group of other genes, among which are genes previously unknown to be related to mitochondrial function.

Knockout mice for the mitochondrial transcription factor A (TFAM) can be considered primary OXPHOS mice, because TFAM has a direct role in the regulation of OXPHOS gene expression. Using a conditional knockout approach, three distinct TFAM knockout mice have been created (Larsson et al., 1998; Wang et al., 1999): one for skeletal and cardiac muscle; one for cardiac muscle alone; and one for pancreatic β-cells. TFAM is essential for

(probably from the electrogenic exchange of ATP for ADP) and can import the hundreds of other proteins needed for non‐OXPHOS functions. The pioneering cybrid work using the selectable markers was limited in the sense that endogenous mtDNAs were also present; that is, the cybrids were heteroplasmic. The Rho0 cell approach allowed creation of

The use of individual patient's cells in tissue cultures enables the study of specific defects. With respect to cell type, myoblasts are most likely to express the phenotype observed in muscle, but it is generally not feasible to derive myoblasts for each diagnostic muscle biopsy, because most of the muscle tissue is used up for enzymatic, pathological and molecular workup. Moreover, myoblasts are not representative of some liver-specific phenotypes. An alternative to myoblasts, are fibroblasts, which are much easily obtained during a muscle biopsy or after (Robinson, 1996). Fibroblasts cultures are in general, the most obtainable and renewable source of cells for both diagnosis and research. The major drawback with fibroblasts in culture is that they sometimes fail to maintain the diseased phenotype. This is especially true for fibroblast cultures derived from tissue specific forms of mitochondrial diseases. Nevertheless, many patients do express mitochondrial dysfunction in primary fibroblasts albeit the defect is sometimes unmasked only under stressful growth conditions in culture media, devoid of glucose or serum (Iuso et al., 2006; Robinson, 1996; Taanman et al., 2003). Therefore, patient's fibroblast harboring nuclear encoded mutations can be a suitable tool to study OXPHOS diseases and a platform for the search for treatments by small molecules,

Despite some obvious limitations, our ability to mimic human disease in animal models is undoubtedly one of the most important technological breakthroughs in modern genetics. Since the first knockout mice with impaired OXPHOS were generated in 1995 (reviewed in (Smeitink et al., 2001) and (Larsson & Rustin, 2001)), eight others have been described. Classical knockout (KO) technology has been achieved for the manganese superoxide dismutase gene (SOD2) and the ANT1. These mice can be considered as secondary OXPHOS-deficient mice because the genes are only indirectly related to the OXPHOS system. SOD2 is an oxygen radical scavenger in the mitochondrial matrix, which acts as a first line of defense against the superoxide that is produced as a by-product of OXPHOS (Li et al., 1995). To gain further insight into the effects of the ANT1 mutation in particular, study the regulation of nuclear and mitochondrial genes in the skeletal muscle of mice KO of ANT1 (Murdock et al., 1999) revealed upregulation of 17 genes that fall into four categories: nuclear and mitochondrial genes that encode OXPHOS components; mitochondrial tRNA and ribosomal RNA genes; genes involved in intermediary metabolism; and an eclectic group of other genes, among which are genes previously unknown to be related to

Knockout mice for the mitochondrial transcription factor A (TFAM) can be considered primary OXPHOS mice, because TFAM has a direct role in the regulation of OXPHOS gene expression. Using a conditional knockout approach, three distinct TFAM knockout mice have been created (Larsson et al., 1998; Wang et al., 1999): one for skeletal and cardiac muscle; one for cardiac muscle alone; and one for pancreatic β-cells. TFAM is essential for

homoplasmic or heteroplasmic cells, depending on the mtDNA donor cell(s) used.

using individual approaches tailored to a specific defect.

**3.2 Human fibroblasts** 

**3.3 Mouse models** 

mitochondrial function.

mitochondrial biogenesis and embryonic development, and the conditional knockouts have indicated that the OXPHOS system is crucial for normal heart function and insulin secretion. Five mouse models that were specifically designed to mimic isolated complex I deficiency in humans involve the NDUFS4 gene. This gene constitutes a mutational hotspot in humans. Four models are KO or conditioned KO for NDUFS4, the fifth one corresponds to a point mutation in NDUFS4. The whole-body and neuron-targeted NDUFS4 KO mice displayed small size and displayed weight loss. This was accompanied by ataxia, blindness, hearing loss, loss of motor skills and death from a fatal encephalomyopathy. The Purkinje cell specific KO mice only manifested mild behavioral and neuropathological abnormalities. Homozygote point mutation NDUFS4 mice were not viable, demonstrating that the presence of mutated NDUFS4 protein leads to a much more severe phenotype than complete absence of NDUFS4 (reviewed in (Roestenberg et al., 2011)).

Two mouse models for Friedreich ataxia have also been created (Puccio et al., 2001). Like the ANT1- and SOD2-deficient mice, these mice can also be considered as secondary OXPHOSdeficient mice. The frataxin-deficient mammals showed time-dependent iron accumulation and will allow the detailed study of the mechanism of frataxin involvement in iron metabolism and iron–sulphur biogenesis.

Finally, Jun-Ichi Hayashi's group, using a completely different approach, generated mice that carry large-scale mtDNA deletions (Inoue et al., 2000). Synaptosomes from mouse brains with naturally occurring somatic mtDNA mutations were fused with Rho0 cells. Each fusion event introduced a variable number of mutant and wild-type mtDNAs, which then repopulate the Rho0 cell, creating a cybrid cell line. Enucleated cybrid cells were fused to donor embryos and implanted in pseudo pregnant females. In this way, they generate heteroplasmic founder female animals in which mtDNA deletion transmission was obtained for three generations (Inoue et al., 2000).
