163rd ENMC International Workshop: Nucleoid and nucleotide biology in syndromes of mitochondrial DNA depletion myopathy 12–14 December 2008, Naarden, The Netherlands

2.1. Diagnosis 

The pattern of inheritance of MDS is autosomal recessive and onset is largely during infancy and childhood. Clinical syndromes [4], [5], [6] and general diagnostic approaches [7], [8] have been summarized recently, and the genes involved are listed in Table 1. Alpers disease due to mutations in the alpha subunit of the mitochondrial gamma polymerase (POLG1) was discussed in the 155th workshop (2007), and the main focus of the current workshop was the remaining 8 causes, because of their potential for therapeutic advances. Most patients with MDS present in childhood with severe symptoms of myopathy, neuropathy, hepatopathy, CNS involvement (movement disorders being common and severe epilepsy being a major feature of childhood presentations of POLG1) and less commonly nephropathies. Because they are relatively heterogeneous, biochemical and molecular analysis plays an important role in diagnosis. For instance, myoneurogastrointestinal encephalomyopathy (MNGIE) due to thymidine phosphorylase (TP) deficiency usually presents in middle childhood or young adults and can readily be diagnosed by measurement of blood and urine thymidine levels. Nevertheless, MH reported that ∼98% of MNGIE patients are initially misdiagnosed, frequently as anorexia nervosa, and this probably underlies the apparent regional variation in prevalence. While gastrointestinal symptoms may be prominent in MNGIE, some patients present with chronic inflammatory demyelinating polyneuropathy and minimal gastrointestinal features, even being well nourished.

In most of these disorders, quantitative reduction in mtDNA content of affected tissues, including one or more of muscle, liver and brain, is characteristic. There may also be low levels of heteroplasmy for point mutations [9] and/or multiple mtDNA deletions in MNGIE. Mutations in nine different genes are known to cause mtDNA depletion syndrome (Table 1), accounting for 10–80% of cases of mtDNA depletion syndrome. There is considerable uncertainty within the mitochondrial disease community about the prevalence of MDS, the prevalence of the commonest group (Alpers due to POLG1 mutations) is said to be about 1 in 100,000 [10]. The proportion where the molecular defect can be identified will depend in part on the phenotypic definition, which varies from centre to centre. Mitochondrial DNA copy number is notoriously difficult to quantify accurately. We suspect that the apparent regional differences are partly technical. Encouragingly, centres are increasingly aware of the need to use age-specific normal ranges [11]. However, the presence of significant numbers of ribonucleotides in non-replicating as well as replicating mtDNA [12] is a confounding factor that few have considered: many groups have their own methods of extraction and processing DNA for comparing the mtDNA:nuclear DNA ratio. It has been suggested that it is critical to relax or cut the circles of mtDNA to gain an accurate estimate of copy number [13], yet we have not found this to be the case (Holt and colleagues, unpublished experiments).

There is a clear need to evaluate whether differences in the isolation procedures between laboratories have an effect on the copy number measurements. While we await a universal robust method of sample preparation, it is important to standardize both tissue and extraction methods for patients and controls within individual laboratories. Uncertainty about the accuracy of mtDNA copy number estimates has doubtless played a part in making several groups conservative in their diagnoses of MDS. Many confine it to patients in whom there is either a reduction in mtDNA copy number to 20–30% of mean control values or a mutation in a known causative nuclear gene. Making this threshold level less conservative would have advantages: in some disorders (such as SCS deficiency due to SUCLA2 mutations) the mtDNA depletion is mild, and potentially less important in pathogenesis than other biochemical abnormalities that result from the TCA cycle block, including impaired conversion of propionyl CoA into succinyl CoA, manifest as elevated urinary methyl malonic acid. On the other hand, lowering the threshold has the disadvantage that moderate decreases in mtDNA copy number can occur secondary to other pathologies [11].

2.2. Underlying genetic causes 

Considerable progress has been made in defining the causal genes. Identifying a mutation in a causative gene that has previously been implicated in MDS is clinically very useful. A good database already exits for POLG1 [14]. The participants agreed to collate a table of mutations identified in causative genes for publication (reference to be added). This meeting emphasized that the development of rational therapies underpinned by a firm understanding of the fundamental science has led to significant progress. There is particular potential in the area of nucleotide imbalance because our understanding of mitochondrial DNA replication is far from complete. This suggests potential interventions that might limit deleterious consequences of the disorder, while we await a comprehensive model of mtDNA replication.

3.1. Background 

Cells maintain pools of nucleotide precursors in the cytosol and in mitochondria, and in some cases the biosynthetic enzymes are duplicated in the two compartments. Nevertheless, largely uncharacterized transport proteins enable transfer of some forms of nucleotide precursor from one cellular compartment to another. Several forms of mitochondrial disease are the result of patients lacking a defined component of the pathway, such as the mitochondrial thymidine kinase (TK2) or deoxyguanosine kinase (dGUOK), or the cytosolic p53-inducible small subunit of ribonucleotide reductase p53R2. Also mutations of catabolic enzymes such as cytosolic thymidine phosphorylase may destabilize the deoxynucleotide pools and the mt genome. Vera Bianchi has developed an assay to measure cytosolic and mitochondrial nucleotide levels that will help to clarify the mechanism of disease. This shows that the nucleotide pools are 10- to 20-fold greater in proliferating as opposed to quiescent cells [15]. Furthermore, in proliferating cells synthesis of deoxynucleotides in the cytosol is the main source of mtDNA precursors with a marginal contribution of intramitochondrial salvage of deoxynucleosides. J.P. presented preliminary data suggesting that mtDNA replication in cultured cells is partly linked to the cell cycle, potentially driven by the relatively high dNTP levels during S phase. However, quiescent cells are the most relevant to mitochondrial disease as these manifests primarily in solid tissues, such as muscle. Hence dNTP synthesis in quiescent cells via the salvage pathway becomes relevant to muscle because it is post mitotic. However, contrary to previous knowledge, p53R2-dependent ribonucleotide reduction occurs in non proliferating cells independently of DNA damage, albeit at 50- to 100-fold lower levels than typical S-phase specific de novo synthesis of deoxynucleotides [16]. Thus in postmitotic cells and tissues deoxynucleotide pools result from the interplay of the de novo and salvage pathways of deoxynucleotide synthesis and the regulatory action of catabolic enzymes.

3.2. Attempts to improve nucleotide balance in MNGIE 

The excess of thymidine resulting from TP def in MNGIE can be modeled in cell cultures by supplementing the medium. This produces a significant [15] but small [3] increase in mitochondrial (and cytosolic) dTTP which may underlie the specific pattern of point mutations detailed by Michio Hirano. This “next nucleotide” effect [17] is effectively a ‘run-on thymidine insertion effect’ whereby higher levels of thymidine facilitate T to C transitions at increased frequency after a series of adenine nucleotides. In addition, excess extracellular thymidine triggers ATP dissipation by futile cycles that buffer dTTP pool buildup [15], possibly aggravating the energy deficit in cells with mtDNA defects. The identification of excess thymidine as the fundamental underlying problem in both cellular and animal [18] models of MNGIE indicates a clear target for disease therapy. TP ablation in the mouse does not result in any overt pathology. However, when uridine phosphorylase (UP) is also knocked out, the mice develop leukoencephalopathy and myopathy, but not the gastrointestinal features that are a hallmark of MNGIE. mtDNA depletion with respiratory chain dysfunction is present in brain only, but multiple mtDNA deletions and point mutations that are the characteristic of MNGIE are absent. This double knockout (TP(−/−)UP(−/−)) mouse model will be very useful both for modeling the neurogenic (but not the gastrointestinal) components of MNGIE [18] and for developing rational therapies. In MNGIE patients, renal dialysis does not appear to be of sustained benefit, perhaps because only a modest proportion of the excess thymidine and related metabolites is removed [19]. Altering the upstream supply of substrate might be of benefit, possibilities including low pyrimidine diet and inhibitors of dihydroorotate dehydrogenase, such as Leflunomide. The latter is currently used to treat rheumatoid arthritis but might affect mitochondrial function at the level of Coenzyme Q.

3.3. Attempts to improve nucleotide balance in dGK deficiency 

On the other hand, mutations in the DGUOK gene cause a deficiency syndrome, by impairing the activity of dGK reducing synthesis of dGMP and ultimately of dGTP. This has been demonstrated by J.W.T., who supplemented cells with phosphorylated nucleotide precursors (dRMP) to restore mtDNA copy number in cultured cells derived from dGK deficient patients. A.S.R. also showed that nucleotide precursors are able to restore mtDNA in dGK deficient cells cultured in the laboratory [20]; these experiments used unphosphorylated precursors that more readily cross the plasma membrane.

Is this a viable therapeutic strategy? Nucleotides precursors can be given orally, and are tasteless in drinking water; hence delivery is not a problem. However, nucleotide imbalance is known to be mutagenic for nuclear DNA synthesis and so cancer could result. Moreover, nucleotide supplementation could replace an imbalance in one nucleotide precursor for an imbalance in the other three, which could be as detrimental to mtDNA synthesis as the defect the treatment was designed to correct. Nevertheless, the idea is worthwhile pursuing as relatively low doses that achieved a modest increase mtDNA copy number could well produce a marked improvement in phenotype. This is because of the well-recognised ‘threshold effect’ in mtDNA diseases. Mitochondrial DNA is typically present in hundreds or thousands of copies of per cell and pathology manifests only once the mutant load exceeds a threshold. The precise values for the threshold depend both on the specific mutation and the cell type, but for one of the most common mutations it appears to be around 50% in muscle but potentially higher for cochlear and islets [21]. The extent of mtDNA depletion required to cause disease is uncertain, and mosaicism is a potentially confounding factor; nevertheless most researchers would only expect symptoms to result when the residual copy number is in less than 50% of normal. Therefore, an intervention producing an increase from say 20% to 50% of normal mtDNA copy number could be expected to have clear benefits for the patient. Modest increases in dGTP levels might therefore be beneficial; large increases in dGTP need to be avoided as they would risk creating an insufficiency of the other nucleotides and could well be mutagenic.

A dGUOK mouse model would be valuable for trials, although none is currently available, there are plans to address this gap in the market. TP and TK2 KO mouse are already available (M.H. and A.K., respectively). Knocking out mitochondrial thymidine kinase (TK2) in mice causes death within a few weeks of birth, preceded by cachexia, muscle wasting and a failure to develop adipose tissue. mtDNA depletion and respiratory chain dysfunction are apparent in multiple tissues [22]. Nucleotide supplementation has not yet been tested in the TK2 knockout mouse because depletion is not apparent in tissue culture. Nevertheless, muscle is preferentially affected because cytosolic TK1 is downregulated in quiescent cells and tissues. It is possible that up-regulating the activity of p53RNR and hence de novo nucleotide synthesis might be beneficial.

3.4. Other approaches to therapy in nucleotide deficiencies 

Nucleotide supplementation is far from the only treatment conceived for MNGIE and allied defects of mitochondrial nucleotide metabolism. Enzyme replacement therapy is a strategy that was proposed by S.R., and one for which lysosomal storage disorders provide a useful paradigm. M.H. demonstrated proof of principle for enzyme replacement in MNGIE by transiently improving thymidine levels following platelet transfusion [23]. It is thus a potential candidate for approaches such as carrier erythrocyte encapsulated thymidine phosphorylase therapy [24], but this is still in its infancy.