Human Reproduction Update Advance Access originally published online on September 30, 2005
Human Reproduction Update 2006 12(2):119-136; doi:10.1093/humupd/dmi042
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The transmission of OXPHOS disease and methods to prevent this
1 Department of Genetics and Cell Biology, 2 Research Institute GROW, 3 Department of Ethics and Philosophy, University of Maastricht, Maastricht and 4 Department of Child Neurology, Erasmus MC, University Medical Center, Rotterdam, The Netherlands
5 To whom correspondence should be addressed at: Department of Genetics and Cell Biology, University of Maastricht, P.O. Box 616, 6200 MD Maastricht, The Netherlands. E-mail: bert.smeets{at}molcelb.unimaas.nl
Submitted on May 6, 2005; resubmitted on August 26, 2005; accepted on September 7, 2005
 |
Abstract |
|---|
TOP
Abstract
Mitochondrial disordersDiseases owing to defects of oxidative phosphorylation (OXPHOS) affect approximately 1 in 8000 individuals. Clinical manifestations can be extremely variable and range from single-affected tissues to multisystemic syndromes. In general, tissues with a high energy demand, like brain, heart and muscle, are affected. The OXPHOS system is under dual genetic control, and mutations in both nuclear and mitochondrial genes can cause OXPHOS diseases. The expression and segregation of mitochondrial DNA (mtDNA) mutations is different from nuclear gene defects. The mtDNA mutations can be either homoplasmic or heteroplasmic and in the latter case disease becomes manifest when the mutation exceeds a tissue-specific threshold. This mutation load can vary between tissues and often an exact correlation between mutation load and phenotypic expression is lacking. The transmission of mtDNA mutations is exclusively maternal, but the mutation load between embryos can vary tremendously because of a segregational bottleneck. Diseases by nuclear gene mutations show a normal Mendelian inheritance pattern and often have a more constant clinical manifestation. Given the prevalence and severity of OXPHOS disorders and the lack of adequate therapy, existing and new methods for the prevention of transmission of OXPHOS disorders, like prenatal diagnosis (PND), preimplantation genetic diagnosis (PGD), cytoplasmic transfer (CT) and nuclear transfer (NT), are technically and ethically evaluated.
Key words: mitochondria / OXPHOS disease / PGD / PND / transmission
|
Mitochondrial disorders |
|---|
Mitochondrial disorders are a group of diseases and syndromes commonly defined by lack of energy owing to defects in oxidative phosphorylation (OXPHOS) (Zeviani and Di Donato, 2004
). They affect at least 1 in 8000 of the general population, making them the most common inherited metabolic disease (Chinnery, 2004). Energy in the form of ATP is produced by the OXPHOS system, which consists of five multiprotein enzyme complexes that release the energy stored in the form of a proton gradient across the inner mitochondrial membrane (Saraste, 1999). Disease manifestations because of OXPHOS defects usually involve tissues with a high energy demand like brain, heart, liver and the renal and endocrine systems (Wallace, 1999). Clinical manifestations of OXPHOS diseases are extremely variable and range from a single-affected tissue, like the loss of vision in Leber’s hereditary optic neuropathy (LHON), to multisystemic syndromes like Leigh syndrome (subacute necrotizing encephalomyelopathy, LS), mitochondrial encephalopathy, lactic acidosis and stroke-like episodes (MELAS), neuropathy, ataxia and retinitis pigmentosa (NARP) and myoclonic epilepsy with ragged red fibres (MERRF). Table I lists several syndromes and symptoms associated with OXPHOS disease. Involvement of the central nervous system, skeletal muscle or both is seen in many mitochondrial syndromes. A frequent symptom in paediatric patients is developmental delay and failure to thrive. Symptoms can present in just a single tissue or organ, but a multiorgan involvement in a patient or affected relatives is more common. When at least two organ systems unexplained by other diseases are involved in a single person or in affected (maternal) relatives, then an OXPHOS disease should be considered. Clinicians should be aware that apparently unrelated symptoms might have a common genetic cause.
|
|
Mitochondrial DNA |
|---|
The first description of a circular DNA structure located in the mitochondria dates from 40 years ago (Nass, 1966
). Several unique characteristics discriminate mitochondrial from nuclear DNA.
- The mitochondrial DNA (mtDNA) is a multicopy genome. A cell contains hundreds of mitochondria, and each mitochondrion contains five to ten copies of mtDNA (Goto, 2001). Dependent on the tissue and energy demand, each cell contains between 500 and 10 000 mtDNA molecules, except for mature oocytes which contain between 100 000 and 600 000 mtDNA molecules (Reynier et al., 2001). Oocytes store mitochondria to deal with the lack of mtDNA replication during the first cleavage stages of the embryo (Schaefer et al., 2001).
- In a cell, all mtDNA molecules can be identical (homoplasmy), or two types of mtDNA molecules, that differ in sequence, in the same cell, tissue or even in the same organelle can coexist (heteroplasmy) (Holt et al., 1988; Lightowlers et al., 1997).
- The mtDNA is transmitted entirely through the maternal line.
- The mtDNA is a double-stranded circle (Figure 1) of 16 569 bp with a genetic code different from the nuclear DNA (Fernandez-Silva et al., 2003). The mtDNA encodes 37 genes, of which 13 genes encode OXPHOS subunits [complex I (7), III (1), IV (3) and V (2)] and 22 transfer RNA (tRNA) and 2 ribosomal RNA (rRNA) genes required for mitochondrial translation (Clayton, 1991; Wallace et al., 1995). Approximately 6% of the mtDNA is noncoding, located predominantly in the D-loop and involved in the replication and transcription of the mtDNA (Berdanier and Everts, 2001). The mtDNA is compact, it contains no introns, several overlapping genes and incomplete termination codons (Lightowlers et al., 1997).
|
MtDNA molecules are packed in somatic cells as nucleoids in which six to ten molecules form a group with several different proteins (Jacobs et al., 2000; Iborra et al., 2004
; Legros et al., 2004). These nucleoids are not static entities, and mtDNA molecules exchange between nucleoids. The nucleoids are attached to the inner mitochondrial membrane near the OXPHOS system, where reactive oxygen species (ROS) are produced (Richter et al., 1988). Because of the lack of histones and other protective proteins and an ineffective repair mechanism, the mtDNA mutates 10–16 times more frequently as the nuclear DNA, and because of the lack of introns the mutations have a high probability of affecting genes and being pathogenic (Larsson and Clayton, 1995; Treem and Sokol, 1998). |
Replication, transcription and translation of the mtDNA |
|---|
Replication of the mtDNA is termed ‘relaxed’, because it is not connected to the cell cycle, and there is a constant degradation and production of mtDNA (Chinnery and Samuels, 1999
). Replication of mtDNA takes place in post-mitotic terminally differentiated cells. In most cell types, two possible mechanisms for the replication of the mtDNA exist (Bowmaker et al., 2003; Holt and Jacobs, 2003; Reyes et al., 2005). The strand displacement mechanism involves unidirectional initiation from the origin of replication of the ‘heavy’ H-strand (OH) located in the D-loop region of the mtDNA molecule (Figure 1). The replication of this leading strand initiates the synthesis of the lagging strand from the light L-strand origin of replication (OL) (Shadel and Clayton, 1997; Bogenhagen and Clayton, 2003). Alternatively, strand-coupled replication of the mtDNA implies initiation of lagging strand synthesis at multiple sites probably involving the synthesis of short Okazaki fragments (Holt et al., 2000; Bogenhagen and Clayton, 2003). The original strand displacement mechanism is probably the main replication method in cells which are in a steady-state level, whereas the strand-coupled model seems to be predominant in cells recovering after depletion and in cells in need of accelerating mtDNA synthesis (Holt et al., 2000; Fish et al., 2004). The mtDNA is synthesized by a mitochondrial-specific polymerase, DNA polymerase gamma (POLG), which requires additional factors like Twinkle [a ring helicase (Spelbrink et al., 2001)] and mitochondrial topoisomerases I and II responsible for the removal and introduction of supercoils in the mtDNA, respectively (Kosovsky and Soslau, 1993; Zhang et al., 2001). The mechanism regulating mtDNA replication is still not completely understood. Tfam, a limiting factor, and the size of the nucleoside pool are known to play an important role in the regulation of the mtDNA copy number (Ekstrand et al., 2004; Kanki et al., 2004; Kang and Hamasaki, 2005), but other factors also exist (Kaukonen et al., 2000; Brown and Clayton, 2002).
Transcription of the mtDNA requires mtRNA polymerase, mitochondrial transcription factor A (Tfam) and B1 or B2 (TFB1M or TFB2M) and several other transacting factors (Gaspari et al., 2004; Kang and Hamasaki, 2005). L-Strand transcription is initiated at the L-strand promoter (LSP) and results in a single polycistronic precursor RNA. The H-strand is transcribed by two overlapping units starting at two different initiation sites HSP1 and HSP2 (Fernandez-Silva et al., 2003). Transcription can be regulated at the level of initiation, termination, by the mitochondrial transcription termination factor (mTERF) (Asin-Cayuela et al., 2005) or both. Autonomous regulation of the mtDNA transcription occurs as in isolated mitochondria, the transcription of mtDNA continues for several hours (Enriquez et al., 1996). External signals, which play a role in the transcription regulation include, e.g. ATP levels in the cells and thyroid hormones (Enriquez et al., 1996; Weitzel et al., 2003).
In humans, mitochondrial translation occurs at the mitochondrial ribosomes (Sasarman et al., 2002), composed of a small ribosomal subunit (the 12S rRNA subunit encoded by the mtDNA and 29 nuclear encoded proteins) and a large ribosomal subunit (the 16S rRNA subunit encoded by the mtDNA and 48 nuclear encoded proteins) (Koc et al., 2001a,b). Additional factors are initiation factors [IF2 and IF3 (Ma and Spremulli, 1996; Koc and Spremulli, 2002)], elongation factors [EFTu (Ling et al., 1997), EFTs (Xin et al., 1995), EFG1 (Gao et al., 2001) and EFG2 (Lochmuller et al., 1999; Hammarsund et al., 2001)] and release factors [RF1 (Zhang and Spremulli, 1998)].
|
Biochemical investigations in OXPHOS disease |
|---|
In general, lactate (cell redox state, normal <20) and alanine levels are increased. Histochemical studies of skeletal muscle with accumulation of abnormal mitochondria under the sarcolemmal membrane in muscle fibres (RRF) or cytochrome oxidase (COX) negative fibres confirm mitochondrial dysfunction. Electron microscopy may provide additional information. Biochemical studies carried out in skeletal muscle or cultured skin fibroblasts or in any other (preferably affected) available tissue can determine enzyme deficiencies in one or more of the OXPHOS enzyme complexes (van den Heuvel and Smeitink, 2001). Spectrophotometric methods or blue native polyacrylamide gel electrophoresis combined with histochemistry (BN–PAGE) can both be applied to determine the activity of the individual OXPHOS complexes or combinations of complexes (Munnich and Rustin, 2001
; Van Coster et al., 2001). These biochemical measurements are preferably performed in fresh muscle specimens or other fresh tissues clinically expressing the disease, as frozen muscle or cultured fibroblasts do not always present the enzymatic deficiencies. Some difficulties are associated with the biochemical analysis. Normal variation in enzyme activity is high, and therefore the frequently detected moderate decreases in activity remain inconclusive. Furthermore, substantial variation exists in normal activity range as determined by different centres, because of the use of different protocols and the lack of widely accepted diagnostic criteria (Thorburn et al., 2004). A classification scheme has been developed by Bernier et al. (2002) including clinical features and enzyme activities found in several groups of patients. |
Genetic causes of OXPHOS disease |
|---|
OXPHOS diseases can be caused by mutations in the nuclear and mtDNA. Nuclear OXPHOS mutations can be classified as (i) gene defects altering the stability of mtDNA, (ii) gene defects in structural components or assembly factors of the OXPHOS complexes, (iii) defects in nonprotein components of the respiratory chain, like CoQ10 or taffazzin and (iv) gene defects in proteins indirectly related to OXPHOS (Chinnery, 2003
; Zeviani and Di Donato, 2004). OXPHOS diseases caused by nuclear gene mutations usually follow a Mendelian inheritance pattern. Disease causing mutations in the mtDNA can be large rearrangements or point mutations or a reduced copy number (mtDNA depletion).
Large-scale rearrangements are usually single deletions. Since 1988 (Holt et al., 1988), over 200 different mtDNA deletions have been reported, associated with several, different OXPHOS diseases. Three main clinical phenotypes are Kearns–Sayre syndrome (KSS), chronic progressive external ophthalmoplegia (PEO) and Pearson syndrome (Table I). The vast majority of deletions reported are flanked by short-repeat sequences ranging from 3 to 14 bp in length (Mita et al., 1990; Ota et al., 1994). No minimal area of overlap exists between the different deletions, but always at least one tRNA is removed (Tang et al., 2000). The severity of the disease and the age of onset are partly dependent on the amount and tissue distribution of the mtDNA rearrangement and the presence of deletion dimers or partially duplicated mtDNA molecules (Poulton and Holt, 1994; Rotig et al., 1995; Jacobs et al., 2004).
Point mutations in the mtDNA can be pathogenic or neutral. Neutral polymorphisms are common and based on a combination of specific polymorphisms; the mtDNA can be classified into haplogroups. Over 150 pathogenic point mutations in the mtDNA that affect protein-coding genes or RNA genes have been reported since 1988 (Wallace et al., 1988). Most pathogenic point mutations are heteroplasmic, but homoplasmic disease causing point mutations in the mtDNA have been described as well. The clinical phenotype of homoplasmic mutations (Table II) is generally restricted to a single tissue. Penetrance is often incomplete and other factors like nuclear-encoded proteins, epigenetic factors, environment or lifestyle [tobacco smoking (Tsao et al., 1999)] and mtDNA haplogroups (Brown et al., 2002) are likely to be involved (Guan et al., 2001).
|
Heteroplasmic point mutations in protein encoding and in RNA genes are more often pathogenic (Table II). Many mutations are infrequent or even private, presenting in a single family. All mutations display clinical heterogeneity (Sparaco et al., 2003), but this is most evident for the common m.3243A>G mutation (Table II). This variable phenotypic expression cannot be explained by the heteroplasmy level only, and so nuclear genes may be involved (Dunbar et al., 1995; Jacobs and Holt, 2000; Torroni et al., 2003). The threshold at which ATP production decreases is dependent on the tissue and mutation analysed. It appears to be lower in those tissues with a higher energy demand, such as brain and muscle (Larsson and Clayton, 1995). The existence of such a threshold implies that in the normal situation there is an overcapacity of the OXPHOS system (Rossignol et al., 2003), required to deal with an increased energy demand. This can also be considered a protective mechanism against deleterious mutations, which inevitably will accumulate during life.
Depletion and multiple mtDNA deletions
MtDNA depletion is a reduction in copy number of mtDNA molecules, which can be the consequence of a nuclear gene defect. MtDNA-depletion syndrome (MDS, Table I) with or without multiple mtDNA deletions is a severe autosomal recessive genetic disease caused by a mutation in one of the genes involved in mtDNA synthesis or nucleotide metabolism. Other mutations are detected in genes involved in mtDNA replication and the maintenance of the mitochondrial dNTP pool. Defects in comparable genes (ANT1, Twinkle, POLG1) are involved in multiple mtDNA deletions, which can present with or without depletion. Especially, mutations in the POLG gene have a clinically heterogeneous presentation, and both autosomal dominant and autosomal recessive families have been reported.
Inherited mtDNA mutations are usually present in all or most of the human tissues, but somatic mutations occur as well. Cytochrome b mutations have been described in muscle of patients only (Andreu et al., 1999). Age-related ROS damage is the most common source of acquired somatic mtDNA mutations. Over 200 different deletions and several point mutations have been found in the mtDNA that accumulate during ageing, especially in ageing muscle (Cottrell and Turnbull, 2000; Wei and Lee, 2002), in humans, but also in other species like monkeys (Lee et al., 1993; Schwarze et al., 1995), mice (Tanhauser and Laipis, 1995; Khaidakov et al., 2003) and nematodes (Melov et al., 1995). The mutant load of these individual mutations usually does not exceed 1%; but the total number of mtDNA mutations can be of such that the mitochondrial respiration and OXPHOS is significantly impaired (Hayakawa et al., 1996; Liu et al., 1998). A direct relation between acquired mtDNA mutations and ageing has been shown in a mouse model with a deficient polymerase gamma leading to deletions and point mutations in the mtDNA. These mtDNA-mutator mice show a reduced lifespan and premature onset of ageing problems like hair loss, osteoporosis anaemia and reduced fertility (Trifunovic et al., 2004). Depletion of the mtDNA can also be acquired and, e.g. pharmacologically induced by antiviral nucleoside analogues, as used in HIV therapy (Kakuda, 2000).
|
Treatment of OXPHOS disease |
|---|
Despite extensive studies on use of various pharmacological agents and vitamin supplements, there is still no cure for OXPHOS disease. Pharmacological therapy mainly relies on the administration of artificial electron acceptors, metabolites and cofactors or oxygen radical scavengers (Dimauro et al., 2004
). The administration of these factors can have a beneficial effect in some cases, but the effect is often transient. Novel strategies are being developed, directed at manipulating the level of heteroplasmy in the cell (Chinnery and Turnbull, 2001; Chinnery, 2004). These techniques aim at lowering the level of mutant mtDNA by selectively inhibiting the replication of mutant mtDNA by sequence-specific peptide nucleic acids (PNAs) or by the removal of mutated mtDNA by means of restriction enzymes. Alternative strategies attempt to treat the disease at the biochemical level by supplying cells with the normal mitochondrial proteins. Both these strategies encounter problems when executed in isolated organelle models with respect to the specificity and delivery of the product (Taylor et al., 2001). Another novel strategy is to redesign mitochondrial genes for expression from the nucleus and import normal copies of the redesigned gene from the cytosol into the mitochondria. The same can be done with allotopic expression of tRNAs. For the allotropic expression of both mitochondrial proteins and tRNAs, the correctly engineered genes must be delivered, recombined into the nucleus and expressed in a large number of cells to be a viable therapeutic approach (Smith et al., 2004).
Physical exercise can also be important to prevent disease manifestations. Most patients with mitochondrial disease are inactive because of exercise intolerance or fear for muscle damage, in spite of the fact that aerobic training increases work and oxidation capacity in these patients (Taivassalo et al., 2001; Taivassalo and Haller, 2004). Questions remain on the (long-term) effect of exercise on the mutant load, which may rise during life (Chinnery, 2004). Until a definite cure is developed, patients can only be given support and some limited therapy aimed at improving the quality of life. Palliative therapy is directed at preventing, e.g., the complications of diabetes mellitus and cardiomyopathy and surgical correction of ptosis and cataracts (Dimauro et al., 2004).
|
MtDNA segregation and transmission |
|---|
The mtDNA is transmitted through the maternal line via the mitochondria contained in the ooplasm. Maternal transmission is also a hallmark of mtDNA-related diseases. Mature human oocytes contain between 100 000 and 600 000 mitochondria and mtDNA copies (Reynier et al., 2001; Poulton and Marchington, 2002). This is in contrast to sperm cells which have been reported to contain between 10 and 700 copies mtDNA (Hecht et al., 1984; Shitara et al., 2000
; Diez-Sanchez et al., 2003; May-Panloup et al., 2003). The mtDNA content of the spermatozoon decreases five- to six-fold during the spermatogenesis, probably because of a down-regulation of the mitochondrial Tfam (Larsson et al., 1997; Rantanen and Larsson, 2000; Diez-Sanchez et al., 2003). During spermatid development, ubiquitin binds to the mitochondria, which makes the sperm mitochondria prone to proteolysis (Sutovsky, 2003), resulting in the loss of paternal mtDNA molecules (Shitara et al., 1998; Sutovsky et al., 2003). In another study, t-tpis, a testis-specific translocator, belonging to the translocator of mitochondrial outer membrane (TOM) complex, has been identified as a sperm mitochondria-specific factor, which incorporates an elimination factor present in the oocyte. The elimination factor is not yet identified, but it probably activates an endonuclease system. The ubiquination process is thought to follow the selective digestion of sperm mtDNA by endonucleases. Elimination of sperm mitochondria in the mouse can be inhibited by treatment with anti-tpis and (Hayashida et al., 2005). Recently, transmission of paternal mtDNA was detected in skeletal muscle of a patient (Schwartz and Vissing, 2002), but this is an infrequent phenomenon (Filosto et al., 2003; Johns, 2003; Schwartz and Vissing, 2003; Taylor et al., 2003; Schwartz and Vissing, 2004). Paternal transmission has also been studied in ICSI and IVF embryos and offspring. In these cases, low amounts of paternal mtDNA were detected in 16 of the 32 abnormal polyploid embryos (St John et al., 2000) but not in offspring normal embryos (Danan et al., 1999; Marchington et al., 2002).
Correct functioning and intactness of the mitochondria is vital for sperm motility. OXPHOS inhibitors decrease sperm motility (Ruiz-Pesini et al., 2000; St John et al., 2005), which suggests that mutations affecting mitochondrial functioning could affect sperm motility. The m.3243A>G mtDNA mutation shows a higher mutation level in semen fraction with a lower motility (Spiropoulos et al., 2002), and analysis of semen from men with lower semen quality revealed a higher incidence of homoplasmic base changes in the mtDNA especially at two locations, nt 9055 and nt 11719 (Holyoake et al., 2001). Kao et al. (1998) observed a higher incidence of especially the 4977 bp ‘common’ mtDNA deletion, in semen with a lower motility, but this was not confirmed by others (Cummins et al., 1998; St John et al., 2001). Multiple mtDNA deletions have been observed in both normozoospermic and oligozoospermic men, but as the semen quality diminishes the number of multiple deletions accumulates. A deviant number of CAG repeats (normally 10) in the polymerase gamma gene has been associated with unexplained male infertility (Rovio et al., 2001; Jensen et al., 2004). This association was not confirmed by others (Krausz et al., 2004; Aknin-Seifer et al., 2005). Pathogenic POLG mutations have however been associated with hypofertility in both males and females (Ferrari et al., 2005), and premature menopause has been found in many females suffering from CPEO caused by POLG mutations (Luoma et al., 2004), probably because of a link with steroid hormone genesis (Bose et al., 2002). These data are confirmed by the mutator mice, carrying a proofreading-deficient polymerase gamma, which show reduced fertility of both male and female mice (Trifunovic et al., 2004).
During cell division, mitochondria are randomly divided (Rotig and Munnich, 2003), and in heteroplasmic cells this can lead to a shift in the proportion of mutant mtDNA in the daughter cells. A loss of mutations is observed in fast-dividing tissues, probably because of a selection against cells containing high mutation loads. An example is the average decrease of 1% per year of the m.3243A>G mutation in blood of patients (Rahman et al., 2001). Increased ROS production is a critical factor triggering mtDNA replication, but also increasing mtDNA damage, eventually leading to apoptosis. In post-mitotic tissues, accumulation of mtDNA deletions and point mutations has been observed (Larsson et al., 1990; Weber et al., 1997). This proliferation only takes place in cells containing high amounts of mutant mtDNA, and because of this heteroplasmy percentage in tissues as a whole increases and variation in mutation load between muscle fibres develops (Chinnery et al., 2002). In case of deletions, the replicative advantage of the smaller molecule also adds to the accumulation of the mutated mtDNA in tissues and cells (Diaz et al., 2002).
Polymorphism and mutations in oocytes
Mutations in oocytes have been described as part of the transmission of pathogenic familial mutations and as de novo events. The first group is important for the recurrence risk of mtDNA disease in families and carriers (see next paragraph), the second could potentially explain the occurrence of new disease cases. Deletions in the mtDNA have been reported in 40–60% of unfertilized oocytes or oocytes that failed to develop into mature metaphase II oocytes, although usually in very low mutation percentages (Chen et al., 1995; Keefe et al., 1995; Brenner et al., 1998; Reynier et al., 1998; Barritt et al., 1999; Hsieh et al., 2002). Because none of the donating couples showed symptoms of mtDNA deletion syndromes, these mutations probably arose in the oocyte. Recently, we screened the entire mtDNA in oocytes for predominantly heteroplasmic point mutations and found that over 25% of the oocytes contained point mutations. The mutation percentages varied from very low levels (<1%) to high levels (>50%) with most oocytes containing low level mutation percentages (<30%) (Jacobs et al., in preparation). Under the assumption that at least 10% of the point mutations in the mtDNA will be pathogenic, this would mean that more than 5% of the oocytes harbour a possible pathogenic mutation in the mtDNA. Mostly, these mutations are present in very low levels. Some percentages can be above the threshold of expression, and these de novo mutations can have a direct phenotypic effect (De Coo et al., 1996; Degoul et al., 1997; Maassen et al., 2002; Thorburn, 2004). The low level mutations can get lost by cell division, but also fixed during life by random genetic drift, which has been observed in rapidly dividing colonic crypt cells (Taylor et al., 2003) and cancer cells (Carew and Huang, 2002). This also means that a very low level of mtDNA mutations in the oocyte can, because of relaxed replication, accumulate during life and might predispose for diseases, like Alzheimer and Parkinson’s disease, which are associated with mtDNA mutations (Chinnery et al., 2002; Coskun et al., 2004).
It is unlikely that mutations in the oocyte in general influence the fertilizability as carriers of mtDNA mutation do not present with fertility problems, and children with a high mutation load are born (Moilanen and Majamaa, 2001). However, oocytes can accumulate mutations in an age-dependant manner. The m.414T>G point mutation is present in 40% of the oocytes of women aged 
37 years in contrast to 4% of the oocytes of women aged <37 years, which could be associated with reproductive senescing (Barritt et al., 2000). Reynier et al. (2001) have shown a lowered number of mitochondria in oocytes from patients with fertilization failure owing to unknown causes, and a lower number of mitochondria is found in ageing oocytes (de Bruin et al., 2004). This means that the number of mitochondria in itself is important and not necessarily the ATP production by the OXPHOS system during embryo development. Therefore, acquired mtDNA mutations affecting mtDNA replication might affect the fertility.
Segregation of mtDNA diseases in families
The segregation of mtDNA disease in families is not straightforward and is highly dependent on the nature and amount of the mtDNA mutation. A woman carrying an mtDNA mutation will transmit a variable amount of this mutation to her offspring. The percentage heteroplasmy of point mutations in the offspring is related to the mutation percentage in the mother (Chinnery et al., 1998), although extreme shifts in mutation percentages occur (White et al., 1999; Carelli et al., 2002). Only a few studies report on the inheritance of heteroplasmic mtDNA mutations (Chinnery et al., 2000; Wong et al., 2002), and it appears that mutations, like the m.8363G>A, m.3460G>A and m.8993T>C, are in general randomly transmitted to offspring, although in some cases skewing in favour of the mutation can be observed (Larsson et al., 1992; Chinnery et al., 2000; Hurvitz et al., 2002; Wong et al., 2002). Transmission of the m.8344A>G, the m.3243A>G and m.8993T>G mutations is possibly not completely random, when comparing blood levels in mother and child (Chinnery et al., 2000). The mutation percentage of the m.8344A>G mutation is lower in the offspring and of the m.3243A>G and m.8993T>G mutations higher than expected by random transmission only (White et al., 1999a,b; Chinnery et al., 2000; Wong et al., 2002). However, the number of reported transmissions is small, a selection bias is likely, because analysis is performed after discovery of an offspring with clinical symptoms, and the age of sampling differs between mother and child.
Analysis of 82 oocytes collected from a woman carrying the m.3243A>G mutation with a mutation load of 18% in muscle and 7% in leukocytes revealed a binomial distribution pattern. The mutation percentage in the oocytes ranged from 0 to 45% (mean 12.6%), which was a random segregation pattern (Brown et al., 2001). Oocytes from a carrier of the m.8993T>G mutation demonstrated an extremely skewed segregation pattern in seven oocytes of a woman with a mutant load of 50% in blood. Six of the seven oocytes contained a mutant load >95%, and the remaining oocyte showed no evidence of the mutation (Blok et al., 1997). It is unclear whether this is a good representation of the entire pool of her oocytes and of other women carrying this mutation.
For most mutations, a relation exists between maternal mutation load and the mutation load in offspring and, therefore, the chance of being affected. This has been extensively studied for the m.8344A>G, m.3243A>G and m.8993T>G/C mutation (Chinnery et al., 1998). Carriers of the m.8344A>G mutation are at risk of affected offspring, if the mutation load in blood is >40%. This risk ranges from 12 (mutation load 40–59%) to 78% (mutation load >80%). For the m.3243A>G mutation, the chance of affected offspring ranges from 25 (mutation load <20% in blood) to 57% (mutation load 40–60%). The risk of affected offspring is therefore substantial even at low mutation levels in the carrier. Finally, for the m.8993T>G/C mutations, the risk of affected offspring rises from 0 (mutation load <20%) to >75% (mutation load 61–80%) (White et al., 1999).
The segregation of large single deletions is different, and these deletions are in general de novo. Chinnery et al. (2004) collected data on 226 families in which a single mtDNA deletion was identified in the proband. Possible other mtDNA rearrangements like mtDNA duplications and deletion dimers, which may affect the transmission (Rotig et al., 1992; Poulton et al., 1993; Ballinger et al., 1994; Shanske et al., 2002), were not taken into account. The overall recurrence risk for disease caused by single mtDNA deletions was estimated at 4.11% (Chinnery et al., 2004). Transmission of mtDNA deletions in the form of duplications has also been observed in mouse strains containing a pathogenic 4696 bp deletion in the mtDNA. After introduction of the deletion, partially duplicated molecules were formed which were transmitted to offspring and caused deletion symptoms (Nakada et al., 2001). MtDNA-deletion disorders can also be caused by nuclear gene mutations, and usually multiple deletions are observed which are transmitted in a Mendelian way (Kaukonen et al., 2000; Spelbrink et al., 2001; Van Goethem et al., 2001).
In the 1980s, a study on the segregation of mtDNA in Holstein cows revealed a rapid shift in the mtDNA genotype within two generational transitions (Hauswirth and Laipis, 1982). This shift has been confirmed several times in these cows (Ashley et al., 1989; Koehler et al., 1991), in other species like mice [heteroplasmic New Zealand Black/BINJ progeny (Meirelles and Smith, 1997)] and in humans for the homopolymeric tract heteroplasmy located between nt 303 and 315 of the mtDNA (Lutz et al., 2000). This has lead to the identification of the ‘mtDNA bottleneck’ (Figure 2), which is a restriction in the number of mtDNA molecules to be transmitted followed by an amplification of these founder molecules (Howell et al., 1992). The exclusive maternal transmission of mtDNA, the high mutation rate and the lack of a good repair mechanism and recombination would lead to decay of the mtDNA [Muller’s ratchet (Muller, 1964; Hoekstra, 2000)]. The stringent bottleneck has an evolutionary advantage as a sort of reset and acts to maintain a healthy mtDNA by filtering out mutations and minimizing heteroplasmy (Cummins, 1998, 2001). Because this filtering happens very early during the development, the chance to preserve age-related mutations in the early oocyte is small, although the low amount of mtDNA copies per mitochondria in the early developmental stages of the oocytes renders these oocytes vulnerable for mutational events (Keefe et al., 1995).
|
When the mitochondrial bottleneck exactly occurs during oocyte or embryo development and what the size is, is not yet clear (Poulton et al., 1998). Early during the first developmental stages of oocytes, the number of mitochondria and mtDNA molecules is reduced, and the lowest number of mitochondria (<10) is found in the early primordial germ cells (PGCs) of a 3-week-old embryo. The number of mitochondria is estimated from published electron micrographs of PGCs (Jansen and de Boer, 1998). It cannot be excluded that in the embryonic germ cell line, a week earlier, an even lower number of mitochondria is present (Jansen, 2000). The mtDNA copy number is unknown in PGCs, but in oocytes usually only one mtDNA molecule per mitochondrion is observed (Michaels et al., 1982; Chen et al., 1995). In mice and frogs, there is no mtDNA synthesis during embryogenesis until the stage of gastrulation (Larsson et al., 1998; Jansen, 2000) except for a small period during the one- to two-cell stage of mouse preimplantation development were there is some mtDNA turnover. The mtDNA content of the embryo does, however, not increase during this time (McConnell and Petrie, 2004; Thundathil et al., 2005). In humans, no mtDNA synthesis, measured by BrdU incorporation, is observed until the late morulae and blastocyst stage (McConnell, personal communication). This suggests that in humans most mitochondria remain haploid during the first developmental stages (Jansen and de Boer, 1998). The mean number of mitochondria and mtDNA molecules increases from 10 in the PGC to about 200 in the oogonium and eventually to 100 000–600 000 in the mature oocyte (Jansen, 2000). Segregation of the mtDNA during embryogenesis has been studied in mouse models by Jenuth et al. (1996), in which BALB/c cytoplasm was introduced in NZB/BINJ oocytes. The mtDNA variants remain evenly distributed in the developing fetal tissues, and no evidence is found for an additional bottleneck during embryogenesis (Jenuth et al., 1996; Meirelles and Smith, 1997), although events during embryonal development still can influence the final heteroplasmy percentage (Meirelles et al., 2001). From these studies, it appears that the major component of the bottleneck occurs between the PGC and the primary oocyte stage.
The bottleneck has considerable implications for a carrier of mtDNA mutations, and the mutation load can vary largely in both ways among her oocytes. The exact size of the bottleneck is hard to determine and may vary among individuals (Brown, 1997). Several studies have attempted to calculate the number of mtDNA units inherited through the bottleneck in cows, humans and mice (Howell et al., 1992; Bendall et al., 1996; Blok et al., 1997; Jenuth et al., 1997; Marchington et al., 1998; Brown et al., 2001). A repeated selection model, which attempts to take the number of cell divisions of oogenesis into account, and a single-selection model which proposes the bottleneck as a one time sampling of mtDNA molecules from a large pool have been applied (Poulton et al., 1998). The repeated selection model appears to represent the physiology more closely but assumes an identical sampling of mtDNA molecules every cell division, approximately 15, during oogenesis. The single-selection model assumes that the bottleneck occurs only once, that replication is equal from all templates and that the levels of heteroplasmy relate to the proportions in oocytes (Bendall et al., 1996). It has become clear that one common bottleneck size does not exist and that it will vary between meioses within and between different women. The bottleneck size using the single-selection model is calculated to be 1–30 segregating units (one unit could represent one mtDNA molecule, one nucleoid or one mitochondrion) in contrast to 20–200 units when using a repeated-selection model (Bendall et al., 1996; Poulton et al., 1998).
|
Mouse models for OXPHOS disease |
|---|
Animal models are essential for understanding the pathophysiological mechanisms of OXPHOS disease and for testing therapeutic interventions, but only few natural models exist [hearing loss in mice (Johnson et al., 2001
)]. Over the last decade, several mouse models have been developed for OXPHOS disease for both nuclear and mtDNA mutations. Only two mouse models with mtDNA mutations exist, the CAP-resistant (CAPR) mice with the m.2433T>C mutation in the 16SrRNA and the mtDNA-deletion mice with a 4.696 bp deletion (Sligh et al., 2000; Wallace, 2001). Disease symptoms were related to the human OXPHOS disease, but for the mtDNA deletions most mice died of renal failure which is uncommon in human deletion patients (Inoue et al., 2000; Sligh et al., 2000; Wallace, 2001). Both these transgenic animal lines demonstrated transmission of the mutated mtDNA to successive generations and can be used to study the inheritance and segregation of pathogenic mtDNA mutations. The CAPR mice transmitted the heteroplasmic mtDNA mutation to some of there progeny in homoplasmic or heteroplasmic state. Progeny, born alive, exhibited growth retardation, myopathy and dilated cardiomyopathy. Most animals died either in utero or within the first day after birth, one animal survived 11 days (Sligh et al., 2000). The mtDNA-deletion mice transmitted the rearranged mtDNA through three successive generations with a tendency to increasing heteroplasmy percentage to a maximum of 90% in muscle of some animals, most likely because of the replication advantage of the smaller mtDNA molecule. A percentage above 90% has not been found and may cause lethality in oocytes or embryos. Severe disease and COX-negative fibres were only found in mice with predominantly (>60%) deleted mtDNA (Inoue et al., 2000).
Several mouse models showing an OXPHOS disease phenotype caused by nuclear mutations have been developed (Wallace, 2001; Zeviani, 2001; Biousse et al., 2002). Mutations were introduced in genes associated with the OXPHOS system, including protein complex genes, radical scavenger genes (Sod2 mutant mouse), transcription factors (Tfam-deficient mouse) and adenine nucleotide translocator genes (Ant1-deficient mouse) (Li et al., 1995; Lebovitz et al., 1996; Graham et al., 1997; Larsson et al., 1998; Wang et al., 1999). These mouse models show different OXPHOS-related symptoms, but fertility is usually normal. This in contrast to the earlier mentioned mutator mice with a proofreading-deficient polymerase gamma. These mice show a premature onset of ageing and a reduced fertility, both males and females. Female reproductivity was nil after the age of 20 weeks, and male fertility was severely reduced probably because of low sperm count and smaller testes size (Trifunovic et al., 2004).
|
How to prevent transmission of mitochondrial disease |
|---|
A definitive diagnosis of mitochondrial disease is needed for prognosis and genetic counselling of patients and their families (Thorburn and Dahl, 2001
). As these disorders cannot be cured, counselling is important to judge the recurrence risk of mitochondrial disease and the options to prevent the transmission of this disease. Refraining from children or adoption is the safest and most reliable method, but this is usually not the first choice. IVF enables prospective parents to opt for using donor oocytes. In some cases, prenatal diagnosis (PND) or preimplantation genetic diagnosis (PGD) is possible, but other, more experimentally, methods are being developed as well (Figure 3). The ethical aspects concerned with these techniques are discussed separately.
|
PND of OXPHOS disease can be performed at the level of the enzyme or at the DNA level. Although the latter is preferable, the genetic defect is often not known for patients with OXPHOS disease, and the recurrence risk for these patients is hard to determine and based on family information only. If an enzyme deficiency is detectable in fibroblasts, then biochemical analysis of amniocytes might be an option, as fibroblasts, chorionic cells and amniocytes have the same embryonic origin (Graff et al., 2002). Biochemical analysis of fetal samples is feasible, although the methods used must be sufficiently sensitive given the low amount of fetal cells that can be obtained (Table III). Another limitation is that only 50% of the patients express the enzymatic defect in fibroblasts and that knowledge on complex assembly and activity during embryonic development is lacking. OXPHOS diseases caused by nuclear gene mutations show a Mendelian mode of inheritance. For known DNA mutations, PND can be offered by direct mutation analysis of a chorionic villus sampling (CVS) and/or amniotic cells. Where only the causing gene and location are known, but not the exact mutation, intragenic or closely linked polymorphic markers are used. DNA diagnostics is more reliable than enzymatic analysis and should be used whenever possible.
|
PND for heteroplasmic point mutations in the mtDNA has its own complexity. Genotype–phenotype correlations are less straightforward and (time-dependent) differences may occur between the tested fetal tissue and the actual embryo. For point mutations in the mtDNA, three criteria have been proposed to allow reliable PND (Poulton and Marchington, 2000; Poulton and Turnbull, 2000). (i) A close correlation between the mutant load and disease severity. (ii) A uniform distribution of mutant mtDNA in all tissues. (iii) No change in mutant load over time. Sufficient data are available for only three mutations (m.8993T>G/C, m.8344A>G and m.3243A>G) to judge these criteria properly. For the m.8993T>G/C and m.8344A>G mutations, PND can be reliably performed, although for each of these a grey zone of inconclusive results exists. For example, a mutant load of <20% for the m.8993T>G would predict healthy offspring, whereas a mutant load of 60% would give a 25% chance of disease (White et al., 1990). The number of data used to calculate these risks for the m.8993T>C mutation is so low that statistically even a mutant load of 0% does not preserve from a severe outcome. The amount of data required to reduce the confidence intervals of these percentages is for most mutations not available. For private mutations or mutations, which have only been reported a few times, PND should be carefully evaluated, based on genotype–phenotype correlations, available number of data and additional experiments (Jacobs et al., 2005). Until now nine prenatal tests were reported, for the m.8993T>G and m.8993T>C, the m.3243A>G and the m.9176T>C mutations (Table III). Also PND for mtDNA rearrangements is becoming an issue, as the recurrence risk for mtDNA-deletion disorders appears to be around 4% (Chinnery et al., 2004), and two PND have been performed (Table II). MDSs are usually caused by nuclear gene defects, but if the causing mutation is not known PND remains a possibility. Amniocytes of children suffering from a mtDNA-depletion disorder have been studied and were found to express the mtDNA depletion (Blake et al., 1999). The exact timing of onset of mtDNA depletion during fetal development is still unknown. A second report describes two cases of mtDNA depletion presenting prenatally with skin oedema and diminished fetal movements at 36 weeks of pregnancy (Arnon et al., 2002).
PGD is an alternative to PND. Oocytes are fertilized in vitro, and cells from the usually eight-cellular embryo are dissected and tested for the presence of a genetic defect. Unaffected embryos are transferred into the uterus (Handyside et al., 1990). PGD avoids the dilemmatic choice of a pregnancy termination, which is an advantage compared with PND. PGD can be performed by sampling either polar bodies (Rechitsky et al., 1999; Briggs et al., 2000) or blastomeres (Holding and Monk, 1989; Handyside et al., 1990). PGD is an option for mitochondrial disease because of nuclear gene defects, but it may also be a solution for mtDNA disease. The high copy number of mtDNA makes the analysis less prone to artefacts like amplification failure and allelic dropout (Thorburn, 2004). PGD for mtDNA mutations is, however, not straightforward with respect to the interpretation of the data. In a heteroplasmic mouse model, the distribution of both genotypes was identical between the ooplasm and polar body of a mature oocyte and between the blastomeres of two-, four-, and six- to eight-cell embryos (Molnar and Shoubridge, 1999; Dean et al., 2003). Both, polar bodies and blastomeres can be analysed, but the efficiency in diagnosing blastomeres is higher (Dean et al., 2003). Analysis of polar bodies might be preferred by consenting couples with a strong reservation against embryo testing, but the lower amount of mtDNA molecules in polar bodies may make the analysis susceptible to allelic dropout and preferential amplification. The criteria for reliable PND also apply for PGD. PGD is especially suited for women with a high mutation load and a high risk of affected offspring (Poulton and Turnbull, 2000). Embryos transferred to the uterus should have a mutant load, which would guarantee a healthy outcome. For some women, this could mean that they might need multiple PGD cycles before a suitable embryo can be identified. A disadvantage of PGD is the need of an IVF procedure as only 20–25% of the IVF cycles results in a pregnancy (Broekmans and Klinkert, 2004). If the IVF/PGD procedure is unsuccessful, the analysis of the embryos could still give valuable information for subsequent PGD cycles or other reproductive choices (Thorburn and Dahl, 2001).
The use of donor oocytes with sperm of the partner is a reliable method to prevent the transmission of OXPHOS disease caused by mtDNA mutations. The use of donor oocytes of maternal relatives is not advisable, because these may carry the same mtDNA mutations even though the mutation is undetectable in blood of the possible donor.
Cytoplasmic transfer (CT), an adaption of the ICSI technique (Cohen et al., 1997, 1998), has been tested on women experiencing repeated embryonic development failure, thought to be caused by depleted ATP levels in these oocytes (Van Blerkom et al., 1995, 2001). This resulted in 13 clinical pregnancies from which one was spontaneously aborted and one selectively aborted in a twin pair, both because of Turner’s syndrome. With this technique, 16 children have been born from which one developed